# POLYAMINES IN PLANT BIOTECHNOLOGY, FOOD NUTRITION AND HUMAN HEALTH

EDITED BY : Rubén Alcázar, Ana Margarida Fortes and Antonio F. Tiburcio PUBLISHED IN : Frontiers in Plant Science and Frontiers in Nutrition

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88963-608-2 DOI 10.3389/978-2-88963-608-2

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

## POLYAMINES IN PLANT BIOTECHNOLOGY, FOOD NUTRITION AND HUMAN HEALTH

Topic Editors: Rubén Alcázar, University of Barcelona, Spain Ana Margarida Fortes, University of Lisbon, Portugal Antonio F. Tiburcio, University of Barcelona, Spain

Citation: Alcázar, R., Fortes A. M., Tiburcio, A. F., eds. (2020). Polyamines in Plant Biotechnology, Food Nutrition and Human Health. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-608-2

# Table of Contents

*06 Editorial: Polyamines in Plant Biotechnology, Food Nutrition, and Human Health*

Rubén Alcázar, Ana Margarida Fortes and Antonio F. Tiburcio

*08 Polyamines – A New Metabolic Switch: Crosstalk With Networks Involving Senescence, Crop Improvement, and Mammalian Cancer Therapy*

Ewa Sobieszczuk-Nowicka, Ewelina Paluch-Lubawa, Autar K. Mattoo, Magdalena Arasimowicz-Jelonek, Per L. Gregersen and Andrzej Pacak

*20 Polyamine Catabolism in Plants: A Universal Process With Diverse Functions*

Wei Wang, Konstantinos Paschalidis, Jian-Can Feng, Jie Song and Ji-Hong Liu

*33 Transcriptional Modulation of Polyamine Metabolism in Fruit Species Under Abiotic and Biotic Stress*

Ana Margarida Fortes, Patricia Agudelo-Romero, Diana Pimentel and Noam Alkan

*42 Polyamines in Halophytes*

Milagros Bueno and María-Pilar Cordovilla

*49 Polyamines and Legumes: Joint Stories of Stress, Nitrogen Fixation and Environment*

Ana Bernardina Menéndez, Pablo Ignacio Calzadilla, Pedro Alfonso Sansberro, Fabiana Daniela Espasandin, Ayelén Gazquez, César Daniel Bordenave, Santiago Javier Maiale, Andrés Alberto Rodríguez, Vanina Giselle Maguire, Maria Paula Campestre, Andrés Garriz, Franco Rubén Rossi, Fernando Matias Romero, Leandro Solmi, Maria Soraya Salloum, Mariela Inés Monteoliva, Julio Humberto Debat and Oscar Adolfo Ruiz

#### *64 Polyamines in Food*

Nelly C. Muñoz-Esparza, M. Luz Latorre-Moratalla, Oriol Comas-Basté, Natalia Toro-Funes, M. Teresa Veciana-Nogués and M. Carmen Vidal-Carou

*75 Polyamines and Gut Microbiota*

Rosanna Tofalo, Simone Cocchi and Giovanna Suzzi

*80 Dietary and Gut Microbiota Polyamines in Obesity- and Age-Related Diseases*

Bruno Ramos-Molina, Maria Isabel Queipo-Ortuño, Ana Lambertos, Francisco J. Tinahones and Rafael Peñafiel

*95 Metabolic Characterization of* Hyoscyamus niger *Ornithine Decarboxylase*

Tengfei Zhao, Changjian Wang, Feng Bai, Siqi Li, Chunxian Yang, Fangyuan Zhang, Ge Bai, Min Chen, Xiaozhong Lan and Zhihua Liao

#### *106 Structural Study of Agmatine Iminohydrolase From* Medicago truncatula*, the Second Enzyme of the Agmatine Route of Putrescine Biosynthesis in Plants*

Bartosz Sekula and Zbigniew Dauter

*120 Spermidine Synthase (SPDS) Undergoes Concerted Structural Rearrangements Upon Ligand Binding – A Case Study of the Two SPDS Isoforms From* Arabidopsis thaliana

Bartosz Sekula and Zbigniew Dauter


Hamed Soren Seifi and Barry J. Shelp

*152 Extracellular Spermine Triggers a Rapid Intracellular Phosphatidic Acid Response in Arabidopsis, Involving PLD*d *Activation and Stimulating Ion Flux*

Xavier Zarza, Lana Shabala, Miki Fujita, Sergey Shabala, Michel A. Haring, Antonio F. Tiburcio and Teun Munnik


Savithri U. Nambeesan, Autar K. Mattoo and Avtar K. Handa

*244 Conservation of Thermospermine Synthase Activity in Vascular and Non-vascular Plants*

Anna Solé-Gil, Jorge Hernández-García, María Pilar López-Gresa, Miguel A. Blázquez and Javier Agustí

*254 Complexity and Conservation of Thermospermine-Responsive uORFs of*  SAC51 *Family Genes in Angiosperms*

Soichi Ishitsuka, Mai Yamamoto, Minaho Miyamoto, Yoshitaka Kuwashiro, Akihiro Imai, Hiroyasu Motose and Taku Takahashi

#### *263 Thermospermine Synthase (*ACL5*) and Diamine Oxidase (*DAO*) Expression is Needed for Zygotic Embryogenesis and Vascular Development in Scots Pine*

Jaana Vuosku, Riina Muilu-Mäkelä, Komlan Avia, Marko Suokas, Johanna Kestilä, Esa Läärä, Hely Häggman, Outi Savolainen and Tytti Sarjala

*278 Compatible and Incompatible Pollen-Styles Interaction in* Pyrus communis *L. Show Different Transglutaminase Features, Polyamine Pattern and Metabolomics Profiles*

Manuela Mandrone, Fabiana Antognoni, Iris Aloisi, Giulia Potente, Ferruccio Poli, Giampiero Cai, Claudia Faleri, Luigi Parrotta and Stefano Del Duca

# Editorial: Polyamines in Plant Biotechnology, Food Nutrition, and Human Health

Rubén Alcázar 1\*, Ana Margarida Fortes <sup>2</sup> and Antonio F. Tiburcio<sup>1</sup>

<sup>1</sup> Department of Biology, Healthcare and Environment, Section of Plant Physiology, Faculty of Pharmacy and Food Sciences, University of Barcelona, Barcelona, Spain, <sup>2</sup> Faculdade de Ciências de Lisboa, Department of Plant Biology, Biosystems and Integrative Sciences Institute, Universidade de Lisboa, Lisbon, Portugal

Keywords: polyamines, agriculture, climate change, health, nutrition, metabolism, plant protection, food

Editorial on the Research Topic

#### Polyamines in Plant Biotechnology, Food Nutrition, and Human Health

Polyamines are small polycations derived from arginine and/or ornithine. These compounds are present in all living organisms and play common and organism-specific functions. Polyamines are present in most food products of plant and animal origin, thus having an impact on human nutrition and health. In this Topic, we aimed to cover both basic and applied research on polyamines in the areas of plant biotechnology, food nutrition, and human health.

In plants, the most abundant polyamines are putrescine (Put), spermidine (Spd), and spermine (Spm). The control of polyamine levels is achieved through regulation of their biosynthesis, catabolism, and transport, which are modulated by the environment. Past and current research on polyamines has investigated basic processes of polyamine homeostasis, as a mean to obtain crops better adapted to climate change. The discovery of polyamine signaling pathways will also help in reaching this major goal. In this Topic, Zhao et al. report on the characterization of the Ornithine decarboxylase (ODC) enzyme from Hyosciamus niger, involved in Put biosynthesis. The ODC enzyme exhibits higher catalytic efficiency than other plant ODCs reported so far, thus providing an ideal candidate gene for polyamine biosynthesis engineering. Sekula and Dauter report on the crystal structure of Agmatine iminohydrolase involved in Put biosynthesis from arginine and identify that the dimeric assembly of monomers is drastically different from the bacterial enzyme. The same authors also obtained crystals of spermidine synthase, involved in Spd formation, and characterized the structures of the two dimeric enzyme isoforms in Arabidopsis (Sekula and Dauter).

Plants also contain thermospermine (tSpm), a less abundant but very relevant polyamine synthesized from Spd by thermospermine syntase (tSPM), in a reaction that is conserved throughout the plant kingdom (Solé-Gil et al.). According to these authors, tSPM might play developmental and/or stress-related roles in addition to the confirmed function in regulating xylem cells maturation, in both non-vascular and vascular plants. Ishitsuka et al. have shown that the response to tSpm is conserved in dicots and monocots and plays a role in translational enhancement that could be eventually applied as a biotechnological tool in animal and fungal systems. The work by Zarza et al. shows that Spm triggers a quick phosphatidic acid response in Arabidopsis, which is mainly mediated by phospholipase D. Their results suggest the participation of phosphatidic acid in Spm perception.

#### Edited and reviewed by:

Chang-Jun Liu, Brookhaven National Laboratory (DOE), United States

> \*Correspondence: Rubén Alcázar ralcazar@ub.edu

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

> Received: 15 January 2020 Accepted: 27 January 2020 Published: 19 February 2020

#### Citation:

Alcázar R, Fortes AM and Tiburcio AF (2020) Editorial: Polyamines in Plant Biotechnology, Food Nutrition, and Human Health. Front. Plant Sci. 11:120. doi: 10.3389/fpls.2020.00120

The cellular content of polyamines is also regulated by degradation mediated by diamine and polyamine oxidases (DAO and PAO). These catabolic processes and their diverse functions in plants have been reviewed by Wang et al. and include the involvement of polyamine catabolism in fruit ripening, senescence, and stress responses. Arabidopsis has four genes homologs of the human histone demethylase LSD1 (LDL1-3 and FLD) that bear a flavin amine oxidase domain, and act differently in the control of flowering time (Martignago et al.). This contribution evidences the role of different epigenetic mechanisms in the control of plant development and defense and their impact on agronomical traits.

The participation of polyamines in many aspects of plant development is well known. In this regard, Nambeesan et al. report that Spd impacts floral organ identity and fruit set in tomato involving GA metabolism and signaling. These authors also suggested that altered polyamine ratios may regulate floral developmental processes. A role for free polyamines in pollination of Pyrus communis flowers is also suggested (Mandrone et al.). Both tSPM synthase and DAO are required for zygotic embryogenesis and vascular development in Scots pine (Vuosku et al.). According to the authors, specific manipulation of polyamine gene expression might provide a way to enhance somatic embryo production in recalcitrant Scots pine lines with important biotechnological applications. Furthermore, the work from Sobieszczuk-Nowicka et al. contributes to a better understanding of the cellular and molecular mechanisms underlying senescence-related cell death.

Numerous studies have reported increased levels of polyamines under conditions of abiotic and biotic stresses. In fact, Spm has been proposed as a plant defense activator to both type of stresses (Seifi and Shelp). Under abiotic stress, Spm promotes abscisic acid (ABA) biosynthesis (Marco et al.) and activates ABA-mediated signaling pathways (Seifi and Shelp). Spm also modulates oxidative/ antioxidant responses (Seo et al.) and promotes transcription of several defense-related genes including some involved in polyamine metabolism (Fortes et al.). In contrast to glycophytes [i.e., pepper; (Piñero et al.)], the levels of Spd/Spm are high in halophytes, thus polyamines have been proposed as useful indicators of plant salinity adaptation (Bueno and Cordovilla).

On the other hand, under biotic stress Spm promotes jasmonic acid-mediated signaling pathways (Seifi and Shelp), modulates oxidative/antioxidant responses (Seo et al.) and promotes transcription of several stress-related genes including PAOs (Menendez et al.). High Spm content in maize genotypes has been implicated in higher resistance to Aspergillus flavus and aflatoxin contamination (Majumdar et al.). So far, the role of Put during defense has remained elusive. In this Topic, Liu et al. report that this polyamine contributes to amplification of pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI), through ROS production leading to enhanced disease resistance against bacterial pathogens. Menendez et al. suggest a link between polyamines and abiotic stress mitigation in legumes by symbiosis with fungi and bacteria. Furthermore, several authors (Bueno and Cordovilla; Fortes et al.,; Sobieszczuk-Nowicka et al.,; Vuosku et al.,; Wang et al.) highlight that manipulation of polyamine contents can be used to increase crop productivity and quality under more sustainable conditions by improving abiotic and biotic stress resilience. Investigations on the specific molecular mechanisms and signaling pathways by which polyamines exert these functions will follow in the near future.

Food is an important source of polyamines for human and animal nutrition. During the neonatal period, polyamine requirements are high. However, de novo biosynthesis decreases with age, which is the reason why polyamine dietary sources acquire a greater importance in aging populations. A wide range of polyamine concentrations can be found in all types of foods. The main polyamine in plant-based products is Spd, while Spm content is higher in animal-derived foods. Muñoz-Esparza et al. review the polyamine contents in breast milk and infant formula, as well as in different food of plant and animal origin. They also provide estimated levels of polyamine intake in different human populations. It is well known that dietary polyamines can affect human and animal health. Thus, exogenous polyamines (either dietary polyamines or produced by the gut microbiota) are able to induce longevity and cardioprotective effects in animals. Exogenous Spd or Spm improve glucose homeostasis and insulin sensitivity, and reduce hepatic fat accumulation in mouse models. Interestingly, increasing evidence indicate that polyamines act in the control of relevant human pathologies including cancer, immunological, neurological, and gastrointestinal diseases. Ramos-Molina et al. review how the diet influences circulating and local polyamine levels, and how the modulation of either polyamine dietary intake or endogenous production by gut microbiota can be used for potential therapeutic purposes.

Overall, this topic provides a good number of examples of basic and applied research on polyamines aiming at improving the quality of life of present and future human populations.

## AUTHOR CONTRIBUTIONS

All authors listed have made substantial, direct, and intellectual contribution to the work and approved it for publication.

#### FUNDING

Research by RA and AT are supported by the Agencia Estatal de Investigación (AEI) and the Fondo Europeo de Desarrollo Regional (FEDER) BFU2017-87742-R grant of the Programa Estatal de Fomento de la Investigación Científica y Técnica de Excelencia (Ministerio de Economía y Competitividad, Spain). Research by AF is supported by project PTDC/ASP-HOR/28485/ 2017 and UID/MULTI/04046/2019 Research Unit grant from FCT, Portugal (to BioISI).

Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Alcázar, Fortes and Tiburcio. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Polyamines – A New Metabolic Switch: Crosstalk With Networks Involving Senescence, Crop Improvement, and Mammalian Cancer Therapy

#### *Ewa Sobieszczuk-Nowicka1 \* , Ewelina Paluch-Lubawa1 , Autar K. Mattoo2 , Magdalena Arasimowicz-Jelonek 3 , Per L. Gregersen4 and Andrzej Pacak5*

*1 Department of Plant Physiology, Faculty of Biology, Institute of Experimental Biology, Adam Mickiewicz University in Poznań, Poznań, Poland, 2 Sustainable Agricultural Systems Laboratory, Henry A. Wallace Beltsville Agricultural Research Center, United States Department of Agriculture, Beltsville, MD, United States, 3 Department of Plant Ecophysiology, Faculty of Biology, Institute of Experimental Biology, Adam Mickiewicz University in Poznań, Poznań, Poland, 4 Department of Molecular Biology and Genetics, Aarhus University, Slagelse, Denmark, 5 Department of Gene Expression, Faculty of Biology, Institute of Molecular Biology and Biotechnology, Adam Mickiewicz University in Poznań, Poznań, Poland*

#### *Edited by:*

*Antonio F. Tiburcio, University of Barcelona, Spain*

#### *Reviewed by:*

*Stefano Del Duca, University of Bologna, Italy Vasileios Fotopoulos, Cyprus University of Technology, Cyprus*

*\*Correspondence:* 

*Ewa Sobieszczuk-Nowicka evaanna@amu.edu.pl*

#### *Specialty section:*

*This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science*

*Received: 15 March 2019 Accepted: 14 June 2019 Published: 03 July 2019*

#### *Citation:*

*Sobieszczuk-Nowicka E, Paluch-Lubawa E, Mattoo AK, Arasimowicz-Jelonek M, Gregersen PL and Pacak A (2019) Polyamines – A New Metabolic Switch: Crosstalk With Networks Involving Senescence, Crop Improvement, and Mammalian Cancer Therapy. Front. Plant Sci. 10:859. doi: 10.3389/fpls.2019.00859*

Polyamines (PAs) are low molecular weight organic cations comprising biogenic amines

that play multiple roles in plant growth and senescence. PA metabolism was found to play a central role in metabolic and genetic reprogramming during dark-induced barley leaf senescence (DILS). Robust PA catabolism can impact the rate of senescence progression in plants. We opine that deciphering senescence-dependent polyaminemediated multidirectional metabolic crosstalks is important to understand regulation and involvement of PAs in plant death and re-mobilization of nutrients during senescence. This will involve optimizing the use of PA biosynthesis inhibitors, robust transgenic approaches to modulate PA biosynthetic and catabolic genes, and developing novel germplasm enriched in pro- and anti-senescence traits to ensure sustained crop productivity. PA-mediated delay of senescence can extend the photosynthesis capacity, thereby increasing grain starch content in malting grains such as barley. On the other hand, accelerating the onset of senescence can lead to increases in mineral and nitrogen content in grains for animal feed. Unraveling the "polyamine metabolic switch" and delineating the roles of PAs in senescence should further our knowledge about autophagy mechanisms involved in plant senescence as well as mammalian systems. It is noteworthy that inhibitors of PA biosynthesis block cell viability in animal model systems (cell tumor lines) to control some cancers, in this instance, proliferative cancer cells were led toward cell death. Likewise, PA *conjugates* work as signal carriers for slow release of regulatory molecule nitric oxide in the targeted cells. Taken together, these and other outcomes provide examples for developing novel therapeutics for human health wellness as well as developing plant resistance/tolerance to stress stimuli.

Keywords: autophagy, cancer therapy, cell death, CRISPR/Cas9, crop improvement, polyamines-nitric oxide conjugates, senescence, transcriptome profiling

**8**

### INTRODUCTION

Senescence is a developmental program that precedes programmed cell death (PCD) in plants and involves re-utilization/re-direction of carbon (C), nitrogen (N), phosphorus (P), and other nutrients including metals for the growth of younger leaves, seeds, and/or grain/fruit of a plant crop (Jones et al., 2012). This fundamental process has attracted investigations from different perspectives to identify the different players involved and determine if senescence process can be delayed or enhanced (Sarwat and Tuteja, 2019). For example, delaying senescence in plants induces tolerance to drought (Rivero et al., 2007). A number of plant hormones, namely, ethylene, jasmonic acid, abscisic acid, salicylic acid, strigolactones, brassinosteroids, and gibberellins are known to play an important role in plant senescence (Wojciechowska et al., 2018). Antisenescence regulators known to impact senescence include nitric oxide (NO) (Tun et al., 2006) and polyamines (PAs) (Mattoo and Sobieszczuk-Nowicka, 2019). PAs have attracted a lot of scientific attention in recent years and evidence in favor of them being hormone-like molecules has accumulated. This review is intended to address the recent progress made in our understanding of the dynamics involved in the regulation of plant senescence by polyamines.

The commonly used polyamines include putrescine (Put), spermidine (Spd), and spermine (Spm) (Handa et al., 2018). PA biosynthesis occurs with decarboxylation of ornithine catalyzed by ornithine decarboxylase (ODC) to form Put (**Figure 1**). Also, amino acid arginine is decarboxylated to agmatine by arginine decarboxylase (ADC), and agmatine in turn is converted to Put *via* N-carbamoyl-Put. Put is then successively aminopropylated to Spd and Spm by Spd synthase (SPDS) and Spm synthase (SPMS), respectively. T-Spm is the product of ACL5 (t-SPMS) in plants (Knott et al., 2007; Handa et al., 2018). The aminopropyl groups are donated by decarboxylated S-adenosylmethionine (dcSAM) whose synthesis is catalyzed by SAM decarboxylase (SAMDC; Cohen, 1998). SAMDC activity represents a common rate-limiting step in PA biosynthesis (Stanley et al., 1989; Thu-Hang et al., 2002). ODC is also a transcriptional target of the c-Myc oncogene (Bello-Fernandez et al., 1993). Unlike plants, animals possess the ODC antizyme, which control the cellular ODC levels (Matsufuji et al., 1995).

PA catabolism is mediated mainly by two classes of amine oxidases (AOs), the diamine oxidase (DAO) and PA oxidase (PAO) (Moschou et al., 2012). In arabidopsis, the AO constitutes a family of several functionally redundant genes (Tavladoraki et al., 2006; Planas-Portell et al., 2013). DAOs mainly oxidize Put and cadaverine, but also Spd and Spm with lower affinity. PAOs oxidize Spd and Spm but not Put (Angelini et al., 2010). The plant apoplastic PAOs catalyze the terminal catabolism of PAs, yielding pyrroline and 1-(3-aminopropyl)pyrrollinium from Spd and Spm, respectively, as well as 1,3-diaminopropane and H2O2 (Cohen, 1998). The plant intracellular (cytoplasmic or peroxisomal) PAOs interconvert Spm to Spd, and Spd to Put, which results also in H2O2 production (Moschou et al., 2008). Intracellular PAOs also oxidize t-Spm (Fincato et al., 2011), but the oxidation products of t-Spm have not been identified.

Animal PAOs also interconvert PAs in the peroxisomes but show preference toward the acetylated PAs produced by the inducible Spd/Spm N1-acetyltransferase (SSAT; Matsui et al., 1981; Pegg et al., 1981; Casero et al., 1991; Casero and Pegg, 1993). In animals, Spm is interconverted by cytoplasmic Spm oxidase independent of the SSAT pathway (SMO; Vujcic et al., 2002). The SMO pathway is evolutionarily conserved between animals and plants, and AtPAO1 represents the plant counterpart of SMO. The interconversion of PAs produces, in parallel to a corresponding PA, 3-aminopropanal or 3-acetamidopropanal. SMO is induced by tumor necrosis factor (TNF), resulting in the production of H2O2 which adds to inflammation, mutagenesis, and subsequently cancer development (Babbar and Casero, 2006).

Under normal conditions, AtPAO1 expression is barely detectable (Tavladoraki et al., 2006). Conditions that lead to AtPAO1 activation have not as yet been determined. In both animals and plants, genetic manipulation of PA catabolism is not an easy task, since there are several potentially redundant genes encoding AOs with similar substrate specificity. To bypass this, one alternative is to develop chemical genetics approach. Several plant DAO and PAO inhibitors have been successfully used in animals to block the catabolism and back conversion of PAs. However, their potency *in vivo* in plant biology remains largely unexplored (Moschou and Roubelakis-Angelakis, 2013).

The function of PAs as cell growth and development regulators has attracted much attention in recent years. PAs are absolutely essential for cellular viability through their role(s) in critical cellular functions, including regulation of nucleic acid and protein synthesis, and macromolecular structural integrity (Kusano and Suzuki, 2015). PA homeostasis is tightly regulated, with an excessive intracellular PAs in certain tissues leading to undesired (Wang and Casero, 2006) and desired phenotypes (Mehta et al., 2002). Thus, PAs are recognized as important ubiquitous bioactive substances with impact on several diverse biological phenomena. PAs may employ signaling mechanisms that are different than those known for the other well studied plant hormones (Mattoo and Sobieszczuk-Nowicka, 2019), since PAs are present in plant cells at higher levels, being effective often in the micro- to millimolar range (Sobieszczuk-Nowicka and Legocka, 2014; Anwar et al., 2015).

Earlier studies that suggested that PAs are anti-senescence in nature have been validated in recent studies (Sobieszczuk-Nowicka, 2017 and references therein). One such validation has come from genetic dissection of leaf senescence models, including dark-induced leaf senescence (DILS) (Sequera-Mutiozabal et al., 2016; Sobieszczuk-Nowicka et al., 2016, 2018).

**Abbreviations:** ADC, Arg decarboxylase; Arg, Arginine; ACC, 1-aminocyclopropane-1-carboxylate; C, Carbon; DAOs, Diamine oxidases; DILS, Dark-induced leaf senescence; N, Nitrogen; NO, Nitric oxide; NONOates, Exposing secondary amines to high pressure of NO results in diazeniumdiolates formation commonly known as "NONOates"; Orn, Ornithine; ODC, Orn decarboxylase; PAs, Polyamines; PAOs, Polyamine oxidases; PAOs(bc), Polyamine back-conversions oxidases; Put, Putrescine; PTS, Polyamine transport system; Rfd, Chl fluorescence decrease ratio called vitality index Rfd = (Fm − Fs)/Fs; Spd, Spermidine; Spm, Spermine; SAM, S-adenosylmethionine; SAMDC, SAM decarboxylase; TOR, Target of Rapamycin kinase signaling pathway.

FIGURE 1 | Plant polyamine metabolism pathways and its inhibition. Putrescine (Put), spermidine (Spd), and spermine (Spm) constitute major PAs that are primary amines with two or more amine groups. The biosynthesis of PAs is well established in plants. Put in many plants is synthesized from arginine (Arg) *via* agmatine catalyzed by Arg decarboxylase (ADC) and from ornithine (Orn) by Orn decarboxylase (ODC) except *Arabidopsis* in whose genome Orn decarboxylase seems not present. Put is thereafter sequentially converted to Spd and Spm/thermo-Spm (T-Spm) through successive addition of aminopropyl groups from S-adenosylmethionine catalyzed by S-adenosylmethionine decarboxylase (SAMDC). The transfer of the aminopropyl groups is catalyzed by Spd and Spm/T-Spm synthases (SPDS/SPMS/TSPMS), respectively. On the other hand, diamine (DAO) and polyamine (PAO) oxidases work in tandem to deaminate each PA, producing, in the process, the signaling molecule hydrogen peroxide (H2O2). Back-conversions from Spm to Put *via* Spd, and Spm to Spd, are catalyzed by PAOs(bc). Blue boxes indicate inhibitors of polyamine metabolism and blue font indicate polyamines and enzymes taking part in PA metabolism. ADC, arginine decarboxylase; ODC, ornithine decarboxylase; SAMDC, S-adenosylmethionine decarboxylase; SPDS/SPMS, Spd and Spm synthases; DAO, diamine oxidases; PAO, polyamine oxidases; AIH, agmatine iminohydrolase; CPA, N-carbamoyl putrescine amidohydrolase. Inhibitors: 1,4-DB – 1,4-diamino-butanone, AG – diaminoguanidine, CHA – cyclohexylammonium sulphate, D-arginine, DFMA – γ-difluoromethylarginine, DFMO – α-difluoromethylornithine, G – guazatine, MGBG -methylglyoxal*bis*-(guanylhydrazone).

These developments indicate that PA catabolism plays a central role in metabolic reprogramming, directing a senescing leaf toward the programmed organ death. Thus, depending upon which direction PA metabolism undertakes, for instance whether toward synthesis/accumulation, or toward their catabolism that generates H2O2, the plant will either grow or senesce, respectively.

### DILS VERSUS DEVELOPMENTAL SENESCENCE MODELS

The genomic resources available for *Arabidopsis* have made it a very attractive model for the identification and functional analysis of senescence-regulated genes (Buchanan-Wollaston et al., 2003, 2005; Breeze et al., 2011). In many plants, such as barley, removal of the developing flowers and pods significantly extends the life of their leaves, while in *Arabidopsis*, male sterile mutants or plants from which the developing bolts are removed do not extend the lifespan of the leaves. Because of these differences, cereal leaves have been used over the years as a model for studying leaf development and senescence (*Zea mays* – Smart et al., 1995; *Oryza sativa* – Lee et al., 2001; *Triticum aestivum* – Uauy et al., 2006; and *Hordeum vulgare* – Kleber-Janke and Krupinska, 1997; Jukanti et al., 2008; Christiansen and Gregersen, 2014; Avila-Ospina et al., 2015; Springer et al., 2015; Wehner et al., 2015; Sobieszczuk-Nowicka et al., 2018). Distinct differences in the senescence program of *Arabidopsis* as compared to that in the monocot plants have been revealed. Senescence in cereals is generally regulated at the level of an individual leaf. Nutrients from the older leaves are remobilized for the younger leaves and eventually for the flag leaf, contributing thereby to the nutrients required for grain development. Cereal leaves have a basal meristem, the leaf tip consists of older cells, and the younger cells are concentrated at the leaf base. Such a cellular organization enables studies on senescence progression easier to differentiate (Gregersen et al., 2008). Nonetheless, the lack of coordinated development of the cells within an individual leaf introduces complexity in studying leaf senescence. Therefore, induced senescence that directs a synchronous process, such as the dark-induced senescence (DILS), has become more in vogue (Buchanan-Wollaston et al., 2005).

DILS is an extreme example of shading which induces senescence in leaves similar to that observed during normal plant development. The DILS model fits well with other important monocot crop plants, e.g., maize and rice, eliminating the confounding factors that overlap with developmental senescence such as bolting or flowering. Early and late events of leaf senescence in the DILS crop model were deciphered to reveal the time limit for dark to light transition in reversing the senescence process (Sobieszczuk-Nowicka et al., 2018). Differences in gene medleys including the hormone-activated signaling pathways, lipid catabolic processes, glutamine catabolic processes, DNA and RNA methylation and carbohydrate metabolic processes between DILS and developmental senescence processes in barley leaves have been revealed (Sobieszczuk-Nowicka et al., 2018). These studies also demonstrated that the DILS program is reversible by re-exposure of the barley plants to light prior to day-7 of dark exposure (Sobieszczuk-Nowicka et al., 2018). The senescence reversal involved regaining of photosynthesis, increase in chlorophyll and reversal of chlorophyll fluorescence vitality index (Rfd), inspite of the activation of macro-autophagyrelated genes. Rfd was found to be an earliest parameter that correlated well with the cessation of photosynthesis prior to micro-autophagy symptoms, chromatin condensation, and initiation of DNA degradation.

#### POLYAMINES AND DILS PROGRAM

That applying PAs to plants prevents their senescence has been a phenomenon known for a long time (Galston et al., 1978; Kaur-Sawhney et al., 1978; Cohen et al., 1979; Apelbaum et al., 1981; Mizrahi et al., 1989; Besford et al., 1993; Legocka and Zajchert, 1999). Our understanding of plant senescence vis a vis PAs has been advanced through developments in molecular tools, genome sequencing, and genetic engineering (for instance, see Del Duca et al., 2014 and references therein; Cai et al., 2015 and references therein; Sobieszczuk-Nowicka et al., 2015, 2016; Sequera-Mutiozabal et al., 2016; Mattoo and Sobieszczuk-Nowicka, 2019). PA biosynthesis, catabolism, conjugation, interconversions, and transport all contribute to PA homeostasis (Angelini et al., 2010 and references therein; Moschou and Roubelakis-Angelakis, 2013). Transformations between individual PAs essentially contribute to darkness-induced barley leaf senescence responses (Sobieszczuk-Nowicka et al., 2016). Transcript levels and corresponding PA catabolic enzymes – DAO and PAO, increase during induced and developmental senescence making them important components of senescencerelated mechanisms (Ioannidis et al., 2014; Sobieszczuk-Nowicka et al., 2016). Moreover, inhibition of PAO activity drastically slows down the senescence-associated chlorophyll loss (Sobieszczuk-Nowicka et al., 2016).

*Arabidopsis* PA back-conversion oxidase mutants have also been utilized for studying dark-induced senescence. In these mutants, conversion of Spm to Spd, and/or Spd to Put, does not occur and their senescence is delayed (Sequera-Mutiozabal et al., 2016). The delayed dark-induced senescence in these mutants was associated with accumulation of Spm levels (Sequera-Mutiozabal et al., 2016). Additionally, during this phase, these plants had a reduced production of reactive oxygen species (ROS) and, interestingly, an increase in the levels of the signaling molecule, nitric oxide (NO). These data suggest that Spm is a "signaling" metabolite, leading to protection against stress through metabolic conversions that involve ascorbate/dehydro-ascorbate redox state transitions, changes in sugar and nitrogen metabolism, cross-talk with ethylene biosynthesis, and mitochondrial electron transport chain modulation (Sequera-Mutiozabal et al., 2016). Thus, metabolic interactions between PAs, particularly Spm, occur with cellular oxidative balance and transport/biosynthesis of amino acids, likely a strategy to cope with damage during senescence.

Plants also respond to environmental factors by secreting Spd to the apoplast where its catabolism leads to H2O2 production like what is also known as a hypersensitive response. Based upon the H2O2 concentration, these cells initiate either a defense response or cell death program (Yoda et al., 2003, 2006; Marina et al., 2008; Moschou et al., 2008). It is also known that high Spd and Spm pools accumulate in the apoplast during DILS, which is associated with a gradual accumulation of apoplastic diaminopropane (PA catabolism intermediate) and H2O2 (Sobieszczuk-Nowicka et al., 2016). The basal Put levels in the apoplastic pool of PAs are an order of magnitude lower, increasing only slightly during senescence. However, Put is a dominant PA in the free PA fraction, accumulating to high levels before decreasing. The decrease in free Put is concomitant with the formation of PCA-soluble Put conjugates that accumulate at high levels in the senescing leaf, indicating that the Put-conjugating enzymes are active in a senescing cell (Sobieszczuk-Nowicka et al., 2016). Senescence-dependent remobilized nitrogen (N) and carbon (C) flow may contribute to PA conjugation, since the expression of respective protein-coding genes also increases (Sobieszczuk-Nowicka et al., 2016). That plant cells sense PAs as organic-N and stimulate turnover of N molecules has been previously substantiated and discussed (Mattoo et al., 2006, 2010).

Physiological and structural changes in barley chloroplasts during DILS occurs in association with PCA-insoluble PA conjugation, modification of chloroplast proteins, and modulation of chloroplast-localized transglutaminases (ChlTGases). TGases catalyze post-translational modification of proteins by establishing covalent linkage of ε-(γ-glutamyl) moiety on PAs (Serafini-Fracassini and Del Duca, 2008). Thus, *in situ* localization and changes in the ChlTGase activity during dark-induced senescence mirror increase in the levels of plastid membrane-bound Put and Spd (Sobieszczuk-Nowicka et al., 2009, 2015). ChlTGase catalyzes binding of [3 H]Put and [3 H]Spd to the photosystem proteins (Sobieszczuk-Nowicka et al., 2015). Substrates of ChlTGases in mature leaves include apoproteins of the chlorophyll a/b antenna complex, LHCII, ATP synthase and pSbS (photosystem II 22 kDa protein), proteins that are essential in energy-dependent quenching and increased thermal dissipation of excessively absorbed light energy in the photosystems (Del Duca et al., 1994; Dondini et al., 2003; Della Mea et al., 2004; Campos et al., 2010). Several stress-responsive proteins detected in the PA-bound fraction only after dark-induced senescence include the antioxidant enzyme peroxiredoxin, heat shock protein, ent-copalyldiphosphate synthase, and IAA-amino acid hydrolase (Wang et al., 2004; Van der Graaff et al., 2006; Iqbal et al., 2011; Cejudo et al., 2012). PAs in concert with TGases are functionally involved in DILS as supported by proteomic analysis and TGase activity/transcript modulation (Sobieszczuk-Nowicka et al., 2009, 2015). Thus, PAs and plant senescence do cross paths.

#### POLYAMINES AND DILS AS A MODEL FOR STUDYING MAMMALIAN CANCER MECHANISMS

Regulation of cell death mechanisms in plants and animals (including humans) with radically different anatomy and physiology highlight PAs as universal bioregulators of this process across kingdoms (Della Mea et al., 2007; Del Duca et al., 2014; Cai et al., 2015; Handa et al., 2018). Thus, delineation of the roles of PAs should lead to a better understanding of the mechanisms in plant senescence homoeostasis causing longevity or cell death. Research into these areas should also provide new knowledge about similar mechanism(s) in mammalian systems at cellular and molecular level.

In animal systems, longevity-promoting regimens, including the natural PAs, have been associated with apoptosis (cancer cell lines) (Madeo et al., 2010). Addition of PA inhibitors to block cell *via*bility in tumor lines and to channel cancer cells toward the path of autophagic death has been considered (Madeo et al., 2010) Longevity-promoting regimens, including caloric restriction and inhibition of TOR (Target of Rapamycin kinase signaling pathway) with rapamycin, resveratrol or the natural polyamine associated with autophagy need to be considered (Madeo et al., 2010; Zabala-Letona et al., 2017). TOR has a central role in sensing cellular nutrition and energy status and regulating cellular metabolism. It is a negative regulator of autophagy in yeast and animals (Dann and Thomas, 2006). TOR role in plants has been documented (Ren et al., 2012) following the report of its negative regulation of autophagy in *Arabidopsis* (**Figure 2**; Liu and Bassham, 2010). In plants, its upstream signals and downstream substrates that control the autophagy pathway still need to be investigated. A role of PAs in TOR regulation has been previously discussed (Ren et al., 2012). A model example of stress-induced selective autophagy is the C and/or N starvation. In such conditions, intensified auto-destruction allows to acquire respiratory substrates and cell survival.

In mammals, autophagy is important in maintaining normal health since it prevents a number of diseases (including cancer). In plants, autophagy participates in circulation of cell components and acts as a quality control mechanism during senescence (Sobieszczuk-Nowicka et al., 2018). Therefore, DILS, may be a good model also for studying autophagy and PCD pathways because it involves defined physiological and cytological transformations: disruption of the nucleus and mitochondria, chromatin condensation, enhanced expression of cysteine proteases, autophagy proteins, and nDNA fragmentation (PCD marker) (**Figure 3**). It is important to note that, in spite of macro-autophagy advancement, degradation symptoms can be reversed by replacing dark conditions with light prior to the time when the process becomes irreversible. Turnover of macromolecules *via* autophagy might be critical for cell homeostasis during DILS. How the autophagy switches between cell survival and cell death is not known! Therefore, how plant cells mechanistically regulate DILS *via* autophagy is important to explore (Sobieszczuk-Nowicka et al., 2018).

PA transport as well as PA biosynthesis, degradation and conjugation play a vital role in the regulation of intracellular PA levels (Fujita and Shinozaki, 2014). PA uptake in plant cells occurs *via* energy-dependent, protein-mediated transport systems. Evidence has accumulated which suggests that paraquat, one of the most widely used herbicides, is transported by the PA transport system in diverse organisms including plants (Fujita and Shinozaki, 2014).

There is evidence that links the signaling molecule NO with PAs in living organisms. In plants, PAs overlap NO metabolism in response to developmental and stress stimuli. NO formation increases in response to exogenous PAs (Tun

Conversely, autophagy is inhibited by the target of rapamycin kinase (TOR), a central cell growth regulator that integrates growth factor and nutrient signals. Under nutrient sufficiency, high TOR activity prevents ATG13 activation by phosphorylating ATG13 Ser 757 and disrupting the interaction between ATG1 and ATG13. TOR, protein kinase A, AMK (AMP activated kinase), and GCN2 (General Control Nonderepressible2) are kinases operating in autophagy signaling pathway. Elongation initiation factor 2α (E2Fa) and the transcription factor GCN4 regulate expression of ATG1 and 13. (B) Potential TOR signaling pathways in *Arabidopsis*. The TOR complex, including TOR, RAPTOR, and LST8 (RAPTOR recruits substrates and presents them to TOR for phosphorylation, and LST8 stabilizes the TOR complex), senses and integrates multiple upstream signals such as nutrient starvation. TOR may serve as a negative regulator of autophagy. Some TOR substrates in plants have been identified, including: AML1 (*Arabidopsis* Mei2-like1, Mei2 is a meiosis signaling molecule that has been suggested to be a potential TOR substrate, also in yeast), EBP1 (ErbB – 3 epidermal growth factor receptor binding protein), S6K (ribosomal p70 S6 kinase), and Tap46 (a regulatory subunits of PP2A (protein 2 phosphatase type 2A), which is phosphorylated by TOR, suggesting that Tap46 is a direct substrate of TOR). These substrates may function to control translation, cell growth, and autophagy (modified from Liu and Bassham, 2012).

FIGURE 3 | Autophagy and dark-induced leaf senescence (DILS) of barley seedlings. (A) Model. (B) Ultrastructure of autophagy symptoms of dark-induced senescing parenchyma cells. Early (days 3 and 7) and late (day 10) events of leaf senescence and time limit for reversal (arrows on the top indicate the point of no return) of the senescence process. During the initial senescence period (day 3 of darkness), tonoplast invagination, presence of small cytoplasmic fragments near or connected with tonoplast and vacuoles, and shrunken protoplasts are apparent. On day 7 of senescence, a few cells show discontinuity of the cell membrane, while by day 10, tonoplast apparently ruptures. Consequently, all the organelles undergo gradual disintegration and localized to the central part of the cell. The cell membrane increasingly loosens and, consequently, the intracellular compartmentation is lost. Cell death during senescence is distinguished by rapidly occurring changes in the chloroplasts, whereas the nucleus and mitochondria are relatively more stable, and their degradation occurs only after the final lytic stage following vacuole tonoplast rupture. In the later stages of cell death, distinguishing specific organelles is not possible. However, the shrinking of the protoplast and deformation of the cell wall are clearly observed (see the micrographs). Autophagy is apparent during ultrastructural observations of senescing parenchyma cells seen as small autophagic bodies inside vacuoles, autophagosomes presence in protoplasts, and tonoplast rupture. The changes, at each stage of DILS, are accompanied by elements of micro-, macro- and megaautophagy. Autophagy role in the metabolic turnover of cell components as one of the mechanisms of quality control of the leaf senescence is discerned. In this process, nutrients such as carbon, nitrogen and phosphorus are released in the course of degradation of proteins, lipids, sugars and nucleic acids, and then transported to younger leaves, ripening fruits, and for seed formation. Metabolism and selective remobilization of macromolecules, which are crucial in the effective performance of the process, are accompanied first by micro- and then macro-autophagy. These studies have emphasized that the efficient regulation of autophagous process is a symptom of the vitality of senescing cells, which must at every stage maintain the ability to keep homeostasis. Bars, 200 nm days 0 and 3; 500 nm days 7 and 10 (top row of B); bars, 200 nm days 0, 3, 7, and 10 (bottom row of B) (modified from Sobieszczuk-Nowicka et al., 2018).

et al., 2006; Arasimowicz-Jelonek et al., 2009; Wimalasekera et al., 2011; Diao et al., 2017). In turn, PAs metabolism can be adjusted in a dose-dependent manner by exogenous NO (Fan et al., 2013; Filippou et al., 2013). Also, L-arginine is a precursor of both PAs and NO, and arginase has been considered as a decisive checkpoint to direct the metabolism of L-arginine to either PAs or NO (Flores et al., 2008). Also, some amino acids (e.g., methionine), which are precursors of PAs, can also affect cellular status of NO (Montilla-Bascón et al., 2017).

Generally, PA uptake is elevated in rapidly proliferating cells. Many tumor cells possess an active energy-dependent PA transport system (PTS), which selectively helps in the accumulation of endogenous PAs and structurally related compounds. Therefore, the idea of attaching cytotoxic drugs to polyamine vectors for selectively targeting cancer cells by utilizing the PTS is worth pursuing (Palmer et al., 2009). Polyamine *conjugates* have been used as signal carriers for slow release of regulatory molecules, such as NO, in targeted cancer cells (Kielbik et al., 2013). From a chemical point of view, exposing secondary amines to NO results in diazeniumdiolate formation, commonly known as "NONOates." These compounds contain the functional group [N(O)NO]− that binds an amine nucleophile adduct (Fleming et al., 2017). Numerous examples of these zwitterionic poly-amine/NO adducts are known (Hrabie et al., 1993). They are a unique group of compounds functioning in biological systems as NO donors and spontaneously release NO in aqueous solution without requiring redox or light activation (Lam et al., 2002). NONOates possess great potential in a variety of biomedical treatments, requiring rapid or prolonged release of NO *in vivo*.

Parameters, including pH, temperature, and chemical nature of the nucleophile affect the decomposition rate of NONOates, which presents a range of 1 min to 1 day under physiological conditions (Fitzhugh and Keefer, 2000). For example, spermine NONOate which is a diazenium diolate NO donor has a unique pattern of NO release (half-life around 39–73 min) which fits well with angiogenesis (Majumder et al., 2014). In general, diazeniumdiolates are useful for the study of tumor biology since they can be used as antineoplastic agents (Huerta, 2015). Also, polyamine/NO adducts were found effective in the induction of ovarian cancer cell death *via* inhibition of signal transducer and activator of transcription 3 (STAT3), serine-threonine protein kinase AKT protein phosphorylation, and downregulation of their cytocolic levels (Kielbik et al., 2013). Spermine NONOate has been adopted in plant research as an NO-releasing compound, documenting interrelationship among NO, cyclic GMP and heme oxygenase-1 in gibberellin-treated wheat aleurone layers (Wu et al., 2013).

An unusual diazeniumdiolate (*N*-nitrosohydroxylamine) was found in rhizosphere bacteria (*Paraburkholderia graminis*). *P. graminis* produces gramibactin which is a siderophore with a diazeniumdiolate ligand system (Hermenau et al., 2018). Gramibactin biosynthesis genes seem conserved in numerous plant-associated bacteria, including rice, wheat, and maize. Thus, plants could benefit not only from the iron that is mobilized by the bacteria, but also from NO released by the unique diazeniumdiolate. Although diazeniumdiolate groups are extremely scarce in nature, they can be an effective system to release NO in targeted cells. It is worth testing if such an outcome could lead to the development of novel therapeutics for human health wellness and plant resistance/tolerance to stress stimuli.

### APPROACHES TO POLYAMINE METABOLISM CROSSTALK WITH THE SENESCENCE NETWORK IN BARLEY

Genetic mechanisms that lead to stress-induced senescence and delineate processes involved in either delaying or accelerating senescence are important to decipher. The growing world population and global climate change have necessitated the development of high yielding and nutritious crops, a theme that has become a central challenge in this century. The impact of early senescence on crop yield and quality demands that crop losses be contained since more than 30% of crop losses occur pre- and post-harvest. Modulating senescence behavior in a versatile species such as barley benefits commercial production in at least two ways: (1) delaying senescence onset and extending the photosynthetic period in malting grains can increase grain starch content; (2) accelerating the onset of senescence increases nitrogen content in grains used for animal feed.

Development of transgenic barley plants that are defective in specific PA metabolic genes can be generated *via* the ubi-overexpression, RNAi approaches or CRISPR/Cas9 (Bartlett et al., 2008; Smedley and Harwood, 2015; Holme et al., 2017). This would allow gaining "anti-aging" or "pro-aging" phenotypes as an important intervention. Lossof-function barley transgenics deficient in PA metabolism also need to be developed. Agrobacterium-mediated transformation has been optimized in barley, conferring highly efficient transfer of foreign DNA into the barley genome within the agrobacterium T-DNA borders (Bartlett et al., 2008). Efficient over-expression of candidate genes (e.g., Holme et al., 2012) or downregulation of gene expression using RNAi techniques (e.g., Carciofi et al., 2012) in barley has been successful. CRISPR/Cas9 gene editing tools can secure efficient delivery of the Cas9 site-directed nuclease to subject the plant (barley) to genome editing (Holme et al., 2017; Lawrenson and Harwood, 2019). Notably, the inserted Cas9 and guide-RNA cassettes can be removed from selected lines after Mendelian segregation in the subsequent generations.

Senescence in barley is a complex process regulated by many diverse factors, PAs being one of them. Previously, modifying the senescence program focused on transcription factors, in particular NAC transcription factors (*No apical meristem, ATAF1/2, cup-shaped cotyledon 2*), which are known key regulators of plant senescence (e.g., Christiansen and Gregersen, 2014). Thus, plants constitutively over-expressing the senescenceassociated NAC gene HvNAC005 showed an early senescence, albeit with stunted plants (Christiansen et al., 2016). In order to manipulate the PA metabolism, there is a range of potential candidate genes, e.g., DAO and PAO, that could be downregulated by targeted knockout using the CRISPR/Cas9 mutation strategy. Designing a strategy to systematically knockout the genes involved in the biosynthesis, degradation and sequestering of PAs could provide an array of plants with different senescence phenotypes. Studies with these lines could help delineate the role(s) of different genes in the homeostasis of PAs in a plant. In order to have plants that can be grown without restrictions under field conditions, one possible alternative is to select promising CRISPR/Cas9 mutations and mimic them by TILLING screening of a classical mutant collection.

Constitutive or inducible overexpression of PA biosynthetic genes (ADC, ODC, SAMDC, SPDS, and SPMS) from different plant and animal sources in *Arabidopsis*, rice, tobacco, tomato, eggplant, and pear have resulted in increasing the endogenous levels of a known PA. This led to enhanced plant tolerance to various abiotic stresses, such as salt, drought, low/high temperature, wounding, ozone, flooding, heavy metals (Cu, Cr, Fe, and Ni), acid stress, and oxidative stresses (Hussain et al., 2011; Moschou et al., 2012; Tavladoraki et al., 2012; Shi et al., 2013). Conversely, knockout mutants of AtADC1/2 with lowered level of Put had decreased tolerance to salt and freezing (Cuevas et al., 2008). Also, knockout mutants of AtSPMS/AtACL5 exhibited less Spm accumulation and had decreased tolerance to salt, drought, and heat stresses (Tavladoraki et al., 2012).

A number of studies have also utilized the inhibitors of PA biosynthesis or catabolism. Some chemicals were identified as inhibitors of PA metabolism (Kaur-Sawhney et al., 2003). Among these, difluoromethylarginine (DFMA) and difluoromethylornithine (DFMO) are reversible inhibitors of ADC and ODC, respectively; methylglyoxal-bis guanylhydrazone (MGBG) is a potent inhibitor of SAMDC; however, it can also inhibit ADC and PAO activity; 1,4-diamino-butanone (1,4-DB) is a Put inhibitor; cyclohexylamine (CHA) is a competitive inhibitor of SPDS; D-arginine is also a PA biosynthetic inhibitor, though less compatible form of arginine compared with L-arginine (Kaur-Sawhney et al., 2003). Guazatine blocks action of PAO, while aminoguanidine inhibits DAO (**Figure 1**). These inhibitors have unspecific roles in PA metabolism. Thus, using them requires caution, particularly when interpreting results obtained from such a pharmacological approach.

Examples of the use of PA supplementation and PA metabolism's blockers in senescence models and the response of these systems to PA metabolism changes have been discussed (Cai et al., 2015; Sobieszczuk-Nowicka, 2017). Generally, it has been considered that the supplementation of exogenous PA or blocking their oxidation has an anti-aging effect, while PA biosynthesis inhibitors that block cell viability in animal model systems (cell tumor lines) to prevent cancers have also been tested (Russell and Snyder, 1968; Upp et al., 1988; Manni et al., 1995; Wallace and Caslake, 2001; Gilmour, 2007; Nowotarski et al., 2013). PAs as promoters of cellular proliferation and growth became a consideration after ODC activity was detected in regenerating rat liver, chick embryo, and various tumors (Russell and Snyder, 1968). The higher levels of PAs in cancerous cells suggest their possible role in tumor formation/growth (Upp et al., 1988; Manni et al., 1995; Wallace and Caslake, 2001; Gilmour, 2007). This rationale advanced the use of an inhibitor of ODC, difluoromethylornithine (DFMO), as a chemopreventive agent to treat cancerous cells (Nowotarski et al., 2013).

#### CONCLUSIONS

All types of stresses, including darkness, limit plant growth and crop productivity. This has more dire consequences in view of the fact that, based on FAO data, the world will need 70% more food to feed the anticipated 9 billion people by 2050. Thus, achieving global food security while reconciling demands of the environment is the greatest challenge faced today by mankind. Thus, it is importantly clear that increasing plant productivity, improving food quality and enhancing agricultural sustainability can no longer be ignored. In this regard, research on the anabolism or catabolism of polyamines plays an important role in reprogramming metabolic switches such that plant leaf senescence can be altered for a particular pro-growth phenotype or where organ death is enhanced without affecting the energy-use efficiency of plants. Likewise, exogenous application of natural or synthetic PAs can help plants to improve their tolerance against a broad spectrum of stress factors which, in turn, should lead to higher plant productivity as well as extend the boundaries of crop cultivation. Details on PA signaling transduction pathway(s) and the crosstalk between PA and other plant regulators/hormones are basically unknown and just beginning to be unraveled. Scientists delving in the PA science arena are already making inroads into PA omics profiling, developing novel germplasm by genetic engineering, and unraveling the interactions between PAs, other hormones and stress-responsive molecules such as NO. Such studies should bring new insights to our understanding of PA-related stress induced-senescence and cell death mechanism(s). Comparison of gene expression units between

metabolic switch for barley leaf senescence is a good model for developing a molecular basis of the process in order to apply such information for developing resilient crops for the future. PA metabolism inhibitors as well as transgenic approaches can be used to over-express and/or silence some of the rate-limiting PA biosynthetic and catabolic genes to test specific barley PA transgenics for their adaptability to leaf senescence phenomenon. PAs may, in future, play a role in reprogramming plant senescence that can be altered for a pro-growth phenotype by exogenously directed application of natural and synthetic PAs. This can help plants to develop tolerance to the broad spectrum of stress factors and thereby lead to higher plant productivity. The delineation of the roles of PAs in senescence should lead also to a better understanding of senescence-related cell death mechanisms and provide new knowledge about PCD in mammalian systems since PAs are universal bioregulators of these processes across kingdoms.

control and transgenic plants using RNA-Seq followed by NGS (Next Generation Sequencing) approaches should provide a broader picture of interconnected gene medleys that lead to senescence and other mechanisms regulated by PAs (**Figure 4**). The delineation of the roles of PAs in senescence should lead to a better understanding of senescence-related cell death mechanisms and provide new knowledge about senescence and PCD also in mammalian systems at cellular and molecular level since PAs are universal bioregulators of these processes across kingdoms. Such an outcome should also contribute in developing new strategies to be applied toward human health wellness.

#### DATA AVAILABILITY

All datasets for this study are included in the manuscript and/ or the supplementary files.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

#### FUNDING

This work was supported by the National Science Centre, Poland (Project Numbers 2018/29/B/NZ9/00734 6 to ES-N and 2017/27/N/NZ9/02135 to EP-L). AM is supported through USDA-ARS intramural Project No: 8042-21000-143-00D. The mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the US Department of Agriculture. USDA is an Equal Employment Opportunity provider.

in gene expression and signalling pathways between developmental and dark/starvation-induced senescence in *Arabidopsis*. *Plant J.* 42, 567–585. doi: 10.1111/j.1365-313X.2005.02399.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Sobieszczuk-Nowicka, Paluch-Lubawa, Mattoo, Arasimowicz-Jelonek, Gregersen and Pacak. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Polyamine Catabolism in Plants: A Universal Process With Diverse Functions

#### *Wei Wang1† , Konstantinos Paschalidis2† , Jian-Can Feng1 , Jie Song3 and Ji-Hong Liu3 \**

*1 College of Horticulture, Henan Agricultural University, Zhengzhou, China, 2 Department of Agriculture, School of Agricultural Sciences, Hellenic Mediterranean University, Heraklion, Greece, 3 Key Laboratory of Horticultural Plant Biology, College of Horticulture and Forestry Sciences, Huazhong Agricultural University, Wuhan, China*

#### *Edited by:*

*Antonio F. Tiburcio, University of Barcelona, Spain*

#### *Reviewed by:*

*Thomas Berberich, Senckenberg Nature Research Society, Germany Fernando Matias Romero, CONICET Institute of Biotechnological Research (IIB-INTECH), Argentina*

*\*Correspondence:* 

*Ji-Hong Liu liujihong@mail.hzau.edu.cn*

*† These authors have contributed equally to this work*

#### *Specialty section:*

*This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science*

*Received: 28 February 2019 Accepted: 12 April 2019 Published: 07 May 2019*

#### *Citation:*

*Wang W, Paschalidis K, Feng J-C, Song J and Liu J-H (2019) Polyamine Catabolism in Plants: A Universal Process With Diverse Functions. Front. Plant Sci. 10:561. doi: 10.3389/fpls.2019.00561*

Polyamine (PA) catabolic processes are performed by copper-containing amine oxidases (CuAOs) and flavin-containing PA oxidases (PAOs). So far, several CuAOs and PAOs have been identified in many plant species. These enzymes exhibit different subcellular localization, substrate specificity, and functional diversity. Since PAs are involved in numerous physiological processes, considerable efforts have been made to explore the functions of plant CuAOs and PAOs during the recent decades. The stress signal transduction pathways usually lead to increase of the intracellular PA levels, which are apoplastically secreted and oxidized by CuAOs and PAOs, with parallel production of hydrogen peroxide (H2O2). Depending on the levels of the generated H2O2, high or low, respectively, either programmed cell death (PCD) occurs or H2O2 is efficiently scavenged by enzymatic/nonenzymatic antioxidant factors that help plants coping with abiotic stress, recruiting different defense mechanisms, as compared to biotic stress. Amine and PA oxidases act further as PA back-converters in peroxisomes, also generating H2O2, possibly by activating Ca2+ permeable channels. Here, the new research data are discussed on the interconnection of PA catabolism with the derived H2O2, together with their signaling roles in developmental processes, such as fruit ripening, senescence, and biotic/abiotic stress reactions, in an effort to elucidate the mechanisms involved in crop adaptation/ survival to adverse environmental conditions and to pathogenic infections.

Keywords: polyamine catabolism, polyamine oxidases, ROS, plant development, fruit ripening and senescence, abiotic and biotic stress

### INTRODUCTION

Polyamines (PAs) are small aliphatic amines present in all living cells. For more than 100 years in biology, they were misunderstood as "ptomaine or food poisoning" substances by toxicologists (Cohen et al., 1981). The largely known PAs in plants are putrescine (Put), spermidine (Spd), and spermine (Spm). In addition, cadaverine (Cad) and thermospermine (t-Spm), a Spm isomer, are also reported to exist in higher plants.

Polyamine homeostasis is determined by PA metabolism, conjugation, interconversion, chemical alteration and transport (Moschou and Roubelakis-Angelakis, 2014; Handa et al., 2018; Nguyen et al., 2018; Tiburcio and Alcazar, 2018; Podlesakova et al., 2019). Biochemical effects of PAs have been unraveled in many physiological processes, primarily in stability and function of proteins

**20**

and nucleic acids (Handa et al., 2018), partly due to their positive charge that enables them to electrostatically interact with polyanionic molecules inside the cell. Polyamines correlate with numerous vital biochemical functions, including protein regulation (Takahashi and Kakehi, 2010; Sayas et al., 2019), regulation of chemiosmosis and photoprotection in chloroplasts (Ioannidis et al., 2016), ATP synthesis (Ioannidis et al., 2006), ion channeling (Pottosin et al., 2014a; Shabala et al., 2016), membrane fluidity (Paschalidis et al., 2010; Bleackley et al., 2014; Shabala et al., 2016; Dorighetto Cogo et al., 2018), and control of N/C balance (Moschou et al., 2012; Gupta et al., 2013; Majumdar et al., 2016). Exogenous PA application enhanced plant tolerance/resistance to several abiotic stress conditions, such as salinity, drought, water logging/flooding, osmotic stress, heavy metals, and extreme temperatures (Liu et al., 2006, 2015; Moschou et al., 2008c, 2012; Paschalidis et al., 2009a; Toumi et al., 2010; Moschou and Roubelakis-Angelakis, 2014; Gupta et al., 2016; Handa et al., 2018; Nguyen et al., 2018; Tiburcio and Alcazar, 2018; Pal et al., 2019; Yin et al., 2019). Polyamine application also enhanced tolerance to a few phytopathogenic infections *in planta*, such as *Alternaria alternata* (Estiarte et al., 2017), *Phytophthora capsici* (Koç, 2015), and *Pseudomonas viridiflava* (Rossi et al., 2015, 2018), and *in vitro*, such as *Fusarium* strains (Wojtasik et al., 2015) and *Sclerotinia sclerotiorum* (Garriz et al., 2003). The increase of host PA levels, either by using transgenic method or treatment with exogenous PAs, strongly decreased *in planta* growth of biotrophic pathogen *Pseudomonas viridiflava*, which was relieved by a PA oxidase (PAO) inhibitor (Marina et al., 2008). However, increase of leaf PA levels, by the same experimental approaches, led to increased necrosis *in planta* due to infection by *Sclerotinia sclerotiorum*, and the PA-induced increase of leaf necrosis after fungal infection was attenuated by inhibiting the activity of DAO and PAO (Marina et al., 2008). There is evidence that exogenous PA application modifies pathogenic responses depending on the strategy of the specific pathogen (Marina et al., 2008; Stes et al., 2011; Valdes-Santiago et al., 2012; Vilas et al., 2018).

Polyamines have crucial roles in a plethora of developmental procedures, including floral initiation and development (Liu et al., 2006, 2015; Liu and Moriguchi, 2007; Tavladoraki et al., 2016; Ahmed et al., 2017), leaf development and senescence (Kusano et al., 2008; Paschalidis et al., 2009b; Sobieszczuk-Nowicka et al., 2015; Sobieszczuk-Nowicka, 2017), fruit development and ripening (Liu et al., 2006; Liu and Moriguchi, 2007; Tsaniklidis et al., 2016; Fortes and Agudelo-Romero, 2018), and abiotic/biotic stress response (Alcazar et al., 2006, 2010; Moschou et al., 2008a; Liu et al., 2015, 2018; Montilla-Bascon et al., 2017).

Cellular PA levels are largely dependent on the dynamic regulation/balance among their biosynthesis, transport, and catabolism interchange. Polyamine biosynthesis has been thoroughly studied in the abovementioned physiological processes and a number of excellent literature reviews refer to their role (Kusano et al., 2008; Paschalidis et al., 2009a; Rangan et al., 2014; Liu et al., 2015; Majumdar et al., 2016; Fortes and Agudelo-Romero, 2018; Handa et al., 2018; Tiburcio and Alcazar, 2018; Wuddineh et al., 2018; Podlesakova et al., 2019). Nevertheless, there is a substantial lack of information on PA catabolism; so far, the enzymes involved in this process and the potential functions of their genes remain poorly characterized. As far as substrate specificity is concerned, it is well known that PAs are catalyzed by two major categories of amine oxidases, copper-containing amine oxidases (CuAOs) and flavin-containing PA oxidases (PAOs) (Cona et al., 2006), with cell type-specific functions in plant tissue/organ differentiation and development (Tavladoraki et al., 2016).

Emerging evidence suggests that PA catabolism plays a critical signaling role in a variety of cellular and developmental processes in all organisms, mediated *via* regulation of their homeostasis in reaction to intercellular and/or intracellular signs, as developmentally generated by abiotic and/or biotic alarms. In an effort to elucidate the underlined biological mechanisms, the latest advances are updated here on the function of CuAOs and PAOs, as sources of bio-reactive products, such as H2O2, during developmental processes with emphasis in fruit ripening and senescence, and, moreover, in abiotic/biotic stress reactions. The present approach might help in unraveling the role/use of the PA catabolic pathway in plants as a focus area for innovative stress resistance/tolerance approaches.

### ADVANCE IN POLYAMINE CATABOLISM RESEARCH

#### Copper-Containing Amine Oxidases in Polyamine Catabolism

Generally, in terms of substrate specificity, CuAOs exhibit strong preference for diamines (Put and Cad), and mainly catalyze their oxidation at primary amino groups, thus generating 4-aminobutanal, H2O2, and ammonia (Alcazar et al., 2010; Moschou et al., 2012). However, it has been demonstrated that some CuAOs in *Arabidopsis* also catalyze the oxidation of Spd (Planas-Portell et al., 2013). Recently, CuAO genes from apple (*Malus domestica*) exhibited different substrate preferences, with MdAO1 displaying elevated catalytic efficiency for 1,3-diaminopropane, Put, and Cad, whereas MdAO2 consumed only aliphatic and aromatic monoamines, comprising 2-phenylethylamine and tyramine (Zarei et al., 2015). Plant CuAOs usually exist at increased levels in dicot plants (Cona et al., 2006). Their genes have been identified in several species, as, for example, *Arabidopsis* (Møller and McPherson, 1998; Planas-Portell et al., 2013), chickpea (Rea et al., 1998), pea (Tipping and McPherson, 1995), tobacco (Paschalidis and Roubelakis-Angelakis, 2005b; Naconsie et al., 2014), apple (Zarei et al., 2015), grapevine (Paschalidis et al., 2009b), and sweet orange (Wang et al., 2017). *Arabidopsis* has at least ten recognized *CuAO* genes, however, only five of them (*AtAO1*, *AtCuAO1*, *AtCuAO2*, *AtCuAO3*, and *AtCuAO8*) have been characterized at protein level (Møller and McPherson, 1998; Planas-Portell et al., 2013; Ghuge et al., 2015; Groβ et al., 2017). The apple genome contains five putative *CuAO* genes with two of them (*MdAO1* and *MdAO2*) being identified at protein level (Zarei et al., 2015) and, recently, eight putative *CuAO* genes were reported in sweet orange (Wang et al., 2017).

As far as subcellular localization is concerned, plant CuAOs are separated into two groups (Zarei et al., 2015). The first group includes CuAOs that are typical extracellular proteins which contain an N-terminal signal peptide. Until now, seven CuAO members of the first group have been reported comprising *Pisum sativum* (PsCuAO), apple (MdAO2), *Arabidopsis* (AtAO1 and AtCuAO1), and sweet orange (CsCuAO4, CsCuAO5, and CsCuAO6) (Tipping and McPherson, 1995; Møller and McPherson, 1998; Planas-Portell et al., 2013; Zarei et al., 2015; Wang et al., 2017). The second group includes CuAOs localized in peroxisomes, containing a C-terminal peroxisomal targeting signal 1 (PTS1). At present, seven CuAO members of the second group have been reported, including two CuAOs from *Arabidopsis* (AtCuAO2 and AtCuAO3), two from tobacco (NtMPO1 and NtCuAO1), one from apple CuAO (MdAO1), and two from sweet orange (CsCuAO2 and CsCuAO3) (Planas-Portell et al., 2013; Naconsie et al., 2014; Zarei et al., 2015; Wang et al., 2017).

#### Polyamine Oxidases as Terminal and Back-Conversion Reaction Types in Polyamine Catabolism

In contrast to CuAO, in terms of substrate specificity, PAOs exhibit strong affinity for Spd, and Spm, as well as their derivatives (Alcazar et al., 2010). According to their functions in PA catabolism and subcellular localization, plant PAOs can be classified into two classes. The first class of PAOs (PA terminal catabolism reaction type) performs the oxidation and decomposition of Spd and Spm producing H2O2, 1,3-diaminopropane (DAP), and 4-aminobutanal (Spd catabolism) or N-(3-aminopropyl)-4 aminobutanal (Spm catabolism) (Cona et al., 2006; Angelini et al., 2010; Moschou et al., 2012; Tavladoraki et al., 2016; Bordenave et al., 2019). On the other hand, the second group (PA back-conversion reaction type) includes PAOs that catalyze the PA back-conversion reactions which convert Spm to Spd and Spd to Put (Moschou et al., 2012; Tavladoraki et al., 2016; Takahashi et al., 2018), in a reverse reaction of PA synthesis and produces 3-aminopropanal and H2O2. Although PAOs occur at high levels in monocot plants (Sebela et al., 2001), until now, *PAO* genes have been characterized in both monocots and dicots, including maize (Tavladoraki et al., 1998; Cervelli et al., 2000, 2006), rice (Ono et al., 2012), barley (Smith and Davies, 1985; Cervelli et al., 2006), *Arabidopsis* (Fincato et al., 2011), tobacco (Paschalidis and Roubelakis-Angelakis, 2005b; Yoda et al., 2006), grapevine (Paschalidis et al., 2009b), poplar (Tuskan et al., 2006), apple (Kitashiba et al., 2006), sweet orange (Wang and Liu, 2015, 2016), *Brachypodium* (Takahashi et al., 2018), tomato (Ono et al., 2012; Chen et al., 2016; Sagor et al., 2017; Hao et al., 2018), and upland cotton (Chen et al., 2015). So far, only six *PAO* genes that belong to the first group have been identified. The best characterized *PAO* gene of the first group is the maize *PAO* gene (*ZmPAO*) (Tavladoraki et al., 1998; Cona et al., 2006) and *PAO* genes from barley (*HvPAO1* and *HvPAO2*), rice (*OsPAO7*), sweet orange (*CsPAO4*), and *Brachypodium* (*BdPAO2*), which are proved to catalyze the PA terminal catabolism (Smith and Davies, 1985; Liu et al., 2014a; Wang et al., 2016; Takahashi et al., 2018). In contrast, most of the identified plant *PAO* genes belong to the second group. All of the five existing *PAO* genes in *Arabidopsis* (*AtPAO1*–*AtPAO5*) catalyze the PA back-conversion reactions (Tavladoraki et al., 2006; Kamada-Nobusada et al., 2008; Moschou et al., 2008c; Fincato et al., 2011; Ahou et al., 2014). In the rice genome, four (*OsPAO1*, *OsPAO3*, *OsPAO4*, and *OsPAO5*) out of seven (*OsPAO1–OsPAO7*) existing *PAO* genes execute the PA back-conversion reactions (Ono et al., 2012; Andronis et al., 2014; Liu et al., 2014b; Zarza et al., 2017). Similarly, in the tomato genome, four (SlPAO2, SlPAO3, SlPAO4, and SlPAO5) out of seven (SlPAO1– SlPAO7) existing PAO genes are suggested to execute the PA back-conversion reactions (Hao et al., 2018). On the other hand, six putative *PAO* genes have been identified in sweet orange and only one of them (*CsPAO3*) is demonstrated to catalyze the PA back-conversion reactions (Wang et al., 2016) and, of the 12 putative PAO genes (*GhPAO1–GhPAO12*) recognized in upland cotton, only one (*GhPAO3*) is verified to be implicated in the back-conversion pathway (Chen et al., 2017). To date, in terms of subcellular localization, all of the reports support that the PA terminal catabolic pathway is specifically activated in the apoplastic compartments (extracellularly), whereas the PA back-conversion pathway mainly occurs in the intracellular space (peroxisomes).

Beyond their functional/subcellular localization, in terms of either the terminal or the back-conversion type, PAOs exhibit further individual substrate specificities. The AtPAO1 only catalyzed the oxidation of Spm, but not Spd (Tavladoraki et al., 2006), while AtPAO3 preferred Spd as substrate instead of Spm (Moschou et al., 2008c). However, the AtPAO2 and the AtPAO4 present similar preference for both Spd and Spm (Fincato et al., 2011). Differently, AtPAO5 only uses t-Spm as its substrate and catalyzes the back-conversion of t-Spm to Spd (Kim et al., 2014). Furthermore, PAOs also exhibit individual reaction conditions, as, for example, they present different optimal pH values and temperature upon catalyzing different substrates. The optimal pH of catalytic activity for AtPAO2 is 7.5 towards both Spd and Spm, while the optimal pH for AtPAO4 catalytic activity towards Spd and Spm is 8.0 and 7.0, respectively (Fincato et al., 2011). In adddition, for CsPAO4 catalytic activity the optimal pH was 7.0 towards Spd and 8.0 towards Spm (Wang and Liu, 2016).

### POLYAMINE CATABOLISM IN PLANT DEVELOPMENT

Increasing studies report that PA catabolism is directly involved in plant development. Several evidence suggests that PA oxidation in the apoplast together with the generated reactive oxygen species (ROS) are involved in programmed cell death (PCD) and xylem differentiation (Corpas et al., 2019; Podlesakova et al., 2019). As early as 1998, Møller and McPherson found that *AtCuAO* localization in root xylem tissues is preceding and overlays with the synthesis of lignin in *Arabidopsis* (Møller and McPherson, 1998), and the PAO-generated apoplastic H2O2 levels considerably contribute to *Zea mays* leaf blade elongation (Rodriguez et al., 2009). In addition, the perturbation of PA catabolism by overexpressing the *ZmPAO* gene, as well as by down-regulating the S-adenosyl methionine decarboxylase (*SAMDC*) gene *via*

RNA interference, in tobacco, promotes vascular cell differentiation and induces PCD in root cap cells (Moschou et al., 2008b; Tisi et al., 2011). Recently, the *AtPAO5* has been reported to participate in the tightly controlled interplay between auxins and cytokinins, which are necessary for proper xylem differentiation (Alabdallah et al., 2017), and to regulate *Arabidopsis* growth through t-Spm oxidase activity (Kim et al., 2014).

Other studies suggest that PAs, along with ROS derived by their oxidation, control ion channeling in plant cells throughout normal and stress conditions, by affecting the plasma membrane ion transporting or acting as second messenger molecules (Pegg, 2014; Pottosin et al., 2014b). It has been reported that the Spd oxidase-produced H2O2 controls pollen plasma membrane hyperpolarization-activated Ca(2+)-penetrable canals and pollen tube growth (Wu et al., 2010). In *Arabidopsis thaliana*, differences in expression patterns are revealed for all of the AtPAO gene family members, as *AtPAO1* was mainly found in the transition area among meristems and elongation root regions, as well as in anther tapetum, and *AtPAO2* was most expressed in the pollen, quiescent center and columella initials, whereas AtPAO3 was predominantly identified in pollen, columella and guard cells. In addition, *AtPAO5* was specifically expressed in the root vascular system and in hypocotyls (Fincato et al., 2012). Moreover, the gene structure of *AtPAO5* was quite different from the other four *AtPAO* genes (Fincato et al., 2011). Its expression was detected during various growth stages, with the highest expression being observed in flowers, especially in sepals (Takahashi et al., 2010). AtPAO5 is classified as a cytosolic Spm oxidase/dehydrogenase protein undergoing proteasomal control (Ahou et al., 2014), that controls *Arabidopsis* growth *via* t-Spm oxidase activity (Kim et al., 2014; Liu et al., 2014d), while the rice OsPAO1 is a functional ortholog of AtPAO5 (Liu et al., 2014d) and the rice OsPAO7 is involved in lignin synthesis in anther cell walls (Liu et al., 2014c).

### POLYAMINE CATABOLISM IN FRUIT RIPENING AND SENESCENCE

Fruits usually keep higher PA levels at early developmental stages and are followed by a continuing decrease thereafter, especially at ripening stage (Fortes and Agudelo-Romero, 2018). This phenomenon has been reported in both climacteric and non-climacteric fruits, such as apple (Biasi et al., 1988), avocado (Kushad et al., 1988), peach (Liu and Moriguchi, 2007; Ziosi et al., 2009), mango (Malik and Singh, 2004), olive (Gomez-Jimenez et al., 2010), tobacco (Paschalidis and Roubelakis-Angelakis, 2005a), strawberry (Guo et al., 2018), raspberry (Simpson et al., 2017), oil palm (Teh et al., 2014), tomato (Rastogi and Davies, 1991; Tassoni et al., 2006; Mattoo et al., 2007; Van de Poel et al., 2013; Tsaniklidis et al., 2016; Liu et al., 2018), and grapevine (Paschalidis et al., 2009b; Agudelo-Romero et al., 2013; Fortes et al., 2015). As PA contents largely depend on the balance between anabolism and catabolism, it is necessary to unravel this balance during fruit ripening.

Although the precise roles of PA catabolism in fruit ripening are poorly understood, current studies reveal their tight interplay. High expression levels of CuAOs and PAOs during fruit ripening denote the involvement of PAs in associated physiological processes. For example, the free and conjugated PAs dramatically decrease during grape ripening, together with an up-regulation of the CuAO and PAO genes/enzymes and an increase of the H2O2 content (Agudelo-Romero et al., 2013), as well as an increase in γ-aminobutyric acid (GABA), a major product of PA catabolism (Fortes et al., 2015). These data suggest that increased PA oxidation might lead to decrease in PA titers. As PAO-derived ROS usually act as secondary messengers, the up-regulation of CuAOs/PAOs during ripening might establish an adequate ROS source for signaling actions driving to ripening hastening. In peach fruit, a jasmonate-induced ripening delay was closely related to increased PA levels (Ziosi et al., 2009). A few studies have unraveled strong indications for the interactions among PAs, PAO-derived products, and hormones, such as abscisic acid (ABA), cytokinins, auxins, and ethylene, aiming on their coordinated action in signaling pathways of several physiological processes, like fruit ripening and stress response (Podlesakova et al., 2019) (Agudelo-Romero et al., 2013). Inhibition of PA catabolism in grape with guazatine, a potent inhibitor of PAO activity, led to profound changes in amino acids, carbohydrates, and hormonal metabolism (Agudelo-Romero et al., 2013).

### POLYAMINE CATABOLISM IN ABIOTIC STRESS

Increasing evidences have showed that the plant PA catabolism is involved in various abiotic stresses responses, especially in salinity. Previously, it has been reported that GABA generated by CuAO-mediated PA oxidation exerts a substantial role in salinity stress response (Su et al., 2007). The PAOs exerting multifaceted roles on plant growth and salt stress response have been identified in, among others, tobacco, grapevine, sweet orange, tomato, and *Arabidopsis* (Moschou et al., 2008b; Paschalidis et al., 2010; Fincato et al., 2011; Wang and Liu, 2016; Gemes et al., 2017; Hao et al., 2018). Salinity induces tobacco cells to secrete exodus of Spd to the apoplast, where it is oxidized by PAO, thus generating abundant H2O2 and leading to enhanced PCD (Moschou et al., 2008a,b). A *PAO* gene of sweet orange (*CsPAO4*) has been further characterized functioning in PA terminal catabolism and playing an important role against salinity (Wang and Liu, 2016). This *CsPAO4* was overexpressed in tobacco, which significantly promoted the germination of transgenic seeds, while prominently inhibited the vegetative growth and root elongation of transgenic plants under salinity (Wang and Liu, 2016). The PAO activity provided a significant apoplastic production of ROS, which partly contributed to the maize leaf blade elongation under salt stress (Rodriguez et al., 2009). On the other hand, the peroxisomal AtPAO5 loss-of-function mutation in *Arabidopsis thaliana* exhibits constitutively higher t-Spm levels and activates metabolic and transcriptional reprogramming promoting salinity stress protection (Zarza et al., 2017). Moreover, PAO inhibitor treatment significantly decreased the H2O2 and NO production in tomato under salinity (Takacs et al., 2016), which indicates that PAO may contribute to H2O2 and NO production in order to cope with salinity and that the terminal activities of CuAO and PAO might play a role in cell death induction under lethal salt stress. The PA catabolism is also involved in many other abiotic stress responses, among others, in improved thermotolerance in *Nicotiana tabacum* by underexpressing the apoplastic PA oxidase (Mellidou et al., 2017), in aluminuminduced oxidative stress of wheat (Yu et al., 2018), in seleniuminduced H2O2 production in *Brassica rapa* (Wang et al., 2019), and in wound-healing by producing the necessary H2O2 for suberin polyphenolic domain and lignin synthesis catalyzed by peroxidase (Angelini et al., 2008).

#### POLYAMINE CATABOLISM IN PATHOGEN RESPONSE

Plants have developed a series of strategies to thwart pathogen attack (Vilas et al., 2018). The production of ROS is one of the defense responses against pathogen attack. Hydrogen peroxide may act either as an antimicrobial means preventing pathogen from growing or contributing as a signaling molecule, which induces the activation of protecting genes (Corpas et al., 2019).

As the terminal catabolism of PAs is followed by the generation of H2O2, PA catabolism is, thus, involved in pathogen defense response. The ornithine decarboxylase (ODC) activity increased 20-fold during the hypersensitive response (HR) to tobacco mosaic virus (TMV) infection; however, the levels of Put, Spd, and Spm were not greatly altered, as expected (Negrel et al., 1984), while the activities of arginine decarboxylase (ADC), ornithine decarboxylase (ODC), and CuAO were all obviously increased (Marini et al., 2001). In addition, the PAO expression level and PA titers were also increased in tobacco plants resistant to TMV (Yoda et al., 2003, 2006), suggesting that both PA biosynthesis and catabolism are activated in the host during pathogen infection, where appropriate.

The H2O2, resulted from increased activities of CuAO and PAO, might be the cause for the HR observed in barley after powdery mildew infection (Cowley and Walters, 2002). The increase of host PA levels limited bacterial growth, while inhibition of the PAO host enzymes increased the infection (Marina et al., 2008). It has been reported that DAO and PAO activities might play role in promoted defense against biotrophic or hemibiotrophic pathogens. However, these activities enhanced the infection of necrotrophic pathogens (Marina et al., 2008; Yoda et al., 2009; Moschou et al., 2009a). Similarly, the accumulation and further oxidation of free PAs was detected in the apoplast of tobacco leaves during tobacco defense against infection by microorganisms with contrasting pathogenesis strategies (Marina et al., 2008). This response affected the pathogen's ability to colonize host tissues and was detrimental for plant defense against necrotrophic pathogens, but it might be beneficial for plant defense against biotrophic pathogens because the former fed on necrotic tissue while the latter depended on living tissue for successful host colonization (Marina et al., 2008). Therefore, apoplastic PAs were suggested to play significant roles in plant-pathogen interactions and lead to significant changes in host susceptibility to different kinds of pathogens through regulation of host PA levels, particularly in the leaf apoplast (Marina et al., 2008). Similarly, tobacco plants overexpressing a *ZmPAO* unraveled a preinduced disease tolerance against the biotrophic bacterium *Pseudomonas syringae* pv *tabaci* and the hemibiotrophic oomycete *Phytophthora parasitica var nicotianae* (Moschou et al., 2009a), showing a critical role for a PAO-generated H2O2 apoplastic barrier for these fungi and bacteria. The PA catabolism also contributed to a resistance state through modulation of the immune response in grapevine following osmotic stress and/or after *Botrytis cinerea* infection (Hatmi et al., 2018). The pretreatment of stressed berries with appropriate inhibitors of DAO and PAO further increased PA level and greatly lowered defense responses, leading to higher susceptibility to *B. cinerea* (Hatmi et al., 2018). It is evident that the host PA apoplastic catabolism and the mediated H2O2 accumulation play an important signaling role in plant-pathogen interactions. However, the specific mechanisms of PA catabolism against plant resistance to pathogens are often more complicated. Further research is needed to clarify the exact role of PA catabolism in biotic stress resistance, in an effort to help plants cope with adverse environmental conditions and survive.

### POLYAMINE CATABOLISM AND H2O2 IN ABIOTIC AND BIOTIC STRESS RESPONSES

Stress conditions are accompanied by ROS accumulation and induce a composite signaling system recognized by endogenous plant cell sensors and transferred *via* secondary messengers to kinases, which lead to differentiations in gene expressions and related metabolites by means of the corresponding transcription factors in a plethora of processes identified as stress responses (Skopelitis et al., 2006; Waszczak et al., 2018).

In addition to several pathways, as, for example, photorespiration and electron transferring in chloroplasts and mitochondria, ROS are produced by apoplastic enzymes or enzymes that have different subcellular localization (Moschou and Roubelakis-Angelakis, 2014; Waszczak et al., 2018; Bordenave et al., 2019). The NADPH oxidase (Papadakis et al., 2005; Papadakis and Roubelakis-Angelakis, 2005; Andronis et al., 2014; Gemes et al., 2016), peroxidases (Papadakis et al., 2005; Paschalidis and Roubelakis-Angelakis, 2005b), oxalate oxidase (Angelini et al., 2008), xanthine dehydrogenase (Zarepour et al., 2010), and PAOs (Paschalidis and Roubelakis-Angelakis, 2005b; Moschou et al., 2008a,b,c; Paschalidis et al., 2009a, 2010; Takahashi et al., 2010; Gupta et al., 2016; Tavladoraki et al., 2016; Hao et al., 2018; Wu et al., 2018; Corpas et al., 2019) are included in these pathways, depending on each specific occasion.

Polyamines, as key compounds in plant physiology, are involved in this stress-signaling scheme, playing essential roles in the control of plant stress tolerance (Moschou and Roubelakis-Angelakis, 2014). Furthermore, numerous protein kinases are transcriptionally or posttranscriptionally regulated by PAs (Moschou and Roubelakis-Angelakis, 2014). Almost 3.5 centuries since their discovery – 1,678 in human semen – PAs still remain fundamental research interests, as they are widely implicated in a plethora of developmental and stress signaling responses. Proteomic and transcriptomic analyses on the PA-stress interplay and identification of over- or underexpressed key related genes, among others, ADC, ODC, SAMDC, Spd synthase (SPDS), Spm synthase (SPMS), CuAOs, and PAOs (Liu et al., 2006, 2015; Liu and Moriguchi, 2007; Tanou et al., 2014; Corpas et al., 2019) may offer a new insight into the molecular mechanisms controlling stress responses. Polyamines partially reversed the NaCl-induced phenotypic and physiological disturbances and systematically up-regulated the expression of PA biosynthesis (ADC, SAMDC, SPDS, and SPMS) and catabolism (DAO and PAO) genes, reprograming the oxidative and nitrosative status and the proteome of citrus plants exposed to salinity stress (Tanou et al., 2014). Recent transcriptomic analyses of the effect of Spd or norspermidine on *Arabidopsis* indicate up-regulation of the response to heat stress and denatured proteins, inhibiting protein ubiquitylation, both *in vivo* and *in vitro*, and this interferes with protein degradation by the proteasome, a situation known to deplete cells of amino acids (Sayas et al., 2019). Furthermore, by *in situ* RNA–RNA hybridization approaches, the spatial contribution of ODC1, 2; ADC2; and CuAO gene transcripts has been largely elucidated in developing tomato fruits in order to decode the potential connection of PA anabolism/catabolism to developmental processes, like fruit ripening (Tsaniklidis et al., 2016).

Polyamines may further alleviate the unfavorable stress effects by activating the antioxidant machinery (Podlesakova et al., 2019). Spd and Spm, and to a lesser extent, Put, inhibit NADPHoxidase, whereas Put prevents the induction of PCD (Papadakis and Roubelakis-Angelakis, 2005; Andronis and Roubelakis-Angelakis, 2010; Andronis et al., 2014). Abiotic and biotic stress may cause radical alterations in PA metabolism. Several model systems, like *Arabidopsis thaliana*, have helped in deciphering the role of PAs and elucidating their metabolic paths (Gupta et al., 2016). The preservation of an appropriate balance of the PA catabolic pathways with the H2O2 dual role under normal and stress conditions has helped in illuminating the plant adaptation mechanisms (Paschalidis and Roubelakis-Angelakis, 2005b; Paschalidis et al., 2010; Gupta et al., 2016). ABA is an upstream signal for the induction of the polyamine catabolic pathway in the apoplast of grapevine, thus, amine oxidases are producing H2O2 which signals stomatal closure (Paschalidis et al., 2010). When the titers of H2O2 are below a threshold, expression of tolerance effector genes is induced, while when it exceeds this threshold, the PCD syndrome is induced (Paschalidis et al., 2010). Polyamines also increase nitric oxide and ROS in guard cells of *Arabidopsis thaliana* during stomatal closure (Agurla et al., 2018) and during growth inhibition in *Triticum aestivum* L seedlings (Recalde et al., 2018). In addition, the redox gradient across plasma membranes may play an essential role in climate changes, as a redox signaling regulator (Gupta et al., 2016).

Plant life and stress go hand-to-hand. During growth, in order to overcome abiotic stress conditions, plants develop a remarkable organ/tissue/age-specific PA-related phenotypic plasticity (Paschalidis and Roubelakis-Angelakis, 2005a). Under favorable conditions, a balanced hypogeous and hypergeous PA homeostasis is critical to allow constant water/nutrient uptake and photosynthetic flux, respectively. For example, PA genes/metabolites may contribute to an accurate adaptation of the shift between advancement in cell cycle/cell division, that pushes the growth of very young root/shoot primordia toward cell expansion, differentiation, and lignification (Paschalidis and Roubelakis-Angelakis, 2005b; Paschalidis et al., 2009a). On the contrary, plant growth under abiotic stress might, among other effects, wound/wilt the leaf surface or increase evaporation, rendering plant susceptibility. In this case, plants constantly examine whether or not the environmental signals are favorable for their development/growth, and might redirect a PA-associated phenotypic plasticity, involving Η2Ο2, the product of PA catabolism, either for growth or for stress adaptation, e.g., *via* spermidine-mediated stomatal closure (Paschalidis et al., 2009b, 2010). It is also specified that the seriousness/type of reaction (s) to (a)biotic stress is a cell/tissue/ organ/age-specific route, related to PA catabolism (Paschalidis et al., 2009b, 2010). The assessment of the antioxidant genes/ machinery, along with the photosynthetic factors, the intracellular cation titers, and the PA interplay in over/underexpressing ZmPAO plants under prolonged/varying salinity (Gemes et al., 2016, 2017) and heat (Mellidou et al., 2017) stress have highlighted a plant ontogenetic stage-specific role for PA oxidase and Η2Ο2 during plant developmental reactions to (a)biotic stress conditions.

During abiotic stress conditions, PAs (mainly Spd) are secreted in the apoplast and oxidized by PAOs (they refer to both CuAOs and PAOs, but, for simplicity, they are depicted only as PAOs, throughout the model presentation) (**Figure 1**), resulting in PA catabolism intermediates. The level of PAO-mediated Spd oxidation results in: (①) moderate apoplastic PAO oxidizing Spd at a small percentage producing modest (beneficial) H2O2 (and 1,3-diaminopropane) contents, that act as signaling molecules, inducing a ROS-dependent protective pathway, thus triggering abiotic stress tolerance reactions; (②) high apoplastic PAO, over a specific threshold, oxidizing Spd considerably faster, producing high (harmful) H2O2 levels, and resulting in down-regulation of pro-survival genes and execution of a specific PCD pathway in plants under abiotic stress conditions (**Figure 1**; Moschou et al., 2008b; Moschou and Roubelakis-Angelakis, 2014; Gupta et al., 2016; Corpas et al., 2019).

A possible scenario below may be postulated in order to explain the stress signaling/defense. Abiotic stress induces the production of intracellular and extracellular H2O2, higher PAs, and second messengers like Ca2+ (Wu et al., 2010). The higher PA levels, when oxidized, generate additional H2O2 that activates the plant antioxidant machinery. Indeed, under salt stress conditions, with increased levels of endogenous PAs induced by exogenously applied Spd, PAO activity is further enhanced, thus contributing to H2O2 accumulation, subsequently inducing enhanced antioxidant defense, which is helpful for growth (Wu et al., 2018). A cold-responsive ethylene-responsive factor from *Medicago falcata* was demonstrated to confer cold tolerance by upregulating polyamine turnover, antioxidant protection, and proline accumulation (Zhuo et al., 2018). ABA endogenous contents are also activated by stress,

transferred by several cellular biochemical pathways. Abiotic and biotic stresses result in ROS production. The stress-signaling pathway gives also rise to intracellular PAs, which are secreted/oxidized in the apoplast by PAOs in order to supply H2O2 and several N compounds. Hydrogen peroxide and N molecules may involve further reactions, including, among others, mitogen-activated protein kinases (MAPKs) and Oxidative Signal Inducible 1 (OXI1) pathways (Rentel et al., 2004; Moschou et al., 2009a,b, 2012; Toumi et al., 2010; Moschou and Roubelakis-Angelakis, 2014; Gupta et al., 2016; Podlesakova et al., 2019). Under abiotic conditions, according to the H2O2 level created: ① when low (H2O2 below a specific threshold), it is powerfully scavenged leading to abiotic defense or ② when high (H2O2 over a specific threshold), it cannot be efficiently scavenged and PCD is caused. The abiotic tolerance stress signal (③) is received by plants generating essential signal molecules like ABA that are involved in an augmentation of PA synthesis rendering tolerance in plants (Toumi et al., 2010). Under biotic stress conditions, mostly Spm is secreted in the apoplast and oxidized by the respective enriched PAO, causing a H2O2 buildup (④, biotic defense) that protects plants from phytopathogenic bacteria. In the scavenging process, antioxidant enzymes are involved, such as ascorbate peroxidase (APX), in a procedure rendering defense reactions. The implication of PA oxidation to H2O2 production is not only a matter of apoplastic or cytoplastic PAOs. Polyamines are also back-converted in peroxisome, with the parallel generation of H2O2 and nitrogenous substances. Peroxisomally produced H2O2 might trigger Ca2+-penetrable canals (Wu et al., 2010; Moschou et al., 2012; Zepeda-Jazo and Pottosin, 2018; Corpas et al., 2019). However, the N compounds generated as a result of the PA back-conversion path are not yet elucidated. Further details are found in the text.

which may trigger ROS-related routes involving PAs (Gupta et al., 2016). A well-organized protection mechanism comprising of PAs, Ca2+, ABA, and H2O2 coordinates an adaptation response of plants to stress (**Figure 1**; Skopelitis et al., 2006; Moschou et al., 2008b; Paschalidis et al., 2010; Toumi et al., 2010; Moschou and Roubelakis-Angelakis, 2014; Gupta et al., 2016; Majumdar et al., 2016; Gemes et al., 2017; Handa et al., 2018).

Under abiotic stress conditions, ABA triggers the PA machinery in tolerant/sensitive grapevine genotypes (Toumi et al., 2010). The abiotic tolerance stress signal (③) is received by plants generating essential signal molecules like ABA that are involved in an augmentation of PA synthesis (**Figure 1**). Tolerant plants showed higher PA synthesis, as compared with the sensitive, giving rise to higher PA levels (Toumi et al., 2010). Regardless the genotype competence in withstanding stress, PAs follow the secretion way and are oxidized in the apoplast by PAOs (Paschalidis et al., 2010; Toumi et al., 2010). In this way, higher intracellular PA titers and higher PA synthesis, together with the apoplastic PAO-derived H2O2, are participating in a "positive feedback loop" helping to maintain homeostasis and enhance tolerance through activation of further defense mechanisms. On the contrary, lower PA titers/anabolism enhance PCD syndrome (Paschalidis et al., 2010; Toumi et al., 2010; Gemes et al., 2017).

This model/hypothesis elucidates the role of mostly the intercellular PAs. In abiotic-induced PCD of down-regulated SAMDC tobacco plants, the cellular Spd and Spm levels were reduced, but, unexpectedly, these plants showed similar, to the wild type, PA levels and oxidation in the apoplast (Moschou et al., 2008b). The plants with silenced SAMDC unravel a PA-dependent trade-off between growth and tolerance reactions (Mellidou et al., 2016) and the stimulation of the ADC pathway acts as a positive feedback loop to maintain the PA homeostasis (Toumi et al., 2010).

A biotic stress-induced increase in PAO gene/enzyme occurred in overexpressing PAO tobacco plants infected by *Pseudomonas syringae* pv *tabaci* (Moschou et al., 2009a). Under biotic stress conditions, mostly Spm is secreted in the apoplast and oxidized by the respective enriched PAO, causing a H2O2 buildup (④, biotic defense) that protects plants from phytopathogenic bacteria (**Figure 1**). In this context, overexpressing PAO plants reveal a preinduced tolerance against diseases, such as the biotrophic bacterium *Pseudomonas syringae* pv *tabaci* and the hemibiotrophic oomycete *Phytophthora parasitica* var *nicotianae* (Moschou et al., 2009a). PAO and DAO activities promote defense against biotrophic or hemibiotrophic pathogens and, by contrast, these activities favor the spread of the lesions provoked by necrotrophic pathogens (Marina et al., 2008; Yoda et al., 2009; Moschou et al., 2009a). Oxidation of others polyamines, such as t-Spm, is also involved in response to pathogenic bacteria, increasing *Arabidopsis* resistance to *Pseudomonas viridiflava* (Marina et al., 2013). This is probably related to the ability of plant PAOs to oxidize t-Spm in a wide range of tissues and organs, as occurs when other PAs such as Spm are accumulated throughout the plant (Marina et al., 2013). In addition to that, pathogens activate their own and the plant PA metabolism during the compatible interaction between tomato and *Pseudomonas syringae* (Vilas et al., 2018). This activation results in the accumulation of Put in whole leaf tissues, as well as in the apoplastic fluids, which is explained by the induction of its synthesis in plant cells and also on the basis of its excretion by bacteria (Vilas et al., 2018). *Ralstonia solanacearum* also produces abundant Put, acting as a virulence metabolite and accelerating wilt disease, possibly reducing ROS in the host (Lowe-Power et al., 2018). The present dual abiotic and biotic stress protection scheme may represent an innovative route for generating tolerant transgenic plants to a variety of environmental and phytopathogenic stress factors.

#### POLYAMINES ACT AS ORTHODOX-CONCERTERS OR STRESS-RELIEVERS

During development, several molecules exist inside common plant tissues in normal environmental and phytopathogenicfree states, concerting an orthodox plant behavior. However, as soon as normal conditions are substituted by stressful ones, these molecules begin to work as stress-relievers. In this work, PAs are suggested to work as such molecules, i.e., as "orthodox-concerters" under normal conditions and as "stressrelievers" under stressful ones. Polyamines have established duties inside plants; however, when they are found in adverse conditions, they may reveal novel functions, not expected until that time. Polyamines, PA oxidases, and the generated H2O2 all have specific roles in sustaining plant developmental procedures, such as fruit ripening and senescence. Furthermore, in this work, the role for the concerted action of PA catabolism and its products, in reaction to both abiotic and biotic stress

#### REFERENCES

Agudelo-Romero, P., Bortolloti, C., Pais, M. S., Tiburcio, A. F., and Fortes, A. M. (2013). Study of polyamines during grape ripening indicate an important role of polyamine catabolism. *Plant Physiol. Biochem.* 67, 105–119. doi: 10.1016/j.plaphy.2013.02.024

are discussed. The PA oxidation will surely remain a fascinating area for scientific examination, as its concerted action with the generated H2O2 is shown to classify specific stressful parameters and build an effective defense device.

#### CONCLUSION

To date, many attempts have been made to investigate the roles of PA catabolism in plant growth, development, fruit ripening, and responses to biotic and abiotic stresses. Therefore, the understanding of the roles played by CuAOs and PAOs in these processes has progressed significantly during the recent decades, especially in rice and *Arabidopsis*. However, many key questions remain unanswered. Firstly, current studies show that the homeostasis regulation of PAs in plants is rather complex. So far, the information about specific regulatory mechanisms in PA biosynthesis and catabolism is very limited. Although it has been revealed, among others, that the transcription of PA biosynthetic genes is regulated by several transcription factors under stress (Paschalidis and Roubelakis-Angelakis, 2005a; Paschalidis et al., 2009a; Sun et al., 2014; Wu et al., 2016; Liu et al., 2018), relatively less information is available on the transcriptional regulation of PA catabolism. Secondly, although many members of CuAOs and PAOs involved in PA back-conversion pathway have been identified, the explicit role of the PA back-conversion reactions in PA homeostasis and associated physiological processes remains obscure. Last, but not least, although a dual signaling role for PA catabolism and the generated H2O2 under abiotic and/or biotic plant stress conditions has been revealed, further study will enable researchers to better elucidate this role by using new era technology.

#### AUTHOR CONTRIBUTIONS

WW, KP, J-CF, and JS wrote the paper. J-HL conceived the work and finalized the MS.

#### FUNDING

This work was financially supported by National Key Research and Development Program of China (2018YFD1000300), National Natural Science Foundation of China (31320103908), and Hubei Provincial Natural Science Foundation for Innovative Group (2017CFA018), Major Science and Technology Project in Henan Province (151100110900) and Scientific and Technological Project of Henan Province(182102110480).


for spermidine in the timing of flowering and other developmental response pathways. *Plant Sci.* 258, 146–155. doi: 10.1016/j.plantsci.2016.12.002


of grapevine berries to gray mold by priming polyamine accumulation. *Front. Plant Sci.* 9:1010. doi: 10.3389/fpls.2018.01010


interactions: the pathway is regulated at the post-transcriptional level. *Front. Plant Sci.* 7:78. doi: 10.3389/fpls.2016.00078


a cell wall copper amine oxidase in chickpea seedlings. *FEBS Lett.* 437, 177–182. doi: 10.1016/S0014-5793(98)01219-8


Zepeda-Jazo, I., and Pottosin, I. (2018). Methods related to polyamine control of cation transport across plant membranes. *Methods Mol. Biol.* 1694, 257–276. doi: 10.1007/978-1-4939-7398-9\_23

Zhuo, C., Liang, L., Zhao, Y., Guo, Z., and Lu, S. (2018). A cold responsive ethylene responsive factor from Medicago falcata confers cold tolerance by up-regulation of polyamine turnover, antioxidant protection, and proline accumulation. *Plant Cell Environ.* 41, 2021–2032. doi: 10.1111/pce.13114

Ziosi, V., Bregoli, A. M., Fregola, F., Costa, G., and Torrigiani, P. (2009). Jasmonateinduced ripening delay is associated with up-regulation of polyamine levels in peach fruit. *J. Plant Physiol.* 166, 938–946. doi: 10.1016/j.jplph.2008.11.014

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Wang, Paschalidis, Feng, Song and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Transcriptional Modulation of Polyamine Metabolism in Fruit Species Under Abiotic and Biotic Stress

*Ana Margarida Fortes1 \*, Patricia Agudelo-Romero2,3,4 \*, Diana Pimentel1 and Noam Alkan5*

*1 Faculdade de Ciências de Lisboa, Department of Plant Biology, Biosystems and Integrative Sciences Institute, Universidade de Lisboa, Lisbon, Portugal, 2 School of Molecular Science, The University of Western Australia, Perth, WA, Australia, 3 ARC Centre of Excellence in Plant Energy Biology, The University of Western Australia, Perth, WA, Australia, 4 Telethon Kids Institute, University of Western Australia, Nedlands, WA, Australia, 5 Department of Postharvest Science of Fresh Produce, Agricultural Research Organization, Volcani Center, Rishon LeZion, Israel*

#### *Edited by:*

*Deyu Xie, North Carolina State University, United States*

#### *Reviewed by:*

*Laura Valdes-Santiago, Instituto Tecnológico Superior de Irapuato, Mexico Andrés Gárriz, National Council for Scientific and Technical Research (CONICET), Argentina*

#### *\*Correspondence:*

*Ana Margarida Fortes amfortes@fc.ul.pt Patricia Agudelo-Romero patricia.agudeloromero@ telethonkids.org.au*

#### *Specialty section:*

*This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science*

*Received: 04 March 2019 Accepted: 06 June 2019 Published: 02 July 2019*

#### *Citation:*

*Fortes AM, Agudelo-Romero P, Pimentel D and Alkan N (2019) Transcriptional Modulation of Polyamine Metabolism in Fruit Species Under Abiotic and Biotic Stress. Front. Plant Sci. 10:816. doi: 10.3389/fpls.2019.00816*

Polyamines are growth regulators that have been widely implicated in abiotic and biotic stresses. They are also associated with fruit set, ripening, and regulation of fruit qualityrelated traits. Modulation of their content confers fruit resilience, with polyamine application generally inhibiting postharvest decay. Changes in the content of free and conjugated polyamines in response to stress are highly dependent on the type of abiotic stress applied or the lifestyle of the pathogen. Recent studies suggest that exogenous application of polyamines or modulation of polyamine content by gene editing can confer tolerance to multiple abiotic and biotic stresses simultaneously. In this review, we explore data on polyamine synthesis and catabolism in fruit related to pre- and postharvest stresses. Studies of mutant plants, priming of stress responses, and treatments with polyamines and polyamine inhibitors indicate that these growth regulators can be manipulated to increase fruit productivity with reduced use of pesticides and therefore, under more sustainable conditions.

Keywords: abiotic stress, biotic stress, fruit ripening, grape, polyamine, tomato

### INTRODUCTION

Polyamines (PAs) are small aliphatic amines that regulate various cellular functions. These compounds present at least two amino groups; the diamine putrescine (Put), the triamine spermidine (Spd), and the tetraamine spermine (Spm) are the most common PAs in plants (Mattoo et al., 2010). PAs often occur as free molecular bases, but they can also be covalently associated with small molecules, namely, phenolic acids (conjugated forms), and with various macromolecules such as proteins (bound forms) (Mattoo et al., 2010).

PAs are growth regulators that have been implicated in abiotic and biotic stresses (Cona et al., 2006; Cuevas et al., 2008; Marina et al., 2008; Alcázar et al., 2010; Gonzalez et al., 2011; Nambeesan et al., 2012; Agudelo-Romero et al., 2014; Minocha et al., 2014; Pál et al., 2015), as well as in plant morphogenesis and development (Applewhite et al., 2000; Fortes et al., 2010, 2011; Tiburcio et al., 2014; Jancewicz et al., 2016), senescence (Pandey et al., 2000;

Sobieszczuk-Nowicka, 2017), and fruit development and ripening (Mattoo and Handa, 2008; Agudelo-Romero et al., 2013, 2014; Tavladoraki et al., 2016). Several publications have suggested that the role played by these growth regulators in plant–microbe interactions is either exerted directly by PAs functioning as signaling molecules or mediated through the products of their catabolism together with jasmonic acid (JA), abscisic acid (ABA), salicylic acid (SA), auxins, cytokinins, and ethylene (Jiménez-Bremont et al., 2014).

In plants, Put is synthesized through arginine decarboxylase (ADC) and/or ornithine decarboxylase (ODC) (Alcázar et al., 2010). Arginase hydrolyzes arginine to urea and ornithine. Conversion of Put to Spd and Spm requires the activity of Spd synthase and Spm synthase, respectively. Thermospermine synthase is involved in the synthesis of the tetraamine thermospermine (Jiménez-Bremont et al., 2014). *S*-adenosylmethionine decarboxylase (SAMDC) carries out a rate-limiting step in the biosynthesis of decarboxylated SAM, which donates the aminopropyl moiety for the biosynthesis of these PAs. Catabolism of PAs involves diamine oxidases (CuAOs) and polyamine oxidases (PAOs). Intracellular levels of PAs are mostly regulated by anabolic and catabolic processes, as well as by their transport and conjugation with phenolic compounds, mainly hydroxycinnamic acids (Jiménez-Bremont et al., 2014; Fortes and Agudelo-Romero, 2018). The transport of PAs into different cell compartments is a crucial step in regulating the intracellular levels of these free forms, thereby interfering with cellular processes. However, only a few PA transporters have been characterized (Jiménez-Bremont et al., 2014). In addition, PAs have been connected to metabolic pathways involving ethylene, γ-aminobutyric acid (GABA), nitric oxide, the Krebs cycle, and ABA (Alcázar et al., 2010).

In this mini-review, we will examine recent data focusing on the modulation of PA metabolism in plants and fruit under pre- and postharvest abiotic stresses and during interactions with pathogens.

#### REPROGRAMMING OF POLYAMINE SYNTHESIS, CATABOLISM, AND CONJUGATION IS INVOLVED IN ABIOTIC AND BIOTIC STRESS RESPONSES

#### Transcriptional Modulation of Polyamine Metabolism Under Individual Stresses

Several datasets have been obtained in tomato and grape related to transcriptome reprogramming under abiotic and biotic stresses (**Figures 1, 2**). It is clear that extensive modulation of PA metabolism occurs in both leaves and fruit in response to drought, salt, heat, and a variety of pathogens, such as viruses, fungi, and bacteria. Responses to abiotic stress involve mainly upregulation of genes encoding enzymes involved in PA biosynthesis, namely, ADC and SAMDC. However, *SAMDC* was found to be downregulated in grapevine under water and salt stresses. In this respect, it is interesting to note that different functional roles have been observed with the genetic divergence of the *Arabidopsis thaliana SAMDC* gene family (Majumdar et al., 2017). The gene encoding thermospermine synthase was upregulated in tomato under water stress and upon calcium treatment. In tomato, *ODC* was upregulated in response to calcium and downregulated in ABA-treated plants and in those exposed to water stress. This gene seemed to be more modulated upon biotic stress. The gene encoding arginase was downregulated in response to several abiotic stresses in tomato, suggesting that synthesis of PAs occurs preferentially through ADC. Interestingly, transcriptional modulation of genes involved in PA catabolism is strongly dependent on the type of abiotic stress applied and likely involves activity reprogramming of different isoenzymes, as suggested by the differential expression of several genes.

On the other hand, changes in PA metabolism in response to biotic stress seem to involve mainly inhibition of PA catabolism *via* downregulation of genes encoding CuAO and PAO. The gene encoding Spd synthase was downregulated in tomato infected with the fungus *Pseudomonas syringae*, and a virus. In addition, genes involved in PA biosynthesis (encoding arginase, ODC, and ADC) were upregulated (**Figure 2**). These results suggest activation of PA metabolism in response to biotic stresses such as bacteria and fungi, but less pronounced involvement in response to viruses.

#### Polyamine Metabolism Under Individual and Multiple Stresses Using Transgenic, Biochemical, and Physiological Approaches

Transgenic tomato plants overexpressing the human *SAMDC* gene had higher PA content than the wild type. This transgenic tomato line showed increased resistance to two important fungal pathogens, *Fusarium oxysporum* causing Fusarium wilt and *Alternaria solani* causing early blight, as well as tolerance to multiple abiotic stresses, such as salinity, drought, cold, and high temperatures (Hazarika and Rajam, 2011). Interestingly, these transgenic plants also showed higher conversion of free Put and Spd to conjugated PAs when infected with pathogens (Hazarika and Rajam, 2011). In fact, conjugated PAs have been shown to have antimicrobial properties (Walters, 2003). These PAs contribute to cell-wall strengthening, thereby protecting it against the activity of microbial hydrolytic enzymes [reviewed by Fortes and Agudelo-Romero (2018)].

Similarly, transgenic eggplants overexpressing the oat *ADC* gene acquired resistance to Fusarium wilt disease (Prabhavathi and Rajam, 2007). These plants showed increased ADC activity and accumulation of PAs, particularly the conjugated forms of Put and Spm. Since CuAO activity was also enhanced, it was suggested that the acquisition of resistance might involve both PA biosynthesis and degradation.

Other studies that did not involve transgenic approaches also revealed the importance of PAs in resistance to biotic stress. One study demonstrated that NH4 + induces resistance to *P. syringae* pv*. tomato DC3000* (*Pst*), by causing mild toxicity in tomato plants and inducing basal H2O2, ABA, and Put

FIGURE 1 | guazatine treatment, and light. Tomato heatmaps were generated using RNAseq (SL2.50 genome) and microarray (GPL4741) approaches. Grape heatmaps were generated using two microarray platforms: GrapeGen (GPL11004) and GeneChip (GPL1320). RNAseq data were downloaded from the Sequence Read Archive repository (SRA; https://www.ncbi.nlm.nih.gov/sra) and microarray data were downloaded from the Gene Expression Omnibus repository (GEO; https://www.ncbi.nlm.nih.gov/geo/) using the GEOquery R library.

accumulation (Fernández-Crespo et al., 2015). Treatment with inhibitors of Put accumulation showed that Put plays a role in resistance to *Pst* in tomato plants (Fernández-Crespo et al., 2015). On the other hand, the increase in H2O2, leading to a strong and rapid oxidative burst, was partially attributed to the activity of CuAOs. Export of PAs to the apoplast is a common source of H2O2 in abiotic stress and in host- and nonhost hypersensitive responses during pathogen infection (Yoda et al., 2009; Pottosin et al., 2014). Oxidation of PAs can generate an oxidative burst, leading to induction of defenseresponse genes and the hypersensitive response (Cona et al., 2006; Yoda et al., 2009). The activation of PA metabolism by NH4 + supplementation was suggested to be mediated by ABA-dependent signaling pathways (Toumi et al., 2010). Similarly, NH4 <sup>+</sup> application and *Colletotrichum coccodes* inoculation of tomato fruit induced NADPH oxidase, leading to an oxidative burst, activation of the SA pathway, and upregulation of *ODC* and other PA-related genes, ultimately resulting in programmed cell death (Alkan et al., 2009, 2012).

Under conditions of abiotic and biotic stress, PA oxidation might alter cell redox homeostasis and modulate hormone signaling (Moschou et al., 2008; Pál et al., 2015; Seifi and Shelp, 2019). However, PA synthesis and oxidation control its homeostasis and can lead to PA excess or deficiency, which could lead to susceptibility to stress (Aziz et al., 1999; Nambeesan et al., 2012; Hatmi et al., 2015). Thus, fine-tuning of PAs could be important to coordinating stress responses (Pál et al., 2015).

Interestingly, drought stress was also described to prime the immune response of grapevine against *Botrytis cinerea* infection *via* modulation of PA biosynthesis and catabolism. Drought stress led to upregulation of the genes encoding ADC, CuAOs, and PAOs and their corresponding enzymes' activities (Hatmi et al., 2015). Plants from a *B. cinerea*-tolerant cultivar subjected to drought stress exhibited significantly higher Put accumulation and a decrease in Spd and Spm levels as compared to plants from the sensitive cultivar. Hatmi et al. (2015) indicated that PA synthesis and oxidation and increased contents of some PA-related amino acids, together with increased content of stilbenes and upregulation of immune response-related genes were involved in the increased tolerance to *B. cinerea*. CuAO and PAO inhibitors attenuated drought-induced defense responses and enhanced disease susceptibility in grapevine. Furthermore, grapes treated with guazatine, a potent inhibitor of PAO activity, showed downregulation of genes encoding CuAOs (**Figure 1**) and the pathogenesis-related protein 1 precursor (*PR1* gene) involved in the biotic stress response (Agudelo-Romero et al., 2014).

Upon infection with *B. cinerea*, grape fruit from a highly susceptible cultivar presented upregulation of *CuAO* and *ODC* at an early stage of ripening together with increased transcription of stilbene synthases involved in stilbene synthesis (Agudelo-Romero et al., 2015). However, with the onset of ripening, these genes were no longer upregulated in infected berries but others involved in PA metabolism such as *Spm synthase* (**Figure 2**). Similarly, accumulation of Spd in transgenic tomato (overexpressing the yeast Spd synthase gene) was associated with weakened ethylene-induced defense responses, thereby increasing the fruit's susceptibility to *B. cinerea* (Nambeesan et al., 2012). Osmotic stress also induced PA accumulation and inhibited the defense response in ripe berries after *B. cinerea* infection (Hatmi et al., 2018). In fact, the plant's response to individual stresses may differ from that to multiple stresses, which could lead to opposite effects on PA metabolism. However, grapevine plants exposed to osmotic stress that induced PA oxidation and later inoculated with *B. cinerea* showed a great reduction in CuAO and PAO activities, consistent with the enhanced levels of PAs and impaired defense responses to the fungus (Hatmi et al., 2014).

It is not known whether fungi can reprogram PA metabolism in fruit, or how this may differ depending on genotype, ripening stage and fruit susceptibility. However, interactions between microbial effectors and plant enzymes involved in PA metabolism have been reported [reviewed by Jiménez-Bremont et al. (2014)]. Microbes can also produce PAs and alter plant physiology and resilience to stress (Kim et al., 2013; Jiménez-Bremont et al., 2014). In one study, the effector AvrBsT of *Xanthomonas campestris* pv. *vesicatoria* (Xcv) interacted with ADC, leading to induction of a hypersensitive response in bell pepper fruit (Kim et al., 2013). In bell pepper, the gene *CaADC1* is constitutively expressed in stems, roots, flowers, and fruit, but not leaves. However, *CaADC1* was highly induced in leaves during avirulent (incompatible) Xcv infection compared to the mock control or virulent (compatible) Xcv infection. Silencing of *CaADC1* in bell pepper leaves significantly compromised nitric oxide and H2O2 accumulation as well as cell-death induction, leading to enhanced avirulent Xcv growth during infection (Kim et al., 2013). Based on these findings, the authors suggested that *CaADC1* acts as a key defense and cell-death regulator *via* mediation of PA metabolism.

### EXOGENOUS APPLICATION OF POLYAMINES AFFECTS FRUIT RESISTANCE TO ABIOTIC AND BIOTIC STRESSES

Chilling injury results in a significant increase in PA levels in many fruit (Serrano et al., 1997, 1998; González-Aguilar et al., 2000; Rodriguez et al., 2001), as does mechanical damage (Valero et al., 1998; Martínez-Romero et al., 1999, 2000; Pérez-Vicente et al., 2002). This suggests that PAs protect fruit

FIGURE 2 | Expression of genes involved in polyamine metabolism in tomato (A) and grapevine (B) under biotic stresses: *Colletotrichum gloeosporioides*, *Pseudomonas syringae*, *Sclerotinia sclerotiorum*, potato spindle tuber viroid (PSTVd), tobacco rattle virus (TRV), *Botrytis cinerea*, *Colletotrichum coccodes*, *Phytophthora infestans*, *Pyrenochaeta lycopersici*, *Alternaria solani*, *Ralstonia solanacearum*, spider mite, Bois noir, and grapevine leafroll-associated virus 3 (GLRaV-3). Tomato heatmaps were generated using RNAseq (SL2.50 genome) and microarray (GPL4741) approaches. Grape heatmaps were generated using three microarray platforms: NimbleGen (GPL17894), GrapeGen (GPL11004) and GeneChip (GPL1320). RNAseq data were downloaded from the Sequence Read Archive repository (SRA; https://www.ncbi.nlm. nih.gov/sra) and microarray data were downloaded from the Gene Expression Omnibus repository (GEO; https://www.ncbi.nlm.nih.gov/geo/) using the GEOquery R library.

from abiotic stresses, due to their ability to maintain membrane integrity and possibly activate the JA-related defense pathway (Radhakrishnan and Lee, 2013; Tanou et al., 2014). Indeed, postharvest application of PAs (1 mM Put or Spd) alleviated chilling injury of apricot during storage at 1°C (Koushesh saba et al., 2012). Similarly, postharvest application of 1–2 mM Put or 1 mM Spd reduced chilling injury of pomegranate stored at 2°C (Mirdehghan et al., 2007; Barman et al., 2011).

During growth and ripening of both climacteric and nonclimacteric fruit, the natural level of PAs changes (Fortes and Agudelo-Romero, 2018): during the early phase of fruit growth and cell division, PA levels are high; during fruit ripening and senescence, PA levels usually decline, with a few exceptions [Liu et al., 2006; reviewed by Valero et al. (2002)]. During fruit ripening and senescence, there is crosstalk between ethylene and PAs (Pandey et al., 2000; Valero et al., 2002) as SAM is a common precursor for both growth regulators. Therefore, ethylene and PAs may induce or delay fruit ripening and senescence in opposite manners. Indeed, postharvest application of PAs to fruit led to inhibition of ethylene emission. However, the percentage of inhibition was dependent on the ethylene climacteric peak: as the fruit emitted less ethylene, higher inhibition was observed [reviewed by Valero et al. (2002)]. Thus, one of the main effects of PA application is inhibition of fruit ripening, affecting color change, decreasing fruit softening, and delaying ethylene emission and respiration (Valero et al., 2002).

Polyamines may also delay fruit softening by attaching to pectin elements in the cell wall, resulting in increased fruit firmness. This binding blocks cell wall-degrading enzymes' access to the cell-wall matrix (Valero et al., 1999). On the other hand, a delay in fruit softening and ripening is strongly correlated with increased fruit resistance to fungal pathogens and reduced postharvest decay (Cantu et al., 2008). Polyamines also function as signaling molecules that interact with JA, ABA, and SA (Jiménez-Bremont et al., 2014)—hormones that activate broad defense responses against pathogens in fruit (Alkan and Fortes, 2015).

In accordance with the PAs' effects on fruit ripening, softening, and defense hormones, a number of studies have shown that in most cases, postharvest application of PAs inhibits fruit ripening and softening while reducing postharvest decay. One of the most significant effects of PA infiltration after harvest was its contribution to fruit firmness in apple (Kramer et al., 1991), strawberry (Ponappa et al., 1993), apricot, peach (Martínez-Romero et al., 2002; Valero et al., 2002), and lemon (Martínez-Romero et al., 1999). Preharvest treatments were similarly effective at increasing fruit firmness (Bregoli et al., 2002).

Postharvest application of about 0.5–1 mM PA inhibited the ripening of plum, kiwi and mango fruit, as reflected by reduced ethylene emission, respiration, and inhibition of softening, thereby prolonging shelf life (Serrano et al., 2003; Petkou et al., 2004; Malik and Zora, 2005). Similarly, preharvest application of PAs inhibited fruit ripening and the expression of genes involved in ethylene synthesis in nectarine and plum (Torrigiani et al., 2004; Khan et al., 2007). In plum, preharvest treatments inhibited ripening and softening better than postharvest treatments. In addition, an increased concentration of 0.3–2 mM Put applied postharvest inhibited strawberry ripening (firmness, total soluble solids, *and* titratable acidity) and was correlated to a delay in rotting caused by fungal pathogens (Khosroshahi et al., 2007). Postharvest application of 100 mg/L Put, Spd, or Spm inhibited decay accumulation in mandarin as well (Zheng and Zhang, 2004). Similarly, overexpression of the yeast Spd synthase gene in tomato led to the accumulation of PAs, reduced ripening and softening, and consequently, reduced decay symptoms (Nambeesan et al., 2010). However, a high concentration of PAs could result in fruit injuries, such as black spot in apples (Kramer et al., 1991), and in the induction of postharvest decay caused by *B. cinerea* in grapes (Nambeesan et al., 2012).

### CONCLUSIONS AND PERSPECTIVES

Many of the studies conducted to date have indicated that PA synthesis, conjugation, and catabolism play important roles in abiotic and biotic stress responses in fruit. However, the analysis of PA-metabolism reprogramming by biotic stress is complicated by the fact that both plants and microbes can synthetize PAs. Changes in gene expression might be due to plant defense responses induced against the pathogen, or they might be triggered by the pathogen's virulence mechanisms which may differ according to the pathogen's lifestyle. Biotrophs might benefit from reduced PA oxidation, whereas necrotrophs exploit the generation of an oxidative burst (due to increased PA catabolism) for their pathogenicity (Jiménez-Bremont et al., 2014). What is clear, however, is that the accumulation of free and conjugated PAs—under tight regulation by mechanisms controlling PA biosynthesis and catabolism—plays an important role in host–pathogen interactions, which involve the oxidative stress response, strengthening of the fruit cell wall, and modulation of ABAand ethylene-related pathways.

Moreover, the stress response might differ with plant species and their metabolism, tissue-specific gene expression, and the interactions among other growth regulators and defense-signaling pathways. The molecular mechanisms regulating these processes need to be elucidated through the use of transgenic and mutant plant lines. However, this is a challenge for herbaceous and woody fruit species, for which transgenesis protocols have not yet been established. Studies in these plants have been mostly performed by testing treatments with PAs and inhibitors of PA metabolism.

Some PAs that were scarcely investigated in the past are receiving more attention today, namely, thermospermine and caldopentamine (Jiménez-Bremont et al., 2014), and products of PA catabolism such as GABA and 1,3 diaminopropane. Studying these may reveal new mechanisms involved in tolerance to abiotic and biotic stresses. Another level of complexity is added when one considers the role of PAs in epigenetic regulation, a topic that has been little explored. PAs can interact with chromatin, eventually leading to epigenetic modifications of DNA and histones [reviewed by Jiménez-Bremont et al. (2014)]; this opens new and exciting frontiers for research focusing on how PA metabolism affects fruit resilience. In addition, the study of PA metabolism in fruit ripening has highlighted the possible application of these natural polycations for the control of ripening and postharvest decay.

Many challenges still remain in PA research toward increasing plants' tolerance to stresses, in particular in fruit species. Manipulation of PA levels by either modulating their biosynthesis or catabolism or eventually conjugating them with other compounds may contribute in the future to obtaining better

#### REFERENCES


yields under more sustainable conditions with reduced application of fungicides.

### AUTHOR CONTRIBUTIONS

AF designed the mini-review. AF and NA wrote the manuscript with some inputs from DP. PA-R generated the heatmaps.

#### FUNDING

This work was supported by a UID/MULTI/04046/2019 Research Unit grant from FCT, Portugal (to BioISI). Additional funding was provided by the Portuguese Foundation for Science and Technology (FCT Investigator IF/00169/2015, Fellowship PD/ BD/114385/2016) and is integrated in the COST (European Cooperation in Science and Technology) Action CA17111.

#### ACKNOWLEDGMENTS

We would like to thank Dr. Caparrós-Martín for his valuable help with the figures.

pomegranate fruit quality during cold storage. *Sci. Hortic.* 130, 795–800. doi: 10.1016/j.scienta.2011.09.005


and decay in pepper fruit. *Postharvest Biol. Technol.* 18, 19–26. doi: 10.1016/ S0925-5214(99)00054-X


on reducing mechanical damage and polyamines and abscisic acid levels during lemon storage. *J. Sci. Food Agric.* 79, 1589–1595. doi: 10.1002/(SIC I)1097-0010(199909)79:12<1589::AID-JSFA403>3.0.CO;2-J


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Fortes, Agudelo-Romero, Pimentel and Alkan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Polyamines in Halophytes

#### Milagros Bueno\* and María-Pilar Cordovilla

Laboratory of Plant Physiology, Department of Animal Biology, Plant Biology and Ecology, Faculty of Experimental Sciences, University of Jaén, Jaén, Spain

Polyamines (PAs) are related to many aspects of the plant's life cycle, including responses to biotic and abiotic stress. On the other hand, halophytic plants are useful models for studying salt tolerance mechanisms related to the adaptive strategies that these plants present in adverse environments. Furthermore, some halophytes have high economic value, being recommended instead of glycophytes as alternative agricultural crops in salt-affected coastal zones or saline farmlands. In recent years, the understanding of the role of PAs in salt-tolerant plants has greatly advanced. This mini review reports on the advances in the knowledge of PAs and their participation in achieving better salt tolerance in 10 halophytes. PAs are associated with responses to heavy metals in phytoremediation processes using certain salttolerant species (Atriplex atacamensis, A. halimus, Inula chrithmoides, and Kosteletzkya pentacarpos). In crops with exceptional nutritional properties such as Chenopodium quinoa, PAs may be useful markers of salt-tolerant genotypes. The signaling and protection mechanisms of PAs have been investigated in depth in the extreme halophyte Mesembryanthemum crystallinum and Thellungiella spp., enabling genetic manipulation of PA biosynthesis. In Prosopis strombulifera, different biochemical and physiological responses have been reported, depending on the type of salt (NaCl, Na2SO4). Increases in spermidine and spermine have been positively associated with stress tolerance as these compounds provide protection in Cymodocea nodosa, and Solanum chilense, respectively. In addition, abscisic acid and salicylic acid can improve the beneficial effect of PAs in these plants. Therefore, these results indicate the great potential of PAs and their contribution to stress tolerance.

Keywords: antioxidant system, extremophiles, ion sequestration, saline markers, wetlands species, xerohalophytes

#### INTRODUCTION

Halophytes comprise a group of plants able to complete their life cycle under saline environmental conditions of around 200 mM NaCl or even more (Flowers and Colmer, 2008; Golldack et al., 2014). Representing roughly 1% of the total world flora, they are distributed mainly in arid and wetlands saline areas, as well as in temperate zones (Gul et al., 2013; Kumari et al., 2015). These plants provide useful models to understand the mechanisms of adaptation to saline stress (Hurst et al., 2004; Arbona et al., 2010). The salt-tolerance strategies include: accumulation or exclusion of ions, control and compartmentalization of ion uptake, maintenance of osmotic balance (Shabala and Mackay, 2011; Bose et al., 2014), biosynthesis of compatible solutes (Flowers and Colmer, 2015; Slama et al., 2015), shift in the photosynthetic pathway (Uzilday et al., 2015), activation and synthesis of antioxidant compounds (Ozgur et al., 2013), induction and modulation of plant hormones (Gupta and Huang, 2014; Fahad et al., 2015), and the modulation of polyamines (PAs)

#### Edited by:

Rubén Alcázar, University of Barcelona, Spain

#### Reviewed by:

Stefania Biondi, University of Bologna, Italy Oscar A. Ruiz, CONICET Institute of Biotechnological Research (IIB-INTECH), Argentina Francisco Marco, University of Valencia, Spain

#### \*Correspondence:

Milagros Bueno mbueno@ujaen.es

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 26 December 2018 Accepted: 22 March 2019 Published: 09 April 2019

#### Citation:

Bueno M and Cordovilla M-P (2019) Polyamines in Halophytes. Front. Plant Sci. 10:439. doi: 10.3389/fpls.2019.00439

**42**

(Shi and Chan, 2014; Liu et al., 2015). The use of mutants, transgenic plants, or exogenous PAs has revealed the key role of these compounds in multiple responses to salt stress (Liu et al., 2014; Parvin et al., 2014; Pathak et al., 2014).

Polyamines are polybasic aliphatic amines, widely distributed in all living organisms and involved in multiple cellular functions (Kusano et al., 2008; Hussain et al., 2011; Gupta et al., 2013; Tiburcio et al., 2014). The most commonly occurring PAs, found in halophytes, are the diamines putrescine (Put) and cadaverine (Cad), the triamine spermidine (Spd), and the tetramine spermine (Spm) (Shevyakova et al., 2006b; Kuznetsov et al., 2007). These compounds may serve as bioindicators of stresstolerant lines (Simon-Sarkadi et al., 2007), stabilize biological membranes, regulate ion homeostasis, delay senescence processes (Lutts et al., 2013), regulate membrane transport through a direct interaction with plasma membrane or vacuolar transporters (Pottosin and Shabala, 2014), protect photosynthetic tissues (Malliarakis et al., 2015), regulate the antioxidant system and free-radical machinery (Sudhakar et al., 2015), participate in abiotic stress signaling (Pál et al., 2015), and protect against metal toxicity (Ghabriche et al., 2017; Zhou et al., 2018a,b). Furthermore, PAs also activate genes for stress response and interact with other metabolic pathways by establishing hormonal cross-talk (Pál et al., 2015; Llanes et al., 2018). Most studies on PAs have been made in glycophytes. The present review attempts to consolidate our understanding of the effects of PAs and their role in the physiology of 10 halophytes (**Table 1**), helping to elucidate the potential mechanisms of these growth regulators in halophytic plants.

#### HALOPHYTES AS MODEL PLANTS TO STUDY POLYAMINES

Shevyakova et al. (2006b) were the first to investigate the free PA content in Mesembryanthemum crystallinum L. (the common ice plant) a C3-CAM (crassulacean acid metabolism) plant (Adams et al., 1998; Winter and Holtum, 2007), under severe salinity conditions. The dynamics of Spm accumulation resembled that of phosphoenolpyruvate carboxylase (PEPC), a key enzyme of the water-saving CAM pathway; this could indicate an indirect involvement of PAs in plant adaptation to stress. The gradual increase in Spm could be due to several causes: the conversion of Spd into Spm, the release of free Spm (active) from its conjugates, changes in the biosynthetic rate, and the degradation and/or competition between PAs and ethylene based on alterations in sulfur metabolism, necessary for S-adenosyl methionine (SAM) synthesis (a precursor of both PAs and ethylene). Shevyakova et al. (2006a) and Stetsenko et al. (2009) showed the accumulation of free and conjugated forms of PAs, although the latter predominated under saline conditions. However, for Cad, conjugated forms underwent a transition into the free form. These results indicate considerable quantitative organ-specific changes in the pool of free and conjugated forms, regulated by the combined action of H2O2-peroxidase and the oxidative degradation of PAs. 1.3-diaminopropane (Dap) was located in roots and was related with protection of membranes. TABLE 1 | List of revised halophytic species where Polyamines have been studied.


Kuznetsov et al. (2007) suggested that accumulation of Cad could compensate for the Put deficit, acting as a DNA protector and a free-radical scavenger. Shevyakova et al. (2011) also demonstrated an antioxidant role of PAs. Spd diminished the expression of ferritin gene and consequently the production of reactive oxygen species (ROS). Shevyakova et al. (2013) found a protective effect of ABA related to the regulation of proline (Pro), PAs, and cytokinin contents. Under saline conditions, ABA treatment increased free PA content, in roots, whereas that in leaves, could regulate conjugate forms, thus controlling the PA biological functions. ABA may also be related to the transport of Cad from roots (decreased levels) to leaves (increased levels). The mechanisms of signaling and protection of PAs in M. crystallinum were reported by Surówka et al. (2016). These researchers studied

(under exogenous application of H2O2) the diurnal rhythm cycle of antioxidants and osmotic compounds (transition C3-CAM), which could be responsible for the maintenance of the waterpotential gradient, ROS homeostasis, and the prevention of oxidative damage. Among these components, PAs (accumulation of Put and Spd but lower amounts of Spm and agmatine) showed a daily pattern of accumulation, similar to malate, suggesting a metabolic connection between the two routes. On the other hand, the accumulation of these amines could be related to mitigate oxidative stress, and maintenance maximum photosystem II (PS II) photochemical efficiency (Fv/Fm) (Kotakis et al., 2014). These mechanisms regulate cellular metabolism and trigger signaling cascades leading to stress acclimation.

Radyukina et al. (2007) studying a new model plant Thellungiella halophila, found a high capability of accumulation of Na<sup>+</sup> and Cl<sup>−</sup> ions, in their roots and leaves, under progressing salinity. In addition, they observed Pro accumulation due to enhanced expression of biosynthetic genes, accumulation of Spd and later Spm, increased expression of SAM synthetase and Spd synthase genes, and high constitutive levels of peroxidase that palliates the severity of oxidative stress. Later, Zhou et al. (2015) studied a novel Spd synthase gene (EsSPDS1), which was cloned and characterized from the halophytic plant Eutrema salsugineum. The expression of this gene was induced by polyethylene glycol (PEG), NaCl and ABA treatments. In transgenic tobacco plants, EsSPDS1-overexpressing lines exhibited drought tolerance via the reduction of water loss and the increase of sugar and Pro, favoring osmoregulation, stabilizing protein structure, and reducing oxidative potential. Overexpression of EsSPDS1 lowered the levels of malondialdehyde (MDA), diminished ion leakage, and also activated the antioxidant enzymes superoxide dismutase (SOD) and catalase (CAT), which mitigated oxidative stress. These data suggest that the activation of the EsSPDS1 gene could be an option for developing drought-tolerant plants. Ping Lee et al. (2016), working with diverse accessions of Thellungiella spp. collected from different geographical origins in many saline habitats, found that all accessions survived and grew with up to 700 mM NaCl. Salt stress induced the accumulation of high amounts of osmolytes. With respect to PAs, Put levels increased linearly with salt concentrations in all accessions. These researchers indicated a pre-adaptation to salt stress and a stronger metabolic reaction (e.g., accumulation of osmolytes) compared to its close relative Arabidopsis thaliana, a glycophyte. The protective role of Put in response to UV-B irradiation, in leaves of T. salsuginea and other species, was investigated by Radyukina et al. (2017), who concluded that the contribution of bound and free PAs is determined by the species studied and was organ-specific (roots versus leaves).

#### POLYAMINES IN HALOPHYTIC CROP SPECIES

Most of the species for human consumption belong to glycophytes (salt-sensitive plants), so halophytes can be used as an alternative to traditional agricultural crops, especially in countries with severe climatic conditions (Panta et al., 2014). Quinoa (Chenopodium quinoa Willd.) is a halophytic Andean plant, belonging to the Amaranthaceae, with remarkable tolerance to abiotic stress (Bosque-Sánchez et al., 2003; Martínez et al., 2009). This plant has excellent nutritional properties, such as a high protein content and excellent amino acid composition (Repo-Carrasco et al., 2003; Aloisi et al., 2016). Orsini et al. (2011), studying a quinoa accession (BO78) from southern Chile, demonstrated an inverse relationship of Put with tissue levels of Na<sup>+</sup> or K+, thus relating this diamine to the maintenance of a suitable cation/anion balance. A more complete study on four quinoa genotypes, located in coastal-central and southern Chile along a latitudinal gradient, was undertaken by Ruiz-Carrasco et al. (2011), comparing their growth, physiological, and molecular responses, at seedling stage, under saline conditions. With respect to PA levels, all salt-treated quinoa genotypes followed the same pattern, i.e., a sharp decline in Put levels, accompanied by an increase in the ratio of (Spd + Spm) to Put, Spd and Spm being more closely related to vital physiological processes. These researchers concluded that salttolerant genotypes might have high levels of both Pro and PAs (Spd/Spm) and could offer a protective function, being useful indicators of salinity adaptation, and potential genotypes for crop improvement. Pottosin and Shabala (2014) and Pottosin et al. (2014) found that PAs were related to the regulation of slow-(SV) and fast-(FV) vacuolar cation channels, as well as vacuolar and (PM) <sup>+</sup>H pumps and Ca+<sup>2</sup> pumps of the plasma membrane. Activation of Ca+<sup>2</sup> efflux by PAs and contrasting effects of PAs on net H<sup>+</sup> fluxes and membrane potential could contribute to modulate transport processes across the plasma membrane in stress conditions. More recently, Ruiz et al. (2017) studied the PA content and expression of genes involved in PA metabolism in R49 (salares ecotype and northern Chile) and Villarrica (VR, coastal lowlands ecotype, and southern Chile), in response to saline stress. A salt-induced increase was found in the (Spd + Spm): Put ratio, in VR the increase in the ratio occurs later than in R49, although it was significantly higher, distinguishing quinoa genotypes from different habitats. At the gene expression level, it was confirmed that lower Put levels may have resulted from diamine oxidase (DAO) up-regulation, arginine decarboxylase (ADC1) down-regulation, and (ADC2) up-regulation under saline conditions. In both landraces, the higher expression of Spm synthase (SPMS) relative to control was consistent with Spm accumulation. In addition, these authors suggested that the PA response found in both landraces could be a consequence of the early up-regulation of 9-Cis-epoxycarotenoid dioxygenase (NCED), a key enzyme in ABA biosynthesis, which preceded that of the PA biosynthesis genes, and the increase of ABA levels. Pál et al. (2015) have proposed a positive feedback loop in the response to abiotic stress between ABA and PAs.

In recent years, tomato has lost notable growth and yield, especially in arid and semi-arid zones due to intensive irrigation and soil salinization (Munns and Tester, 2008; Gharbi et al., 2017). Comparisons of the glycophyte Solanum lycopersicum and its wild relative halophyte S. chilense have revealed different ways of tolerating stress, the higher tolerance being related to enhanced levels of salicylic acid (SA), ethylene, and PAs

(Gharbi et al., 2016). These researchers have shown an outstanding osmotic adjustment in leaves and greater ethylene production in salt-treated plants of S. chilense, correlated with increases, in both shoot and root, of K<sup>+</sup> concentrations, as occurs in other species such as Arabidopsis (Yang et al., 2013). With respect to its biosynthesis, ethylene overproduction proved to be related to 1-aminocyclopropane-1-carboxylic acid synthase (ACCS2) gene induction, which has not been reported in S. lycopersicum. Jiang et al. (2013) showed that ethylene may contribute to Na+/K<sup>+</sup> homeostasis, under saline stress. Endogenous SA increased in the halophyte in relation to greater stomatal conductance (gs), improving the metabolism under stressful conditions. Exogenous SA application and salt may act in synergy on osmolyte synthesis, and a positive linear correlation was recorded between SA content and ethylene synthesis for S. chilense. PA levels proved consistently higher in the halophyte than in the glycophyte. Saline stress lowered Put levels and reduced the expression of genes coding for ornithine decarboxylase ODC, ADC1, and ADC2 in the halophyte, whereas NaCl increased the expression of those genes in S. lycopersicum. Conversely, the tetraamine Spm was only increased in S. chilense. Hu et al. (2012) considered high Spd and Spm content to have protective functions in the glycophyte tomato. The application of SA raised PA levels in the aerial part of S. chilense (increased ADC1 and ADC2 gene expression) so that Put production was partly used for Spd synthesis, but this treatment had no impact in S. lycopersicum. However, when the plants were treated with NaCl + SA, the induction of gene expression was not correlated with the accumulation of products, implying a post-transcriptional regulation of PA metabolism. Finally, NaCl slightly induced the expression of the SAM decarboxylase (SAMDC) gene in S. chilense, while a higher expression was registered in S. lycopersicum, suggesting differential responses to salinity in the two species.

### POLYAMINES AS RELATED TO THE POTENTIAL USE OF HALOPHYTES IN PHYTOREMEDIATION

Two of the halophytes most widely studied for their high levels of resistance to salinity stress and drought (Jordan et al., 2002; Martínez et al., 2005) have been the xero-halophytes Atriplex atacamensis Phil., a native shrub from Atacama Desert (Chile), and A. halimus, a species of Mediterranean areas (Amaranthaceae family). These tolerate salinity by excreting ions (Cl<sup>−</sup> and Na+) through the trichomes located on the surface of their leaves, known as salt bladders, which allow the expulsion of certain amounts of heavy metals (Ben Hassine et al., 2009; Vromman et al., 2011). Given their high biomass production, and capacity to tolerate elevated concentrations of toxic ions (Lutts et al., 2004; Manousaki and Kalogerakis, 2009; Vromman et al., 2011), these could be promising species for phytoremediation. Data reported by Ben Hassine et al. (2009) in A. halimus, suggest that free PAs, mainly Spd and Spm, may be involved in the excretion processes and in the regulation of ion fluxes through salt bladders. These researchers suggested that the increase in endogenous PAs (interacting with ATPase or ion channels involved in Na<sup>+</sup> and Cl<sup>−</sup> fluxes between epidermis and basal cells of bladders) could reduce Na<sup>+</sup> influx into mesophyll cells, thereby re-directing Na<sup>+</sup> flux toward salt bladders. In addition, ABA contributes to PA synthesis and the conversion from bound and conjugated to free soluble forms of PAs, thus favoring salt excretion. Lefèvre et al. (2009, 2010) also found high levels of Put and Spd in a water-stress resistant cell line of A. halimus, when tissues were maintained on a medium containing cadmium (Cd), whereas Spm and Dap were higher in the drought stresssensitive line. PAs are effective in decreasing ionic toxicity, reducing oxidative stress, and restricting the harmful impact of Cd. On the other hand, A. atacamensis is able to grow with high external arsenic (As) concentrations in their natural habitat. Vromman et al. (2011), studying the root and leaves of this halophyte, found an increase in free soluble PAs in As-exposed plants. Specifically, Spd and Spm increased at the expense of Put. These researchers hypothesize that these polycationic molecules may assist in arsenate sequestration in the stressed tissues, and therefore should be tested in field trials for the phytomanagement of As-contaminated sites.

Another halophyte common in Mediterranean zones is Inula chrithmoides, which belongs to the Asteraceae family. It is a perennial coastal species found on sea cliffs and in salt marshes. Abdel-Wahhab et al. (2008) found medicinal applications by analyzing their metabolite production. This species is found in areas contaminated by Cd, nickel, and salt (Zurayk et al., 2001). In absence of Cd, Ghabriche et al. (2017) have shown that NaCl (100 mM) strongly boosted plant growth and improved the Na+: K <sup>+</sup> ratio in shoot of I. chrithmoides. Meanwhile, in Cd-treated plants, NaCl protected this species from ionic toxicity and helped reduce Cd absorption and translocation. PAs increased in response to both NaCl and CdCl<sup>2</sup> (mainly the bound PA fraction) in leaves and roots, and this increase could be related to protective functions of PAs on endogenous cellular structures. In rice, Li et al. (2013) found that Cd decreased the activities of glutamine synthetase and glutamate synthase, enzymes of N assimilation pathway, and therefore increased NH<sup>4</sup> <sup>+</sup> and arginine content (a precursor of Put). Thus, Ghabriche et al. (2017) proposed that the Cd-induced alteration of the N cycle was the main factor responsible for altering PA levels in I. chrithmoides.

The halophyte Kosteletzkya pentacarpos (formerly K. virginica) is a perennial wetland species (Malvaceae family) and a promising plant for saline agriculture (He et al., 2003). This plant grows in coastal areas contaminated by industrial activity (Lutts and Lefèvre, 2015). Zhou et al. (2018a,b) found that salt and heavy metals altered the endogenous PA content, providing protection under these stress conditions. Heavy metals augmented ethylene synthesis, but NaCl depressed it in plants exposed to Cd or the combined treatment (Cd + Zn) but not with Zn. PA concentrations were altered by Cd (Spd and Spm decreased concomitantly with Put accumulation). These researchers concluded that PA profile could be the result of ethylene overproduction in Cd-treated plants because SAM is a precursor of ethylene, Spd, and Spm. Inhibition of ethylene synthesis increased Spd and Spm level and reduced Cd accumulation.

#### POLYAMINES IN EUHALOPHYTES

In Cymodocea nodosa, a seagrass, PAs are distributed in all plant tissues, but mainly the apical section of the rhizome (Marián et al., 2000). Zarranz Elso et al. (2012) showed a decline in PAs during embryo development (free PAs), and an increase during germination, indicating the key role of these growth regulators in the first steps of the zygotic and germination stages. Exogenous Spd helps improve Fv/Fm-values, exerting a protective role in maintaining the photosynthetic apparatus. Ioannidis and Kotzabasis (2007), Hamdani et al. (2011), and Malliarakis et al. (2015) found that application of PAs (Spd and Spm) could restore PSII efficiency, via electrostatic interactions with extrinsic and intrinsic, and PSII thylakoid proteins, thereby stabilizing them. These results indicate that C. nodosa requires elevated PA levels to survive in high salinity. Prosopis strombulifera, a common spiny shrub, belonging to the family Fabaceae (subfamily Mimosoideae) is distributed in arid regions (of North and South America). Reginato (2009) and Reginato et al. (2012) considered this plant to be a true

#### REFERENCES


"euhalophyte" for NaCl, being less tolerant to Na2SO4, although a partial improvement was found with two salt mixtures. Put accumulation exerted a beneficial effect in NaCl-treated plants, stimulating growth and the antioxidative defense system. However, low Put levels, in leaves, were correlated with inhibition of shoot growth, associated with Na2SO<sup>4</sup> toxicity. Cad and Dap were related with stress symptoms, and possibly, the activities of PA oxidases.

#### AUTHOR CONTRIBUTIONS

MB and M-PC revised the Bibliography. MB wrote the manuscript.

#### FUNDING

This research was supported by Spanish Ministry of Science and Innovation (CGL2006-08830, "The involvement of growth regulators in plant responses to salinity").



with iron and different intensity of oxidative stress. Russ. J. Plant Physiol. 58, 768–775. doi: 10.1134/S1021443711050219


in chloroplasts of the extreme halophyte Eutrema parvulum (Thellungiella parvula) under salinity. Ann. Bot. 115, 449–463. doi: 10.1093/aob/mcu184


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Bueno and Cordovilla. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Polyamines and Legumes: Joint Stories of Stress, Nitrogen Fixation and Environment

*Ana Bernardina Menéndez1,2, Pablo Ignacio Calzadilla1†, Pedro Alfonso Sansberro3, Fabiana Daniela Espasandin3, Ayelén Gazquez1, César Daniel Bordenave1, Santiago Javier Maiale1, Andrés Alberto Rodríguez1, Vanina Giselle Maguire1, Maria Paula Campestre1, Andrés Garriz1, Franco Rubén Rossi1, Fernando Matias Romero1, Leandro Solmi1, Maria Soraya Salloum4, Mariela Inés Monteoliva4, Julio Humberto Debat5 and Oscar Adolfo Ruiz1,4\**

#### Edited by:

Ana Margarida Fortes, University of Lisbon, Portugal

#### Reviewed by:

Susana Araújo, New University of Lisbon, Portugal Magda Pál, Hungarian Academy of Sciences, Hungary

#### \*Correspondence:

Oscar Adolfo Ruiz ruiz@intech.gov.ar; oaruiz@conicet.gov.ar

#### †Present address:

Pablo Ignacio Calzadilla, Institut de Biologie Intégrative de la Cellule (I2BC), CNRS, CEA, Université Paris-Sud, Université Paris-Saclay, Paris, France

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 26 February 2019 Accepted: 11 October 2019 Published: 04 November 2019

#### Citation:

Menéndez AB, Calzadilla PI, Sansberro PA, Espasandin FD, Gazquez A, Bordenave CD, Maiale SJ, Rodríguez AA, Maguire VG, Campestre MP, Garriz A, Rossi FR, Romero FM, Solmi L, Salloum MS, Monteoliva MI, Debat JH and Ruiz OA (2019) Polyamines and Legumes: Joint Stories of Stress, Nitrogen Fixation and Environment. Front. Plant Sci. 10:1415. doi: 10.3389/fpls.2019.01415

1 Instituto Tecnológico de Chascomús (INTECH), UNSAM-CONICET, Chascomús, Argentina, 2 Departamento de Biodiversidad y Biología Experimental, Facultad de Ciencias Exactas y Naturales, UBA-CONICET, Buenos Aires, Argentina, 3 Instituto de Botánica del Nordeste (IBONE), UNNE-CONICET, Corrientes, Argentina, 4 Instituto de Fisiología y Recursos Genéticos Vegetales (IFRGV) Ing "Victorio S Trippi," Instituto Nacional de Tecnología Agropecuaria (INTA), Córdoba, Argentina, 5 Instituto de Patología Vegetal (IPAVE) Ing "Sergio Nome," Instituto Nacional de Tecnología Agropecuaria (INTA), Córdoba, Argentina

Polyamines (PAs) are natural aliphatic amines involved in many physiological processes in almost all living organisms, including responses to abiotic stresses and microbial interactions. On other hand, the family Leguminosae constitutes an economically and ecologically key botanical group for humans, being also regarded as the most important protein source for livestock. This review presents the profuse evidence that relates changes in PAs levels during responses to biotic and abiotic stresses in model and cultivable species within Leguminosae and examines the unreviewed information regarding their potential roles in the functioning of symbiotic interactions with nitrogen-fixing bacteria and arbuscular mycorrhizae in this family. As linking plant physiological behavior with "big data" available in "omics" is an essential step to improve our understanding of legumes responses to global change, we also examined integrative MultiOmics approaches available to decrypt the interface legumes-PAs-abiotic and biotic stress interactions. These approaches are expected to accelerate the identification of stress tolerant phenotypes and the design of new biotechnological strategies to increase their yield and adaptation to marginal environments, making better use of available plant genetic resources.

#### Keywords: legume, plant polyamines, plant stress and adaptation, symbionts, constrained environments

### INTRODUCTION

Polyamines (PAs) are organic polycations, acknowledged as regulators of plant growth, development and stress responses, being putrescine (Put), spermidine (Spd), and spermine (Spm) the most related to this physiological role (Cohen, 1998). A high number of metabolites and enzymes participate in PAs metabolism (see Calzadilla et al., 2014 for an extensive description). The diamine Put can be synthesized directly from ornithine by the enzyme ornithine decarboxylase (ODC, EC 4.1.1.17) or indirectly, *via* a series of intermediates following decarboxylation of arginine by arginine decarboxylase (ADC, EC 4.1.1.19) (**Figure 1**). In turn, Spd and Spm are synthesized from Put by successive additions of aminopropyl groups provided by decarboxylated S-adenosylmethionine (SAM), a metabolite derived

1 **49**

FIGURE 1 | Polyamine metabolism. Biosynthetic pathways for PAs and related metabolites are indicated by continuous lines. Dashed lines show the metabolites from the catabolism of the Pas. Abbreviations: Put, putrescine; Cad, Cadaverine; Spd, spermidine; Spm, spermine; tSpm, thermospermine; SAM, S-adenosylmethionine; dcSAM, decarboxylated S-adenosylmethionine; ACC, aminocyclopropane carboxylic acid; P5C, glutamate-5-semialdehyde. Numbers refer to enzymes: 1, arginine decarboxylase (ADC, EC 4.1.1.19); 2, agmatine iminohydrolase (EC 3.5.3.12); 3, N-carbamoylputrescine amido-hydrolase (EC 3.5.1.53); 4, ornithine decarboxylase (ODC, EC 4.1.1.17); 5, spermidine synthase (SPDS, EC 2.5.1.16); 6, spermine synthase (SPMS, EC 2.5.1.22); 7, thermospermine synthase; 8, L-lysine decarboxylase (EC 4.1.1.18); 9, SAM synthetase; 10, SAM decarboxylase (SAMDC, EC 4.1.1.50); 11, ACC synthase (EC 4.4.1.14); 12, ACC oxidase (EC 1.14.17.4); 12, back-conversion polyamine oxidase (non-specific polyamine oxidase, EC 1.5.3.17); 13, terminal catabolism polyamine oxidase (propane-1,3 diamine-forming, EC 1.5.3.14); 14, diamine oxidase (DAO, EC 1.4.3.6); 15, GAD (glutamate decarboxylase, EC 4.1.1.15); 16, P5CDH (Δ1-pyrroline-5-carboxylate dehydrogenase, EC 1.5.1.12); 17,P5CS (L-glutamate γ-semialdehyde dehydrogenase, EC 1.2.1.88); 18, P5CR (Δ1-pyrroline-5-carboxylate reductase, EC 1.5.1.2); 19, ProDH (proline dehydrogenase, EC 1.5.99.8); 20, OAT (ornithine δ-aminotransferase, EC 2.6.1.13); 21, Glutamate to Ornithine subpathway (five steps subpathway catalyzed by EC 2.3.1.1, EC 2.7.2.8, EC 1.2.1.38, EC 2.6.1.11 and EC 3.5.1.16).

from the S-adenosylmethionine decarboxylase (SAMDC, EC 4.1.1.50) activity. The aminopropyl additions to Put are catalyzed by the aminopropyl-transferases Spd (EC 2.5.1.16) and Spm synthases (EC 2.5.1.22). Both ADC and ODC pathways occur in higher plants and bacteria (**Figure 1**).

PAs are present in cells as free and bound forms, in variable amounts, depending on the species and developmental stage (Jiménez-Bremont et al., 2007; Alcázar et al., 2010; Hussain et al., 2011 ). The free forms of PAs show water-soluble properties and therefore, are readily translocated within cells. They may cause conformational stabilization/destabilization of DNA, RNA, chromatin, and proteins due to their ability to form electrostatic linkages with negatively charged molecules (Alcázar et al., 2010; Wimalasekera et al., 2011). PAs can stabilize membranes or nucleic acids, by binding to their negative surfaces (Galston and Sawhney, 1990; Kusano et al., 2008). Although the H2O2 derived from PAs catabolism contributes to reactive oxygen species (ROS) (Gonzalez et al., 2011), PAs can also act as ROS scavengers and activate antioxidant enzymes (Pottosin et al., 2014a). In addition, PAs display effects on vacuolar channels and cation transport in Menéndez et al. Polyamines and Legumes

plants (Pottosin and Muñiz, 2002; Pottosin and Shabala, 2014; Zepeda-Jazo and Pottosin, 2018). Notwithstanding the precise molecular mechanisms by which PAs control plant responses to abiotic stress remain unknown, several aspects about their apparently clashing roles in the development have been reviewed the last years (e.g., Minocha et al., 2014; Singh et al., 2018; Chen et al., 2019a). These aspects include the involvement of PAs signaling in direct interactions with different metabolic routes like intricate hormonal cross-talks, the nitric oxide formation and the modulation of ion channel activities, and Ca2+ homeostasis (Tun et al., 2006; Yamasaki and Cohen, 2006; Alcázar et al., 2010; Bitrián et al., 2012; Minocha et al., 2014; Singh et al., 2018; Podlešáková et al., 2019).

The family *Leguminosae*, containing close to 770 genera and over 19,500 species (LPWG, 2017), is the third largest Angiosperms family in terms of species numbers after *Asteraceae* and *Orchidaceae*. A considerable number of features make legumes an excellent model system to study the different aspects of PAs metabolism. Legumes are critical components of natural and agricultural ecosystems (Escaray et al., 2012). Indeed, some legumes such as soybean (*Glycine max* L. Merr.) and peanut (*Arachis hypogaea*) are food crops of primary economic importance for livestock and human consumption, as their seeds are rich in proteins, carbohydrates, and oils (Duranti and Gius, 1997; Graham and Vance, 2003). In addition, legumes are characterized by their ability to establish symbiotic interactions with nitrogen fixation bacteria (NFB) and arbuscular mycorrhizal fungi (AMF), which help at plant nutrition and adaptation to soils offering different environmental constraints (Escaray et al., 2012; Gibson et al., 2008).

The ability of legumes to establish symbiotic interactions with nitrogen-fixing bacteria and arbuscular mycorrhizae fungi gives to some legume species that can exploit their molecular machinery "pioneer" attributes, with better competition in nutrient-poor soils and higher adaptation to restricted environments. However, many legume crops may be affected by several biotic and abiotic stresses, whereby maintaining their yields safe from adverse environmental conditions is probably one of the biggest challenges facing modern agriculture. Therefore, the obtaining of vigorous genotypes with higher tolerance to abiotic and biotic stresses has turned an increasingly important biotechnological target. At this scenario, PAs can play an important role, and genetic manipulation of crop plants with genes encoding PAs biosynthetic pathway enzymes is envisioned as a strategy to achieve plants with improved stress tolerance and symbiotic performance.

In order to support legume crops yields and to understand their limitations, biological, physiological, and diverse omics studies have been carried out in the last 80 years. More recently, genomic tools (genome sequences, expressed sequence tags, oligonucleotide, and cDNA microarrays) have emerged, along with comprehensive databases such as the Legume Information System (http://www.comparative-legumes.org). In particular, genomic sequencing of *Medicago truncatula* (Young et al., 2011) and *Lotus japonicus* (Sato et al., 2008), differing in their patterns of root nodule formation (Barker et al., 1990; Handberg and Stougaard, 1992), and also of soybean (Schmutz et al., 2010), pigeon pea (Varshney et al., 2012), and chickpea (Varshney et al., 2013) have provided a solid framework to explore PAs metabolism for legume crop improvement.

The recent developments in legumes research are fundamental to sustain food security at a global level. However, the current paradigm in plant science is characterized by a disconnection of ecophysiology and "omics," which have been established in parallel, with only exceptional cross-talks for the past 20 years. A new field has been proposed in order to capitalize the concurrent advances of both areas into a single discipline: "ecophysiolomics" (Flexas and Gago, 2018). This multidisciplinary approach would require joining forces, equipment, and abilities in the context of user-friendly integrative bioinformatics resources and concurrent shared research protocols. This utopic collaborative environment is essential to advance in the understanding of legumes at levels ranging from cellular to agroecosystem scales. Then breeders may be able to use this understanding and translate it into practices or biotechnological tools.

This review explores the contribution made by studies on legume species on the basic knowledge of PAs metabolism, their role in tolerance to biotic and abiotic stresses, and the establishment of mutualistic relationships relevant to the physiology of plants and the environment. As significant future challenges are the development and implementation of progressive methods of genetic improvement oriented to develop varieties of legumes that have genetic recovery capacity against environmental stresses, this review will also get a glimpse of the state-of-the-art toolkit landscape in legume research.

#### Polyamines and Abiotic Stress in Legumes Drought Stress

Exogenous PAs application has shown to mitigate drought stress in several legumes such as *Phaseolus vulgaris* (Torabian et al., 2018a; Torabian et al., 2018b), *Trifolium repens* (Zhang et al., 2018), and *Vigna radiata* (Farhangi-Abriz et al., 2017), whereas the mechanisms involved in the PAs-mediated alleviation of drought include the crosstalk with several phytohormones, the improvement of plant water status, stress signaling, antioxidant biosynthesis, melatonin production, and DNA protection.

Plant acclimation strategies to water deficit are based on short-term osmotic responses (Tardieu et al., 2018), and abscisic acid (ABA) and other phytohormones are acknowledged by their involvement in fast tolerance response to dehydration (Ullah et al., 2018). Several studies carried out with ABA-deficient mutants of non-legume species confirmed that ABA mediates the drought upregulation of *ADC2*, *SPDS1*, and *SPMS* genes at the transcriptional level (Alcázar et al., 2010). Legume-base studies have also provided significant pieces of evidence pointing to PAs playing a role in the regulation of short-term osmotic responses by interacting with phytohormones. For example, Espasandin et al. (2014) showed that the overexpression of the oat *ADC2* gene in *Lotus tenuis* plants, using a heterologous oat *ADC* gene under the control of a drought/ABA-inducible promoter *RD29A*, increased Put content in shoots of drought-stressed plants. These authors revealed that Put controls ABA biosynthesis in response to drought by modulating the expression of 9-cis-epoxycarotenoid dioxygenase (NCED). Drought increased the expression of oat ADC, total ADC activity, and Put content with only minor variations in Spd and Spm. The wild-type plants showed relatively smaller changes in PAs metabolism upon exposure to water shortage. Concomitantly, the higher Put content in transgenic lines significantly increased the NCED expression, suggesting the possibility of transcriptional regulation of ABA synthesis by Put. All these results underpin the theory that Put and ABA are assimilated in a positive feedback loop in response to osmotic stress (Singh et al., 2018). In contrast, Spd application had no significant effect on ABA accumulation in *T. repens*, although it led to increased GA and cytokinin (CTK) content as well as decreased indole-3-acetic acid (IAA) content under water deficit condition (Li et al., 2016a). In this species, IAA-PAs crosstalk was involved in the improvement of antioxidant defense and osmotic adjustment conferring plant tolerance to water stress (Li et al., 2018a). Moreover, crosstalk with ABA, IAA, and CKs is predictable since these phytohormones modulate root architecture to maintain the water homeostasis to cope with the long-term water shortage (Khan et al., 2016). In effect, in drought-stressed *L. tenuis* (Espasandin et al., 2014), *Cicer arietinum, G. max* (Nayyar et al., 2005), and *Medicago sativa* (Zeid and Shedeed, 2006), the rise in Put level either by overexpression of genes related to PAs biosynthesis, or exogenous application promoted root development. These effects may also be related to the involvement of PAs in the control of cell division and differentiation, which plays an essential role in the root apex and during lateral root formation (Couée et al., 2004). Further proteomic analysis in *T. repens* (Li et al., 2018b) revealed that PAs could be one key adaptive response against drought stress, through the regulation of growth, ribosome, amino acid and energy metabolism, and antioxidant reactions.

Drought induces oxidative stress due to an imbalance between ROS formation and scavenging (Fariduddin et al., 2013). The excessive ROS production, in turn, decreases the membrane fluidity and damages membrane proteins inactivating the related receptors, enzymes, and ion channels (Gill and Tuteja, 2010). PAs, mainly Put and Spm, are responsible for the scavenging of ROS and can indirectly affect the activities of the involved enzymes including catalase, peroxidases, and superoxide dismutase (Alcázar et al., 2010; Minocha et al., 2014; Sánchez-Rodríguez et al., 2016). A body of evidence obtained from experiments using water-stressed legumes has also contributed to support that PAs could act as a signal molecule or as antioxidants during the stress response. PAs treatment increases the activities of antioxidant enzymes and reduces the oxidative damages in *C. arietinum* (Nayyar and Chander, 2004). Likewise, the exogenous application of Spm and Spd regulates antioxidant defense system by increasing the reduced glutathione concentration or catalase activity in drought-stressed *G. max* (Radhakrishnan and Lee, 2013) and *T. repens* (Li et al., 2014a). Furthermore, PAs synthesis and oxidation have shown to improve H2O2-induced antioxidant protection in *M. sativa* (Guo et al., 2014).

Chloroplasts employ several strategies to cope with energy imbalances and prevent ROS formation, which was recently reviewed by Vanlerberghe et al. (2016). In such circumstances, Spm plays an essential role in protecting the photosynthetic apparatus by interaction with photosystem II and light harvesting complex (LCH) proteins, preserving the integrity of the thylakoid membranes structure (Hamdani et al., 2011), helping in the maintenance of the photosynthetic activity. Drought-sensitive *P. vulgaris* plants presented significant decreases in the contents of all PAs associated with thylakoids isolated from plants growing in sorbitol and salt conditions (Legocka and Sobieszczuk-Nowicka, 2012), suggesting that thylakoid-associated PAs would be good markers of plant stress tolerance. Also, Spm application to *G. max* leaves reduced osmotic stress-induced losses in chlorophyll, carotenoid, and protein levels (Radhakrishnan and Lee, 2013).

It is known that PAs actively participate in stress signaling through an intricate crosstalk with several signal molecules (Marco et al., 2011; Shi and Chan, 2014). Nitric oxide (NO) is a key signaling molecule that can also be induced by PAs. In *T. repens*, Spd played a role in drought stress-activated pathways associated with NO release, which mediated antioxidant defense, contributing to drought tolerance in this plant (Peng et al., 2016), whereas in *Vicia faba*, NO accumulation proved necessary for ABA-induced closure of stomata (Garcia-Mata and Lamattina, 2002). On other hand, hydrogen sulfide (H2S) is currently regarded as a novel gaseous signaling molecule in plants during environmental stress response (Li et al., 2019). In *T. repens*, dehydration or exogenous application of Spd caused a quick H2S accumulation, followed by significant improvement of antioxidant activities and increased transcript levels of several transcription factors and genes encoding antioxidant enzymes, all associated with dehydration tolerance. Further analyses using NO and H2S scavengers led to the notion that Spd-induced H2O2 could be an upstream signal molecule of NO and H2S, whereas Spd-induced H2S might act as the downstream signaling of NO in *T. repens* leaves.

Drought perturbs photosynthesis due to CO2 limitations resulting from stomatal closure, and biochemical restrictions associated with the accumulation of reducing power (Pinheiro and Chaves, 2010). PAs are shown to regulate ion channel and Ca2+ homeostasis, and it is known that changes of free Ca2+ in the cytoplasm of guard cells are responsible for stomatal movement (Allen and Sanders, 1996; Peiter et al., 2005). Several works performed on legumes have also suggested that the interaction between PA-induced H2O2 and Ca2+ signaling plays a role in stomata movement. In *V. faba,* diamine oxidase (DAO) catalyses the degradation of Put to produce H2O2, thereby elevating the Ca2+ level in guard cells (An et al., 2008). PAs-regulated tolerance to water stress in *T. repens* was associated with antioxidant defenses and dehydrins *via* their involvement in the Ca2+ messenger system and H2O2 signaling pathways (Li et al., 2015a). The constitutive overexpression of oat arginine decarboxylase 2 (*ADC2*) gene increases the net CO2 assimilation rate in *M. truncatula* stressed leaves at the expenses of an increase in the stomatal conductance and transpiration (Duque et al., 2016). Contrarily, the *ADC2* overexpression driven by the stress-inducible *RD29A* promoter improved drought tolerance in *L. tenuis* plants subjected to a gradual decrease in water availability, by inducing stomatal closure to reduce transpiration (Espasandin et al., 2014). As the intensity of stress increases, PAs promote osmoregulation and preserves the leaf relative water content by inducing the accumulation of proline, an amino acid playing a highly beneficial role in plants exposed to osmotic stress conditions (Hayat et al., 2012; in *C. arietinum, G. max*, and *L. tenuis*). Furthermore, pretreatment of *T. repens* seeds by soaking with Spd 30 μM for 90 min increased α- and β-amylase activities accelerating the starch metabolism during germination under low soil water content (Li et al., 2014b). Likewise, pre-treatment of 7-day-old *T. repens* plants with Spd 500 μM for 7 days speed up the water-soluble sugar, sucrose, fructose, sorbitol, and dehydrins accumulation in drought-stressed leaves (Li et al., 2015b). Also, γ-aminobutyric acid (GABA), a product of Put catabolism by DAO and terminal catabolism of Spd (Xing et al., 2007) may act as an osmolyte to reduce the loss of cellular water and also protects the plant from stress by regulating cell pH (Podlešáková et al., 2019). In fact, treatment with exogenous GABA led to improved drought tolerance of *T. repens*, associated with a positive regulation in the GABA-shunt and PAs metabolism (Yong et al., 2017).

Legumes as model species have facilitated the study of unique biological mechanisms used by plants in response to stress. *M. truncatula* has been valuable to reveal the role of *MtSPDS* and *MtSPMS* genes encoding Spd synthase and Spm/ Spd synthase, respectively, as part of the molecular mechanisms underlying DNA damage response in legumes (Pagano et al., 2019). In *M. sativa*, melatonin pre-treatment exerted, through PAs modulation, a protective effect on plants against waterlogging (Zhang et al., 2014). Interestingly, melatonin improved tolerance to salt and drought stresses in *G. max* and *V. faba* as well (Wei et al., 2014; Dawood and El-Awadi, 2015). It would be worthwhile to test whether melatonin also induces the development of lateral root primordia through the stimulation of polyamine oxydase (*PAO1*) expression as it was recently shown for *Solanum lycopersicum* (Chen et al., 2019b). Finally, physiological and iTRAQ-based proteomic analyses on Spd-treated *T. repens* explained Spd-induced physiological effects associated with improved drought tolerance through the higher abundance of differential expressed proteins involved in protein (ribosomal and chaperone proteins) and amino acids biosynthesis, in carbon and energy metabolisms, in antioxidants (ascorbate peroxidase, glutathione peroxidase, and dehydrins), and in GA and ABA signaling pathways (Li et al., 2016b).

#### Salt Stress

Salinity is a severe problem for crops worldwide (Flowers, 2004), affecting around 800 million ha. (FAO, 2008). Salt stress disturbs plants in a two-phases way; the first involves the reduction of shoot growth due to the osmotic stress caused by high salt concentration in the rhizosphere, and the second is driven by the accumulation of toxic ions (Munns and Tester, 2008). Supplementation of salt affected plants with exogenous PAs led to improved toxicity symptoms and plant growth in several legume species such as *V. radiata* (Nahar et al., 2016a) and *G. max* (Zhang et al., 2014).

Experiments using *L. tenuis* (formerly *Lotus glaber*) as a model to test the hypothesis that free Spd and Spm are biochemical indicators of the salt stress response, have shown that salt induced a decrease and an increase of free Spd and Spm, respectively (Sanchez et al., 2005). These results suggest the lack of a relationship between the salt induced reduction of growth rate and Spd content, while Spm might be related to stress signaling. Several studies using "omics" techniques have revealed that salinity modulates the expression of genes involved in PAs metabolism. In *C. arietinum* (chickpea) roots, it was shown that salt induced the up-regulation of *ADC* and *SAMDC* (Molina, 2008). Sánchez et al. (2011) analyzed the contrasting responses to salinity of six *Lotus* species by using comparative ionomics, transcriptomics, and metabolomics. These authors found that many salt-elicited genes of PAs metabolism showed a similar gene-expression profile in sensitive and tolerant species. These shared transcripts included many genes previously implicated in plant stress such as enzymes of PAs biosynthesis and catabolism, proline oxidase, polyamine oxidase, SAMDC, and Spm synthase. Also, recent metabolomic studies revealed that salinity increases the content of different free PAs in legumes, in *Prosopis strombulifera* leaves (Llanes et al., 2016) and *G. max* roots (Jiao et al., 2018). Interestingly, *Pr. strombulifera* showed an increase of cadaverine (Cad), an uncommon diamine that characterizes the legume family (Jancewicz et al., 2016).

The fact that salinity induces osmotic stress and redox imbalances implies that some stress symptoms and mechanisms for their mitigation are shared with drought, including improvement of plant water status, stress signaling, and synthesis of antioxidants (see previous section). In this regard, several studies on legume species have contributed to reveal possible roles of PAs on key physiological responses of plants to salt stress. In NaCl-treated *Pr. strombulifera*, Put accumulation was related to the antioxidant defense system in this species (Reginato et al., 2012). In *V. radiata*, supplementation with exogenous Put resulted in better seedling growth, associated with enhanced glutathione and ascorbate contents, increased activities of antioxidant enzymes and glyoxalase enzyme, and reduced cellular Na+ (Nahar et al., 2016a). GABA is also an important intermediate involved in ROS scavenging under abiotic stress and has been proposed that it contributes to stress protection (Bouche and Fromm, 2004; Liu et al., 2011). NaCl (100 mM) stress induced higher GABA accumulation in *G. max* through DAO activity stimulation, whereas GABA levels were reduced concomitantly to PAs increment during stress recovery (Xing et al., 2007). A regulatory role of GABA on PAs genes the expression of emerged from salt-treated plants of the shrub *Caragana intermedia* (Shi et al., 2010), and *V. faba* plants grown under hypoxia (Yang et al., 2013; Yang et al., 2018). Using quantitative profiling, Dias et al. (2015) detected a decrease of Put level in tolerant and sensitive chickpea genotypes subjected to stress, whereas the sensitive genotype also had reduced GABA. These results were confirmed by a transcriptomic analysis (Liu et al., 2019), indicating that the expression of key genes of the GABA shunt pathway and polyamine degradation was positively induced in soybean leaves under saline stress. Interestingly, Ca regulated GABA metabolism pathways in germinating soybean under NaCl stress, by changing the contribution ratio of GABA shunt and polyamine degradation pathway for GABA formation (Yin et al., 2014). Taken together, these results strongly support the idea that sustaining GABA levels could be a major strategy to cope with salinity in legumes.

It is well-known that both PAs and the osmolyte proline possess a common precursor: glutamate. Glutamate can be directly converted to proline through the glutamate Δ1 -pyrroline-5-carboxylate pathway or indirectly to PAs *via* its acetylation in ornithine and arginine (Hayat et al., 2012). This connection between PAs and proline metabolisms would imply that considerable stress-induced changes in the pool of PAs could cause a shift in the proline synthesis pathway. Effectively, in 2-week-old soybean seedlings subjected to NaCl, Su and Bai (2008) found a negative correlation between proline accumulation and endogenous Put content. However, in longterm salt-stressed *L. glaber*, PAs and proline accumulations were not correlated (Sanchez et al., 2005). Such a divergence is not surprising, since it was recently found in wheat that production of proline was partly regulated independently, and not in an antagonistic manner from the PAs synthesis (Pál et al., 2018).

Other works have addressed the salt-specific, ionic homeostasis response, through the regulation of non-selective cation channels. These protein channels are known to be PAs targets (Liu et al., 2000) and their blockage by PAs led to the prevention of the salt-induced K+ efflux from *Pisum sativum* mesophyll cells (Shabala et al., 2007). In the last species, PAs were also shown to interact with ROS to alter intracellular Ca2+ homeostasis by modulating both Ca2+ influx and efflux transport systems at the root cell plasma membrane (Zepeda-Jazo et al., 2011). Pottosin et al. (2012) found a synergism between PAs and ROS in the induction of passive Ca2+ and K+ fluxes in roots, which would impact K+ homeostasis and Ca2+ signaling under stress. Also, in *P. sativum* roots, PAs caused plasma membrane depolarization, activated Ca2+, and modulated H+-ATPase pump activity (Pottosin et al., 2014b). Taken together, this data suggests a possible link between PAs and Ca2+ homeostasis, and stress responses in legumes, which deserves further attention.

Most of the studies mentioned above have focused on the effects caused by salt after many hours or days of stress treatment. Geilfus et al. (2015) performed an extensive metabolomic analysis of the fast responses to moderate NaCl stress in *V. faba*. The metabolite profile revealed a rapid reduction in the content of leaf Spd, a PA that is especially relevant for H2O2 production during its catabolism by PAO. This fast reduction of leaf Spd was suggested to contribute to the excessive ROS production observed in these plants, which started simultaneously 45 min after NaCl treatment. However, authors did not report whether that early oxidative burst served as a beneficial event under NaCl stress or caused oxidative damage. A hint that PAs catabolism would be beneficial for plant growth was provided by Campestre et al. (2011). These authors treated 7 day-old *G. max* seedlings with NaCl demonstrated that ROS generated as a consequence of PAs catabolism participate in the hypocotyl elongation of stressed plants, as apoplastic ROS promote the leaf elongation under salinity (Rodríguez et al., 2009).

A significant number of additional works have analyzed the salt-induced changes in the PAs profiles of different legumes like *G. max* (Zhang et al., 2014), *P. vulgaris* (Zapata et al., 2008; Shevyakova et al., 2013; López-Gómez et al., 2014, Talaat, 2015; López-Gómez et al., 2016) and *L. tenuis* (Maiale et al., 2004; Sanchez et al., 2005; Sannazzaro et al., 2007). Taken together, these studies performed on legumes indicate that PAs metabolism is involved in many physiological processes affected by salinity, through specific mechanisms which contribute to counteract the effects of salinity.

#### Heavy Metals and Extreme Environments

Heavy metals-derived soil pollution is one of the most serious worldwide environmental problems (Ruttens et al., 2006), which poses a health risk to humans and animals through the food chain or contaminated drinking water (Granero and Domingo, 2002). In plants, heavy metal toxicity may cause chlorosis, necrosis and several alterations in plant phenotype (Benzarti et al., 2008). One of the symptoms of metal phytotoxicity is oxidative stress, so plant defense system includes a battery of diverse antioxidants (Hossain et al., 2012). Although Lin and Kao, (1999) stated almost 20 years ago that Put accumulation may be part of the syndrome of copper toxicity, more recent works indicated that PAs biosynthesis in the presence of heavy metals such as Zn, Cu, Cd, Mn, Pb, Fe, and Al could be exerting an antioxidant function by protecting the tissues from the metals-induced oxidative damage, although the precise mechanism of protection still needs to be elucidated (Wolff et al., 1995; Franchin et al., 2007; Groppa et al., 2007).

Heavy metal stress induced the accumulation of Spd in *Cajanus cajan* (Radadiya et al., 2016) and of Put, Spm, and Spd in *V. radiata* (Choudhary and Singh, 2000). In *G. max*, the involvement of PAs seedlings response to cadmium stress was revealed by the induction of *SAMDC* after 3 and 24 h of Cd treatment (Chmielowska-Bąk et al., 2013). Also, in *V. radiata*, exogenous Spm application reduced content, accumulation, and translocation of Cd to different plant organs, which consequently reduced ROS production and oxidative damage, thus preventing chlorophyll degradation (Nahar et al., 2016b).

Increased solubilized Al may result from soil acidification (common in tropical and subtropical regions; Hoekenga et al., 2003; Rahman et al., 2018), limiting crop production (Kochian et al., 2004). Aluminum toxicity induces oxidative damage by overproducing reactive oxygen species (ROS; H2O2 and O2 •−). Exogenous Spd induced the protection of photosynthetic pigment and improved growth performances of *V. radiata* against Al stress, by regulating proline, and activating enzymatic and nonenzymatic antioxidant defenses (Nahar et al., 2017). Favoring NO over H2O2 production by the application of ascorbate (a H2O2 scavenger), a higher expression of the *ADC* gene and increased PAs biosynthesis and GABA were associated with improved Pb tolerance in *Prosopis farcta* (Zafari et al, 2017).

On other hand, evolution has allowed plants to adapt to extreme environments, including severe cold, high salinity, drought conditions, intense heat, acid soils, and desert environments (Oh et al., 2013). Plants that inhabit those environments and can grow optimally at or near those extreme ranges are usually called extremophiles and harbor a range of mechanisms that help them to withstand these extreme conditions. The literature addressing PAs metabolism in legumes using species that naturally occur in these environments is null, but some information is available from studies performed on common legume crops cultivated under heat conditions and high heavy metal levels. Heat stress induced accumulation of PAs in heat-tolerant *A. hypogaea*, a phenomenon already observed on cell cultures of heat tolerant plants and absent on susceptible ones (Königshofer and Lechner, 2002; Raval et al., 2018). A protective role of PAs in this condition has also been demonstrated in *V. radiata* and *C. cajan* under heat stress. Application of either 0.1 or 1 mM of exogenous Put, Spd, or Spm in *V. radiata* seedlings resulted in an enhancement of the thermal protection, as shown by growth parameters (Basra et al., 1997). Also, exogenous Put or Spm application (0.5 mM) to *C. cajan* seeds resulted in a higher germination percentage, and when applied on the seedlings, it resulted in the accumulation of proline (Silva et al., 2015). The protective role of the *de novo* PAs synthesis was revealed by treating *V. radiata* seedlings with inhibitors of PAs biosynthesis [2 mM and 4 mM of either difluoromethylornithine (DFMO) or DFMA], which rendered them vulnerable to heat-shock (lower root, hypocotyl and whole seedling length), being the inhibitory effect reversed by exogenous Put (1 mM; Basra et al., 1997). Overall, PAs accumulation seems to protect legumes against extreme environments and heavy metal toxicity. Reports suggest that the protective effect would be mainly through modulation of the redox metabolism and osmo-protection. However, further research is needed in order to unravel the connection between PAs pathways and the observed mitigation effects.

#### Polyamines and Biotic Interactions in Legumes

#### Plant Pathogens and Herbivores

Accumulating evidence indicates that PAs play an essential role in maintaining cell viability during biotic stress, but they also participate in the elicitation of plant defense responses, either functioning as signaling molecules or rather enabling the generation of ROS through their oxidation by PAO. Part of the information supporting the important role played by PAs oxidation during plant response to microbes (reviewed by Jiménez Bremont et al., 2014) has emerged from studies using legume species. These studies suggest that Put oxidation might play a critical role in plant defense against fungal pathogens.

PAs metabolism is tightly regulated during plant-pathogenic interactions (Jiménez Bremont et al., 2014, Rossi et al., 2018). Transcriptomic data indicate that the regulation of PAs metabolism occurring upon pathogen recognition by legume plants is an intricate and complicated process that obeys to genotype, plant growth stage, and the kind of pathogen involved. For instance, in an anthracnose-resistant line of *P. vulgaris*, Spm synthase, and *SAMDC* were downregulated in the first stages of infection by the hemi-biotrophic fungus *Colletotrichum lindemuthianum,* compared to a susceptible line (Padder et al., 2016). However, in later stages *ADC, ODC*, and Spd/Spm synthase transcripts were up-regulated, while those encoding hydroxycinnamoyl transferases and N-acetyl transferases (enzymes involved in conjugation of PAs to hydroxycinnamic acid and acetyl groups, respectively) were down-regulated, indicating that PAs conjugation has no relation to plant resistance. Nevertheless, PAs metabolism is not always directly linked to plant resistance. In this sense, in tolerant and susceptible lines of *L. japonicus* and *M. truncatula* confronted with the bacteria *Pseudomonas syringae*, PAs synthesis, and degradation were equally up-regulated (Bordenave et al., 2013; Nemchinov et al., 2017). These discrepancies challenge the understanding of the real contribution of PAs to plant resistance. For instance, Spd and Spm synthase activities were also up-regulated in soybean against the cyst nematode *Heterodera glycines* (Wan et al., 2015); *SAMDC* and *ADC* transcripts were down-regulated, and the *PAO* gene family was up-regulated in response to the Asian soybean rust *Phakopsora pachyrhizi* Sydow (Panthee et al., 2009). In turn, *ODC* and Spm synthase down-regulation was reported in *L. sativus* in response to *Ascochyta lathyri* (Almeida et al., 2015) and the *M. truncatula–Phymatotrichopsis omnivore* interaction (Uppalapati et al., 2009), respectively. Thus, further research is required for a deeper understanding of the connection between the regulation of PAs homeostasis and plant biotic stress tolerance.

It is known that defense mechanisms deployed by plants against pathogens depend on the coordinated activation of signaling pathways involving the production of hormones such as SA, JA, and Et. Besides, some reports have demonstrated a clear connection between PAs and hormone metabolism. For instance, Ozawa et al. (2009) demonstrated that the treatment of lima bean (*P. lunatus*) with PAs (particularly Spm), led to an increment on JA levels, which in turn promote the production of volatile terpenoids capable of protecting plants against herbivores. Moreover, co-treatment with Spm and JA led to a higher terpenoids production, which has a high potential as a strategy for herbivores control. JA is also produced because of tissue damage and the attack of pathogenic fungi. Chickpea plants treated with JA provoked a remarkable induction in *DAO* expression and conversely, antagonists of JA (such as SA and ABA) repressed the expression of this gene (Rea et al., 2002). In turn, chickpea plants treated with inhibitors of the Spd synthesis (such as cycloheximide) showed higher levels of Et, which seems to be a consequence of the accelerated SAM production and induction of enzymes participating in Et biosynthesis (Gallardo et al., 1994; Gallardo et al., 1995). These data demonstrate a bidirectional relationship between PAs and defense signaling pathways mediated by hormones.

#### Root Symbiosis

#### *Interactions With Arbuscular Mycorrhizal Fungi*

Most lineages of terrestrial plants form symbiotic associations with fungi called arbuscular mycorrhizae (AM) belonging to the *phylum Glomeromycota* (Bonfante and Genre, 2008; Wang et al., 2010). AM fungal root colonization requires the mutual recognition of both organisms involved in the symbiosis (Gadkar et al., 2001; Vierheilig and Piche, 2002), the penetration of the root, and the invasion of the cortical cells to form arbuscules: highly branched fungal structures that facilitate the exchange of nutrients between symbionts (Gutjahr and Parniske, 2013).

Several works using legumes have provided evidences indicating that PAs directly stimulate root colonization by AM fungi. This stimulation would occur through at least two different mechanisms: 1) stimulating mycelial growth and adhesion, and 2) inhibiting ethylene (Et) production in roots. Regarding the first mechanism, it was shown that exogenously supplied PAs to the growth medium of *P. sativum* increased the root colonization with *Glomus intraradices* (El-Ghachtouli et al., 1995). Complementarily, the application of DFMO inhibited colonization by *Glomus mosseae* in *P. sativum* roots, and the inhibition was reverted by simultaneous application of Put (El-Ghachtouli et al., 1996). Also, a positive correlation was found between polyamine chain length and their stimulation of fungal development (El-Ghachtouli et al., 1995). Another approach was used in *G. max* roots, where PAs increased in AM-roots. In these roots, silencing arginine decarboxylase gene (*GmADC*) had a negative effect on mycorrhizal colonization, also affecting the normal development of the plant (Salloum et al., 2018). Interestingly, the silencing of *GmDAO* in the same experimental system, promoted arbuscule formation (Salloum et al., 2018). All these studies successfully demonstrated the importance of PAs in the stimulation of mycorrhizal colonization. Additionally, PAs might directly interact with pectinases of the fungi, increasing adhesion or penetration to the plant cell wall, as observed in other plant-fungi interactions (Charnay et al., 1992; Nogales et al., 2008). Exogenous Et applied to both roots growth medium or leaves, have negatively affected AM infection in *M. sativa* (Azcon-Aguilar et al., 1981). Since Spm and Spd has been proved to block Et synthesis in apple fruits (Apelbaum et al., 1981; Mattoo et al., 2018), it is possible that part of the PAs effect could be mediated by the reduction of Et levels in root tissues. The hypothesis about Et inhibiting root colonization has been tested in brz (E107) *P. sativum* mutants, where higher Et levels were correlated with lower mycorrhizal colonization (Foo et al., 2016; Morales Vela et al., 2007). Interestingly, in *ein2* (Et insensitive) mutant plants treated with Ethephon, mycorrhizal colonization was not reduced as in the wild type, although PAs were not study in these mutants. In consequence, the role of PAs reducing root colonization by Et has been barely suggested. Further research needs to be done to confirm if Et is required to the role of PAs in AM-*ein2* pea mutants treated or not with Ethephon.

As previously discuss, PAs are involved in stress tolerances responses, and in the case of mycorrhizal interaction, PAs could be mediating mitigation effects. A few works have addressed the putative role of PAs on the AM-induced mitigation of plant abiotic stress in legumes. In mycorrhizal *Trigonella foenum-graecum*, a reduction of salt-induced damage of the cell membrane ultrastructure was attributed to a higher PAs (and osmolyte) concentration (Evelin et al., 2013). The inoculation of *V. faba* with selected AM fungi *Funneliformis mosseae* (syn. *Glomus mosseae*), *Rhizophagus intraradices* (syn. *Glomus intraradices*), and *Claroideoglomus etunicatum* (syn. *G. etunicatum*) caused amelioration of the negative effects produced by NaCl (Abeer et al., 2014). In the last work, AM fungi stimulate increases in PAs at a higher extent to that induced by NaCl itself. The higher PAs increase was interpreted by authors as a proof of the protective role of these phytoconstituents against salt stress. Likewise, in *L. tenuis* plants mycorrhized with *Glomus intraradices* under salt stress, a higher content of total free PAs, compared to non-AM ones was reported by Sannazzaro et al. (2007). Since PAs have been proposed as candidates for the regulation of root development under saline situations, authors suggested that the better shape to cope with salt stress displayed by AM *L. tenuis* plants was related to the higher polyamine levels registered.

AM colonization may also increase tolerance of legumes to heavy metals. Attenuation of Pb toxicity in AMF-associated *Calopogonium mucunoides* was associated to a change in amino acids composition favoring metabolic pathways not related to protein, but to PAs biosynthesis (Souza et al., 2014).

As it was pointed out in previous sections PAs play a role in abiotic stress tolerance by regulating water status, ion homeostasis, photosynthesis, and redox status in plant tissues. However, the specific mechanisms linking PAs to abiotic stress mitigation in legumes by AM fungi are not well understood. In this sense, transcriptomic and metabolomic analysis in plants with silenced *ADC* and *DAO*, interacting with AM fungi under abiotic stress could be a good starting point for new research hypothesis. Importantly, in contrast to *Arabidopsis* and other plant families, legumes will allow the study of PAs roles in the triple interaction with rhizobia and AM fungi.

#### *Interaction With Rhizobia*

Legumes may establish symbiotic associations with 98 species of NFB (Weir, 2011). A new plant organ, the symbiotic root nodule, (Brewin, 1991; Hadri et al., 1998), which hosts bacteria in an optimized environment for fixing atmospheric dinitrogen is formed during the interaction (Jones et al., 2007; Wagner, 2012). One of the first evidences that bacterial PAs could be involved in the nodulation ability of the plant was provided by Ferraioli et al. (2001). These authors studied the response of *P. vulgaris* roots to the inoculation with an *N-acetyl-gammaglutamyl phosphate reductase* mutant strain of *Rhizobium etli* (unable to grow with ammonium as the sole nitrogen source), and revealed that the early root responses to rhizobial infection were absent with the arginine auxotrophous, but present with the wild-type parent. PAs regulate nodule metabolism mainly through their action on plasma membrane proteins. In soybean nodules, the addition of 200 μM Spd and Put inhibited by 37 and 54% the H+-ATPase activity, and both inward and outward ammonium channels, showing that high PAs levels have potential to reduce nitrogen supply to the plant *in vivo* (Whitehead et al., 2001). In addition, the *M. truncatula*-*Sinorhizobium meliloti* symbiotic system provided evidence that the H2O2 produced by PAs catabolism plays a role in the inhibition of the symbiosis establishment (Hidalgo-Castellanos et al., 2019).

Previous information regarding the role of PAs during early infection stages, their effect (as well as that of GABA) on the regulation of nodule development and efficiency for nitrogen fixation, as well as the expression of genes related with PAs metabolism during the interaction was reviewed by Jiménez-Bremont et al. (2014). That information was centered on *L. japonicus, Galega orientalis, M. sativa*, and *M. truncatula.* More recently, Becerra-Rivera and Dunn (2019) have thoroughly reviewed the types and levels of PAs contents in nodules, their biosynthetic pathways, and the influence of polyamine on traits that are important for the bacterial-host interaction, such as growth capacity, abiotic stress resistance, motility, EPS production, and biofilm formation, addressing some possible intervening mechanisms. These authors also described current knowledge on polyamine synthesis and regulation in rhizobia.

Some works regarding the relationships among the symbiosis with rhizobia, abiotic stresses, and PAs levels were left aside by the mentioned reviews. For example, salinity induced increased PAs levels in *M. sativa/R. meliloti* (Goicoechea et al., 1998) and *L. tenuis/Mesorhizobium tianshanense* (Echeverria et al., 2013). Also, *A. hypogaea* inoculated with *Bradyrhizobium* sp. SEMIA6144 presented a higher (Spd + Spm)/Put relationship in leaves, concomitantly with ameliorated drought symptoms, compared with non-inoculated ones, suggesting that this condition favored tolerance to water deficit (Cesari et al., 2019). In fact, the increase in the relationship (Spd + Spm)/ Put has been used as an indicator of plant tolerance to abiotic stresses on other species (Zapata et al., 2004). However, the mechanism involved in the PAs-mediated improvement of tolerance in symbiotic plants is not well understood yet. In this regard, some clues could be obtained by focusing future research on the mutual relationships among salt-induced changes of PAs, and metabolites intervening in carbon and nitrogen metabolisms in nodules, and the entire plant. Glutamate occupies a central position in amino acid plant metabolism, and it is the precursor of arginine, ornithine (both PAs precursors), and proline (**Figure 1**). In turn, arginine (along with asparagine) is a key storage compound in higher plants (Forde and Lea, 2007). Moreover, the group of reactions from glutamate to proline, ornithine, arginine, PAs, and GABA constitutes major pathways for carbon and nitrogen assimilation and partitioning (Rawsthornea et al, 1980; Naliwajski and Skłdowska, 2018). On other hand, there is strong evidence that the stress affects the activity of the enzymes involved in the glutamate metabolism (Forde and Lea, 2007). Therefore, in order to unravel the relationship among rhizobial inoculation, higher PAs levels, and plant tolerance to stress, it would be helpful to compare the regulation of carbon flow in nitrogen metabolism pathways associated with salt-induced alteration of PAs levels, between nodulated and control plants.

#### Integrative Tools At the Spotlight: Multiomics Approaches to Decrypt the Interface Legumes—Polyamines-Abiotic and Biotic Stress Interactions

Omics such as transcriptomics and metabolomics have had essential roles in identifying how plant-microbe associations could deviate from classical outcomes in specific conditions (Romero et al., 2017; Rosenberg and Zilber-Rosenberg, 2018). Based on these high-throughput sequencing technologies, an emerging role of PAs in transcriptional regulation and translational modulation of the stress response has been proposed (discussed in Tiburcio et al., 2014). The use of omics is essential to address key questions, such as the specific role of PAs in signaling during abiotic stress (Pál et al., 2015). Transcriptomic assessment of gene expression, perhaps the most prevalent approach to investigate the specific effects of plant coping with stress or development, should not only be restricted to mRNA. This technique has been extensively used to assess the global transcriptional response of PAs transgenic over-expressers during abiotic stress (reviewed in Marco et al., 2011). Nevertheless, other RNA species, such as small RNAs have been reported to play important roles in legume symbiosis, nitrogen fixation, and general plant development. In addition, microRNAs could have specific effects associated with modulation of transcription factors *via* translational arrest. In this scenario, mRNA transcriptomic would fail to identify the biological effects of miRNAs (Hussain et al., 2018).

The development of gene networks based on transcriptomics data has scaled reductionist approaches associated with target sequencing traditional practices. Gene networks based predictions have successfully been applied to survey the uptake, translocation, remobilization, and general regulation of N metabolism in model and crop species (Fukushima and Kusano, 2014). Predictions from transcriptomic data can be weighed with additional complementing technologies. For instance, in soybean, differentially regulated proteins were identified by integrated proteomics and metabolomics during hormone treatment presenting a supported model of altered flavonoid and isoflavonoid metabolism upon Et and ABA treatment (Gupta et al., 2018). Another interesting example of complementation of transcriptomics and metabolomics data in legumes was reported during phosphate deficiency in different plant organs (reviewed in Abdelrahman et al., 2018). It is interesting to point out that several of the databases used to assess legume data are restricted to model species, and there is a need to extend these resources to new crops, which are valuable tools to evaluate in-depth omics data (Bagati et al., 2018). In this context, the envisioning of legume molecular targets to be used in biotechnological applications would be more realistic, and the role of PAs probably more deeply understood. Finally, we call to revisit a long-standing proposal, a renewal of the PAs research landscape based on holistic approaches such as system biology (Montanez et al., 2007). Holistic system biology approaches will pave the way to the understanding of the gene-to-metabolite networks that define legume and PAs metabolisms interrelationships.

## AUTHOR CONTRIBUTIONS

PS and FE wrote the relationships of polyamines with water stress; SM and AR with saline stress; and AnG, FMR, FRR, and LS with biotic stress. The interrelationships between symbionts and polyamines were written by MM, MS, AM, VM, and MC. AyG and CB are the authors of the revision of the roles of polyamines in plants in extreme environments. The design and general coordination of the review work was carried out by AM, PC, JUD, and OR.

### REFERENCES


on acidic soils: current status and opportunities. *Int. J. Mol. Sci.* 19, 3073. doi: 10.3390/ijms19103073


Wagner, S. C. (2012). Biological nitrogen fixation. *Nat. Educ. Knowl.* 3, 15.


damage in roots of vegetable soybean. *J. Integr. Agric.* 13, 349–357. doi: 10.1016/ S2095-3119(13)60405-0

**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Menéndez, Calzadilla, Sansberro, Espasandin, Gazquez, Bordenave, Maiale, Rodríguez, Maguire, Campestre, Garriz, Rossi, Romero, Solmi, Salloum, Monteoliva, Debat and Ruiz. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Polyamines in Food

Nelly C. Muñoz-Esparza1,2,3, M. Luz Latorre-Moratalla1,2,3, Oriol Comas-Basté1,2,3 , Natalia Toro-Funes <sup>4</sup> , M. Teresa Veciana-Nogués 1,2,3 and M. Carmen Vidal-Carou1,2,3 \*

<sup>1</sup> Department of Nutrition, Food Sciences and Gastronomy, Faculty of Pharmacy and Food Sciences, University of Barcelona (UB), Barcelona, Spain, <sup>2</sup> Research Institute of Nutrition and Food Safety of the University of Barcelona (INSA·UB), Barcelona, Spain, <sup>3</sup> Catalonian Reference Network on Food Technology (XaRTA), Barcelona, Spain, <sup>4</sup> Eurecat, Technological Unit of Nutrition and Health, Technology Centre of Catalonia, Reus, Spain

The polyamines spermine, spermidine, and putrescine are involved in various biological processes, notably in cell proliferation and differentiation, and also have antioxidant properties. Dietary polyamines have important implications in human health, mainly in the intestinal maturation and in the differentiation and development of immune system. The antioxidant and anti-inflammatory effect of polyamine can also play an important role in the prevention of chronic diseases such as cardiovascular diseases. In addition to endogenous synthesis, food is an important source of polyamines. Although there are no recommendations for polyamine daily intake, it is known that in stages of rapid cell growth (i.e., in the neonatal period), polyamine requirements are high. Additionally, de novo synthesis of polyamines tends to decrease with age, which is why their dietary sources acquire a greater importance in an aging population. Polyamine daily intake differs among to the available estimations, probably due to different dietary patterns and methodologies of data collection. Polyamines can be found in all types of foods in a wide range of concentrations. Spermidine and spermine are naturally present in food whereas putrescine could also have a microbial origin. The main polyamine in plant-based products is spermidine, whereas spermine content is generally higher in animal-derived foods. This article reviews the main implications of polyamines for human health, as well as their content in food and breast milk and infant formula. In addition, the estimated levels of polyamines intake in different populations are provided.

Keywords: spermidine, spermine, putrescine, polyamines, human health, food, breast milk

## INTRODUCTION

In 1678, Antoni van Leeuwenhoeck discovered the presence of crystals in human semen, which 200 years later (1888) were named spermine by A. Landenburg and J. Abel. The chemical structure of spermine and spermidine was determined in 1924 (1). The polyamines spermidine (N-(3-aminopropyl)-1,4-butane diamine), spermine (N,N-bis (3-aminopropyl)-1,4-butane diamine), and putrescine (1,4-butane diamine) have a low molecular weight and are characterized by having two or more amino groups. They are found in all living cells, including in microorganisms, plants, and animals. Due to their structure (**Figure 1**), polyamines are relatively stable compounds, capable of resisting acidic and alkaline conditions and they can establish hydrogen bond with hydroxyl solvents such as water and alcohol (2–6). In the organism, at physiological pH they are completely protonated and strongly bound to polyanionic macromolecules such as DNA and RNA (2, 7, 8). On the other hand, polyamines can also be found in food of both animal and plant origin. An important source of polyamines for humans is breast milk and infant formula (2, 4).

#### Edited by:

Ana Margarida Fortes, Faculty of Sciences, University of Lisbon, Portugal

#### Reviewed by:

Jouko Vepsäläinen, University of Eastern Finland, Finland Keiko Kashiwagi, Chiba Institute of Science, Japan

#### \*Correspondence:

M. Carmen Vidal-Carou mcvidal@ub.edu

#### Specialty section:

This article was submitted to Nutrition and Food Science Technology, a section of the journal Frontiers in Nutrition

Received: 10 April 2019 Accepted: 28 June 2019 Published: 11 July 2019

#### Citation:

Muñoz-Esparza NC, Latorre-Moratalla ML, Comas-Basté O, Toro-Funes N, Veciana-Nogués MT and Vidal-Carou MC (2019) Polyamines in Food. Front. Nutr. 6:108. doi: 10.3389/fnut.2019.00108

**64**

#### Polyamines and Health

Polyamines play an essential role in cell growth and proliferation, the stabilization of negative charges of DNA, RNA transcription, protein synthesis, the regulation of the immune response, apoptosis, the regulation of ion channels, particularly by blocking potassium channels, and as antioxidants (2, 4, 5, 7, 9–12).

The antioxidant activity of polyamines mainly affects membrane lipids and nucleic acids. Spermine is the polyamine with the strongest antioxidant properties, associated with its higher number of positive charges. The main mechanism of polyamine antioxidant action is metal chelation, which prevents the formation of hydroperoxides and delays the generation of secondary oxidation compounds (13–16). It has also been proposed that polyamines can eliminate free radicals, especially in lipophilic media (14, 16).

#### Polyamine Homeostasis

The de novo synthesis of polyamines in the organism begins with the formation of putrescine from the amino acid ornithine, catalyzed by the enzyme ornithine decarboxylase (ODC) (**Figure 2**). Putrescine is converted to spermidine by spermidine synthase through the addition of a propylamine group derived from the decarboxylation of S-adenosyl-methionine. Subsequently, spermidine is transformed into spermine by spermine synthase, which adds a second propylamine group (2, 4, 7, 12, 17).

The interconversion of polyamines is a cyclic process that controls their turnover and regulates intracellular homeostasis (**Figure 2**). This process begins with the acetylation of any of the three polyamines, which is catalyzed by an N-acetyltransferase enzyme with the participation of acetyl coenzyme-A. Subsequently, the enzyme polyamine oxidase (PAO) removes a propylamine group, and putrescine is obtained from the acetylated metabolite of spermidine, or spermidine from the acetylated metabolite of spermine (2, 10, 12, 17, 18).

The elimination of polyamines from the organism is carried out by the oxidative deamination of a primary amino group, mainly by the action of diamine oxidase (DAO) and PAO. Both enzymes can act on polyamines and their acetylated derivatives (2, 4, 7, 10, 17).

Besides endogenous synthesis, polyamines also have an exogenous origin, mainly food and breast milk (2). In addition, gut microbiota is also described as a source of polyamines, mainly forming in the large intestine (2, 19, 20). Some recent studies have been linked different intestinal microbial species with the synthesis of these compounds (20). However, more information is still needed on the capability to form polyamines of the gut microbiota and the corresponding biosynthetic pathways. Finally, intestinal and pancreatic secretions and catabolism products of intestinal cells also contribute to the polyamines in the gut (2). Polyamines are absorbed in the duodenum and in the first portion of the jejunum by various mechanisms, including transcellular (through passive diffusion and transporters) and paracellular pathways (2, 4, 21). Polyamines are partly metabolized in the intestinal wall before reaching the blood circulation, and those that pass into the circulation are distributed throughout the organism and captured by the tissues, where they can undergo interconversion reactions.

The highest concentrations of polyamines are found in the intestine, thymus and liver (2, 4). A diet enriched with polyamines raises plasma levels in experimental animals and humans (22).

#### Potential Effects of Polyamines Postnatal Stage

Several studies describe the importance of polyamines in humans, especially in the early stages of life. It is known that during rapid cell growth, particularly in the neonatal stage, the need for polyamines increases (4, 5, 21, 23, 24). Requirements are also higher after surgery or during periods of wound-healing and aging (2, 23, 25).

Polyamines (spermine and spermidine) promote the proliferation and maturation of the gastrointestinal tract and are involved in the differentiation and development of the immune system (5, 21, 25–31). In addition, due to their antioxidant properties, these compounds can participate in the regulation of the inflammatory response (12, 22).

Several studies have demonstrated that oral administration of polyamines in mice induces early postnatal maturation of the intestine and acts in the repair of the intestinal mucosa and in the immune and inflammatory response. Spermine and spermidine modified protein expression and the activity of disaccharidases and accelerated postnatal intestinal maturation, producing morphological changes in the intestinal epithelium and mucosal permeability (29). They also participate in the maturation of associated organs such as the liver and pancreas. In another study with mice, the oral administration of polyamines, mainly spermidine, was found to promote the early maturation of glycoprotein fucosylation. A dose of 10 µmol/day of each polyamine increased the activity of α-1,2-fucosyltransferase and α-L-fucosidase and induced the synthesis of α-1,2-fucoprotein (32, 33). The authors of this study suggest that postnatal changes in the fucosylation of intestinal glycoproteins could be related mainly to the intake of polyamines, especially spermidine and spermine. Another study showed that oral administration of spermine in mice increases the activity of alkaline phosphatase and disaccharidase, and subsequently alters intestinal maturation (34). The administration of spermine and spermidine in newborn rats increased the intestinal weight and length and accelerated its maturation (21). Regarding the immune response at the intestinal level, various studies in animals have indicated that the oral administration of spermine and spermidine in the postnatal period improves the maturation of the intestinal immune cells and increases the levels of immunoglobulin A in the villi and crypts of the intestine (21, 29).

In humans it is widely reported that breast milk enhances the maturation of immune cells and decreases intestinal permeability to antigenic macromolecules, reducing the risk of food hypersensitivity in the infant (21, 25, 26, 29, 35, 36).

#### Aging

In the aging process, the cellular levels of spermine and spermidine and the enzymatic activity of ODC tend to decrease

(17, 37, 38). Enrichment of the diet with polyamines during this stage can reduce the risk of age-associated pathologies and promote longevity (39, 40). In a study in aging mice, a diet with high levels of spermine and spermidine (374 and 1,540 nmol/g, respectively) increased the concentrations of these compounds in the blood and reduced levels of pro-inflammatory markers, age-associated DNA methylation, renal glomerular atrophy and mortality (39). It has also been observed that spermidine increases autophagy, which involves the removal of damaged proteins and organelles from cells, thus inhibiting the aging process (6, 41–43). In a follow-up study of a cohort of 829 participants during 20 years, spermidine showed the strongest inverse relation with mortality among 146 nutrients investigated. This effect was dose-dependent, and the authors explain that spermidine effectively induced autophagy and can reduce the acetylation of histones, which are critical processes for cell homeostasis in aging. In this sense, a diet rich in spermidine, mainly from foods of vegetable origin (green pepper, wheat grain, mushrooms, etc.), was associated with a decrease in the risk of all-cause mortality in the general community (42).

#### Cardiovascular Disease

The antioxidant and anti-inflammatory effects attributed to polyamines can play an important role in the prevention of chronic inflammatory pathologies, such as cardiovascular diseases (22). A higher intake of spermidine has been correlated with a lower incidence of cardiovascular diseases and a decrease in blood pressure and heart failure (44). It is likely that the anti-inflammatory role of polyamines in the prevention and treatment of cardiovascular disease is similar to that of polyunsaturated fatty acids (PUFA 3-n) and statins (22, 45). In animal studies, mainly in aging mice, spermidine has been shown to decrease age-induced arterial stiffness and oxidative damage of endothelial cells (44). In addition, 6 week supplementation of spermine and spermidine in mice reversed age-associated changes in myocardial morphology (myocardial fibrosis) and inhibited cellular apoptosis of the heart (46).

#### Diabetes

Glycation has an important role in the development of diabetes complications, so compounds that can counteract this reaction are desirable. Due to their chemical structure, polyamines could function as antiglycan agents, delaying the accumulation of advanced glycation end-products (AGEs) (7, 47). This effect would be due to the interaction between the free amino groups of polyamines and the highly reactive carbonyl compounds (10, 18). In vitro studies have demonstrated that the millimolar concentrations of spermine present in the cell nucleus can protect DNA and histones from glycation (18).

On the other hand, some authors have observed a higher PAO activity in children with diabetes mellitus type 1, which could induce an increased production of free radicals and subsequent oxidative damage (47). Therefore, more studies are needed to clarify the role of polyamines in diabetes and establish recommended levels of polyamine intake for the diabetic population.

#### Cancer

Elevated levels of polyamines in cancer patients are associated with tumor growth (48, 49). A deregulation in polyamines biosynthesis, mainly due to an increase in the activity of the ODC enzyme, leads high intracellular polyamine content in cancer cells (12, 39, 45, 49). Therefore, controlling polyamine synthesis could be useful in antineoplastic therapy. According to different experimental studies and clinical trials, the combined treatment using difluoro-methylornithine (DFMO), a potent and irreversible inhibitor of ODC, with polyamine transport inhibitor drugs or non-steroidal anti-inflammatory drugs (NSAIDs), efficiently reduced the carcinogenesis by inhibiting polyamines synthesis and stimulating polyamines catabolism and export (48, 49).

An increase of the acetylated metabolites of polyamines has been observed in urine or blood in patients suffering cancer disease. The rise of acetylated polyamines in urine may be explained by an increase of cellular polyamines, an increase of the SSAT activity, a major excretion of acetylated metabolites from cells or by a decrease of their oxidative degradation by PAO enzyme, although the molecular mechanisms are not wellelucidated (50). The development of more sensitive metabolomic techniques in the last decade has allowed detailed polyamine metabolic profiles to be associated with certain types of cancer (48). In fact, increased levels of acetylated polyamines in urine or blood, particularly, N1,N12-diacetylspermine, N1,N8 acetylspermidine, N1-acetylspermine, and N8-acetylspermidine have been found in patients with ovarian, prostate, colorectal, pancreatic, breast and lung cancers. Among them, N1,N12 diacetylspermine has been extensively described as the most effective urinary biomarker for several types of cancer and to monitor tumor's progression (48, 51, 52).

Despite advances in understanding the role of polyamines in cancer, more research is required on the molecular basis in which polyamines participate. Determining how to optimally intervene in polyamine metabolism and function could lead to therapeutic benefits in cancer treatment.

#### Polyamines in Food

Polyamines are found in foods of both animal and plant origin, either in a free or conjugated form. Conjugated polyamines are found in plant-derived foods mainly linked to phenolic compounds (4, 24). In foods, spermidine and spermine are primarily naturally present, coming from raw plant and animal tissues, whereas putrescine may also be formed by the activity of fermentative or contaminating microorganisms (12, 53). It has also been described that spermidine and spermine may partly have a bacterial origin, especially in fermented products (12, 54, 55). Therefore, processing and storage conditions can influence the total content of polyamines.

#### Breast Milk and Infant Formula

The first dietary exposure to polyamines is through breast milk. **Table 1** shows the contents of polyamines in breast milk and infant formula reported in the literature, with all results expressed in nmol/ml to facilitate comparison. All the studies reviewed agree that the content and profile of these compounds can vary depending on factors such as genetics, the lactation phase, and the age, nutritional status and dietary intake of the mother.

The major polyamines in breast milk are spermidine and spermine and their contents differ considerably, with coefficients of variation >68 and 53%, respectively. Spermine values are generally higher, except in two studies by the same author, in which higher values are reported for spermidine (11, 25). As indicated in **Table 1**, the breast milk analyzed in different studies corresponds to different phases of lactation, which could contribute to the high variability observed. In this sense, some authors have described that the polyamine content tends to decrease over the course of lactation (26, 56). Additionally, two studies found higher polyamine contents in the milk of mothers of preterm infants compared to full-term (25, 31). Also, as preliminary information it should be noted that milk from obese mothers was found to contain fewer polyamines than milk from those with normal weight (11).

In infant formula the variability among results of different studies is even higher than for breast milk, with coefficients


<sup>1</sup>Milk from mothers of normal weight.

<sup>2</sup>Milk from obese mothers.

<sup>a</sup>Breast milk 2 months postpartum.

<sup>b</sup>Breast milk 1 month postpartum

<sup>c</sup>Breast milk 1 week postpartum.

nd, not detected.

of variation of 89% for putrescine, 116% for spermidine, and 160% for spermine. Despite this variability, it can be extrapolated that the polyamine content and profiles in infant formula differ from those of breast milk. For example, the major polyamine in infant formula is putrescine, its content usually higher than in breast milk, whereas spermidine and spermine levels tend to be lower. When first and follow-on formula are compared, no differences can be observed in the mean contents of polyamines. Likewise, the few available data on polyamines in infant formulas for premature babies do not allow to observe differences with other types of formulas.

The available data on polyamine content in breast milk and infant formula are scarce and, in some cases, outdated. More studies are needed to clarify whether the variability observed both in breast milk and infant formula is due to the use of different analytical methodologies or to other factors that have not been sufficiently investigated.

#### Food of Plant Origin

Polyamines are ubiquitous in foods of plant origin, although their content and distribution vary depending on the type of food (**Table 2**). Spermidine, present in all plant-derived foods, is generally the predominant polyamine. The food categories with the highest contents of spermidine and spermine are cereals, legumes and soy derivatives. Wheat germ and soybeans stand out in particular, with respective values of 2,437 and 1,425 nmol/g for spermidine and 722 nmol/g and 341 nmol/g for spermine (37, 59). Mushrooms, peas, hazelnuts, pistachios, spinach, broccoli, cauliflower and green beans also contain significant amounts of both polyamines. The lowest levels are found in the fruit category. For example, in apples, pears, cherries, oranges or tangerines, reported values for spermidine are lower than 21 nmol/g and <1.98 nmol/g for spermine.

Like spermidine, putrescine is found in virtually all foods of plant origin, and is particularly abundant in fruits and vegetables, notably citrus fruits (1,554 nmol/g) and green peppers (794 nmol/g) (9, 61). There are also high amounts of putrescine in wheat germ (705 nmol/g) and soybean sprouts (507 nmol/g) (37, 70).

The variability in polyamine contents in plant-derived products can be due to different factors, including their origin, growing conditions, harvesting, or storage. In this sense, different stress situations of the plant could affect the polyamine content. For example, polyamine levels in plants can increase in response to stress brought by high or low cultivation temperatures or drought (71). Studies show that the application of polyamines pre- and post-cultivation can compensate for the negative effects of cold or drought, thereby favoring germination, plant growth or survival (72–74). Another factor that could explain the high levels of putrescine in some vegetables, such as spinach and peas, is the presence of spoilage bacteria, mainly Enterobacteriaceae and Clostridium spp., which can form putrescine from its amino acid precursor ornithine by their amino acid-decarboxylase activity (12, 62, 75).

#### Food of Animal Origin

In animal-derived foods, like those of plant origin, the contents of polyamines are extremely variable (**Table 3**). Meat and its derivatives may contain high levels of spermidine and spermine, particularly the latter. Spermine values >148 nmol/g have been described in samples of beef, pork, chicken, cured ham, and sausages, without significant differences between fresh meats and derivatives (37, 63, 76, 77). In fish and its derivatives, the contents of spermine and spermidine are generally lower than in meat products, but clearly higher than in milk and eggs, where their levels are low. In most cheeses the values of spermine and spermidine are <10 and 69 nmol/g, respectively, with the exception of a blue cheese with a very high spermidine content (262 nmol/g) (37).

In fresh products of animal origin (meat, fish, milk, and eggs) the putrescine contents are generally lower than in plant-derived foods. However, the highest levels of putrescine are found in products subjected to a fermentation process involving potentially aminogenic microorganisms. The wide range of putrescine contents could also be explained by the TABLE 2 | Ranges of average polyamine content (nmol/g) in foods of plant origin.


nd, not detected.

TABLE 3 | Ranges of average polyamine content (nmol/g) in foods of animal origin.


nd, not detected.

decarboxylase activity of spoilage bacteria. Studies show that the hygienic state of raw materials has an important influence on the formation of putrescine and other amines during the elaboration of different food products. For example, a greater accumulation of amines was reported in dry-fermented sausages when these were produced from raw materials of low microbial quality (78). This factor could also be responsible for increasing putrescine levels in long-maturing cheeses for whose manufacture the use of raw milk is an authorized practice. In this sense, the previous thermal treatment of milk is a useful tool, not only to guarantee the absence of pathogenic microorganisms but also to avoid the formation of putrescine and other biogenic amines, as it decreases a) the load of spoilage microorganisms with amino acid-decarboxylase capacity; b) the presence of free amino acid precursors by delaying proteolysis during ripening; and c) levels of the thermolabile pyridoxal phosphate, a necessary cofactor of the amino acid-decarboxylase enzyme (83).

#### Effects of Culinary Treatment

Culinary treatment can potentially decrease the polyamine content in foods by two possible mechanisms: (a) transfer to the cooking water or (b) due to the high temperatures reached in some types of cooking. The few studies evaluating the effect of culinary treatment on polyamines report variable results, depending on the type of cooking and the food studied. Polyamine contents after the boiling of certain vegetables (spinach, cauliflower, and potatoes) were significantly reduced by transfer to the cooking water, especially putrescine, as this is the most water-soluble polyamine. However, the same cooking process did not induce losses in other types of food (peppers, peas, and asparagus) (84). Another study found no significant differences in polyamine levels between raw and boiled vegetables (carrots, broccoli, cauliflower, and potatoes), although the low number of samples analyzed (two per food type) was a limiting factor (9). In meat subjected to a cooking process involving a large amount of water (stewing and boiling), no significant losses of spermidine and spermine were observed either (23, 53). In the case of some cooking techniques that involved higher temperatures (53) described that roasting, grilling, or frying produced losses of up to 60% of spermidine and spermine in chicken meat.

#### Antioxidant Potential of Polyamines in Food

Studies of the antioxidant role of polyamines in food are scarce compared to those in biological substrates. The protective effect of polyamines against oxidation when added to a lipid matrix has been demonstrated in vitro, mainly acting as metal chelators. A concentration-dependent antioxidant capacity was reported for spermine and spermidine (13). Later, Toro Funes et al. (16) also described an antioxidant effect for each of these polyamines at a wide range of concentrations (from 30 to 1,250µg/mL). Specifically, spermine and spermidine delay the formation of peroxides and secondary oxidation compounds, the effects of spermine being greater due to a higher number of amino groups. In addition, these two studies showed that antioxidant activity of both polyamines is equal to or even higher than that of some antioxidant additives commonly used in foods, such as octyl gallate, alpha-tocopherol, ascorbyl palmitate, or tert-butylhydroquinone, among others.

Foods with high contents of polyamines, such as wheat germ, soya, mushroom, or citrus fruits, could be used as natural antioxidant ingredients in the form of powdered concentrates or polyamine-rich extracts. Prior to the use of these extracts or concentrates of polyamine-rich foods as natural antioxidants, effective and safe doses would need to be determined.

#### Analysis of Polyamines in Food

The analytical methodologies to determine polyamines in food are mainly based on the chromatographic separation coupled with distinct detection techniques due to their high resolution, sensitivity and versatility. Gas chromatography, thin-layer chromatography and high-performance liquid chromatography have been applied for the analysis of polyamines in food (85–87). Concretely, high or ultra high-performance liquid chromatography with ion-exchange columns or reverse-phase columns to separate polyamines are the most frequently reported techniques in the literature (88).

Different detection techniques coupled to chromatographic separation systems have been described such as UV, fluorescence and mass spectrometry. Polyamines have low absorption coefficients or quantum yields and require derivatization when the method involve UV or fluorescent detection. Chemical derivatization of these compounds can be carried out with a variety of reagents, mostly 5-dimethylamino-1 naphtalene-sulfonyl chloride (dansyl chloride) that forms stable compounds after reaction with both primary and secondary amino groups and o-phthaldialdehyde (OPA), which reacts rapidly (i.e., 30 seg) with primary amines. Amine derivatives can be formed before (pre-column), during (on-column) or after (post-column) the chromatographic separation. Prederivatization comprises a series of time- consuming manual steps and may introduce imprecision to the overall analytical procedure. Post-column derivatization has the advantage that it is automatically performed online, thereby avoiding sample manipulation and shortening the time required for the analysis (89). In recent years, the determination of polyamines through liquid chromatography coupled to mass spectrometry (MS) or tandem mass spectrometry (MS/MS) has emerged as an alternative analytical technique, very specific and sensitive and without the need of derivatization (52, 87, 88, 90).

Electrochemical sensors or biosensors are an alternative to the analytical procedures described above, being less expensive, less time-consuming, and analytically simpler, especially for routine screenings. Electrochemical biosensors usually consist on immobilized amino-oxidases, which catalyze the oxidative deamination of polyamines present in foods, and a working electrode that detects the production or the consumption of the redox species produced by the enzymatic activity. Different electrochemical sensors developed for the rapid determination of polyamines in food showed low detection limits and good selectivity toward these compounds (86).

#### Polyamine Intake

The daily intake of polyamines has been estimated for different European countries, Japan and the United States (**Table 4**). The mean polyamine intake in the European adult population was estimated as 354 µmol/day, with differences among the member states, being lowest in the United Kingdom and highest in the countries of the Mediterranean area, Italy and Spain (91). Subsequent studies carried out in Mediterranean countries, such as Spain and Turkey, have estimated much lower intake values for these populations (94, 95), which could be partly related to a decrease in the consumption of plant-derived foods due to the progressive abandonment of the traditional Mediterranean diet observed in the last 20 years (96). The polyamine intake estimates for the adult population of Japan and the United States lie between the European mean and the values corresponding to the Mediterranean area. The only study estimating the intake in an adolescent population was carried out in Sweden (93) and the results were very similar to those previously reported for the Swedish adult population (91).

The differences between intake estimates can be attributed not only to the different dietary patterns of each population, but also to the age group studied, the methodology of data collection and/or to the variability in food polyamine content. For example, the food consumption data used to estimate polyamine intake was obtained from published national surveys (Japan and Spain), a frequency-of-consumption questionnaire (United States), a 7 day food record (Sweden) and a 24 h dietary recall (Turkey). In some studies, the data on polyamine content were obtained from analyses carried out specifically for the intake estimation studies (63, 64, 95), whereas others used data already published in the literature (93, 94).



<sup>a</sup>European Union: United Kingdom, Italy, Spain, Finland, Sweden, and the Netherlands.

<sup>b</sup>J-NNS: Nationwide nutrition survey in Japan.

<sup>c</sup>Spanish national dietary survey in adults, elderly and pregnant women.

All the studies agree that the polyamine contributing most to the total intake is putrescine, mainly from the consumption of fruits and vegetables, or in Japan also from cereals and soy sauce. Fruits, vegetables and cereals are also the main sources of spermidine. The main origin of dietary spermine is meat and fish, except in Sweden, where it is vegetables and cereals.

At present there are no official recommendations for the daily intake of polyamines, but some suggestions have been made. Atiya Ali et al. (93) proposed an intake around of 540 µmol/day, taking into account the guidelines of a healthy diet that promotes a high consumption of fruits, vegetables and cereals (93). This estimate is two to three times higher than the intakes reported in the studies reviewed.

### CONCLUSIONS

There is extensive knowledge about the physiological functions of polyamines and their importance for human health. Several studies indicate the importance of dietary polyamines at different stages and situations of life, such as in the postnatal period or aging, when requirements are higher. In addition, the antioxidant and anti-inflammatory effects described for polyamines can play an important role in the prevention of chronic conditions such as cardiovascular diseases and diabetes. On the other hand, cancer is associated with high levels of polyamines, brought about by an alteration in their homeostasis.

The contents of polyamines in food, even within the same type, are highly variable. Breast milk provides the first dietary exposure to these compounds. Despite the scarcity and variability of available data, the content and profile of polyamines in breast milk are clearly different from those observed in infant formula. Among plant-derived foods, cereals, legumes and soybean derivatives are the categories with the highest contents of spermidine and spermine, whereas the highest putrescine levels are found in vegetables and fruits, especially citrus fruits. In animal-derived foods, meat and derivatives have the highest polyamine contents, with the exception of some cheeses. A range of factors could be responsible for the high variability in the polyamine content in food, notably origin and conditions of cultivation of plants, as well as the conditions of processing and storage. The wide range of putrescine contents could be also explained by the decarboxylase activity of spoilage or fermentative bacteria.

Polyamines have been associated with a high antioxidant activity in foods matrices, especially spermine. Therefore, foods rich in polyamines such as wheat germ, soybean, mushroom, or citrus fruits, in the form of extracts or concentrated powders, could be used as natural antioxidant ingredients. Such application will require previous studies to determine safety and effective dosage.

The few studies estimating polyamine intake have published highly variable results. This inconsistency could be attributed not only to the different diets of the studied populations, but also to methodological differences that could be related to the absence of consensus guidelines for the estimation of polyamine consumption. There are currently no official recommendations for daily polyamines intake, although some authors have proposed levels well above the intake estimates made in different countries. The dietary polyamine requirements in the different age groups should also be establish in order to be able to define a rich or low diet in polyamines.

## AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

### REFERENCES


### ACKNOWLEDGMENTS

NM-E is a recipient of a doctoral fellowship from the University of Guadalajara, Mexico.

immune system maturation. Dev Comp Immunol. (2010) 34:210–8. doi: 10.1016/j.dci.2009.10.001


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Muñoz-Esparza, Latorre-Moratalla, Comas-Basté, Toro-Funes, Veciana-Nogués and Vidal-Carou. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Polyamines and Gut Microbiota

Rosanna Tofalo<sup>1</sup> \*, Simone Cocchi <sup>2</sup> and Giovanna Suzzi <sup>1</sup>

<sup>1</sup> Faculty of Bioscience and Technology for Food, Agriculture and Environment, University of Teramo, Teramo, Italy, <sup>2</sup> Farmacie Comunali di Romano di Lombardia, Bergamo, Italy

Keywords: polyamines, gut microbiome, human health, probiotics, fermented foods

The microbiota of gut is the community of microbes living in an individual's gastrointestinal tract. Several bacterial genera and species act in a concerted manner to establish metabolic interactions with the host (1). Although there is a general high interest in the study of metabolite flow across the microbe-host, at present, only some studies are targeting specific metabolites produced by intestinal microbiota such as polyamines (PAs) (2).

#### POLYAMINES AND GUT MICROBIOTA

Polyamines can be defined as small polycationic molecules with a wide array of biological functions including gene regulation, stress resistance, cell proliferation and differentiation, and are associated to both eukaryotic and prokaryotic cells (3).

In human cells spermine, spermidine, and putrescine are the main PAs. Putrescine is produced in the cytoplasm of cells by decarboxylation of ornithine catalyzed by the enzyme ornithine decarboxylase (ODC). Spermine and spermidine are synthesized by S-adenosyl-methionine decarboxylase (AdoMetDC) and a transferase enzyme, catalyzing the transfer of the aminopropyl group to the primary amine group of putrescine or spermidine, respectively (4). The ingested food is the major source of PAs in the lumen, and the upper parts of intestine adsorb the majority of these compounds for growth processes throughout the body (5). The gut microbiota is considered the main responsible of PAs level in the lower part of intestine (6). Polyamines in the colonic lumen are transferred into the bloodstream via the colonic mucosa (7). Intracellular PAs levels are regulated by endogenous biosynthesis, degradation and exogenous transport. Both endocytic and solute carrier-dependent mechanisms have been described for polyamine uptake in the gut lumen (8). In eukaryotic cells they are involved in several physiological functions since they are able to bind to several anionic macromolecules such as DNA, RNA, proteins, and acidic phospholipids (9). The PAs involvement in maintaining chromatin structure and membrane stability and regulating ion-channels and scavenging free radicals has also been reported (10), as well as their role as second messengers in protein and nucleic acid synthesis for normal cell division and growth (11). In particular, cellular PAs availability contributes to tissue homeostasis of the gastrointestinal mucosa, the rates of epithelial cell division and apoptosis, by modulating the expression of various growthrelated genes (12). In general, PAs are involved in several important cellular processes and their disregulation can affect growth, aging and several diseases such as cancer, neurodegeneration and metabolic disorders (13). To maintain good intracellular PAs contents, biosynthetic and catabolic processes are activated and highly regulated. For example, a high intracellular PAs levels are related with cell growth, whereas the inhibition of ODC decreases cellular PAs (12). On the contrary, its overexpression induces an increased level of PAs in human gut, a result that has been related with gastrointestinal cancers (14).

As regards bacteria, new putative phyla (134) other than the traditional ones (30) have been identified using culture-independent metagenomic sequencing and single-cell sequencing (15). However, the studies of PAs distribution in bacteria have been limited to culturable species and few bacterial species have been studied (16).

#### Edited by:

Antonio F. Tiburcio, University of Barcelona, Spain

#### Reviewed by:

Syed Srfraz Shah, Forman Christian College, Pakistan Rafael Peñafiel, University of Murcia, Spain

> \*Correspondence: Rosanna Tofalo rtofalo@unite.it

#### Specialty section:

This article was submitted to Nutrition and Food Science Technology, a section of the journal Frontiers in Nutrition

Received: 29 November 2018 Accepted: 01 February 2019 Published: 25 February 2019

#### Citation:

Tofalo R, Cocchi S and Suzzi G (2019) Polyamines and Gut Microbiota. Front. Nutr. 6:16. doi: 10.3389/fnut.2019.00016

**75**

The bacterial PAs include spermidine, homospermidine, norspermidine, putrescine, cadaverine, and 1,3-diaminopropane with putrescine and spermidine being the most common PA (17, 18). Furthermore, because of high bacterial diversity, some microorganisms produce sym-homospemidine rather than spermidine or produce only a diamine and some bacteria do not produce PAs, such as Staphylococcus aureus (19).

Different bacterial species, up to 1,000, constitute the intestinal microbiota. This community of microorganisms (bacteria, archaea, fungi, protozoa, viruses) is responsible for the metabolism of non- digested food components and it can supply to the host nutrients such as amino acids and vitamins and other biologically active substances (20). In general, this microbial consortium is subject to fluctuations due to different factors such as environment, diet, disease states and many others (21). The microorganisms colonizing the gut can contribute to the overall health of the host or be pathogenic, invading the host, and causing diseases under certain conditions (22).

The majority of the studies on gut microbiota are focused on bacteria, even if all the biota plays important roles in host health and disease (23). The high-throughput sequencing techniques based on the amplification of the 16S rRNA identified more than 120 different prokaryotic phyla with only 31 phyla included cultured species (24). Moreover, the majority of species that constitute the gut microbiota belong mainly to four phyla: Firmicutes, Bacteroidetes, Actinobacteria, Proteobacteria, and among these Firmicutes and Bacteroidetes are dominant phyla in agreement with the high-throughput sequences carried out the last 10 years. The dominant species belong to the families Bacillaceae, Enterobacteriaceae, Corynebacteriaceae, and Bacteroidaceae, with a prevalence of anaerobic species, uncultured yet (25). Culturomics is a new strategy that could improve the study of microorganisms of the human gut microbiota (26).

As regards Archaea, non-methanogenic and methanogenic species are present in human gut microbiota, with the latter producing methane during anaerobic fermentations. These species belong to Euryarchaeota phylum. A very small fungal community is present in the human gut with three major phyla: Ascomycota 63%, Basidiomycota 32%, and Zygomycota 3%.

Putrescine, cadaverine, spermidine, and spermine are the main PAs encountered in bacteria (**Figure 1**). Their synthesis is highly regulated at molecular level through a concerted biosynthesis and uptake mechanisms, as well as by degradation and efflux processes. Their production relies on the presence of amino acidic precursor or other intermediates which are then converted into functional PAs (28). Besides de novo synthesis pathways, PAs uptake can be controlled through specific transport systems. They are highly conserved among bacteria. The best-known examples are two ABC transporters described in Escherichia coli that are specific for either putrescine or spermidine and two antiporters, exchanging putrescine for ornithine and lysine for cadaverine (29).

Recent studies highlighted the involvement of PAs in bacterial pathogenesis. A clear example is Shigella spp. an intracellular pathogen associated to enteric syndrome in humans (30). For instance, due to mutations and deletions, cadaverine is lost from Shigella spp., improving the pathogenicity process, because cadaverine has a protective effect on intestinal mucosa from enterotoxins. Spermidine accumulation increases Shigella resistance to oxidative stress and its survival in macrophages [for a review see (28)].

### CONTRIBUTION OF GUT MICROBIOTA TO PAs FORMATION

Almost all foods contain PAs; they are abundant in soybeans, mushrooms, wheat germ, beef, pork, chicken livers, oranges, turban shell viscera, and green tea leaves. A great part of PAs introduced by foods is absorbed in the small intestine, whereas microbiota produce these compounds in great amounts in the large bowel (2). Little is known about the production and degradation of biogenic amines (BAs) by gut microbiota and in particular PAs. Recently, isolates from the human gut, belonging to many different species, were found to produce and degrade BAs at different levels depending on the strains (16). Putrescine and spermidine, important metabolites of intestinal bacteria, are present in the intestinal lumen in concentrations ranging from 0.5 to 1 mM in healthy humans (6). Gut microorganisms can synthesize putrescine, spermine and spermidine, that are present at millimolar concentrations, and play a major role in providing PAs for the high demand of these compounds in intestine. Bacteria use PAs for cell to cell communication, cellular signals and cell differentiation and the bacterial metabolism of these compounds determines the PAs intestinal content. The main studies were performed in E. coli, even if it is a minor microbial component in the human intestine and its PAs biosynthetic pathway seems to be different from those present in dominant gut microbiota (31). Few studies report data on metabolites produced by intestinal microbiota and their functions, in particular short fatty acids (32) and PAs (33, 34). In addition, Bacteroides spp. and Fusobacterium spp. can synthesize putrescine and spermidine in vitro and in vivo (35). Recently Nakamura et al. (20) found that in colonic lumen putrescine is produced by different bacteria from collective biosynthetic pathways depending on a complex exchange of metabolites.

The environmental stimuli can modulate the gut microbiota metabolism as well as the absorption and release of PAs. Noack et al. (36) reported that indigestible polysaccharides pass into the large intestine and are fermented with the production of shortchain fatty acids and lower pH, that can modify the intestinal microbiological metabolism and composition, and stimulate intestinal PAs content synthesis. In general, the fermentable carbohydrates present in the large bowel contribute to increase the bacterial PAs formation with consequent beneficial effects on the gut mucosa. In addition, by using in silico analysis, novel PAs biosynthetic and transport proteins have been found. There are few studies on PAs biosynthetic pathway carried out by dominant intestinal microorganisms. In fact, great part of gut bacteria utilizes carboxyspermidine dehydrogenase and carboxyspermidine decarboxylase (CASDC) for spermidine biosynthesis, whereas E. coli utilizes S-adenosylmethionine decarboxilase and spermidine synthase (8). The species in

the genus Bacteroides, which is predominant in intestine of humans with 20 of 56 most abundant species, harbors CASDC homologs (31) that are essential for spermidine biosynthesis contributing to the normal bacterial growth (37). Sugiyama et al. (8) evaluated the capacity of 32 bacterial species dominant in human gut to produce PAs in the cell and in supernatants, suggesting the presence of new genes and transporters. As many colonic microbial species do not possess complete synthetic pathways to produce PAs (38), it is possible to suppose the existence of metabolic interactions among bacterial species in the gut. Kitada et al. (39) showed that putrescine concentration produced by a mixed culture of different microbial species from gut microbiota was higher than that obtained with the single cultures. They demonstrated that a mixed culture of E. coli and Enterococcus faecalis produced the highest quantities of putrescine when the pH of the medium drops to neutral, suggesting the involvement of bacterial acid resistance system (40). A new pathway for putrescine formation was identified, from arginine to agmatine, with the cooperation of these two species. In presence of low pH, the acid resistance system of E. coli produces agmatine from arginine, and an arginine agmatine antiporter exchanges extracellular arginine for the intracellular end product of decarboxylation, agmatine (41). Enterococcus faecalis, through an agmatine/putrescine antiporter, metabolizes the agmatine to putrescine by agmatine deiminase pathway, yielding ATP, CO2, and NH<sup>3</sup> (42). The presence of other bacteria, such as Bifidobacterium spp. producing acid compounds in gut, favors the induction of this new pathway for putrescine production (39, 43, 44). In fact, many intestinal species do not possess a complete synthetic pathway for putrescine production from arginine (45) and therefore, it is possible to suppose the existence of a metabolic pathway spanning multiple bacterial species in the gut (8, 39). However, the knowledge about the contribution of gut microbiota to PAs formation is scarce and not sufficient.

### THE NEXT-GENERATION PROBIOTIC BACTERIA AND POLYAMINES

There is an enormous amount of research on probiotics and their beneficial impact on human health. Probiotics are defined by Boirivant and Strober (46) as "live microorganisms that, when administrated in adequate amounts, confer a health benefit on the host." The main sources of probiotics are gut or some fermented foods, such as kefir grains and yogurts, with Lactobacillus spp. and Bifidobacterium spp. being the most used microorganisms. Saccharomyces boulardii, Bacillus spp., E. coli, enterococci, and Weissella spp. are also included. With the development of new methodologies, a new era in probiotic research is started and the new probiotics are referred to next generation probiotics. In fact, there is an increasing interest in the use of gut commensal bacteria as potential probiotics, such as the genera Bacteroides, Clostridium, Bifidobacterium, and Faecalibacterium that predominate in the human gut microbiome (47). The mechanisms of probiotics activity are not clearly understood, even if many studies have been carried out (48). The potential biological effects of probiotics are characterized by an extremely diverse range, such as the new functional activities that are currently studied (49).

The species of colonic PAs-producing bacteria are several and different. The PAs concentration in the gut depends on the high or low presence of PAs-producing bacteria and also of PAs-absorbing bacteria. However, the presence of probiotics can increase the concentration of PAs in intestinal lumen as reported after the consumption of yogurt added with probiotics such as Bifidobacterium animalis subsp. lactis LKM512 (34). The consumption of yogurt containing the probiotic strain B. animalis subsp. lactis LKM512, increases the PAs concentration in human gut, favoring several positive effects for improving intestinal health, increasing lifespan and quality of life (33, 50, 51). PAs have been associated with cancer risk and represent a specific marker for neoplastic proliferation. The administration of probiotic Lactobacillus rhamnosus strain GG has been found to affect the synthesis of PAs in gut and the proliferation rates of gastric cell cancer. A relationship between PAs biosynthesis and probiotic action in carcinogenesis and cancer growth was found (52).

The consumption of probiotic strain B. animalis subsp. lactis LKM512, colonizes the colon and alters the intestinal microbiota, producing PAs. This alteration in intestinal microbiota favors some bacteria and suppress others, such as Enterobacteriaceae species, and Enterococcus spp. The produced PAs induce maintenance and/or recovery of intestinal barrier function and other beneficial activities such as longevity (53). The activation of PAs biosynthesis is performed by indigenous gut microbiota stimulated by environmental acidification induced by Bifidobacterium. In fact, these microorganisms do not possess enzymes involved in PAs biosynthesis (39).

A study carried out with a cocktail of probiotics, administered for 60 days, enhanced the PAs biosynthesis

#### REFERENCES


in canine inflamed colonic mucosa, regulating PAs levels (54). The administration of mixed probiotic cultures of Lactobacillus spp. strains has been described to induce positive health effects (55). Therefore, the positive effects have been proved in live and dead probiotic preparations [for a review see Adams (56)].

#### CONCLUSION

Microbiota-generated metabolites are an essential part of human physiology and are generated through microorganism– microorganism and host–microorganism interactions, with profound effects on human health and disease. Among the metabolites generated by bacteria in human gut PAs exhibit various beneficial effects, such as increased longevity, recovery of injured mucosa, and favorable effects on cognitive function. However, there is limited knowledge of how microorganisms interact with each other to synthesize metabolites in gut such as PAs. To obtain these tools it will be important to analyse the individual species and strains within these communities including uncultured microorganisms. Future researches on next-generation probiotics and/or mixed cultures of probiotic species should be investigated in order to better understand human health problems in the intestinal tract and find new strategies to face them. PAs modulation by gut microbiota and probiotic consortia could be a good strategy to achieve beneficial effects for human health.

#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.


multiple indigenous gut bacterial strategies. Sci. Adv. (2018) 4:eaat0062. doi: 10.1126/sciadv.aat0062


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Tofalo, Cocchi and Suzzi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Dietary and Gut Microbiota Polyamines in Obesity- and Age-Related Diseases

Bruno Ramos-Molina1,2 \* † , Maria Isabel Queipo-Ortuño2,3†, Ana Lambertos 4,5 , Francisco J. Tinahones 1,2 and Rafael Peñafiel 4,5 \*

#### Edited by:

Antonio F. Tiburcio, University of Barcelona, Spain

#### Reviewed by:

Chaim Kahana, Weizmann Institute of Science, Israel Jie Yin, Institute of Subtropical Agriculture (CAS), China Miguel Ángel Medina, Universidad de Málaga, Spain

#### \*Correspondence:

Bruno Ramos-Molina bruno.ramos@ibima.eu Rafael Peñafiel rapegar@um.es

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Nutrition and Food Science Technology, a section of the journal Frontiers in Nutrition

Received: 21 December 2018 Accepted: 20 February 2019 Published: 14 March 2019

#### Citation:

Ramos-Molina B, Queipo-Ortuño MI, Lambertos A, Tinahones FJ and Peñafiel R (2019) Dietary and Gut Microbiota Polyamines in Obesity- and Age-Related Diseases. Front. Nutr. 6:24. doi: 10.3389/fnut.2019.00024 <sup>1</sup> Department of Endocrinology and Nutrition, Virgen de la Victoria University Hospital, Institute of Biomedical Research of Malaga, University and Malaga, Malaga, Spain, <sup>2</sup> CIBER Physiopathology of Obesity and Nutrition (CIBERobn), Institute of Health Carlos III (ISCIII), Madrid, Spain, <sup>3</sup> Department of Medical Oncology, Virgen de la Victoria University Hospital, Institute of Biomedical Research of Malaga, University and Malaga, Malaga, Spain, <sup>4</sup> Department of Biochemistry and Molecular Biology B and Immunology, Faculty of Medicine, University of Murcia, Murcia, Spain, <sup>5</sup> Biomedical Research Institute of Murcia (IMIB), Murcia, Spain

The polyamines putrescine, spermidine, and spermine are widely distributed polycationic compounds essential for cellular functions. Intracellular polyamine pools are tightly regulated by a complex regulatory mechanism involving de novo biosynthesis, catabolism, and transport across the plasma membrane. In mammals, both the production of polyamines and their uptake from the extracellular space are controlled by a set of proteins named antizymes and antizyme inhibitors. Dysregulation of polyamine levels has been implicated in a variety of human pathologies, especially cancer. Additionally, decreases in the intracellular and circulating polyamine levels during aging have been reported. The differences in the polyamine content existing among tissues are mainly due to the endogenous polyamine metabolism. In addition, a part of the tissue polyamines has its origin in the diet or their production by the intestinal microbiome. Emerging evidence has suggested that exogenous polyamines (either orally administrated or synthetized by the gut microbiota) are able to induce longevity in mice, and that spermidine supplementation exerts cardioprotective effects in animal models. Furthermore, the administration of either spermidine or spermine has been shown to be effective for improving glucose homeostasis and insulin sensitivity and reducing adiposity and hepatic fat accumulation in diet-induced obesity mouse models. The exogenous addition of agmatine, a cationic molecule produced through arginine decarboxylation by bacteria and plants, also exerts significant effects on glucose metabolism in obese models, as well as cardioprotective effects. In this review, we will discuss some aspects of polyamine metabolism and transport, how diet can affect circulating and local polyamine levels, and how the modulation of either polyamine intake or polyamine production by gut microbiota can be used for potential therapeutic purposes.

Keywords: polyamines, diet, gut microbiota, metabolism, aging, obesity

### INTRODUCTION

Polyamines are small aliphatic amines that are present in all living organisms from bacteria to human beings. They can act as polycations since they are positively charged at physiological pH. The major polyamines in mammalian cells are spermidine, spermine, and their precursor the diamine putrescine (1, 2) (**Figure 1**). In addition of these molecules, microorganisms, and plants also synthesize other types of polyamines such as cadaverine, agmatine, and thermospermine (3–5) (**Figure 1**), and even minority polyamines (thermine, caldine, thermospermine, etc.) have been detected in extreme microorganisms (6). Agmatine and cadaverine are also present in very low amount in certain mammalian tissues (7, 8). Although polyamines were discovered in the seventeenth century, major advances in their metabolism and functions were achieved in the second part of the last century (2, 9–11). Numerous experiments have shown that due to their polycationic nature, polyamines can readily bind to negatively charged biomolecules including DNA, RNA, proteins, and phospholipids, modulating in many cases the function of these macromolecules (12, 13). When polyamine metabolism was pharmacologically or genetically altered, many relevant biochemical, and cellular processes resulted affected, including translation, transcription, signal transduction, cell proliferation, and differentiation, apoptosis, or cell stress response (14–19). However, the molecular mechanisms by which polyamines exert their biological effects are only partially understood (20).

In humans, only two genetic diseases affecting enzymes of the biosynthetic pathway of polyamines have been described so far (21, 22). However, several lines of evidence have implicated a dysregulation of the polyamine system in hyperproliferative and neurodegenerative diseases (23–27). The alteration of polyamine metabolism in several types of cancer has sustained the interest in using the polyamine pathway as a target for anticancer therapy (28, 29). In this regard, it is clear that the limitation of the cellular polyamine content has a potential interest in cancer chemoprevention or in the treatment of other pathologies such as certain infectious diseases and type 1 diabetes (28, 30, 31). On the other hand, experiments conducted in different cellular and animal models have revealed that the exogenous administration of spermidine or other natural polyamines may exert beneficial effects by affecting processes such as cellular stress, chronic inflammation, or dysregulated lipid or glucose metabolism (32– 34). In this review, we will discuss different aspects related with the homeostasis of polyamines in mammalian tissues, including the relevance of the polyamines generated by the gut microbiota, how the endogenous levels of polyamines can be affected by the administration of exogenous polyamines, and the impact that these treatments may exert on the evolution of the biochemical changes associated with aging or obesity-related metabolic diseases.

### POLYAMINE FUNCTIONS

Polyamines are essential for life. In fact, the inactivation of the genes that control the biosynthesis of putrescine or spermidine are embryo-lethal in mice (35, 36). This is not surprising due to the multiple essential cellular actions that have being ascribed to polyamines. Thus, owing to their general interaction with nucleic acids, they can affect many processes in which DNA, RNA, or proteins participate as substrates (13, 37, 38). Particularly interesting is the relationship between polyamines and reactive oxygen species (ROS). Although polyamine catabolism may generate potentially toxic products such as H2O<sup>2</sup> and polyamine derived aldehydes associated to pathophysiological consequences (39, 40), numerous in vitro and in vivo experiments have suggested that spermine and spermidine may act as scavengers of ROS, and then protecting DNA from oxidative damage (41–43). This double-edged role of polyamines appears to be dependent of certain factors (44). One of these factors in in vitro studies could be the use of animal serum in the cell culture medium, which contains amino oxidases that can oxidize exogenously administrated polyamines and generate ROS, resulting in cell toxicity independently of the action of the polyamine itself. Interestingly, a recent work demonstrated that in the presence of human serum, polyamine administration to the culture medium does not increase ROS production and does not affect cell viability as in the case of the same experiment in presence of either bovine or horse serum (45). Importantly, studies showing a polyamine-dependent cell toxicity in human cell lines in presence of significant amounts of bovine/horse serum should be reevaluated with human serum to corroborate that toxicity could be due to the production of oxidized polyamine-derived products by the action of serum polyamine oxidases and not to a toxic effect of the polyamines per se.

### POLYAMINE CONTENT AND ITS REGULATION

#### Cellular Polyamine Levels and Distribution

In general, spermidine and spermine are the most abundant polyamines in mammalian cells. In addition, their absolute values and the spermidine/spermine ratio depend on the type of cell and tissue (46, 47), and their tissue concentrations are affected by several factors including age (46–49). In mammalian cells, polyamines are present at relatively high concentration compared with other biogenic amines. Total polyamine concentration is at the mM range. However, it has been estimated that the free intracellular concentration of each polyamine is much lower (7–15% of total for spermidine, and 2–5% for spermine), due to the fact that a high percentage of total polyamines are bound by ionic interactions to nucleic acids, proteins, and other negatively charged molecules in the cell (12, 50). According to these data, only a small fraction of the cellular polyamines appears to be metabolically active. In addition, another unsettled aspect of polyamines is related to the subcellular distribution of these molecules inside mammalian cells. In rat liver mitochondria, spermidine, and spermine are present at estimated concentrations of 1.21 and 0.62 mM, respectively (51). Polyamines are also stored in vesicles of secretory cells and neurons (52–54). The intracellular levels of polyamines depend on several factors: polyamine biosynthesis, catabolism, uptake, and efflux. It is generally accepted that a high

proportion of the cellular polyamines have an endogenous origin, although in some cases the exogenous supply (i.e., dietary and gut microbiota-derived polyamines) may also significantly affect the polyamine content. Regarding circulating polyamines, several studies have shown that they are present at the low µM range in the blood of both humans and mice (55, 56). It is not clear whether blood polyamine levels are age-dependent (48).

### Polyamine Biosynthesis and Catabolism

In mammalian cells, the polyamine biosynthetic pathway consists in four steps catalyzed by enzymes mainly located in the cytosol (**Figure 2**). In this process, L-ornithine, and Sadenosyl methionine (AdoMet) are used as substrates. The diamine putrescine is synthesized by ornithine decarboxylase (ODC), a key enzyme that presents a complex regulation (57). This diamine can be converted into spermidine by the addition of an aminopropyl moiety donated by decarboxylated S-adenosyl methionine (dcAdoMet) through the action of spermidine synthase (5, 58). The tetra-amine spermine is synthesized by another aminopropyl transferase (spermine synthase) through the incorporation of another aminopropyl group to the amino butyl end of spermidine (5, 59). The second key-rate enzyme in the synthesis of spermidine and spermine is S-adenosyl methionine decarboxylase (AdoMetDC, AMD1), which produces dcAdoMet used as aminopropyl donor for the formation of both polyamines (60). Other biochemical route that participates in the regulation of the intracellular levels of polyamines is the so-called back-conversion pathway of polyamines (**Figure 2**). Spermidine/spermine N1 acetyltransferase (SSAT) is a cytosolic enzyme that plays a key role in reducing intracellular polyamine levels, through the acetylation of spermine and spermidine to produce in first instance acetyl-spermine or acetyl-spermidine (61, 62), which are either excreted from the cell or oxidized to spermidine or putrescine, respectively, by an acetyl polyamine oxidase (PAOX) (63). PAOX is a FAD-dependent enzyme located in peroxisomes that generates, apart from the corresponding polyamine, the potentially toxic byproducts 3-acetamidopropanal and H2O<sup>2</sup> (40). Spermine oxidase (SMOX), another FAD-containing enzyme, is capable of directly oxidizing spermine to spermidine, producing a 3-aminopropanal and H2O<sup>2</sup> (64, 65). Due to its localization outside of peroxisomes, the overexpression of the enzyme may enhance the oxidative damage by both increasing H2O<sup>2</sup> production and decreasing spermine levels (40, 66).

Agmatine (decarboxylated arginine) is an aminoguanidine structurally related to polyamines that behaves as a dication at physiological pH. Although agmatine is synthesized in bacteria and plants by arginine decarboxylase (ADC) (**Figure 2**), its biosynthesis in mammalian cells is controversial. In fact, the product of the initially reported ADC clone (67) was later demonstrated to be devoid of ADC activity by various independent groups. Instead, it encoded an ornithine decarboxylase antizyme inhibitor (AZIN2) (68, 69). Nowadays, multiple pharmacological effects of agmatine have been reported with potential therapeutic interest (70, 71), but to our knowledge clinical applications have not been yet implemented.

#### Polyamine Transport

Polyamines can enter and exit the cells through different type of carriers. Whereas, in bacteria and fungi, multiple polyamine transporters have been fully characterized (72), in mammalian cells lesser is known about the molecular components related with polyamine transport (73). It is generally accepted that in these cells the polyamine uptake activity is affected by the polyamine requirements of the cells. Thus, enhanced polyamine uptake is characteristic of cells with either high proliferating activity or in which cellular polyamine levels have been depleted by inhibitors of the biosynthetic pathway (74). Although the kinetics and some biochemical properties of polyamine transport have been studied in different types of cell lines and cell mutants, many questions about the "polyamine transport system" are still unsolved. It is possible that the polyamine carriers belong to the solute transport family (SLC) that contains about 400 annotated members. Different experiments of cell transfection by genes of the organic cation transporter (OCT) family or the cationic amino acid transporter (CAT) family have identified genes that could participate in the polyamine uptake system, but not a clear picture has emerged from all these studies (73, 75–78). On the other hand, polyamine export has also been detected in a variety of mammalian cells, having been identified SLC3A2

as a component of a diamine exporter that could participate in the excretion of putrescine and acetylated spermidine (79) (**Figure 2**). Apart from the polyamine transport system, endocytic pathways for the uptake of circulating polyamines have been described. In one model, polyamines bind electrostatically to the heparan sulfate chains of glypican-1, a proteoglycan of the cell surface, and then are taken into the cells through endocytosis (80). In another model, polyamines interact with non-defined polyamine binding protein(s) and the complex is internalized by caveolar endocytosis by a process that is negatively regulated by caveolin-1 (81).

### Regulation of Polyamine Levels by Antizymes and Antizyme Inhibitors

The control of the polyamine homeostasis in mammals is crucial for maintaining cellular functions, and dysregulation of the cellular polyamine levels has been associated to diverse pathological conditions (20). Antizymes (AZs) are regulatory proteins involved in the control of intracellular polyamine levels through the modulation of both polyamine biosynthesis and uptake. AZs affect the polyamine biosynthetic route by interacting with the ODC monomer and preventing the formation of active ODC homodimers, and by stimulating the degradation of ODC by the proteasome without ubiquitination (82). They also affect the import of extracellular polyamines by inhibiting the plasma membrane polyamine transport system by a still unknown mechanism (83).

The negative effects that AZs exert on both the biosynthesis of intracellular polyamines and the polyamine uptake can be abrogated by the action of antizyme inhibitors (AZINs), proteins homologous to ODC but devoid of enzymatic activity. AZINs are able to bind AZs even more efficiently than ODC, releasing ODC from the inhibitory ODC-AZ complex (84, 85). AZINs are also able to enhance the uptake of extracellular polyamines, likely by negating the inhibitory action of AZs on the polyamine transport system (86). In addition, AZs and AZINs can also modulate the uptake of agmatine by mammalian cells (87).

### ABSORPTION OF POLYAMINES IN THE GASTROINTESTINAL TRACT

The polyamines present in the lumen of the gastrointestinal tract may have different origin: food, intestinal microbiota, pancreatic-biliary secretions, and intestinal death cells. Although a precise quantitation of the contribution of each process to the whole polyamine pool is non-existent, it is believed that dietary polyamines is the major source of luminal polyamines in humans and animals (88, 89). Polyamine levels have been analyzed in hundreds of food items by different groups (49, 88, 90–93). In general, fruits and cheese are rich in putrescine, whereas vegetables and meat products contain high levels of spermidine and spermine, respectively (94). Despite the high variability of the polyamine content in foods, it has been estimated that a standard human diet provides hundreds of micromoles of polyamines per day (88), with small differences between the major polyamines when it is calculated from diets from different countries (95).

The polyamine levels in the intestinal lumen change over time after a meal, from millimolar levels immediately after the ingestion, to much lower values in the fasting period (96). Luminal polyamines can be taken up mainly by the small intestine. Experiments using radiolabeled polyamines administered intragastrically to rats showed that polyamines can be absorbed and distributed to different tissues (88). This distribution was not uniform, since polyamines were accumulated preferentially in those tissues stimulated to proliferate (88). The first barrier for the uptake of polyamines from the intestinal lumen is formed by the epithelial cells of the intestinal mucosa. In isolated enterocytes, as in other mammalian cells in culture, polyamines can be taken up by different specific polyamine transporters, as already commented in the polyamine transport section. The in vivo polyamine uptake by the intestinal cells is more complex due to the existence of different polyamine transporters in the apical and basolateral membranes, as shown by studies using brush-border and basolateral membrane vesicles of the enterocyte (97). According with experimental data, luminal polyamines could be taken by enterocytes by transport across the apical membrane and extruded across the basolateral membrane by low affinity transporters to the systemic circulation (96). It was also hypothesized that the majority of luminal polyamines could be passively absorbed via the paracellular route (96). Whereas, most of spermidine and spermine taken up by the intestinal cells are not metabolized in these cells, a variable proportion of putrescine is transformed into other compounds including spermidine, γ-aminobutyric acid (GABA) and succinate (88, 98). In the small intestine of rats, putrescine can be transformed into succinate acting as a source of instant energy (99). The absorption of polyamines appears to be rapid, since experiments using an ex vivo rat model revealed that values about 70% of the <sup>14</sup>C-polyamines administered to the jejunal lumen were found in the portal vein, after 10 min of polyamine administration (100). Most of the studies on luminal polyamine uptake and their distribution through the body have been based on the acute administration of a low dose of labeled polyamines to rats. Recently, as described in other section, many studies have reported beneficial effects of a prolonged oral administration of either spermidine or spermine to rodents (101–104). However, in most studies tissue polyamine levels were not reported. In mouse models, prolonged administration of polyamine-rich diets have been seen to increase blood levels of spermidine and/or spermine (56, 105, 106). In aged mice spermidine levels significantly increased in blood (107) and liver (101) after supplementation of the drinking water with 3 mM spermidine for 6 months. In line with this, a 28-day oral supplementation of adult mice with 50 mg/kg of spermidine resulted in a significant increase of spermidine in whole blood and heart (but not in brain) of females, but not in males (106). In humans it has been shown that a prolonged intake (2 months) of polyamine-rich products such as natto (fermented soy) produces a significant rise in the levels of spermine (but not spermidine) in blood (56). More recently, the results of a clinical trial using spermidine supplements in older human subjects have been reported, showing no differences in blood polyamine levels between controls and spermidine-supplemented individuals at 3 months of follow-up (106)s.

A part of the absorbed luminal polyamines remains in the intestinal cells. This is not surprising since the intestinal tissue is one of the most rapidly proliferating tissues, and in general the polyamine uptake of intestinal cells is associated with the proliferative stage. Dietary polyamines might be important for the development of the digestive tract, and also for the maintenance of the adult digestive tract (108). Moreover, polyamine reservoirs are not only responsible for maintaining the rapid turnover and high proliferation rates of the intestinal epithelial cells but also for enhancing the integrity of the intestinal barrier. Thus, polyamines are able to stimulate the production of intercellular junction proteins, such as occludin, zonula occludens 1, and E-cadherin, which are essentials to regulate the paracellular permeability and reinforcing epithelial barrier function (109).

Furthermore, intestinal polyamine pools are necessary for the postnatal development of the gastrointestinal tract. These data were confirmed in a study using pup rats, where the polyamine administration was able to induce the production of mucus and secretory IgA in the small intestine, while rats fed with a polyamine-deficient diet developed intestinal mucosal hypoplasia (110). In addition, the oral administration of polyamines to suckling rats accelerated the maturation process affecting intestine, liver and pancreas. Interestingly, the precocious maturation of the intestine only was achieved when spermine was given orally and not when other routes of administration were used, suggesting that this process requires the interaction of the polyamine with the luminal side of the mucosa (111). On the other hand, although it has been postulated that the polyamines present in the human milk, could contribute to prevent food allergic processes in neonates, more studies appear to be necessary (112).

As commented above, antizymes and antizyme inhibitors play a relevant role in cellular polyamine uptake. However, the influence of these proteins on intestinal polyamine absorption is uncertain. Both ODC and AZ1 mRNA transcripts are abundantly expressed in small intestine (113). Treatment of the intestinal epithelial cell line IEC-6 with 10µM of spermidine or spermine induced the synthesis of the antizyme protein negatively affecting polyamine uptake (114). In this cell line AZ1 is induced in the absence of amino acids, through a mechanism involving mTORC1 (115). Other studies using a human colon adenocarcinoma cell line (Caco-2) revealed that amino acid supplementation also modulates antizyme induction

(116), whereas amino acid restriction increased spermidine uptake by Caco-2 cells, by an antizyme-independent mechanism (117). The levels of AZ mRNA have been analyzed in cells isolated from jejunal crypt-villus axis. Whereas, AZ mRNA levels were high in cells from the small intestinal crypts, the message fell to near undetectable levels in cells of the villus tip (118). However, since the expression of AZ protein was not analyzed, and it is well-known that the translation of the AZ mRNA is stimulated by polyamines, no firm conclusion about polyamine uptake though the crypt-villus axis could be established. Another uncertain question is how antizyme inhibitors may affect gut polyamine absorption. In comparison to ODC and AZ1, mRNA expression levels of AZ2, AZIN1, and AZIN2 in mouse intestine are low (113). Whereas, in normal cells AZIN1 stimulates both polyamine biosynthesis and uptake by binding to AZs, an edited form of AZIN1 with higher affinity to AZs has aroused more attention because of its contribution to carcinogenesis (119). In this regard, AZIN1 RNA editing levels were found significantly elevated in colorectal cancer tissues when compared with corresponding normal mucosa (120). It is conceivable that tissues with enhanced AZIN1 activity could accumulate higher levels of polyamines than normal ones, due to the loss of the feedback effect of polyamines by the blockade of AZ function.

#### POLYAMINES AND GUT MICROBIOTA

The quantity and diversity of microbial species in the gut increase longitudinally from the stomach to the colon, being the colon where the most dense and metabolically active community exist (121). The gut microbiome produces a diverse metabolite repertoire that may harm or benefit the host. These gut microbial metabolites include short-chain fatty acids, polyphenols, vitamins, tryptophan catabolites, and polyamines (122, 123). A recent study has provided data on polyamine biosynthesis and transport in the dominant human gut bacteria (124). As described above the high levels of polyamines present in the intestinal tract may be originated from the diet or produced de novo by host cells and intestinal bacteria (125) (**Figure 3**). Nevertheless, it is thought that the greatest amounts of the polyamines present in the lower parts of the intestinal tract could be mostly synthesized by intestinal microbiota. A metabolomics study in mice revealed that the intestinal luminal levels of putrescine and spermidine, but not of spermine, are mainly dependent on colonic microbiota (122). In addition, experiments in rats using different dietary interventions demonstrated that the oral administration of fermentable fibers and fructans stimulates certain intestinal microbial species (i.e., bacteroides and fusobacteria) to synthetize large amounts of polyamines in the large intestine (126–128). It is believed that polyamines synthesized by the colonic microflora are transported to the proximal gut via the portal circulation and biliary tree (129), although the contribution of colonic microflora to circulating levels of polyamines in the host is presently unknown. It is likely that part of the polyamines produced by the colonic microbiota is excreted in the feces. In this regard, it was reported that human fecal polyamine concentration may be linked to fecal microbiota (130).

Unlike the host polyamine metabolism, gut microorganisms can produce polyamines using constitutive or inducible forms of amino acid decarboxylase enzymes. Ornithine, lysine, and arginine decarboxylase activities have been detected in bacteria and archea (5). In addition, spermidine can be produced from putrescine and dcAdoMet by spermidine synthase, although some microorganisms can synthesize this polyamine from putrescine using L-aspartate β-semialdehyde as aminopropyl donor (5). This alternative biosynthetic route of spermidine is dominant in human gut microbiota (131). Arginine decarboxylation is also the dominant pathway for diamine biosynthesis in the most abundant species of the human intestinal microbiome (132). Interestingly, oral administration of arginine to mice after treatment with antibiotics did not result in increased putrescine levels in the intestinal tract, indicating that putrescine could be predominantly produced by intestinal bacteria (133). In addition, it has been demonstrated by using stable isotope labeled arginine that colonic luminal putrescine was produced by intestinal bacteria from arginine metabolism in a dose-dependent manner (133). On the other hand, the arginine deiminase pathway (an anaerobic route for arginine degradation) is common in gut commensal bacteria such as Enterococcus, Streptococcus, Clostridium, Lactococcus, and Lactobacillus species. However, whereas some of these species cannot produce putrescine because of the lack of essential polyamine biosynthetic enzymes, they acquire it from other bacterial species through a collective biosynthetic pathway catalyzed by enzymes derived from multiple bacterial species and from several extracellular intermediates (134).

The metabolism of polyamines has a central role in the regulation of systemic and mucosal adaptive immunity. Arginine is an important modulator of the immunometabolism of macrophages and T cells able to affect their effector functions (125). Thus, the metabolism of dietary arginine by the gut microbiome might have effects on the immune system (135). In this regard, mice colonized with specific microbial communities significantly contain lower amounts of arginine in their intestines compared to germ-free mice, which has been attributed to the microbiome capacity of metabolizing important amounts of arginine from the diet in the gut and, therefore, regulating the availability of arginine for the immune system (136). Furthermore, polyamines inhibit the production of inflammatory cytokines and have antioxidant effects. In fact, exogenous polyamines can decrease the release cytokines that promotes epithelial repair and barrier function. Similarly, both spermine and histamine inhibit the activation of the NLRP6 inflammasome, which is a protein complex expressed by epithelial cells that can regulate IL-18 secretion (137, 138). On the other hand, a previous study using healthy mice, described that the presence of the probiotic strain Bifidobacterium animalis subsp. lactis LKM512 can induce resistance to oxidative stress and promote longevity, which was dependent on enhanced microbial polyamine synthesis (139).

Dysregulated polyamine metabolism has been suggested to be involved in the tumorigenesis of colorectal cancer (CRC)

and other tumors (29). A metabolomics screen comparing paired colon cancer and normal tissue samples from patients with CRC revealed that bacteria biofilm formation, even in the normal colon tissue, was associated with increased colonic epithelial cell proliferation and host-enhanced polyamine metabolism (140). In addition, bacteria-generated polyamines in biofilms may contribute to the inflammation and proliferation of colon cancer (141). Following antibiotic treatment, resected colorectal cancer tissues harbored disrupted bacterial biofilms and lowered N<sup>1</sup> ,N12-diacetylspermine tissue concentrations compare to biofilm-negative colon cancer tissues, suggesting that gut microbes can induce an increase of host generated N<sup>1</sup> ,N12-diacetylspermine (141). In tumor-bearing mice, the administration of antibiotics that reduces the microbial flora of the gastrointestinal tract enhanced the cytostatic effect of difluoromethyl ornithine (DFMO), a potent inhibitor of endogenous polyamine synthesis (142, 143), suggesting that reduction of bacterial polyamine synthesis together with the inhibition of the polyamine biosynthesis route should be considered as an effective antitumoral strategy.

### POLYAMINES IN OBESITY AND RELATED METABOLIC DISORDERS

In several transgenic mouse models, the dysregulation of the polyamine metabolism has been shown to have an impact in the regulation of the glucose, lipid, and energy homeostasis (144– 149). In fact, emerging evidence suggests that increased levels of polyamines in white adipose tissue, liver or skeletal muscle could stimulates energy expenditure and confer resistance to diet-induced obesity and non-alcoholic fatty liver disease (147, 148). In addition, a number of studies using different obesity animal models have reported altered levels of polyamines in adipose tissue (150), liver (151, 152), pancreatic islets (153), and urine (154). In addition, blood polyamines have been shown to be significantly higher in obese children compared to non-obese controls (155). Remarkably, polyamine metabolism has been involved in adipogenesis (156–158), suggesting that increased polyamine levels may be implicated in adipose tissue expandability during obesity. Particularly, in 3T3-L1 preadipocytes both spermidine and spermine are essential factors at early stages of adipocyte differentiation as they modulate the expression levels of transcriptional factors implicated in the regulation of adipogenesis (18, 157, 159).

In diet-induced obesity mouse models, a high-dose daily administration of either spermidine or spermine has been shown to be an effective strategy for weight loss and improvement of the glycemic status (103, 160, 161). For instance, spermidine supplementation resulted in a significant decrease in body weight, increased glucose tolerance and insulin sensitivity and ameliorated hepatic steatosis in high-fat diet-induced obese mice (161). Furthermore, treatment with exogenous spermine has been shown to be effective to decrease body weight and fasting glucose and to improve glucose tolerance in diet-induced obese mice (160). In cultured rat adipocytes, spermine has been shown to enhance glucose transport and the conversion of glucose into triacylglycerols (162), and to increase the ability of insulin to bind to its own receptor (163). Finally, spermidine administration was able to prevent lipid accumulation and necrotic core formation in vascular smooth muscle cells through the induction of cholesterol efflux in an experimental model of atherosclerosis (164).

Despite the fact that several studies have indicated that polyamine metabolism could be dysregulated in obesity, the role of polyamines in type 2 diabetes (T2D), which is frequently associated with obesity, has been less explored. In pancreatic islets, polyamines are mainly located in the secretory granules of the β cells (52), where they have been implicated in proinsulin biosynthesis and insulin secretion (165). Islet polyamine levels diminished with age and obesity (153), suggesting that alterations in the intracellular levels of polyamines might affect β cell function. Remarkably, transgenic mice overexpressing the polyamine catabolic enzyme spermidine/spermine acetyltransferase (SSAT), which contain higher levels of putrescine and reduced levels of spermidine and spermine in the pancreatic islets, display impaired glucosestimulated insulin secretion (146), suggesting that disturbed levels of polyamines may be involved in the physiopathology of T2D. In this regard, it has been recently reported an association between the serum levels of polyamines and T2D in a cohort of patients with metabolic syndrome (166). In particular, it was found that serum putrescine levels were significantly elevated in diabetic subjects and that correlated with glycosylated hemoglobin, and that serum spermine was closely associated with insulin levels (166). In animal models of diabetes, the polyamine biosynthetic enzyme ODC has been described to be upregulated in rat diabetic kidneys, resulting in increased levels of putrescine but not other polyamines (167).

On the other hand, exogenous administration of spermidine in rats with pharmacologically induced diabetes results in an improvement of glycaemia and a concomitant reduction of glycosylated HbA1c levels (168). Furthermore, spermine administration has been shown to be effective to protect β cells against the diabetogenic effect of alloxan (169). In diabetic rats, spermine administration did not affect hyperglycemia but improved the lipid profile and lead to a reduction in the formation of advanced glycation end-products (170). Finally, spermidine could improve endothelial dysfunction in diabetes by restoring nitric oxide production in an autophagy-dependent fashion (171).

Like polyamines, agmatine has significant metabolic effects when exogenously administered in animal models. For instance, the chronic administration of agmatine resulted in a reduction body weight gain and fat depots in rats fed a high-fat diet (172). Remarkably, this study further demonstrated that the effects of long-term agmatine consumption on body weight and adiposity could be due to the regulation of fat oxidation and gluconeogenesis (172). Additionally, recent studies have reported that exogenous agmatine administration significantly reduces atherosclerotic lesions in ApoE knockout mice (173), although the mechanism by which agmatine exerts these antiatherosclerotic effect remains to be elucidated.

Numerous works have demonstrated that both oxidative stress and chronic inflammation are factors attributed to obesity that can contribute to the development of metabolic syndrome. Polyamines can act of modulators of oxidative stress and inflammation (174). However, the possible impact polyamines in these processes in the context of obesity have not been explored yet. Gut microbiota is another well-known factor involved in the pathogenesis of obesity and T2D. In fact, recent insight suggests that an altered composition and diversity of gut microbiota could play an important role in the development of metabolic disorders. Gut microbiota dysbiosis is characterized by metabolic endotoxemia, abnormal incretin secretion and production of bacterial metabolites that can influence energy and glucose metabolism (175). As described above polyamines can be produced by several bacterial strains in the gut, although the potential therapeutic use of polyamines derived from the gut microbiota in the treatment of metabolic disorders remains to be investigated.

### DIETARY POLYAMINES IN AGING AND CANCER

Several studies have shown that the exogenous addition of polyamines, especially spermidine, increases the lifespan in yeast, nematodes, and fruit flies (101, 176). In rodents, a progressive decrease in the tissue polyamine content has been observed with age (49, 177), and the chronic administration of spermidine extends lifespan (107, 178, 179). Likewise, in animal models, supplementation with polyamines could have a beneficial effect on several age-related disorders including memory decline, neuroinflammation, and cardiovascular disease (107, 180–182). Several in vitro studies have shown that the administration of spermidine in cell lines of neuronal origin inhibits the process of cellular senescence (183, 184). It has been proposed that the mechanism by which polyamines produces these beneficial anti-aging effects could be through the activation of autophagy (34, 101), either through the inhibition of the activity of the acetyltransferase EP300, which is involved in the modulation of autophagic flux (185–187), or the stabilization of pro-autophagic factors such as MAP1S (179).

In addition to the exogenous supply of synthetic polyamines, several studies have shown that an increase in the concentration of gut polyamines through the administration of probiotics inhibits cellular senescence and stimulates longevity in mice (133, 139). In humans, recent epidemiological studies have associated a higher consumption of dietary spermidine with a decrease in cardiovascular events and lower mortality (107, 188). Remarkably, many food products included in the Mediterranean diet, which is known to be associated with increased longevity (189), are rich in polyamines (94, 190). However, whether a greater adherence to a Mediterranean dietary pattern increases circulating polyamine levels remains unknown.

On the other hand, in parallel with the progressive reduction of tissue polyamine levels, many works have reported alterations in the DNA methylation status associated with age (191, 192). Noteworthy, the polyamine pathway is connected to the methionine cycle through the action of AdoMetDC, which is able to decarboxylate AdoMet. In turn, AdoMet is the major donor of methyl groups for nucleic acids or other macromolecules. Thus, it has been proposed that lower spermidine and spermine levels can lead to reduced DNA methylation levels by inducing the accumulation of dcAdoMet, which it could be able to inhibit DNA methyltransferase (DNMT) activity (193). In accordance, in vitro experiments consisting in the pharmacological or genetic depletion of polyamine pools in human cell lines resulted in changes in DNA methylation (105, 194, 195). Moreover, abnormalities in the global DNA methylation status associated with aging can be reversed by exogenous polyamine supply (105), supporting the idea that the reduction of tissue polyamines during aging is closely related to alterations in DNA methylation. Remarkably, an abnormal DNA methylation profile has been related not only with aging, but also with other pathologies including cancer and metabolic disorders (196, 197). Interestingly, adherence to a Mediterranean dietary pattern has been described to be associated with changes in the DNA methylation status (198, 199). However, whether these effects are mediated by polyamines or other metabolites related to one carbon metabolism remain to be elucidated.

Despite the fact that emerging evidence is proposing polyamines as therapeutic option for aging-related human events including cardiovascular disease or memory decline, the pro-carcinogenic potential of polyamines should be carefully considered when polyamine-rich diets are therapeutically recommended. It is widely accepted that polyamine production is abnormally stimulated in different types of human tumors by induction of the polyamine biosynthetic pathway, which could be beneficial for tumor progression (200). Many efforts have been directed to limit the polyamine pools in the tumor by administrating chemical inhibitors of the polyamine biosynthetic enzyme ODC (i.e., DFMO), although this therapeutic option frequently has failed due to the capacity of tumor cells of replacing endogenously synthetized polyamines by extracellular polyamines from the circulation. Hence, restriction of dietary polyamines has been proposed to enhance the effects of the inhibitors of the polyamine synthesis. In CRC patients with advanced adenoma it has been proposed that the limitation of dietary polyamines could be an adjunctive strategy to avoid recurrence after 3 years of treatment using polyamine-inhibitory drugs (201). By contrast, it has been suggested that dietary polyamines might decrease the risk of CRC in postmenopausal women (202). This is supported by experiments in mice that demonstrated that polyamine intake could be an effective strategy to prevent colon tumors after 1,2-demethylhydrazine administration (105). In rats, whereas the treatment with a low spermidine diet can promote the induction of mammary tumors, a high spermidine diet may suppress tumorigenesis (203). Additionally, it has been recently reported that polyamine accumulation could have toxic effects for renal cell carcinoma and that the tumor cells avoid polyamine toxicity by inducing arginase 2 (204). Therefore, although polyamines have been classically considered as oncometabolites by its procarcinogenic actions once the tumor is formed, these new studies could indicate that dietary polyamines could be considered for cancer prevention or even for cancer treatment depending on the cancer type in a dose-dependent fashion. However, because of the insufficient number of studies performed in vivo with exogenously administrated polyamines in mouse cancer models, caution is necessary before considering the ingestion of polyamines as an anticancer strategy in humans.

#### CONCLUDING REMARKS

Although it is well-known that the polyamine content in mammalian tissues is tightly controlled through a complex regulatory network involving the modulation of the biosynthesis, catabolism, and transport pathways, the tissue and circulating levels of polyamines are altered in a number of pathological conditions. Whereas, numerous experiments have demonstrated that inhibitors of the polyamine biosynthetic pathway drastically decrease the tissue levels of putrescine and spermidine, less is known about the effect of either the restriction or the administration of exogenous polyamines on tissue and circulating polyamine levels. Interestingly, recent studies have revealed that oral spermidine supplementation extended the lifespan and has cardioprotective effects in different animal models, but the detailed molecular mechanisms by which spermidine exert these effects are only partially understood. In humans, nutritional studies have described a relationship between spermidine-rich foods and lower cardiovascular events and reduced mortality, but in both cases, no evidence about the tissue or circulating levels of spermidine has been reported. More recently, chronic interventions of elderly subjects with controlled amounts of spermidine have revealed increased levels of this polyamine in blood after several months of treatment. Other studies have demonstrated that prolonged consumption of natto, a traditional Japanese fermented food rich in polyamines, significantly increases spermine, but not spermidine in blood of healthy young volunteers. Because of the beneficial effects of polyamine intake recently reported, it would be interesting to know how the dietary polyamine supplementation might affect polyamine levels in different organs and tissues and why this variation induces all the reported effects.

On the other hand, emerging body of evidence has revealed that in diet-induced obesity models the daily administration of high doses of polyamines (either spermidine or spermine) can reduce body weight and adiposity and can positively regulate glucose homeostasis, suggesting that exogenous polyamines can impact peripheral organs with relevant implications in lipid and glucose metabolism such as adipose tissue and liver (**Figure 3**). In pharmacological models of T2D, polyamines could also exert protective effects in pancreatic β cells. These effects, however, have not been proven after the oral intake of low concentrations of polyamines with demonstrated effects on longevity or heart protection. Therefore, the impact of a chronic low-dose oral polyamine intake should be performed in models of obesity or T2D before considering polyamines as therapeutic option for these metabolic disorders.

#### REFERENCES


### AUTHOR CONTRIBUTIONS

All authors: conceptualization, writing review, and editing; RP, BR-M, and MQ-O: writing original draft; RP and BR-M funding.

#### FUNDING

This study was supported by research grants from the Andalusian Health Public System (grant number PI-0096-2017) and by the Spanish Ministry of Economy and Competitiveness, SAF2011- 29051 (with European Community FEDER support) and by Seneca Foundation (Autonomous Community of Murcia), 19875/GERM/15. MQ-O was supported by the Miguel Servet Type II program (CPII18/00003) from the Institute of Health Carlos III (ISCIII), and co-financed by the European Regional Development Fund (ERDF), and by the Andalusian Health Public System (C-0030-2018). BR-M was recipient of a Sara Borrell postdoctoral fellowship from the ISCIII (CD16/0003), cofinanced by the European Regional Development Fund (ERDF).

requirement for translation initiation and elongation. J Biol Chem. (2010) 285:12474–81. doi: 10.1074/jbc.M110.106419


growth of Campylobacter jejuni and is the dominant polyamine pathway in human gut microbiota. J Biol Chem. (2011) 286:43301–12. doi: 10.1074/jbc.M111.307835


methylation in peripheral blood. Am J Clin Nutr. (2018) 108:611–21. doi: 10.1093/ajcn/nqy119


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer MA declared a shared affiliation, with no collaboration, with several of the authors, BR-M, MQ-O, FT, to the handling editor at time of review.

Copyright © 2019 Ramos-Molina, Queipo-Ortuño, Lambertos, Tinahones and Peñafiel. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Metabolic Characterization of Hyoscyamus niger Ornithine Decarboxylase

Tengfei Zhao<sup>1</sup> , Changjian Wang<sup>1</sup> , Feng Bai<sup>1</sup> , Siqi Li<sup>1</sup> , Chunxian Yang<sup>1</sup> , Fangyuan Zhang<sup>1</sup> , Ge Bai<sup>2</sup> , Min Chen<sup>3</sup> , Xiaozhong Lan<sup>4</sup> and Zhihua Liao<sup>1</sup> \*

<sup>1</sup> Key Laboratory of Eco-Environments in Three Gorges Reservoir Region (Ministry of Education), Chongqing Engineering Research Centre for Sweet Potato, TAAHC-SWU Medicinal Plant Joint R&D Centre, School of Life Sciences, Southwest University, Chongqing, China, <sup>2</sup> Tobacco Breeding and Biotechnology Research Center, Yunnan Academy of Tobacco Agricultural Sciences, Key Laboratory of Tobacco Biotechnological Breeding, National Tobacco Genetic Engineering Research Center, Kunming, China, <sup>3</sup> College of Pharmaceutical Sciences, Key Laboratory of Luminescent and Real-Time Analytical Chemistry (Ministry of Education), Southwest University, Chongqing, China, <sup>4</sup> TAAHC-SWU Medicinal Plant Joint R&D Centre, Xizang Agricultural and Husbandry College, Nyingchi of Tibet, China

#### Edited by:

Antonio F. Tiburcio, University of Barcelona, Spain

#### Reviewed by:

Francisco Marco, University of Valencia, Spain Subhash C. Minocha, University of New Hampshire, United States Ana Margarida Fortes, Universidade de Lisboa, Portugal

\*Correspondence:

Zhihua Liao zhliao@swu.edu.cn; zhihualiao@163.com

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 07 September 2018 Accepted: 11 February 2019 Published: 27 February 2019

#### Citation:

Zhao T, Wang C, Bai F, Li S, Yang C, Zhang F, Bai G, Chen M, Lan X and Liao Z (2019) Metabolic Characterization of Hyoscyamus niger Ornithine Decarboxylase. Front. Plant Sci. 10:229. doi: 10.3389/fpls.2019.00229 Ornithine decarboxylase (ODC) catalyzes ornithine decarboxylation to yield putrescine, a key precursor of polyamines, and tropane alkaloids (TAs). Here, to investigate in depth the role of ODC in polyamine/TA biosynthesis and to provide a candidate gene for engineering polyamine/TA production, the ODC gene (HnODC) was characterized from Hyoscyamus niger, a TA-producing plant. Our phylogenetic analysis revealed that HnODC was clustered with ODC enzymes of plants. Experimental work showed HnODC highly expressed in H. niger roots and induced by methyl jasmonate (MeJA). In the MeJA treatment, the production of both putrescine and N-methylputrescine were markedly promoted in roots, while contents of putrescine, spermidine, and spermine were all significantly increased in leaves. By contrast, MeJA did not significantly change the production of either hyoscyamine or scopolamine in H. niger plants. Building on these results, the 50-kDa His-tagged HnODC proteins were purified for enzymatic assays. When ornithine was fed to HnODC, the putrescine product was detected by HPLC, indicating HnODC catalyzed ornithine to form putrescine. Finally, we also investigated the enzymatic kinetics of HnODC. Its Km, Vmax, and Kcat values for ornithine were respectively 2.62 ± 0.11 mM, 1.87 ± 0.023 nmol min−<sup>1</sup> µg <sup>−</sup><sup>1</sup> and 1.57 ± 0.015 s−<sup>1</sup> , at pH 8.0 and at 30◦C. The HnODC enzyme displays a much higher catalytic efficiency than most reported plant ODCs, suggesting it may be an ideal candidate gene for engineering polyamine/TA biosynthesis.

Keywords: biosynthesis, Hyoscyamus niger, ornithine decarboxylase, polyamine, tropane alkaloids

## INTRODUCTION

Polyamines, including putrescine, spermidine, and spermine, are involved in many important biological processes of plants, such as their growth, development, and adaption to biotic and abiotic stresses (Kusano et al., 2008; Kusano and Suzuki, 2015; Aloisi et al., 2016). Moreover, putrescine is essential for the synthesis of polyamines and putrescine-derived alkaloids (Jirschitzka et al., 2012), because forming putrescine is the first step in the polyamines biosynthetic pathway (**Figure 1**), providing a ley precursor for spermine and spermidine (Kusano and Suzuki, 2015). Putrescine can become methylated to form N-methylputrescine, a key intermediate compound of nicotine and

**95**

pharmaceutical tropane alkaloids (TAs) (Biastoff et al., 2009). Among medicinal plants belonging to the Solanaceae family, such as Hyoscyamus niger, Atropa belladonna, and Datura species, all produce pharmaceutical TAs, including hyoscyamine and scopolamine which are widely used as anticholinergic reagents.

In many plants, putrescine is synthesized directly from ornithine by ornithine decarboxylase (ODC), or indirectly from arginine by arginine decarboxylase (ADC) (**Figure 1**). That putrescine is ubiquitous in plants is perhaps not surprising, given its crucial functioning in plant metabolism (Tiburcio et al., 1997). Tissue localization of ornithine/arginine decarboxylases suggests that ODC could be the main enzyme responsible for the synthesis of putrescine in plant roots (Wang et al., 2000; Delis et al., 2005). For solanaceous plants reported on to date, all their precursors of TAs are synthesized in roots and then transferred aboveground, to the plants' aerial parts (Kanegae et al., 1994; Suzuki et al., 1999; Bedewitz et al., 2014). Hence, we may speculate that ODC rather than ADC participates in the biosynthesis of putrescinederived alkaloids. To resolve this issue clearly requires further experimental investigation.

Ornithine decarboxylase is a rate-limiting enzyme in the biosynthesis of ornithine-derived metabolites (Bunsupa et al., 2016). More specifically, it tightly regulates putrescine production that dynamically affects the biosynthesis of polyamines and TAs (**Figure 1**). Because of its involvement in polyamine biosynthesis, the function of ODC has been well-studied and understood in animals and bacteria. With respect to plants, however, most studies of plant ODC, focused on its regulation on polyamine biosynthesis under stressful conditions (Akiyama and Jin, 2007; Pál et al., 2015; Krasuska et al., 2017). By tracing labeled ornithine, earlier work indicated that ornithine was used for TA biosynthesis (Hashimoto et al., 1989; Nyman, 1994), yet TA production was interrupted when Hyoscyamus albus plants were treated with difluoromethylornithine (DFMO), a specific inhibitor of ODC (Nyman, 1994). Although ODC undoubtedly participates in TA biosynthesis, its exact role in this process, especially at molecular and biochemical levels, remains largely unknown.

To our best knowledge, in plants, only the ODC proteins of Nicotiana glutinosa (NgODC) and Erythroxylum coca (EcODC) have been characterized for their enzymatic kinetics, by using purified proteins. The NgODC enzyme is associated with the biosynthesis of nicotine and has a very low catalytic activity (Lee and Cho, 2001); in contrast, the purified recombinant EcODC exhibits much higher catalytic efficiency than does NgODC (**Table 1**) (Docimo et al., 2012). Biochemical characterization of ODC proteins from tobacco and coca tree has fostered a richer understanding of their contribution to the regulation of nicotine and cocaine production, respectively. In solanaceous plants that produce TAs, Datura stramonium was the only species with its ODC gene (DsODC) cloned and characterized. In that work, DsODC was highly expressed D. stramonium roots and crude protein extracts from E. coli expressing DsODC demonstrated the ODC activity, but without a characterization of its kinetics (Michael et al., 1996). Therefore, it is valuable to further study the ODC roles in TA and polyamine biosynthesis in TA-producing plants.

As a plant species well known for producing TAs, especially scopolamine, Hyoscyamus niger is also widely used for studying their biosynthesis. To date, several TA biosynthesis enzymes have been robustly characterized from H. niger as well as other plants species (**Figure 1**). These enzymes include putrescine N-methyltransferase (Liu et al., 2005; Kai et al., 2009b; Geng et al., 2018), tropinone reductase I (Nakajima et al., 1993b; Kai et al., 2009a; Qiang et al., 2016), tropinone reductase II (Hashimoto et al., 1992; Nakajima et al., 1993a,b), CYP80F1 (Li et al., 2006), and hyoscyamine 6β-hydroxylase (Matsuda et al., 1991; Li et al., 2012). Since their identification, the associated TA-biosynthesis genes have been applied to engineer TA biosynthesis in plants via the overexpression method (Zhang et al., 2004; Wang et al., 2011; Zhao et al., 2017). In this context, it is thus very important to distinguish those enzymes with higher catalytic activities to better facilitate the metabolic engineering of metabolite biosynthesis. However, very few plant ODC enzymes are ever studied in great detail for their enzymatic kinetics by using purified recombinant proteins. This gap in knowledge means that most results of ODC studied in TA-producing plants are preliminary.

To better understand ODC's role in the biosynthesis of TAs and polyamines, the ODC gene (HnODC) was isolated from H. niger. Tissue profiling of HnODC was analyzed by using quantitative reverse transcriptase PCR. Furthermore, the expression patterns of HnODC, and TA-biosynthesis genes (HnPMT, HnTRI, and HnH6H) were investigated through MeJA treatment. Simultaneously, the ornithine-derive metabolites, including putrescine, spermidine, spermine, N-methylputrescine and two types of TAs (hyoscyamine and scopolamine), were analyzed. Finally, the purified recombinant HnODC was used to analyze its kinetics. Metabolic characterization of HnODC not only revealed its roles in the biosynthesis of TAs and polyamines, but also provided a candidate gene for potential use in TA engineering and polyamine production applications.

#### MATERIALS AND METHODS

#### Plant Materials and MeJA Treatment

Mature seeds of Hyoscyamus niger were harvested from the medicinal plant garden of the Xizang Agricultural and Husbandry College of Nyingchi (Tibet, China) in August 2016, with their taxonomic identity confirmed by Professor Xiaozhong Lan. These seeds were germinated into plantlets in substrate composed of vermiculite:pindstrap moss:perlite (6:3:1) and grown at 25 ± 1 ◦C under an 16 h-light/8 h-dark conditions. Once the plantlets reached 10 cm in height, their roots and leaves were respectively harvested for the tissue profile analysis of HnODC, HnADC1, HnADC2, and TA-biosynthesis genes, including HnPMT, HnTRI, and HnH6H. To determine whether MeJA influenced the expression levels of these genes and the metabolism of polyamines and TAs, the 10-cm-tall H. niger plants were treated with 100 µM of MeJA for 0, 1, 6, 12, and 24 h. Each duration had three replicate plants, from which the roots were harvested for RNA isolation and metabolite analysis. The leaves collected from same plants were used for metabolite detection. Plant material treated with a solution lacking MeJA for 24 h


served as the control. Three or more independent plants per treatment were used in all analyses.

### Gene Cloning and Bioinformatics Analysis

Total RNA was extracted from the H. niger roots with RNAsimple Total RNA Kit, according to the manufacturer's protocols (Tiangen Biotech, Beijing, China). 50–100 mg of material from each plant part was used to extract total RNA. The first-strand cDNA chain was synthesized by using a FastKing RT kit (Tiangen Biotech, Beijing, China). The prepared reaction mixture, with a total volume of 10 µl, contained 2 µl of buffer (DNase solution provided by the FastKing RT kit), 2 µg of total RNA, and ddH2O was incubated at 42◦C for 3 min to remove any potential genomic DNA. Next, 2 µl of King RT buffer, 1 µl of FastKing RT Enzyme Mix, and 2 µl of FQ-RT Primer Mix were added into the reaction mixture; ddH2O was then also added to obtain the final volume of 20 µl. Then, this 20 µl of the RT reaction mixture was incubated at 42◦C for 15 min and at 95◦C for 3 min. All the cDNA samples were diluted 50 times with RNase-free water, after which 8 µl of cDNA solution served as templates for the RT-PCR.

A pair of gene-specific primers, HnODC-F and HnODC-R (**Supplementary Table S1**), was used to isolate the coding sequence of HnODC based on sequenced H. niger transcriptomes (data not published). Amplification reactions were performed in a final volume of 50-µl buffer containing 5 µl of TransTaq HiFi Buffer I (10×) with 20 mM MgSO<sup>4</sup> (TransGen Biotech, Beijing, China), 4 µl of 2.5-mM dNTPs, 1 µl of each primer (10 mM), 2.5 U of TransTaq HiFi DNA polymerase, and 50 ng of template DNA. PCR conditions were set as follows: the templates were denatured at 94◦C for 5 min, followed by 28 cycles (94◦C for 30 s, 56◦C for 30 s, and 72◦C for 90 s), and finally incubated at 72◦C for 8 min. PCR products were purified from 1.0% (w/v) agarose gel with a DNA purification kit (BioFlux, Hangzhou, China), then subcloned into pMD19-T for sequencing by using the following protocol: a reaction mixture that contained 1 µl of PMD19-T vector, 1 µl of DNA fragment, 3 µl of ddH2O, and 5 µl of Solution I was first prepared, then incubated at 16◦C for 30 min and transformed into E. coli. The recombinant plasmid harboring HnODC was extracted from E. coli for sequencing. The sequence of HnODC was confirmed by sequencing it on a 3730 DNA Analyzer (Thermo Fisher Scientific, Waltham, MA, United States) using M13 forward and M13 reverse universal primers, via the Sanger sequencing approach. The BLAST analysis was performed online at the website<sup>1</sup> (Johnson et al., 2008).Then multiple alignments were performed using the ClustalX bioinformatics program (Larkin et al., 2007). A phylogenetic tree was built by the neighborjoining method in MEGA software v.5 (Tamura et al., 2011). Its bootstrapped values were generated from n = 1000 replicates to evaluate the accuracy of the phylogenetic construction.

### Gene Expression Analysis

To analyze the tissue profile of TA-biosynthesis genes HnODC, HnADC1, HnADC2, HnPMT, HnTRI, and HnH6H, their total RNAs were respectively extracted from the leaves and roots of H. niger plants (three biological replicates) according to the methods described above. Likewise, to analyze the expression patterns of these genes under the MeJA treatment, their RNAs were respectively extracted from the roots of plants treated with MeJA and control plants. After reverse-transcription into cDNAs, expression levels of the genes were analyzed by real-time quantitative PCR (qPCR), using the phosphoglycerate kinase gene (PGK) as an internal reference, by following the method of Li et al. (2014). The qPCR kits were purchased from BIO-RAD and the qPCR system was an IQ5 thermocycler (BIO-RAD, Hercules, CA, United States). The 2−11CT method was used to calculate the relative gene expression levels (Livak and Schmittgen, 2001). At least three independent plants were used in this gene expression analysis. All primers were designed with the software tool Beacon Designer (Premier Biosoft International, Palo Alto, CA, United States), based on sequences publicly available from the NCBI GenBank database. Primer specificity was validated by melting profiles and consisted of a single product-specific melting temperature. The associated GenBank accession numbers are HnODC (MK169378), HnPMT (AB018572), HnTRI (D88156), and HnH6H (DQ812529). All the primers used are listed in **Supplementary Table S1**.

#### Analysis of Alkaloids and Polyamines

N-methylputrescine and polyamines (putrescine, spermidine, and spermine) were extracted using the method described by Do et al. (2013). Root and leaf samples (1.00 g fresh weight, FW) from 24-h-MeJA-treated plants and corresponding control plants were homogenized respectively in liquid nitrogen, and then extracted in 4 ml of 0.2 N perchloric acid (PCA) at 4◦C for 1 h. After centrifugation at 16,000 g at 4◦C for 30 min, the supernatant was used to determine the plant content of polyamines. To do this, 1 ml of the supernatant was added with 10 µl of benzoyl chloride. After incubation at 37◦C for 25 min in the dark, the benzoylzed polyamines were extracted with 2 ml of chloroform, dried with nitrogen flow, and then dissolved in 1 ml of methanol. From each ensuing sample, 20 µl were injected into HPLC for metabolite analysis (Flores and Galston, 1982). The detection methods used here were similar to those applied in the enzymatic assays described below. Tropane alkaloid content was quantified by adhering to previously described methodologies (Qiang et al., 2016; Zhao et al., 2017; Geng et al., 2018). For this, 200 mg of dry powder from each plant part was accurately weighed for alkaloid extraction and detection, with at least three independent plants used for metabolite detections.

### Protein Purification and Enzymatic Assay

The HnODC coding region was amplified by using two primers (HnODC-PF/HnODC-PR) containing a restriction site for BamHI and SacI (**Supplementary Table S1**). First, the PCR products of HnODC were purified and digested using BamHI and SacI. Then the purified HnODC coding region harboring the restriction sites of BamHI and SacI was inserted into a pET-28a+ vector, to generate the prokaryotic expression vector. The ligation mixture consisted of 1 µl of T4 DNA ligase, 1 µl of reaction buffer, 2 µl of linearized pET-28a+ vector, and 6 µl of purified HnODC.

<sup>1</sup>https://blast.ncbi.nlm.nih.gov/Blast.cgi

Next, the constructs were introduced to E. coli Rosetta for protein expression. Bacteria were cultured in an LB liquid medium with 50 mg/L of kanamycin and 34 mg/L of chloramphenicol at 37◦C. When the OD value of bacterial cultures had reached 0.6, HnODC expression was induced by adding isopropyl β-D-1 thiogalactopyranoside (IPTG) at a final concentration of 0.5 mM. Bacteria were further cultured at 25◦C for 7 h, then harvested for protein purifications. The recombinant His-tagged HnODC was purified via Ni2+-chelating resin columns using the same methods we reported in a previous study (Qiang et al., 2016).

To perform the enzymatic assays, we followed Docimo et al. (2012). To test the catalytic activity of HnODC, its purified form was tested in a 10-mM Hepes buffer (pH 8.0) containing 1 mM dithiothreitol and 1 mM pyridoxal phosphate (PLP) for 60 min, by using different concentrations of L-ornithine as the substrate. The products (putrescine) from enzymatic assays and authentic putrescine were benzoylated into benzoyl putrescine (as carried out by Geng et al., 2018). The benzoylated samples were used for the HPLC analysis or stored at −80◦C for future use. All benzoylated samples were detected at 234 nm at an oven temperature of 30◦C on a Shimadzu LC-20 HPLC system (Shimadzu Corp., Kyoto, Japan). A YMC-Pack ODS-A column was used (150 × 4.6 mmI. D.S-5 µm, 12 nm) and the flowing phage consisted of methanol:water (41:59) at a flow rate of 1 ml/min throughout the analysis. 20◦µl of sample was injected for analysis. The Michaelis–Menten curve and Lineweaver-Burk plot of the HnODC enzyme were drawn to determine its K<sup>m</sup> and Vmax values, on which calculations of turnover rate (Kcat) and catalytic efficiency (Kcat/Km) were based.

#### RESULTS

### Molecular Cloning and Sequence Analysis of HnODC

The 1293-bp coding sequence of HnODC encoded a 430-aminoacid polypeptide (**Supplementary Figure S1**). The BLASTP analysis indicated HnODC belonged to the superfamily of type III PLP-dependent enzymes, and that it resembled the ODCs in GenBank. As the major binding site of α-DFMO (Coleman et al., 2003), the GPTCD motif was found at the C-regions in all of these ODC and LDC proteins (**Figure 2**). For HnODC, this motif was located at positions 372–376 (**Figure 2**), while the YAVKCN motif was at positions 90–95, and present in all the ODC and LDC sequences (**Figure 2**). Lysine in this motif was postulated to bind to the cofactor pyridoxal-5<sup>0</sup> phosphate (Lee and Cho, 2001). But compared with the ODC enzymes of mammals, all plant ODC enzymes we found lacked the PEST regions required for constitutive and conditional degradation of ODC by the 26S proteasome system (Lee and Cho, 2001). Our phylogenetic analysis showed a distant evolutionary relationship between mammalian ODC enzymes and those from plants (**Figure 3**). All plant L/ODCs and ODCs occupied the same branch, which had two subgroups, and the L/ODCs are mainly from Leguminosae. Furthermore, HnODC showed closer evolutionary relationships with ODC proteins of Solanaceae plants and EcODC (**Figure 4**). HnODC has been deposited in GenBank (accession number MK169378).

#### Gene Expression Analysis

Expression levels of HnODC, HnADC1, HnADC2 and three TAbiosynthesis genes (HnPMT, HnTRI, and HnH6H) were detected in the roots and leaves of H. niger (**Figure 4**). However, all three genes—HnPMT (**Figure 4B**), HnTRI (**Figure 4C**), and HnH6H (**Figure 4D**)—were specifically expressed in the roots of H. niger, consistent with earlier reported findings (Hashimoto and Yamada, 1987; Hashimoto et al., 1992; Geng et al., 2018). Unlike the root-specific expression of those three TA biosynthesis genes, HnODC was expressed in both roots and leaves, but its expression level was still much higher in roots than leaves (**Figure 4A**). HnADC1 was expressed in roots and leaves at similar level, and HnADC2 was expressed in roots and leaves with no significant difference (**Supplementary Figure S2**). However, these six genes responded differently to the MeJA treatment of H. niger plants (**Figure 4**): the transcript levels of HnODC (**Figure 4E**) and HnPMT (**Figure 4F**) were significantly increased, while the expression of HnTRI (**Figure 4G**), HnH6H (**Figure 4H**) and HnADC1/2 (**Supplementary Figure S2**) was unchanged. Across 1–24 h of the MeJA treatment, the expression of HnODC increased up to 6-fold that of the control, while that of HnPMT increased by 6–30 folds. These results indicated HnODC is mainly expressed in roots and induced by MeJA.

#### Metabolite Analysis

Since the above gene expression analysis revealed HnODC was upregulated by MeJA at the transcriptional level, it was reasonable to investigate ornithine-derived metabolite production in H. niger plants treated with MeJA (**Figure 5**). Specifically, we examined polyamines, N-methylputrescine, hyoscyamine, and scopolamine content in their roots and leaves. Putrescine production was significantly promoted in roots (**Figure 5A**) and leaves (**Figure 5B**), where its concentrations were respectively increased by 34.98 and 90.32% over the corresponding control plants. By contrast, the MeJA treatment did not affect spermidine and spermine production in the roots (**Figure 5A**), but it did elevate their concentrations significantly in leaves (**Figure 5B**). Roots of MeJA-treated plants had a 6.14 fold higher N-methylputrescine content (30.14 ± 6.52 nmol/g FW) relative to control roots (4.91 ± 2.54 nmol/g FW) (**Figure 5A**). In leaves of either MeJA-treated or control plants, N-methylputrescine was not detectable (**Figure 5B**). The production of hyoscyamine and scopolamine was also detected in H. niger roots and leaves, for which more scopolamine than hyoscyamine was found, consistent with former's greater abundance than the latter in H. niger (Han Woo et al., 1995). However, hyoscyamine and scopolamine concentrations were not significantly changed in roots and leaves (**Figures 5A,B**) by MeJA, suggesting it did not significantly affect their production.

#### Protein Purification and Enzymatic Assay

To determine its enzymatic activity, HnODC was expressed in E. coli to produce its recombinant proteins (**Figure 6A**).


Hyoscyamus niger (MK169378).

activity on the decarboxylation of ornithine. represent lysine/ornithine decarboxylases confirmed with activity on the decarboxylation of lysine/ornithine. The numbers on the phylogenetic tree are bootstrapped values (based on 1000 repeats). Gene Bank accession numbers are as follows: Datura stramonium (CAA61121); Erythroxylum coca (AEQ02350.1); Homo sapiens (NP\_001274119.1); Nicotiana glutinosa (AAG45222.1); Escherichia coli (BAE77028.1); Saccharomyces cerevisiae (DAA08982.1); Lupinus angustifolius (AB560664); Sophora flavescens (AB561138); Echinosophora koreensis (AB561139); Hyoscyamus niger (MK169378).

The HnODC enzymes could be readily obtained in the supernatants of lysed E. coli. Then, the His-tagged HnODC was purified using Ni2+-chelating resin column through elution with 50 mM of imidazole (**Figure 6B**). The molecular weight of recombinant HnODC was approximately 50 kDa (**Figures 6A,B**), consistent with its calculated molecular weight

FIGURE 4 | Tissue profiles of four TA-biosynthesis genes and their expression patterns in Hyoscyamus niger plants treated with MeJA for 0 to 24 h. Overall expression levels in roots and leaves of (A) HnODC, (B) HnPMT, (C) HnTRI, and (D) HnH6H. Expression of (E) HnODC, (F) HnPMT, (G) HnTRI, and (H) HnH6H according the duration of treatment with methyl jasmonate. Vertical bars are means ± standard errors (n ≥ 3). ∗∗indicates a significant difference at the level of P < 0.01 (t test).

and similar to the molecular weight of other reported plant ODC enzymes. When the substrate, ornithine, was fed to HnODC, the products were successfully detected by HPLC with a retention time of 14.8 min (**Figure 6C**), which agreed with that of the standard (**Figure 6C**). In our negative controls, no product was detected when HnODC was boiled (**Figure 6C**). Together, these results show HnODC did catalyze ornithine to produce putrescine.

Enzymes catalyzing the same reaction in different organisms usually differ from each other in their enzymatic kinetics, such as affinity to substrate and catalytic efficiency. To obtain kinetics information on HnODC behavior, we derived its Km, Vmax,

and Kcat values for ornithine: respectively, 2.62 ± 0.11 mM, 1.87 ± 0.023 nmol min−<sup>1</sup> µg −1 and 1.57 ± 0.015 at pH 8.0 and at 30◦C based on the Michaelis–Menten curve (**Figure 7A**). Notably, the K<sup>m</sup> value of HnODC was higher than that of either EcODC (0.395 mM) or NgODC (0.56 mM) (**Table 1**), suggesting that HnODC had lower affinity to ornithine than EcODC and NgODC (Docimo et al., 2012). Since the Vmax value of HnODC also exceeded that of EcODC and NgODC, its resulting Kcat value likewise greater. The Kcat/K<sup>m</sup> value, which expresses catalytic efficiency, was 599 M−<sup>1</sup> s −1 for HnODC and thus higher corresponding values reported for EcODC (465 M−<sup>1</sup> s −1 ) and NgODC (16.53 M−<sup>1</sup> s −1 ) (**Table 1**). Lineweaver-Burk plot for evaluation of K<sup>m</sup> and Vmax for HnODC was shown in the **Figure 7B**. In sum, the kinetic analysis demonstrated HnODC had a lower affinity to ornithine but a much higher catalytic efficiency than displayed by EcODC and NgODC.

#### DISCUSSION

#### HnODC Efficiently Converted Ornithine to Putrescine

Generally, all ODCs contain the conserved PLP-binding motifs composed of PFYAVKCN, and the GPTCD sequences, both of which are necessary for ODC activity (Coleman et al., 2003). Both of the motifs present in HnODC strongly suggest that it is a functional enzyme which catalyzes the decarboxylation of ornithine. Although the ODC sequences are similar to those of LDC, the phylogenetic analyses are able to distinguish between them. Since ODC and LDC each had its own clade in the phylogenetic tree, it has been suggested that they may have had common ancestors and evolved to different types of enzymes with slight modification (Bunsupa et al., 2012). Furthermore, the relatively high sequence similarity between the HnODC and other plant ODC proteins also suggest that HnODC should have similar functions to the other plant ODC enzymes.

Our biochemical assays confirmed that HnODC catalyzed the decarboxylation of ornithine to produce putrescine. HnODC had a lower affinity to ornithine than do the ODC enzymes of tobacco and coca tree, but its catalytic efficiency was found to be much greater that of ODCs of tobacco and coca tree. Particularly, HnODC showed about 36-fold increase in catalytic efficiency over NgODC. The low catalytic efficiency of tobacco ODC led it to be a limiting enzyme in nicotine biosynthesis, and consequently overexpression of yeast ODC enhanced the production of nicotine in transgenic tobacco (Hamill et al., 1990). In TA-producing plant species, the production of TAs was greatly reduced when ODC was inhibited by DFMO, suggesting that ODC might play a crucial role in TA biosynthesis (Nyman, 1994).

High catalytic efficiency of HnODC facilitated the production of putrescine that entered the biosynthetic pathway of TAs. Due to higher catalytic efficiency of HnODC, it might be a better candidate for engineering the biosynthesis of putrescine-derived metabolites than the reported plant ODC enzymes.

### HnODC Was Highly Expressed in Roots and Up-Regulated by MeJA

Biosynthesis genes involved in the same pathway usually have similar tissue expression patterns. We found that in Hyoscyamus niger plants the TA biosynthesis genes of HnPMT, HnTRI, and HnH6H were expressed almost exclusively in secondary roots. Unlike them, HnODC was expressed in both roots and leaves yet its expression level was much higher in roots than in leaves. The different expression levels of HnODC likely reflect the changing metabolic demands for putrescine by root and leaf organs. Both polyamines and TAs were synthesized in roots and this required more production of its key precursor (putrescine); therefore, a high expression of HnODC in roots matched this requirement. The phytohormone, MeJA, could up-regulate the TA biosynthesis genes in a species-dependent way. HnPMT expression was dramatically elevated by MeJA in hairy root cultures of H. niger (Geng et al., 2018), while it was not affected by MeJA in Atropa belladonna (Li et al., 2014). HnODC was found strongly induced by MeJA in roots and leaves of the H. niger plants; however, neither HnTRI nor HnH6H was changed by MeJA at the transcriptional level. Hence, we conclude that MeJA positively regulated the TA-biosynthesis genes, including HnODC and HnPMT, though it did not regulate the TA-biosynthesis genes, such as TRI and H6H. Considering that ADC enzymes contribute to the putrescine biosynthesis, their expression was also detected. Unlike HnODC and TA biosynthesis genes with high or specific expression in roots, HnADC1/HnADC2 was expressed in roots and leaves, with no significant difference. The two ADC genes were not responsive to MeJA treatment. Gene expression analysis suggested that the increased production of putrescine was mainly caused by the up-regulation of HnODC.

### MeJA-Induced Expression of HnODC Promoted the Production of Putrescine and N-Methylputrescine

Due to the MeJA-elevated expression of HnODC, putrescine the product given by HnODC—production was significantly increased in roots and leaves, consistent with previously reported results that overexpression of ODC enhanced the production of putrescine in transgenic tobacco and rice (Lepri et al., 2001; Kumria and Rajam, 2002). In particular, we found N-methylputrescine production markedly promoted in roots when H. niger plants were treated with MeJA. This increased production could have been caused by both MeJA-induced expression of HnPMT and a greater supply of putrescine provided by MeJA-induced expression of HnODC. In the leaves with or without MeJA treatment, N-methylputrescine production was undetectable, obviously due to the lack of HnPMT expression in leaves; this results also indicates that N-methylputrescine synthesized in roots was hardly translocated to aboveground to leaf parts. Yet MeJA clearly promoted the production of spermidine and spermine in leaves, whereas their respective production in roots was not altered. Biosynthesis genes involved in TA pathway, such as HnPMT, HnTRI, and HnH6H, were not expressed in leaf, suggesting that putrescine was not metabolized into TA biosynthesis in leaf. However, putrescine was able to go into biosynthesis of spermidine and spermine in leaf. The increased putrescine production induced by MeJA resulted in more putrescine available for biosynthesis of spermidine and spermine in leaf. In roots, putrescine entered the biosynthesis of N-methylputrescine, spermidine, and spermine. MeJA-elevated HnPMT expression led to enhanced conversion of putrescine into N-methylputrescine in roots, and consequently the production of spermidine and spermine was at relatively stable levels. For hyoscyamine and scopolamine, the MeJA treatment did not affect their levels in roots and leaves, suggesting that the elevated expression of HnODC and HnPMT were not enough to promote the production of the two pharmaceutical TAs. Previously, overexpression tobacco PMT gene markedly promoted the N-methylputrescine production but did not enhanced the TA accumulation in root cultures of H. niger (Zhang et al., 2004). To conclude, the up-regulation of HnODC at transcriptional level was able to provide the precursors (putrescine and N-methylputrescine) at higher levels for TA biosynthesis and thereby promote polyamine production in H. niger.

## AUTHOR CONTRIBUTIONS

TZ and ZL conceived and designed the study. TZ, CW, and FB performed gene cloning, expression analysis, and biochemical assays. FZ and GB performed bioinformatic analysis. CY and XL managed the plants. SL and MC detected metabolites. TZ and ZL prepared the manuscript. All the authors have read and approved the manuscript.

## FUNDING

This work was supported by the NSFC projects (31770335 and 31370333), the CSTC project (cstc2017jcyjAX0208), and the YNTC Foundation (2016YN22).

## ACKNOWLEDGMENTS

We are grateful to Charlesworth Group Author Services for language polishing.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00229/ full#supplementary-material

### REFERENCES

fpls-10-00229 February 26, 2019 Time: 16:3 # 10



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Zhao, Wang, Bai, Li, Yang, Zhang, Bai, Chen, Lan and Liao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Structural Study of Agmatine Iminohydrolase From *Medicago truncatula*, the Second Enzyme of the Agmatine Route of Putrescine Biosynthesis in Plants

#### *Bartosz Sekula\* and Zbigniew Dauter*

*Synchrotron Radiation Research Section of Macromolecular Crystallography Laboratory, National Cancer Institute, Argonne, IL, United States*

#### *Edited by:*

*Rubén Alcázar, University of Barcelona, Spain*

#### *Reviewed by:*

*Jarrod B. French, Stony Brook University, United States Pedro Carrasco, University of Valencia, Spain*

#### *\*Correspondence:*

*Bartosz Sekula bartosz.sekula@nih.gov; sekula.bartosz@gmail.com*

#### *Specialty section:*

*This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science*

*Received: 28 December 2018 Accepted: 27 February 2019 Published: 28 March 2019*

#### *Citation:*

*Sekula B and Dauter Z (2019) Structural Study of Agmatine Iminohydrolase From Medicago truncatula, the Second Enzyme of the Agmatine Route of Putrescine Biosynthesis in Plants. Front. Plant Sci. 10:320. doi: 10.3389/fpls.2019.00320*

Plants are unique eukaryotes that can produce putrescine (PUT), a basic diamine, from arginine *via* a three-step pathway. This process starts with arginine decarboxylase that converts arginine to agmatine. Then, the consecutive action of two hydrolytic enzymes, agmatine iminohydrolase (AIH) and *N-*carbamoylputrescine amidohydrolase, ultimately produces PUT. An alternative route of PUT biosynthesis requires ornithine decarboxylase that catalyzes direct putrescine biosynthesis. However, some plant species lack this enzyme and rely only on agmatine pathway. The scope of this manuscript concerns the structural characterization of AIH from the model legume plant, *Medicago truncatula*. *Mt*AIH is a homodimer built of two subunits with a characteristic propeller fold, where five αββαβ repeated units are arranged around the fivefold pseudosymmetry axis. Dimeric assembly of this plant AIH, formed by interactions of conserved structural elements from one repeat, is drastically different from that observed in dimeric bacterial AIHs. Additionally, the structural snapshot of *Mt*AIH in complex with 6-aminohexanamide, the reaction product analog, presents the conformation of the enzyme during catalysis. Our structural results show that *Mt*AIH undergoes significant structural rearrangements of the long loop, which closes a tunnel-shaped active site over the course of the catalytic event. This conformational change is also observed in AIH from *Arabidopsis thaliana*, indicating the importance of the closed conformation of the gate-keeping loop for the catalysis of plant AIHs.

Keywords: polyamine biosynthesis, putrescine, beta/alpha propeller fold, penteins, agmatine deiminase, guanidine-modifying enzymes

#### INTRODUCTION

Biosynthesis of putrescine (PUT) starts from arginine (ARG) and follows one of the two pathways which comprise agmatine (AGM) or ornithine (ORN) biotransformation (Michael, 2017). The AGM route is important and widely spread among plants, algae, and prokaryotic organisms (Michael, 2016). First, ARG is decarboxylated to AGM by arginine decarboxylase (ADC). The

**106**

Sekula and Dauter Structures of M*t*AIH

later conversion of AGM to PUT in plants is carried out by agmatine iminohydrolase (AIH) and further by *N-*carbamoylputrescine amidohydrolase (CPA), an octameric protein with a quaternary structure resembling an incomplete left-handed helix (Sekula et al., 2016). *AIH* and *CPA* genes have been acquired by plants through an endosymbiotic gene transfer from the cyanobacterial ancestor of the chloroplast (Illingworth et al., 2003). Actually, plants are unique eukaryotes to biosynthesize PUT *via* the AGM biotransformation, which makes ADC, AIH, and CPA potential targets for herbicide design (Böger and Sandmann, 1989). In some bacteria, AGM-to-PUT conversion is also catalyzed by agmatine ureohydrolase (agmatinase) (Satishchandran and Boyle, 1986) and most of the cyanobacteria use this enzyme instead of AIH and CPA (Fuell et al., 2010). Some Gram-positive bacteria may also obtain PUT in a catabolic pathway which produces ATP from the carbamoyl phosphate obtained from the AGM-to-PUT transformation (Llacer et al., 2007). The second manner of PUT biosynthesis, the ORN route, is dominant for most eukaryotes, including animals and fungi, and involves ORN decarboxylation catalyzed by ornithine decarboxylase (ODC) (Janowitz et al., 2003). Some plant species like *Arabidopsis thaliana* and *Physcomitrella patens* do not have the *ODC* gene (Hanfrey et al., 2001) and they rely only on the AGM-to-PUT bioconversion. Other plants, which have preserved *ODC*, may obtain PUT either from AGM or ORN.

PUT is the starting backbone for larger polyamines (PAs) which are produced by specialized aminopropyltransferases, enzymes which use decarboxylated S-adenosylmethionine as a donor of the aminopropyl group. Therefore, the first transfer of the aminopropyl group to PUT, catalyzed by spermidine synthase (SPDS), yields triamine spermidine (SPD). SPD is the substrate for the second transfer which yields symmetrical spermine or unsymmetrical thermospermine. For a long time, it was elusive whether both tetraamines are biosynthesized in plants, but they are actually formed by two distinct proteins, spermine synthase (SPMS) and thermospermine synthase (TSPS). Aminopropyltransferases are distinguished by several structural features that favor each enzyme toward the specific PA production (Sekula and Dauter, 2018).

PAs are essential for the regulation of various physiological processes which secure the proper growth and development of higher plants (Takano et al., 2012; Jimenez-Bremont et al., 2014; Minocha et al., 2014; Tiburcio et al., 2014; Liu et al., 2015). The cationic character of PAs promotes their interactions with anionic proteins and nucleic acids, thus affecting transcription, translation (Gill and Tuteja, 2010; Igarashi and Kashiwagi, 2010; Tiburcio et al., 2014), and the rate of membrane transport (Pottosin et al., 2014; Pottosin and Shabala, 2014). SPD is an important donor of aminobutyl group for the posttranslational modification of the hypusine-containing translation elongation factor eIF5A in eukaryotes and archaea (Prunetti et al., 2016). PAs can also modulate the activity of antioxidant enzymes and thereby influence the concentration of reactive oxygen species (Radhakrishnan and Lee, 2013; Kamiab et al., 2014; Mostofa et al., 2014). PA accumulation is often related with its protective role for the environmental stress conditions and leads to an increase of stress tolerance of the plant (Capell et al., 2004; Alcazar et al., 2010; Wang et al., 2011; Berberich et al., 2015). Meanwhile, defects of PA biosynthesis pathway result in the retardation, sterility, and other developmental pathologies in plants (Hanzawa et al., 2000).

Herein, we describe the structural characterization of AIH from *Medicago truncatula* (*Mt*AIH), the model legume plant. The enzyme is responsible for the second step of the AGM pathway of PUT biosynthesis, that is the hydrolytic conversion of AGM to *N-*carbamoylputrescine (NCP) with the release of ammonia. Plant AIHs, as well as CPAs, do not contain chloroplasttargeting peptides and they act in the cytoplasm. This is opposite to the first enzyme of the pathway, ADC, which initializes PUT biosynthesis in plastids (Illingworth et al., 2003). The advantage of AGM production outside plastids could be explained by the availability of AGM in the cytoplasm not only for PUT production but also for the biosynthesis of *N*-hydroxycinnamoyl conjugates, which may serve as precursors of defensive compounds (Burhenne et al., 2003). AIH belongs to one of the seven types of guanidine-modifying enzymes (GMEs) (Shirai et al., 2006). It is a member of the pentein superfamily that is characterized by the propeller-like arrangement of five repeated motifs that form a narrow channel with a central, negatively charged core (Hartzoulakis et al., 2007). The conserved catalytic triad of GMEs (His, Asp, and Cys) is responsible for a range of activities, which cover transferase and hydrolytic reactions on the guanidine-containing compounds (Hartzoulakis et al., 2007). Although AIHs from various plant species, including corn (Yanagisawa and Suzuki, 1981), soybean (Park and Cho, 1991), and maize (Yanagisawa, 2001) were isolated, there is no published structural characterization of any plant AIH available, except for unpublished reported entries in the Protein Data Bank (PDB) of AIH from *A. thaliana* (*At*AIH, PDB ID 3H7K, 3H7C, 1VKP, Center for Eukaryotic Structural Genomics).

In this work, we present the high-resolution crystal structure of non-liganded *Mt*AIH and the structure with the reaction product analog—6-aminohexanamide (AHX). This, combined with the in-solution small-angle X-ray scattering analysis, provides data for the characterization of this plant AIH and a detailed comparison of plant AIHs (*Mt*AIH and *At*AIH) with their bacterial orthologs.

#### MATERIALS AND METHODS

#### Cloning, Overexpression, and Purification of *MtAIH*

In order to express and purify *Mt*AIH (UniProt ID G7JT50), we used the protocol which was recently successfully applied in the studies of other plant enzymes (Ruszkowski et al., 2018; Sekula et al., 2018). Briefly, the following primers, forward: TACTTCCAATCCAATGCCCATGGCTTTCACATGCCTGCAG

**Abbreviations:** ADC, Arginine decarboxylase; AHX, 6-aminohexanamide; AIH, Agmatine iminohydrolase; CPA, *N*-carbamoylputrescine amidohydrolase; GME, Guanidine-modifying enzyme; NCP, *N*-carbamoylputrescine; ODC, Ornithine decarboxylase; PA, Polyamine; PUT, Putrescine; SPD, Spermidine; SPM, Spermine.

AAT and reverse: TTATCCACTTCCAATGTTACTAAATGGCTG GTTGTTGCTGAGTGAT and the cDNA from leaves of *M. truncatula* as a template were used in a polymerase chain reaction (PCR) prior to obtaining *Mt*AIH open reading frame (MTR\_4g112810) with encoded protein starting from codon number 11. The incorporation of *Mt*AIH gene into the pMCSG68 vector (Midwest Center for Structural Genomics) was performed according to the ligase-independent cloning (Kim et al., 2011) protocol. The vector introduces an N-terminal His6-tag followed by the Tobacco Etch Virus (TEV) protease cleavage site to the cloned protein and the Ser-Asn-Ala linker that is not cleaved from the expressed protein. In the next step, the BL21 Gold *E. coli* competent cells (Agilent Technologies) were transformed with the vector containing the *Mt*AIH gene. The cells were precultured at 37°C in LB medium with the addition of ampicillin (150 μg/ml) overnight. Next, 1.5% v/v of the culture was used as the inoculum of the fresh LB medium with ampicillin. It was cultured at 37°C until OD600 reached a value 1.0. In the next step, the culture was cooled to 10°C for 2 h and then the protein expression was induced with 0.5 mM of isopropyl-β-Dthiogalactopyranoside (IPTG). The protein overexpression was carried out at 18°C for 16 h. Before pelleting the cells in the centrifuge at 3,500 × g for 30 min, the culture was cooled to 4°C. Cell pellets were resuspended in 35 ml of the binding buffer [50 mM HEPES pH 7.4; 500 mM NaCl; 20 mM imidazole; 1 mM tris(2-carboxyethyl)phosphine, TCEP] and frozen at −80°C. Thawed cells were disrupted by sonication in an ice/water bath for 4 min (bursts of 4 s with 26-s intervals). Then, the cellular debris was pelleted by centrifugation at 25,000 × g for 30 min at 4°C.

The first step of *Mt*AIH purification was performed on a column packed with 5 ml of HisTrap HP resin (GE Healthcare) connected to the Vac-Man laboratory vacuum manifold (Promega). The supernatant was applied to the column and washed five times with 40 ml of the binding buffer. The protein elution was performed with 20 ml of the elution buffer (50 mM HEPES pH 7.4; 500 mM NaCl; 400 mM imidazole; 1 mM TCEP). His6-tagged TEV protease (final concentration of 0.1 mg/ml) was used to cleave the His6-tag from *Mt*AIH. This step was simultaneous to the overnight dialysis at 4°C against the dialysis buffer (50 mM HEPES pH 8.0; 500 mM NaCl; 1 mM TCEP). After dialysis, the sample was applied on HisTrap HP resin to remove the cleaved His6-tag and His6-tagged TEV protease. The final step of the purification of *Mt*AIH was size exclusion chromatography on HiLoad Superdex 200 16/60 column (GE Healthcare) connected to an AKTA FPLC system (Amersham Biosciences). The column was equilibrated in 50 mM HEPES pH 7.4, 100 mM KCl, 50 mM NaCl, and 1 mM TCEP.

#### Crystallization and Data Collection

*Mt*AIH was concentrated with Amicon concentrators (Millipore) to the final concentration of 8 mg/ml, determined by the absorbance measurement at 280 nm with the extinction coefficient of 77,920. The composition of the protein buffer was the same as the buffer used for the size exclusion chromatography. The sample was subjected to crystallization trials with use of Morpheus Screen (Molecular Dimensions) and PEG/Ion Screen (Hampton Research). Unused protein was stored in −80°C in 50-μl aliquots for later use. Crystals of *Mt*AIH were grown in 0.2 M sodium acetate, 20% PEG 3350 at pH 8.0. The *Mt*AIH-AHX complex was obtained by cocrystallization of *Mt*AIH with 10 mM of the ligand in 37th conditions of Morpheus Screen (Molecular Dimensions; 0.12 M Alcohols, 0.1 M Buffer System 1 at pH 6.5, 30% v/v Precipitant Mix 1) diluted with water to 70% of original concentration. Glycerol (25%) was used as a cryoprotectant for the freezing of native crystals. Crystals of *Mt*AIH-AHX were cryoprotected by original 37th conditions of Morpheus Screen. Protein was crystallized by sitting and hanging drop methods.

Diffraction data were collected at SER-CAT 22-ID beamline at the Advanced Photon Source (APS), Argonne National Laboratory, USA. The data were processed with *XDS* (Kabsch, 2010) and scaled using anisotropic diffraction limits with *STARANISO*<sup>1</sup> . The anisotropic cut-off surface for *Mt*AIH data has been determined with best and worst diffraction limits 1.20 and 1.42 Å, respectively. In the case of *Mt*AIH-AHX, the diffraction resolution was truncated between 2.20 and 2.66 Å. **Table 1** provides detailed statistics for spherical and anisotropic truncation. Anisotropic data treatment improved the electron density maps of refined structures. Coordinates and structure factors were deposited in the PDB with the following IDs: 6NIB (*Mt*AIH), 6NIC (*Mt*AIH-AHX).

#### Structure Determination and Refinement

The structure of *Mt*AIH was solved by molecular replacement in *Phaser* (McCoy et al., 2007) with the structure of *At*AIH (PDB ID 1VKP) as a search model. The initial model was rebuilt in *PHENIX AutoBuild* (Terwilliger et al., 2008). Then, the structure was subjected to manual and automatic refinement with *Coot* (Emsley et al., 2010) and *Phenix* (Adams et al., 2010) with anisotropic *B*-factors. Refined structure of unliganded *Mt*AIH was used as a model for determination of the structure of *Mt*AIH-AHX that was refined with isotropic *B-*factors and *TLS* (Winn et al., 2001, 2003) in *Refmac* (Murshudov et al., 2011). The refinement was carried out until the *Rwork* and *Rfree* values (Brunger, 1992), the geometric parameters, and the overall difference electron density maps were satisfactory. Evaluation of the final structures was performed in *PROCHECK* (Laskowski et al., 1993) and *MolProbity* (Chen et al., 2010). The final refinement statistics are given in **Table 1**.

#### Small-Angle X-Ray Scattering Data Collection and Analysis

SAXS data were collected from 5.5 mg/ml *Mt*AIH solution at the BioCAT 18-ID beamline (Fischetti et al., 2004) at APS with in-line size exclusion chromatography (SEC-SAXS) to separate sample from aggregates, thus ensuring optimal sample homogeneity. The sample was loaded on a WTC-015S5 column (Wyatt Technologies) connected to an Infinity II HPLC (Agilent Technologies). The sample after the column was sent to the Agilent UV detector, a Multi-Angle Light Scattering (MALS)

<sup>1</sup> http://staraniso.globalphasing.org/cgi-bin/staraniso.cgi



*1 Best anisotropic diffraction limit cut-off.*

*2 Worst diffraction limit after cut-off is 1.42 Å.*

*3 Worst diffraction limit after cut-off is 2.66 Å.*

*4 ADP, atomic displacement parameter.* 

*Values in parentheses refer to the highest resolution shell.*

detector, and a Dynamic Light Scattering (DLS) detector (DAWN Helios II, Wyatt Technologies), and an RI detector (Optilab T-rEX, Wyatt). Molecular weights and hydrodynamic radii were calculated from the MALS and DLS data respectively using the ASTRA 7 software (Wyatt). Afterward, the sample was sent to the SAXS flow cell, a 1.5-mm quartz capillary. Scattering intensity was recorded at 1.03-Å wavelength at room temperature, with 0.5-s exposures every 2 s on a Pilatus3 1M detector (Dectris) placed 3.5 m from the capillary (collected q-range was 0.004– 0.4 Å−1). Data reduction and analysis were performed by *BioXTAS RAW* 1.5.1 (Hopkins et al., 2017). Frames corresponding to the elution peak of the chromatogram were averaged to maximize the signal-to-noise ratio. Several frames immediately proximal to the sample peak (buffer frames) were averaged and subtracted from the sample scattering to obtain the final SAXS curve (**Figure 1A**). The *Rg* value calculated from the Guinier (**Figure 1B**) and distance distribution analysis (**Figure 1C**) was 30 Å. The calculated maximum dimension of the particle (*D*max) was 96 Å. The *qRg* limits for further calculations were 0.43–1.29.

*Ab initio* envelopes with the restraint of twofold symmetry were calculated in *DAMMIF* (Franke and Svergun, 2009), averaged with *DAMAVER* (Volkov and Svergun, 2003), refined with *DAMMIN* (Svergun, 1999), and filtered with *DAMFILT. SUPCOMB* was used for the superposition of the SAXS envelope with the crystallographic dimer of *Mt*AIH. *DENSS* (Grant, 2018) was used for the calculation of the *ab initio* electron density maps directly from the SAXS data with no prior information about the symmetry of the molecule.

#### Other Software Used

Molecular illustrations were created with UCSF *Chimera* (Pettersen et al., 2004). Ramachandran plot was calculated in *Rampage* (Lovell et al., 2003). Secondary structure was recognized with *ProMotif* (Hutchinson and Thornton, 1996) within the *PDBsum* server (de Beer et al., 2014). Sequence alignments were performed in *CLUSTAL W* (Thompson et al., 1994) and edited in *BioEdit* (Hall, 1999).

## RESULTS AND DISCUSSION

### *Mt*AIH Presents the Pentein **α**/**β** Propeller Fold

AIHs are assigned by the Structural Classification of Proteins (SCOPe) (Fox et al., 2014) to the porphyromonas-type peptidylarginine deiminase family that is a part of the Superfamily of penteins, characterized by a propeller-like arrangement of five αββαβ units which form a narrow channel in the core (Hartzoulakis et al., 2007). Penteins share a conserved group of residues that recognize the guanidine moiety of the substrate—His, Cys, and two acidic, guanidinebinding residues (usually Asp) which serve to catalyze a range of reactions (Linsky and Fast, 2010). The monomer of *Mt*AIH is no different in this matter, i.e., five motif repeats (I–V) are arranged around fivefold pseudosymmetry axis that is aligned with the catalytic tunnel in the core of the protein (**Figures 2A,B**). Class, Architecture, Topology, Homology (CATH) server (Sillitoe et al., 2015) matches *Mt*AIH with the L-arginine/glycine amidinotransferase superfamily that belongs to the Class 3 of alpha beta proteins with the architecture of a five-bladed propeller.

The overall globular shape of the *Mt*AIH monomer resembles pentagonal prism, where the longest helices (η1/α1, α1, α7, α9, α11) are positioned in the imaginary vertices of the pentagon with all β-strands running along the direction marked by these helices. Sixteen strands in *Mt*AIH form five β-sheets, one four-stranded, and four with three strands each. All β-sheets are oriented toward the center of the molecule forming five "blades" of the propeller (**Figure 2B**). Repeat I with its four-stranded β-sheet actually disturbs the overall fivefold pseudosymmetry of the molecule, i.e., it has the additional strand β3 and the helix η2/α2 which are placed outside the pentagonal shape. Moreover, the αββαβ motif of unit I is, in fact, discontinuous and it is fully formed with the complementation of C-termini, more precisely, by α14 and β16 (**Figure 2A**). In the center of the molecule, four short helices (α3, η6, η9, α14), that directly precede internal strands from repeats I–IV, line the surface of the negatively charged central channel, that is the active site. These core helices are placed on N-terminal sides of the inner strands of β sheets from repeats I–IV. Only the inner strand of repeat V (β13) is not directly preceded by a short helix. Instead, the first helix of this repeat (α11) is actually longer than helices that build the active site and it is placed almost outside of the outline of the protein which precludes it from the interactions with the substrate in the active site. Additionally, α11 is flanked by long coils. One of these coils (residues 291–314) covers the active site entrance and plays a crucial role in the substrate recognition (see below for details). *Mt*AIH has a very high structural similarity to the other plant ortholog, *At*AIH (unpublished, PDB ID 3H7K, overall sequence identity is 70%), with the 0.6 Å root mean square deviation of the superposed structures. Both plant AIHs have a similar organization of secondary structure and almost identical architecture of the active site.

FIGURE 2 | The monomer of *Mt*AIH. (A) Topological diagram of *Mt*AIH with a schematic depiction of the five αββαβ units (I–V) that form internal pseudo fivefold symmetry; secondary structure elements, helices (cylinders) and sheets (arrows) are colored in red and cyan, respectively; blue line indicates the position of the second *Mt*AIH subunit from the dimeric assembly. (B) Ribbon representation of the structure of *Mt*AIH monomer in complex with AHX (shown as spheres); the gate-keeping loop over the active site is marked with a purple dotted line.

#### *Mt*AIH Forms Symmetric Dimers

The molecular mass of *Mt*AIH calculated from MALS is 81 kDa, which almost ideally agrees with the molecular weight of two monomers (theoretical mass of the *Mt*AIH construct is 41.2 kDa). The dimeric assembly of *Mt*AIH is also shown by SAXS envelope even though the calculated mass of *Mt*AIH from SAXS was underestimated to be ~60 kDa. The envelope with twofold symmetry restraints (**Figure 3A**) and *ab initio* electron density map that was calculated with no information about the molecule symmetry (**Figure 3B**) clearly correspond to the *Mt*AIH dimer in the crystal lattice where two subunits are related by twofold symmetry. It is worth to note that both crystal structures, *Mt*AIH and *Mt*AIH-AHX, present different crystallographic symmetry (see **Table 1**) with one and four chains in the asymmetric unit, respectively. In the case of the unliganded structure, the dimer is created by a crystallographic twofold axis, while in the *Mt*AIH-AHX complex, there are two almost identical dimers in the asymmetric unit. The comparison of our results with the crystal structure of *At*AIH (**Figure 3C**) clearly shows that *At*AIH also forms dimers analogical to *Mt*AIH.

The results are somewhat contrary to the analysis of the *Mt*AIH crystal structure done with the PISA server (Krissinel and Henrick, 2007). It shows that monomer of *Mt*AIH has the surface area ~14,000 A2 with the biggest interface area ~900 Å2 that is shared with the closest monomer in the crystal lattice. It is about 6.5% of the monomer surface and it was estimated by the PISA server to have no role in the complex formation. However, in both *Mt*AIH crystal forms, this interface area between two subunits is preserved and, taking it together with SAXS and MALS results, it is, in fact, responsible for the formation of *Mt*AIH dimers. The intersubunit interactions on this interface involve 25 residues (the same set from both interacting subunits) where about half of them create 20 hydrogen bonds or salt bridges. The interface residues belong to the β2, β3, and η2/α2 from repeat I and α5 which belongs to the repeat II (**Figure 2B**). Two subunits of *At*AIH in the crystal structure (PDB ID 3H7K) share the analogical interface area and the PISA server analysis does not recognize it as a dimer either. Most of the interacting residues are preserved in both plant orthologs, but *At*AIH evidently presents fewer interacting residues which altogether form only 14 hydrogen bonds.

A dimeric assembly was independently reported for the other plant AIHs, including *At*AIH (Janowitz et al., 2003), AIH from maize (*Zm*AIH) (Yanagisawa, 2001) and rice (*Os*AIH) (Mohan Chaudhuri and Ghosh, 1985). The reported exception (Park and Cho, 1991) is a 70-kDa monomeric protein from soybean that was described to have AIH activity. However, the authors did not provide the sequence of the isolated protein. Also, any record classified as AIH matches to the reported description. The sequence alignment (**Figure 4**) of the dimeric plant AIHs shows that 19 residues (out of the pool of 25 which form the interface in *Mt*AIH) are identical or very similar in all four plant AIHs. A similar extent of conservation applies to polar and hydrophobic residues on the dimer interface. Identical polar positions are: Gln24, Glu58, Thr61, Ser65, Gln68, Arg73, Arg81, Glu84, Ser86 Lys145, Glu150, and Arg151. These residues in *Mt*AIH are involved in 14 hydrogen bonds, that is, 70% of all hydrogen bonds found in the interface analysis.

Analyzing sequence conservation of all plant AIHs (**Figure 5**), there is no obvious highly conserved area around the dimer interface that can be distinguished right away. However, when considering the conservation of particular residues involved in the hydrogen bonding between both subunits, most of them stand out as highly conserved (Val60, Arg73, Arg81, Val82, Glu84, Ser86, Lys145), whereas only three are very variable residues (Asp148, Val149, Arg151). Moreover, hydrophobic

FIGURE 3 | *Mt*AIH dimers. (A) *Ab initio* SAXS envelope (gray mesh) with the superposed crystallographic dimer of *Mt*AIH; (B) *Ab initio* electron density maps of *Mt*AIH calculated by *DENSS* from the SAXS data; contours of the map are as follows: 5σ (red), 3.5σ (yellow), 1.4σ (green), 0.7σ (blue). (C) The shape of the crystallographic dimer in the *At*AIH structure (PDB ID 1VKP).


FIGURE 4 | Sequence alignment of selected AIHs. The alignment was made with the following AIH sequences (*UniProt* accession numbers are given in square brackets, as well as the sequence identity with *Mt*AIH): *Mt*AIH [G7JT50], *At*AIH [Q8GWW7, 70% sequence identity], *Os*AIH [Q01KF3, 65%], *Zm*AIH [C0PHP8, 64%], *Cj*AIH [Q0P9V0, 26%], *Hp*AIH [O24890, 26%], *Pg*AIH [Q7MXM8, 27%], *Ef*AIH [Q837U5, 44%], *Sm*AIH [Q8DW17, 45%]. Sequence positions above the alignment and annotation of the secondary structure elements (α helices and 310 helices, η, are shown as red cylinders and β strands are shown as cyan arrows) correspond to *Mt*AIH. Residues are color-coded by type. Green circles indicate residues that form the active site or participate in the interactions with the bound substrate. Blue and green lines below the alignment indicate residues that form dimer interface in *Mt*AIH and *Hp*AIH, respectively. The gate-keeping loop over the active site of plant AIHs is marked with a purple dotted line.

calculated from the sequence alignment of plant AIHs; the analysis was done in *ConSurf* (Ashkenazy et al., 2016); dimer interface is circled in blue. The pool of 184 plant AIH sequences was selected from the protein sequences classified to agmatine deiminase family (IPR017754) by *InterPro* (Finn et al., 2017); clear outliers where the sequence length was shorter than 340 or longer than 420 were manually excluded from the alignment.

interactions seem to be important for the dimer formation as well—at the center of the interface between dimer mates there is a patch of apolar residues (Trp69, Val83, Ile85, Val149). All of these residues are apolar in plant AIHs.

#### Bacterial AIH Analogs With Various Biological Assemblies

In the PDB, there are several structures of bacterial representatives of AIHs that not necessarily form dimers like plant AIHs (**Figure 6A**). The bacterial AIHs form tetramers, like AIH from *Streptococcus mutans* (*Sm*AIH, PDB ID 2EWO) and *Enterococcus faecalis* (*Ef*AIH, PDB ID 2JER) (Llacer et al., 2007) or monomers like AIH from *Campylobacter jejuni* (*Cj*AIH, PDB ID 6B2W) (Shek et al., 2017). There are also dimeric bacterial AIHs like *Helicobacter pylori* (*Hp*AIH, PDB ID 3HVM) (Jones et al., 2010a) or *Porphyromonas gingivalis* (*Pg*AIH, PDB ID 1ZBR). *Hp*AIH and *Pg*AIH both present analogical dimer interface, however, it is different than the dimer interface of plant AIHs (**Figure 6B**). In these two bacterial dimeric AIHs, the interface residues are from repeats II and III. To be more precise, these residues correspond to the residues from β5, α4, α5, β6, and the loop between α7 and α8 of *Mt*AIH (**Figure 4**). This dimer interface of both bacterial AIHs is even smaller (~700 Å2 , slightly above 5% of the monomer surface) than that of plant AIHs. A closer look at the superposition of *Mt*AIH with bacterial dimeric AIHs (**Figure 6B,** right panel) reveals the bacterial interface to be placed very close to the region of repeat II that in *Mt*AIH forms short helix α4. In plant AIHs, this region is important for the ligand binding and the conformation of α4 shows that it would create severe steric clashes with the dimer mate. These regions of bacterial dimeric AIHs and *Cj*AIH are five residues shorter. On the other hand, a closer look at the dimer interface of plant AIHs (**Figure 6A,** left panel) shows the different orientation of two helices in bacterial dimeric AIHs—η2/α2 and α5 with significantly different residues in these helices.

## Substrate Binding Mode of Plant AIHs

The *Mt*AIH-AHX structure was obtained by cocrystallization and presents the bound AHX in three out of four protein chains that are present in the asymmetric unit. The ligand used for this study structurally differs from the physiological reaction product of *Mt*AIH, NCP, by the methylene which substitutes the secondary amine of NCP (adjacent to the carbamoyl moiety). Therefore, the conformation of the complex is similar to the conformation of the enzyme with the bound product after the reaction, representing a highly probable NCP binding mode (**Figure 7A**). It is worth to note that the *Mt*AIH-AHX structure shows a somewhat dynamic character where relative conformation of the ligand and surrounding residues (especially close to the terminal amine of AHX) is slightly different in each chain. Residues which are closer to the entrance of the active site have the B factor value significantly higher than the average B factor for the structure. The mean B factor value of the AHX (~35 Å2 ) is comparable to the structure average (~34 Å2 ), but it is still higher than the B factor of residues placed deep in the cavity. This can be explained by the nonphysiological character of AHX in comparison to AGM or even NCP; binding of the ligand was enabled due to its high concentration. The other available plant AIH structure which shows the details about the ligand binding mode is the structure of *At*AIH with the reaction intermediate (**Figure 7B,** PDB ID 3H7K, unpublished structure). Therefore, the substrate binding mode of plant AIHs can be described by the analogy of these two AIHs.

The active site of *Mt*AIH is formed as a negatively charged channel (**Figure 8**) that is covered by coil region which links repeat IV and V (residues 291–314) which forms a kind of a lid over the catalytic site (**Figure 2B**). The active site itself is highly conserved among plant AIHs with the exception of Trp125 which in plant species can also be replaced by Tyr. Most likely, this does not drastically alter the shape and character of the channel. In the case of *Mt*AIH, the channel is formed

by side chains of Trp91 and Trp125 where the planes of their indole rings are positioned almost perpendicularly to each other. The character of this region resembles the active site entrance of *Mt*CPA (Sekula et al., 2016), where also Trp residues shape the tunnel which guides to the catalytic Cys residue. In plant AIHs on the other side of the tunnel, there is Gly361 which due to the lack of side chain leaves the necessary void space for the ammonia and water molecules that are important for catalysis (see below for details).

Generally, GMEs bind their substrates with three different modes—1, 2A, 2B (Shirai et al., 2006). Of course, the bound substrates are structurally very diverse and also the orientation of the guanidine moiety placed in the vicinity of the catalytic triad is not always the same. In mode 1, substrates are bound in a way that their terminal parts (more distant from the catalytic triad) interact with residues from repeat IV and V. This promotes a completely different orientation of guanidine moiety at the bottom of the catalytic channel, which is rotated in comparison to the other two binding modes. The bound ligands in modes 2A and 2B interact with residues from repeat II and III. In the case of plant AIHs, the substrate binding mode corresponds to the mode 2 and is more similar to 2A, where the terminal amine group of AGM interacts with residues from repeat II. Therefore, the guanidine moiety of bound AGM reaches the active site bottom with a catalytic triad (in *Mt*AIH these are Cys366, Asp226, and His224) pointing toward Asp226. Polar residues that interact with the AGM guanidine moiety are Asn94, Asn226, and His224. They are responsible for the positioning of the plane of guanidine moiety very similar to the orientation of amide moiety of AHX (**Figure 7A**) so that it is susceptible to the nucleophilic attack from sulfur atom of Cys366 that is placed almost ideally on the line normal to the amide plane of AHX. The terminal amine of AGM placed by the entrance of the channel is stabilized by direct H-bonds with Asp220, Ala360, and a water-mediated H-bond with Asp89, analogous to the interactions observed in the *At*AIH structure (**Figure 7B,**

FIGURE 7 | The catalytic site of plant AIHs. (A) Close-up view of the catalytic site of *Mt*AIH with bound AHX (dark green) in the chain C of the *Mt*AIH-AHX structure; green mesh represents Polder omit maps (contoured at 5σ) calculated in *Phenix* (Liebschner et al., 2017). (B) Imidine covalent intermediate of hydrolyzed AGM (lime green) captured in the crystal structure of *At*AIH (PDB ID 3H7H, unpublished structure). Dashed black lines indicate important hydrogen bonding interactions described in the main text.

(Baker et al., 2001; Dolinsky et al., 2004); dimer interface is circled in blue.

PDB ID 3H7K, unpublished structure). Therefore, AGM that binds within the catalytic site of AIH is stabilized by a network of polar interactions involving every heteroatom in the substrate and by a series of hydrophobic interactions between its trimethylene moiety and the hydrophobic residues in the active site channel that connects the entrance with the catalytic site.

#### Concerted Conformational Rearrangements Upon Ligand Binding

The close vicinity of the *Mt*AIH active site is surrounded by two very flexible regions built mostly by long loops. One is the gate-keeping loop covering the active site (residues 291–314), which is a linker between repeat IV and V (**Figure 2B**). The other concerns the fragment 122–136 which belongs to the repeat II, where α4 is located. The gate-keeping loop is very variable in plant AIHs except for Arg301 and Arg306. Both regions are disordered in the non-liganded *Mt*AIH structure (**Figure 9A**), more precisely fragments between Glu129-Cys134 and Pro300-Tyr-293 were excluded from the structure due to the lack of electron density maps that would show their conformation. On the other hand, in the structure of *Mt*AIH-AHX complex, the electron density clearly shows their position (**Figure 9B**). This feature is also observed in *At*AIH (PDB ID 3H7K) with reaction intermediate, where the conformation of these coiled regions is fully modeled. Altogether, when the ligand is bound in the active site, these two regions come close together to form hydrogen bonds: between carbonyl oxygen of Cys132 and amide nitrogen of Lys299, and between

FIGURE 9 | The gate-keeping loop of *Mt*AIH. (A) The superposition of nonliganded *Mt*AIH structure (blue ribbons) with the structure of *Mt*AIH-AHX complex (yellow semitransparent ribbons); green dots depict last visible residues from the loops 122–136 and 291–314 in the non-liganded *Mt*AIH structure. (B) Close-up view of the 2*Fo–Fc* electron density map contoured at 1σ (blue mesh) for the residues 128–135 and 294–300 in the chain C of *Mt*AIH-AHX structure.

backbone nitrogen of Cys134 and carbonyl oxygen of Gly297. Additionally, the guanidine group of Arg301 from the gatekeeping loop creates H-bonds with Asp220 and Asn35 in close vicinity of the AGM binding site. Therefore, this disorderto-order transition secures the appropriate conformation of the bound substrate before reaction and the opening of the lid loop helps with product release after catalysis. The concerted disorder-to-order transition upon ligand binding was also observed in *Cj*AIH (Shek et al., 2017), however, it concerned different regions. More precisely, in *Cj*AIH, regions that showed conformational change upon substrate binding correspond to residues 122–136 and 212–224 of *Mt*AIH, therefore to the loops which are flanking α4 of repeat II and η9 which links repeats III and IV. The latter fragment in plant AIHs has a different sequence which, together with 18-residues shorter region of 278–314, results in the different recognition of the terminal amine of bound AGM in bacterial AIHs. Moreover, the sequence comparison of the plant AIHs suggests that the concerted conformational change of the gate-keeping coiled regions upon substrate binding can be characteristic for other plant AIHs as well. Likely, this feature can distinguish plant and bacterial orthologs, especially from those which present a shorter loop link between repeat IV and V.

#### Cys366 Forms a Covalent Intermediate With AGM

The catalytic mechanism of guanidine-modifying enzymes is very similar to the cysteine proteases (Shirai et al., 2006). For AIHs it was structurally studied with bacterial orthologs (Llacer et al., 2007; Jones et al., 2010b) and involves the creation of a thioester covalent intermediate.

The reaction starts after binding of AGM when the gatekeeping loop is closed and the sulfur atom of Cys366 is ready to perform a nucleophilic attack on the central carbon of AGM guanidine moiety to form a tetrahedral covalent adduct. Then, His224 (positioned on the other side of the plane of amide moiety of AHX, **Figure 7A**) donates the proton to the closest amine of the intermediate, thus acting as a general acid for the reaction. This leads to the break of the adjacent bond and release of ammonia. Subsequently, ammonia is most likely H-bonded with the OD1 of Asp226 and it can be replaced by a water molecule (most likely the one, which is H-bonded with Asp218 in *Mt*AIH-AHX structure, **Figure 9A**) so the reaction can proceed. This water molecule is presumably moved deeper in the active site to be activated by transferring a proton to the His224/Glu226 charge relay network to form a hydroxide ion, so it can make a nucleophilic attack on the carbon of amidino intermediate to form another tetrahedral intermediate. The most probable position of the hydroxide ion which attacks the central carbon is represented by water in *At*AIH structure (**Figure 7B**). The intermediate collapses to form *N-*carbamoyl putrescine with a planar ureido carbon. The product of enzymatic action of AIH presents the conformation analogical to that of AHX (with an additional hydrogen bond with Asp94). Finally, the gatekeeping loops can be opened to release the product.

#### CONCLUSIONS

The presented work described *Mt*AIH and compared its crystal structures with the other plant dimeric ortholog, AIH from *A. thaliana.* We have cross-validated our results with the reports on different plant AIHs highlighting residues that take part in the formation of AIH dimers in plants. These are residues from β2, β3, and η2/α2 from repeat I and α5 from repeat II. Plant AIHs are characterized by a different dimer interface to that observed in dimeric bacterial AIHs.

The crystallographic snapshots of *Mt*AIH together with deposited *At*AIH structures showed the detailed conformation of the coiled region that during the catalysis form a lid over the active site of plant AIHs. This loop is responsible for the recognition of the terminal amine of the bound AGM and provides necessary stabilization of the ligand in time of the catalytic event. Interestingly, the structural analysis of plant AIHs showed different disorder-to-order transition of the gate-keeping loops to that observed in bacterial orthologs, which shows that substrate recognition mechanism of plant AIHs differentiates them from bacterial AIH orthologs, especially those which present a shorter loop link between repeat IV and V.

#### DATA AVAILABILITY

The datasets generated for this study can be found in Protein Data Bank, 6NIB, 6NIC.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

BS planned and performed the experiments, analyzed the results, and wrote the manuscript. ZD analyzed the results and supervised the work.

#### FUNDING

This project was supported by the Intramural Research Program of the NCI, Center for Cancer Research. Diffraction data were collected at the Advanced Photon Source (APS), Argonne National Laboratory (ANL) at the SER-CAT beamline 22-ID, supported by the U.S. Department of Energy (DOE), Office of Basic Energy Sciences under Contract W-31-109-Eng-38. SAXS research on the 18-ID BioCAT beamline used resources of APS, a DOE Office of Science User Facility operated by ANL (contract DE-AC02-06CH11357), a project supported by grant 9 P41 GM103622 from the National Institute of General Medical Sciences (NIGMS). Use of the PILATUS 3 1M detector was provided by grant 1S10OD018090-01 from NIGMS.

#### ACKNOWLEDGMENTS

The authors are grateful to Srinivas Chakravarthy, BioCAT, for the assistance during SAXS experiments.


tobacco and tomato: effect on ROS elimination. *Biochem. Biophys. Res. Commun.* 413, 10–16. doi: 10.1016/j.bbrc.2011.08.015


Yanagisawa, H., and Suzuki, Y. (1981). Corn agmatine iminohydrolase: purification and properties. *Plant Physiol.* 67, 697–700. doi: 10.1104/pp.67.4.697

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Sekula and Dauter. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Spermidine Synthase (SPDS) Undergoes Concerted Structural Rearrangements Upon Ligand Binding – A Case Study of the Two SPDS Isoforms From Arabidopsis thaliana

#### Bartosz Sekula\* and Zbigniew Dauter

Synchrotron Radiation Research Section, Macromolecular Crystallography Laboratory, National Cancer Institute, Argonne, IL, United States

#### Edited by:

Antonio F. Tiburcio, University of Barcelona, Spain

#### Reviewed by:

Taku Takahashi, Okayama University, Japan Miguel A. Blazquez, Spanish National Research Council (CSIC), Spain

#### \*Correspondence:

Bartosz Sekula bartosz.sekula@nih.gov; sekula.bartosz@gmail.com

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 05 March 2019 Accepted: 11 April 2019 Published: 07 May 2019

#### Citation:

Sekula B and Dauter Z (2019) Spermidine Synthase (SPDS) Undergoes Concerted Structural Rearrangements Upon Ligand Binding – A Case Study of the Two SPDS Isoforms From Arabidopsis thaliana. Front. Plant Sci. 10:555. doi: 10.3389/fpls.2019.00555 Spermidine synthases (SPDSs) catalyze the production of the linear triamine, spermidine, from putrescine. They utilize decarboxylated S-adenosylmethionine (dc-SAM), a universal cofactor of aminopropyltransferases, as a donor of the aminopropyl moiety. In this work, we describe crystal structures of two SPDS isoforms from Arabidopsis thaliana (AtSPDS1 and AtSPDS2). AtSPDS1 and AtSPDS2 are dimeric enzymes that share the fold of the polyamine biosynthesis proteins. Subunits of both isoforms present the characteristic two-domain structure. Smaller, N-terminal domain is built of the two β-sheets, while the C-terminal domain has a Rossmann fold-like topology. The catalytic cleft composed of two main compartments, the dc-SAM binding site and the polyamine groove, is created independently in each AtSPDS subunits at the domain interface. We also provide the structural details about the dc-SAM binding mode and the inhibition of SPDS by a potent competitive inhibitor, cyclohexylamine (CHA). CHA occupies the polyamine binding site of AtSPDS where it is bound at the bottom of the active site with the amine group placed analogously to the substrate. The crystallographic snapshots show in detail the structural rearrangements of AtSPDS1 and AtSPDS2 that are required to stabilize ligands within the active site. The concerted movements are observed in both compartments of the catalytic cleft, where three major parts significantly change their conformation. These are (i) the neighborhood of the glycine-rich region where aminopropyl moiety of dc-SAM is bound, (ii) the very flexible gate region with helix η6, which interacts with both, the adenine moiety of dc-SAM and the bound polyamine or inhibitor, and (iii) the N-terminal β-hairpin, that limits the putrescine binding grove at the bottom of the catalytic site.

Keywords: polyamine biosynthesis, triamine, spermine, spermidine, putrescine, decarboxylated S-adenosylmethionine, cyclohexylamine, aminopropyltransferase

**Abbreviations:** APT, aminopropyltransferase; CHA, cyclohexylamine; dc-SAM, decarboxylated S-adenosylmethionine; PUT, putrescine; SPD, spermidine; SPDS, spermidine synthase; SPMS, spermine synthase; TSPS, thermospermine synthase.

Putrescine (PUT), a linear diamine, is produced by plants from agmatine or ornithine. Agmatine route, which originated in plants by the horizontal gene transfer from the cyanobacterial ancestor of the chloroplast (Michael, 2017), involves three enzymes: arginine decarboxylase (ADC), agmatine iminohydrolase (AIH), and N-carbamoylputrescine amidohydrolase (CPA). The first enzyme of the pathway, pyridoxal 5<sup>0</sup> -phosphate-dependent ADC, converts arginine to agmatine. Later, agmatine is hydrolyzed to N-carbamoylputrescine by AIH, a dimeric enzyme with a 5-bladed propeller fold (Sekula and Dauter, 2019). The last step of the pathway that provides PUT involves the action of an octameric CPA with a characteristic subunit arrangement that resembles an incomplete helix (Sekula et al., 2016). The alternative route for PUT biosynthesis, ornithine pathway, is carried out by ornithine decarboxylase (ODC). However, this enzyme is not common for all plant species and some plants lack ODC relying only on agmatine biotransformation (Hanfrey et al., 2001).

Longer polyamines are produced by aminopropyltransferases (APTs), enzymes that catalyze the transfer of an aminopropyl group from decarboxylated S-adenosylmethionine (dc-SAM), a universal cofactor of APTs, to a polyamine substrate. In plant organisms, different polyamines are produced by distinct specialized APTs (Shao et al., 2012). Thus, spermidine synthase (SPDS) utilizes PUT as the substrate for spermidine (SPD) biosynthesis. Spermine and thermospermine are produced from SPD by spermine synthase (SPMS) and thermospermine synthase (TSPS), respectively. Although both APTs use the same substrate they present significant differences that predestine them to produce distinct products (Sekula and Dauter, 2018). From the group of plant APTs, only the structures of TSPS from Medicago truncatula (MtTSPS, PDB ID 6bq2) (Sekula and Dauter, 2018), and SPDS1 from Arabidopsis thaliana (AtSPDS1, PDB ID 1xj5, Center for Eukaryotic Structural Genomics) are known, however there is no published structural characterization of AtSPDS1.

Polyamines are abundant cationic compounds that play a critical role in the growth and development of plants (Tiburcio et al., 2014). Increased stress tolerance of plants is one of the most important roles assigned to polyamines (Bouchereau et al., 1999). They take part in the regulation of stress signaling pathways (Kasukabe et al., 2004), scavenging of the reactive oxygen species (Radhakrishnan and Lee, 2013; Kamiab et al., 2014; Mostofa et al., 2014) and stabilization of the photosynthetic apparatus (Hu et al., 2014). Levels of the polyamines undergo massive changes in plants upon the biotic stress (Jiménez-Bremont et al., 2014) and their exogenous application may increase the resistance to pathogen infection (Gonzalez et al., 2011). Moreover, polyamines modulate the rate of membrane transport (Pottosin and Shabala, 2014; Pottosin et al., 2014), as well as they interact with nucleic acids and proteins, thus they take part in the regulation of the transcription and translation (Gill and Tuteja, 2010; Igarashi and Kashiwagi, 2010; Tiburcio et al., 2014). Dysfunctions of the polyamine biosynthesis pathway cause growth retardation, sterility, and other pathologies (Hanzawa et al., 2000). Polyamines are used to create conjugates with hydroxycinnamic acids to form hydroxycinnamic acid amides, essential compounds for certain developmental processes (Tiburcio et al., 2014) and precursors of defensive compounds (Burhenne et al., 2003). SPD plays an essential role in the hypusination process. It acts as a donor of the aminobutyl moiety for the posttranslational modification of lysine residue of the translation factor eIF5A. eIF5A participates in translation elongation and in plants is important for the control of flowering time, the aerial and root architecture, root hair growth, and adaptation for challenging growth conditions (Belda-Palazón et al., 2016).

Spermidine synthases most likely originated from the common ancestor before the separation of prokaryotes and eukaryotes. Then, independently in plants, fungi, and animals, they probably duplicated and evolved to acquire SPMS activity (Minguet et al., 2008). Another enzyme that probably originated from SPDS is PUT N-methyltransferase (Hashimoto et al., 1998a). Interestingly, the dc-SAM binding motifs of plant SPDSs are more homologous to this enzyme than to mammalian or bacterial SPDSs (Hashimoto et al., 1998b). On the other hand, TSPS originated in plants from the horizontal gene transfer, similarly to the proteins from the agmatine route. Therefore, it is not surprising that TSPSs show clear divergence from other APTs of the flowering plants (Sekula and Dauter, 2018), while SPDS and SPMS are more similar. A. thaliana, similarly to many other dicots, has two gene paralogs encoding SPDS1 and SPDS2 whose distribution differs across tissues, stage of development and environment (Alcázar et al., 2010). SPDS1, SPDS2, and SPMS in A. thaliana present dual subcellular localization and are localized in the nucleus and cytosol (Belda-Palazon et al., 2012). They can interact with each other to form heteromultimers (Panicot et al., 2002), however, these are found only in the nucleus. SPDSs can be potently inhibited by CHA and its cyclic or aromatic derivatives (Shirahata et al., 1991). The application of CHA leads to accumulation of the free polyamines and may stimulate radicle emergence and the miotic index (Gallardo et al., 1994).

In this work, we present a structural comparison of the two SPDS isoforms from A. thaliana (AtSPDS1 and AtSPDS2). We also discuss the binding mode of dc-SAM, the inhibition of AtSPDS by CHA and the structural rearrangements of AtSPDS upon the ligand binding.

### MATERIALS AND METHODS

#### Cloning, Overexpression, and Purification of AtSPDS1 and AtSPDS2

Complementary DNA (cDNA) of A. thaliana was obtained according to the protocol described earlier (Sekula et al., 2018). The cDNA was used as a template for a polymerase chain reaction in order to isolate AtSPDS1 and AtSPDS2 open reading frames (ORF), which are annotated in the GenBank as AJ251296.1 and AJ251297.1, respectively. Two sets of forward primers were designed in a way to

clone complete ORFs and the truncated ORFs. Complete ORF of AtSPDS1 was isolated with the following primers: TACTTCCAATCCAATGCCATGGACGCTAAAGAAACCTCT GCCA (forward) and TTATCCACTTCCAATGTTATCAATTGG CTTTTGACTCAATGACCTTCTT (reverse). In case of AtSPDS2 these were: TACTTCCAATCCAATGCCATGTCTTCA ACACAAGAAGCGTCTGTTA (forward) and TTATCCA CTTCCAATGTTACTAGTTGGCTTTCGAATCAATCACCTTC (reverse). The second variant of AtSPDS1 was isolated with the use of the same reverse primer and a different forward primer (TACTTCCAATCCAATGCCAAAAAGGAACCTGCTTGTTTC TCCACTG) which allowed to clone the gene starting from codon 34. Forward primer for the isolation of the second AtSPDS2 variant starting from the codon 39 was as follows: TACTTCCAATCCAATGCCAAGGAGCCTTCTTGTATGTCCT CTATTATT. The amplification products were incorporated into a pMCSG68 vector (Midwest Center for Structural Genomics) according to the ligase-independent cloning protocol (Kim et al., 2011). Then, BL21 Gold Escherichia coli competent cells (Agilent Technologies) were transformed with the vectors which carried AtSPDS1 and AtSPDS2 genes and the truncated constructs. The proteins were overexpressed with N-terminal His6-tag followed by the tobacco etch virus (TEV) protease cleavage site and Ser-Asn-Ala linker, which is not cleaved from the expressed proteins. The cells were cultured at 37◦C in LB medium with ampicillin at 150 µg/ml concentration until OD<sup>600</sup> reached value 0.9 and then the culture was cooled to 10◦C for 1 h. The culture was induced with 0.5 mM of isopropyl-β-Dthiogalactopyranoside, and the overexpression was carried out at 18◦C for the next 16 h. Afterward, the cells were cooled to 4 ◦C and were pelleted by centrifugation at 3,500g for 20 min. The supernatant was discarded and 35 ml of the binding buffer [50 mM HEPES pH 7.4; 500 mM NaCl; 20 mM imidazole; 1 mM tris(2-carboxyethyl)phosphine, TCEP] was added to the cell pellets in order to resuspend them before freezing at −80◦C. The cells were then thawed and subjected to sonication in an ice/water bath. The total time of sonication was 4 min and it consisted of 4-s sonication bursts with the intervals of 26 s. Then, after centrifugation at 25,000g for 30 min at 4◦C, the supernatant was separated from the cell debris by decantation and applied on the column packed with 5 ml of HisTrap HP resin (GE Healthcare) which was connected to Vac-Man (Promega). The resin with captured proteins was washed five times with 40 ml of the binding buffer. AtSPDS1 and AtSPDS2, still carrying the His6-tags, were eluted with 20 ml of elution buffer (50 mM HEPES pH 7.4, 500 mM NaCl, 400 mM imidazole, 1 mM TCEP). Then, the portion of His6-tagged TEV protease (final concentration of 0.1 mg/ml) was added to the protein samples. The cleavage of the His6-tags from AtSPDS1 and AtSPDS2 was carried out in parallel to the overnight dialysis at 4◦C against the buffer containing: 50 mM HEPES pH 8.0, 500 mM NaCl, 1 mM TCEP. Then, the samples were applied on the HisTrap HP resin in order to remove cleaved His6-tag and His6-tagged TEV protease. AtSPDS1 and AtSPDS2 were then concentrated with Amicon concentrators (Millipore) and applied on the HiLoad Superdex 200 16/60 column (GE Healthcare) connected to the AKTA FPLC system (Amersham Biosciences). Size-exclusion

chromatography buffer contained 50 mM HEPES pH 7.4, 100 mM KCl, 50 mM NaCl, 1 mM TCEP.

#### Crystallization and Data Collection

Crystallization trials were carried out parallelly for two constructs (full-length and truncated) of both AtSPDS isoforms by the sitting drop method. Crystallization conditions of the full-length apo AtSPDS1 were as follows: 16 mg/ml protein concentration, 0.2 M ammonium sulfate, 0.1 M BIS-TRIS at pH 5.5, 25% polyethylene glycol (PEG) 3350. Crystals were cryoprotected before diffraction experiment by 25% glycerol. Truncated apo AtSPDS2 crystallized from 19 mg/ml in 76th conditions of the MORPHEUS screen (Gorrec, 2009). There was no necessity to use any cryoprotectant before freezing AtSPDS2 crystals. The complex of AtSPDS1 with dc-SAM and CHA was obtained by mixing truncated AtSPDS1 at 23 mg/ml with ligands (10 mM final concentration of each ligand). Crystals were obtained by streak seeding in conditions containing 0.18 M ammonium sulfate, 0.09 M BIS-TRIS at pH 5.5, and 22% PEG 3350. 25% MPD was used as a cryoprotectant. For clarity, the complex of AtSPDS1 with dc-SAM and CHA is further referred to as AtSPDS1-CHA. The concentration of protein samples was determined spectrophotometrically at 280 nm using theoretical molar extinction coefficients calculated in ProtParam (Gasteiger et al., 2005).

The diffraction data were collected at the SER-CAT 22- BM beamline at the Advanced Photon Source (APS), Argonne National Laboratory, United States. The diffraction data were processed with XDS (Kabsch, 2010). Since the diffraction of all crystals demonstrated significant anisotropy, scaling of the data was performed with STARANISO<sup>1</sup> . The anisotropic cutoff surface for AtSPDS1 data has been determined from 1.80 Å (best diffraction limits) to 3.14 Å (worst diffraction limits). In the case of AtSPDS2, the anisotropic diffraction limits were between 2.0 and 2.77 Å. For the AtSPDS1-CHA complex, the diffraction limits were between 1.80 and 2.70 Å. **Table 1** provides detailed statistics for spherical and anisotropic truncation. After anisotropic truncation of the data, the electron density maps of all refined structures were significantly improved. Coordinates and structure factors were deposited in the PDB under the accession numbers 6o63 (AtSPDS1), 6o64 (AtSPDS2), and 6o65 (AtSPDS1-CHA).

#### Structure Determination and Refinement

The previously deposited structure AtSPDS1 (PDB ID: 1xj5, Center for Eukaryotic Structural Genomics, unpublished structure) was used as an initial model for the phase determination of our AtSPDS1 structure. The model was then taken for the subsequent steps of manual and automatic refinement with Coot (Emsley et al., 2010) and Refmac (Murshudov et al., 2011). TLS parameters (Winn et al., 2003) were applied at the later stages of the structure refinement. In the case of AtSPDS2 and AtSPDS1-CHA structures, the coordinates of the dimer from the refined structure of AtSPDS1 were used as a search model in Phaser (McCoy et al., 2007).

<sup>1</sup>http://staraniso.globalphasing.org/cgi-bin/staraniso.cgi



<sup>1</sup>Best anisotropic diffraction limit cut-off. 1aWorst diffraction limit after cut-off is 3.14 Å. 1bWorst diffraction limit after cut-off is 2.77 Å. 1cWorst diffraction limit after cut-off is 2.70 Å. <sup>2</sup>ADP, atomic displacement parameter (B factor). <sup>3</sup>RMSD, rootmean-square deviation. Values in parentheses refer to the highest resolution shell.

The refinement was analogical to that of AtSPDS1. Rwork, Rfree factors (Brunger, 1992) and geometric parameters were controlled during refinement. The quality of refined structures was investigated in PROCHECK (Laskowski et al., 1993) and MolProbity (Chen et al., 2010). The final refinement statistics are given in **Table 1**. Geometrical restraints for CHA were generated in eLBOW (Moriarty et al., 2009).

### Small-Angle X-Ray Scattering Measurement

SAXS data were collected from the samples of the fulllength AtSPDS1 and the truncated construct of AtSPDS2 at 7 and 4.5 mg/ml, respectively. The experiments were carried out at the BioCAT 18-ID beamline (Fischetti et al., 2004) at APS. The sample was applied to the WTC-015S5 column (Wyatt Technologies) coupled to the Infinity II HPLC (Agilent Technologies) system on the in-line size exclusion chromatography (SEC-SAXS) setup. After the column, the sample was analyzed with the Agilent UV detector, a Multi-Angle Light Scattering (MALS) detector and a Dynamic Light Scattering (DLS) detector (DAWN Helios II, Wyatt Technologies), and an RI detector (Optilab T-rEX, Wyatt). Then, it was sent to the SAXS flow cell, a 1.5 mm quartz capillary. The scattering intensity was collected with the exposure 0.5 and 2-s intervals at 1.03 Å wavelength at room temperature on a Pilatus3 1M detector (Dectris). The sample-to-detector distance was 3.5 m and the collected q-range was 0.004–0.4 Å−<sup>1</sup> . BioXTAS RAW 1.5.1 (Hopkins et al., 2017) was used for data reduction and analysis. To increase the signal-to-noise ratio several frames from to the elution peak of the chromatogram were averaged. The subtraction of the buffer signal from the sample scattering was done on the averaged frames directly proximal the sample peak. The Rg value calculated from the Guinier and distance distribution analysis were 29.4 and 29.6 Å for AtSPDS1 and AtSPDS2, respectively. The calculated maximum dimensions of the particles (Dmax) for AtSPDS1 and AtSPDS2 were 100 and 103 Å. The further calculation was performed with the qRg limits for 0.28–1.30 for AtSPDS1 and 0.26–1.30 for AtSPDS2. DAMMIF (Franke and Svergun, 2009), DAMAVER (Volkov and Svergun, 2003), DAMMIN (Svergun, 1999), and DAMFILT were consecutively used for the calculation of the ab initio envelopes, averaging, refinement and filtration. Twofold symmetry restraints were used for the envelope calculations. SAXS envelopes were superposed with the crystallographic dimers in SUPCOMB.

#### Other Software Used

Molecular illustrations were made in UCSF Chimera (Pettersen et al., 2004). The sequence conservation scores were determined with ConSurf (Ashkenazy et al., 2016). The electrostatic potentials were calculated in PDB2PQR and APBS (Baker et al., 2001; Dolinsky et al., 2004). Polder omit maps were calculated in Phenix (Liebschner et al., 2017).

### RESULTS AND DISCUSSION

#### AtSPDS1 and AtSPDS2 Structures

The structures of apo AtSPDS1 and AtSPDS2 present the PEG molecules (absorbed from the crystallization solution) bound within the active site (**Figures 1A,B**). The conformation of this linear ligand mimics PUT, providing the information about the substrate binding mode inside the catalytic pocket. The third determined structure is the crystal complex of AtSPDS1 with two bound compounds, dc-SAM and CHA (**Figure 1C**), which precisely shows the binding mode of the cofactor and the inhibitor of SPDS.

AtSPDS1 and AtSPDS2 share the fold of polyamine biosynthesis proteins with the characteristic two-domain

topology (**Figure 1D**). The N-terminal domain is smaller (about 100 residues). It is built of six β-strands that fold into two β-sheets, the two-stranded β-hairpin and the four-stranded antiparallel β-sheet. Additional β strand (Ser46-Ile48) is formed only in chain H of AtSPDS2 (one of the eight chains in the asymmetric unit) creating three-stranded β-sheet at the N-terminus. In other chains of AtSPDS2 structure and all chains of AtSPDS1, the N-terminus is either more disordered or curved in a way that no additional β-strand is formed. The C-terminal domain has a Rossmann fold–like topology with a core β-sheet built of seven strands (five parallel and two antiparallel strands) that is buried between two helical bundles. The active site of AtSPDS is formed between the N-terminal and C-terminal domains (**Figure 1D**).

the provided sequence positions refer to the AtSPDS2 sequence.

The overall conformation of AtSPDS1 and AtSPDS2, and the distribution of the secondary structure elements are almost identical in these two isoforms (**Figure 1D**). Single chains of AtSPDS1 and AtSPDS2 superposed to each other present about 0.5 Å root-mean-square deviation. Also, the overall sequence conservation of plant SPDSs (**Figure 1E**) is high with many conserved regions that determine common characteristics of SPDSs in the plant kingdom. Only three regions show significantly lower conservation, and these are the N-terminus (about 45 residues), the region around the disordered loop between β13 and α10, and the C-terminal part (**Figure 1E**).

AtSPDS1 and AtSPDS2 share 83% sequence identity, however, when the highly variable and very flexible N-terminal part (up to Met45 of AtSPDS2) is excluded from the alignment, the sequence

identity is almost 90%. AtSPDS1 is six-residues shorter than AtSPDS2 and for the clarity of further analysis, all sequence positions that are identical in both AtSPDSs are denoted with double numbering, e.g., Asp201/205 indicates aspartic acid in position 201 in AtSPDS1 and 205 in AtSPDS2. Four sequence gaps of AtSPDS1 are in the disordered N-terminus. Two of the missing residues are placed in the long loop, which connects β13 and α10 (**Figure 1E**). Recently solved structure of another plant APT, TSPS from M. truncatula (MtTSPS) (Sekula and Dauter, 2018), shows higher order in this region, which in MtTSPS clearly folds into an additional helix.

#### Biological Assembly of AtSPDS1 and AtSPDS2

Both A. thaliana SPDS isoforms are dimers in solution. The estimated molecular weight of the full-length AtSPDS1 calculated from the SAXS results (**Figure 2A**) is 74.5 kDa, which almost ideally matches the theoretical dimer mass (73 kDa). In the case of the truncated construct of AtSPDS2, the SAXS results (**Figure 2B**) also matched the weight of the dimer (67.8 kDa in comparison to the theoretical value of 66.8 kDa). Also, the calculated ab initio envelope of AtSPDS1 clearly corresponds to the AtSPDS1 dimer in the crystal lattice (**Figure 2C**). The SAXS data of AtSPDS2 presented significantly lower signal-tonoise ratio and the calculated AtSPDS2 envelope (not shown) was worse in comparison to the envelope of AtSPDS1, although it resembled the AtSPDS2 crystallographic dimer in terms of its size. Slightly worse SAXS results for AtSPDS2 can be explained by a lower concentration of the protein sample used for the analysis.

All crystal structures are solved in the monoclinic system with P2<sup>1</sup> space group but in different crystal forms (**Table 1**). The asymmetric unit of apo AtSPDS1 contains two dimers (A-B and C-D), while the complex of AtSPDS1-CHA and the apo AtSPDS2 both present four dimers (A-B, C-D, E-F, and G-H) in the asymmetric unit. The dimers of all structures are very similar. The dimer interface of AtSPDS1 and AtSPDS2 involves about 50 surface residues, that is about 15% of the subunit surface. It is created by the interactions of both domains (**Figure 2C**). In the case of the N-terminal domain, residues from the strand β2 and the loops connecting β4-β5 and β6-α1 have a major contribution to the dimer formation. In the C-terminal domains, interface residues are from the loop β11-α8, strand β12, and the two helices, α10 and α11. In the apo AtSPDS1 crystal structure, dimers A-B and C-D (symmetry-related dimer) form an additional extensive buried surface between N-terminal domains of subunits A and C. This covers about 8% of the subunit surface and involves 12 hydrogen bonds which is recognized through PISA server (Krissinel and Henrick, 2007) as the interface important for oligomerization. No similar tight interdimeric interactions are observed in the other presented structures. Additionally, the SAXS results showing dimers of both AtSPDS isoforms suggest that these additional interactions are rather the consequence of tight crystal packing of AtSPDS1. On the other hand, it has been shown that plant APTs may form multiprotein complexes in vivo (Panicot et al., 2002) and the above mentioned interface may be involved in their formation. This somewhat similar tight interaction is actually responsible for the formation of MtTSPS tetramer in the crystal and in solution (Sekula and Dauter, 2018), where the N-terminal β-hairpins form eight-stranded β-barrel.

Most of the characterized APTs are dimers, analogously to AtSPDS1 and AtSPDS2. These include, e.g., E. coli SPDS (EcSPDS, PDB ID: 3o4f) (Zhou et al., 2010) or human SPDS (HsSPDS, PDB ID: 2o05) (Wu et al., 2007). The exceptions are the tetrameric SPDSs, like Thermotoga maritima SPDS (TmSPDS, PDB ID: 1inl) (Korolev et al., 2002), Bacillus subtilis SPDS (BsSPDS, PDB ID: 1iy9), or Helicobacter pylori SPDS (HpSPDS, PDB ID: 2cmg) (Lu et al., 2007). In some conditions, the latter protein can form either dimers or tetramers (Lu et al., 2007). The biological assembly of the tetrameric SPDSs more resembles the tetramer of MtTSPS with the β-barrel formed by the N-terminal β-hairpins of the four subdomains (Sekula and Dauter, 2018) rather than the formation between two dimers observed in the apo crystal structure of AtSPDS1.

### The Active Site of AtSPDS

The large catalytic cavity on the interface between the N-terminal and C-terminal domains is the region of AtSPDS with the highest negative charge. This feature can be easily explained by the necessity of AtSPDS to attract dc-SAM and PUT, both presenting the cationic character. The active site is composed of two main compartments, the dc-SAM binding site and the polyamine binding grove. Cofactor binding site is placed closer to the C-terminal domain, while the substrate site is rather buried in the N-terminal domain.

The dc-SAM binding site stretches alongside the glycine-rich region between β7 and α2 (residues Gly128/132-Gly133/137). Its boundaries are marked by Asp182/186 from one side and Gln107/111 from the other side (**Figure 3A**). Glu151/155, which is responsible for the H-bonding interactions with ribosyl moiety of dc-SAM, virtually divides the dc-SAM binding site into two compartments that accommodate adenosine and aminopropyl moieties, respectively (**Figure 3A**). The bulky compartment which facilitates adenosine moiety of the cofactor is covered by the very flexible region that comprises η6 together with the flanking loops (see below). The adenosine moiety of the dc-SAM is stacked between Leu212/216 and Ile152/156 and it forms three hydrogen bonds with the surrounding residues (**Figure 3A**). Two of these H-bonds are created by the N<sup>6</sup> amine with the carbonyl oxygen atom of Pro208/212 and the OD1 oxygen atom of Asp182/186. The third hydrogen bond is created between the N<sup>1</sup> of dc-SAM and the backbone amide of Gly183/187. The aminopropyl binding site is significantly smaller. The niche is formed by polar residues, Asp131/135, Asp201/205, and Gln107/111 that create three hydrogen bonds with terminal amine of the cofactor's aminopropyl moiety (**Figure 3A**). Asp201/205 is placed in a way to reach and to deprotonate the amine group of PUT before the reaction. Deprotonated PUT can perform the nucleophilic attack on the carbon atom of the dc-SAM aminopropyl moiety. A very similar hydrogen bonds network between dc-SAM and surrounding residues are present in other SPDSs, like human HsSPDS (PDB

FIGURE 3 | The active site of AtSPDS. (A) The binding mode of dc-SAM (black) and CHA (pink) shown in the chain A of AtSPDS1-CHA structure (green); dashed lines indicate hydrogen bonds; residues are numbered accordingly to the sequence positions of AtSPDS1 and AtSPDS2, respectively. (B) Comparison of the binding mode of CHA (pink) in chain A of AtSPDS1 structure (green) with PEG molecule (cyan) bound in chain G of AtSPDS2 structure (yellow).

ID: 2o0l) (Wu et al., 2007). Also, the analogical residues that correspond to Ile152/156 and Leu212/216 are responsible for the stacking of the adenine base of dc-SAM. Plant SPMSs present almost identical highly conserved primary structure of the dc-SAM binding site as AtSPDS. On the other hand, MtTSPS (PDB ID 6bq2) (Sekula and Dauter, 2018) shows some differences

inside the cofactor binding site. The first difference is the residue that corresponds to Glu151/155 of AtSPDS and binds the ribose moiety of dc-SAM. In MtTSPS, and in other plant TSPSs, it is Asp129. The other difference concerns Gln107/111, which in MtTSPS is replaced with His. Also, the adenine base is differently stabilized by the apolar residues. The residue corresponding to Leu212/216 is the same, but the difference concerns the interactions from the other side of the plane of the adenine base. In MtTSPS the function of Leu212/216 takes Leu179, the residue in a position corresponding to Ser202/206 of AtSPDS.

The polyamine binding site is an elongated tunnel-shaped cavity that stretches deep down the N-terminal domain, from Asp201/205 to Trp55/59. Trp55/59, which is a part of the N-terminal β-hairpin, limits the length of the cavity and shapes its bottom wall. The key residues in this part of the active site are easily recognized in the structure of AtSPDS1-CHA, where the polyamine groove is occupied by the inhibitor. CHA is bound close to two perpendicularly situated Tyr residues, Tyr106/110 and Tyr270/274, and its amine group crates three hydrogen bonds with acidic residues at the bottom of the polyamine grove (**Figure 3A**). These are the direct hydrogen bond with Asp208, and the two water-mediated H-bonds with Glu236/240 and Glu50/54. The position of the amine group of CHA overlaps with the PEG molecule that is bound in the apo structures (**Figure 3B**). The bound PEG molecule stretches along the polyamine groove, resembling PUT bound in other SPDS structures. Therefore, it is highly probable that PUT molecule in AtSPDS creates a very similar hydrogen bond network to CHA at the bottom of the pocket. The aliphatic portion of PUT is most likely stabilized by the interactions with two perpendicularly positioned aromatic side chains of Tyr106/110 and Tyr270/274 so that the other end of PUT can be pointed close to the cofactor, where it can be deprotonated by Asp201/205 and initialize the transfer of the aminopropyl moiety.

The ligands in the polyamine grove of AtSPDS do not reach as deep to the bottom of the active site as bound SPD in MtTSPS (PDB ID: 6bq7) (Sekula and Dauter, 2018). The reason that the amine group of SPD in MtTSPS is placed deeper inside the cleft where it creates a direct hydrogen bond with Glu30 (Glu50/54 of AtSPDS) is the necessity of MtTSPS to accommodate longer substrate in order to synthesize TSP. This difference in the polyamine binding mode between SPDS and TSPS is caused by several features of the N-terminal β-hairpin that distinguish these two plant APTs and determine their substrate discriminatory features (Sekula and Dauter, 2018). Additionally, Asp181 of MtTSPS (Asp204/208 in AtSPDS), the residue which is placed in the loop close to the η6 and in AtSPDS H-bonds the amine group of the substrate, in MtTSPS is rotated outside the polyamine grove and interacts with Gln214 instead of the substrate. In comparison to plant SPMSs, SPDSs lack the additional insert (about 20 residues) in the N-terminal β-hairpin (Sekula and Dauter, 2018) which presumably differentiates the shape of the polyamine groove between plant SPDSs and SPMSs, and therefore determines their different specificity.

The binding mode of CHA by AtSPDS1 is very similar to the binding mode of cyclic and aromatic CHA analogs in Trypanosoma cruzi SPDS (PDB ID 4yuw) (Amano et al., 2015)

#### FIGURE 4 | Continued

fpls-10-00555 May 4, 2019 Time: 16:20 # 9

panel; red arrows indicate major movements that occur upon ligand binding; note that the view in the bottom panel has been changed by the application of appropriate rotation (bottom left corner). Charge distribution mapped on the surface representation around the active site (B) in the closed conformation (chain A of AtSPDS1) and (C) in the open conformation (chain H of AtSPDS2); dc-SAM (black) and CHA (violet) are superposed from the AtSPDS1-CHA to indicate the location of the cofactor and the substrate binding sites. Orientation in the panels (B,C) is identical to the bottom view of the panel (A). The calculation of the electrostatic potential assuming pH 7.3 was made in PDB2PQR and APBS (Baker et al., 2001; Dolinsky et al., 2004).

and Plasmodium falciparum SPDS (Pf SPDS, PDB IDs: 4bp3, 4uoe, and 2pt9) (Dufe et al., 2007; Sprenger et al., 2015). In the case of Pf SPDS, the authors observed that the protein requires the stabilization of the flexible region with η6 together with the flanking loops to actually bind the inhibitor (Sprenger et al., 2015). A similar observation was made with HsSPDS where PUT was bound in the active site only when the cofactor was present (Wu et al., 2007). In the case of AtSPDS1 and AtSPDS2, we have also observed a somewhat similar situation. We have tried to soak the apo AtSPDS crystals with CHA or to cocrystallize AtSPDS with CHA alone, but these attempts were unsuccessful. On the other hand, both apo structures, even though there was no ligand in the dc-SAM binding site, presented bound PEG fragment inside the polyamine groove. It is true for all, but one chains of the two apo structures of AtSPDS. All chains present the conformation very similar to that shown in **Figure 1D**, which from now on is referred to as a closed state. One subunit of the AtSPDS2 (chain H) presents the state, where no ligand is bound in the active site in neither the cofactor binding compartment nor in the polyamine groove. The conformation of this chain stands out from the others and presents the possible open conformation of the active site (**Figure 4A**).

#### Conformational Movement of AtSPDS

Similarly to the other SPDS enzymes, also AtSPDS stabilizes upon ligand binding. This feature is even more emphasized when temperature factors of the two very similar (in terms of resolution) structures are compared – apo AtSPDS1 and AtSPDS1-CHA. Apo structure, where no cofactor is bound in the binding site, has an average B factor significantly higher than the complexed structure. Comparison of the chain H of AtSPDS2 in the open conformation with chain A of the AtSPDS1-CHA complex (**Figure 4A**) shows that the AtSPDS adopts two significantly different conformations. Globally, the main differences between the open (without ligands) and closed (with bound ligands) states concern the following regions (**Figure 4A**): η6 together with the flanking loops (residues 203/207-214/218), α8 with the preceding loop (residues 235/239- 252/256), the loop of the N-terminal β-hairpin (residues 51/55- 57/61) and the C-terminus.

In the first-mentioned region, in the closed conformation, the helix η6 is almost perpendicular to the next helix α7 in a way that it entirely covers the cofactor binding site (**Figure 4B**). Moreover, the loop region is curved in a way that Asp204/208 can reach CHA (or PUT) to create hydrogen bond with its

FIGURE 5 | Structural changes of AtSPDS upon ligand binding. The detailed conformational differences between the closed (chain A of AtSPDS1-CHA structure, green) and open (chain H of AtSPDS2 structure, violet) conformations around the cofactor binding site (A) and the polyamine groove (B); red arrows indicate major movements; note that the bound dc-SAM (black) and CHA (pink) are bound only in the chain A of AtSPDS1-CHA; residues are numbered accordingly to the sequence positions of AtSPDS1 and AtSPDS2, respectively.

amine group and to stabilize the substrate during the catalysis. In the second region in the closed state, helix α8 is positioned parallelly to the β13 strand of the core β-sheet. The C-terminus is quite well structured and visible in the electron density map up to Ser334/337. Also, the loop of the N-terminal β-hairpin is positioned close to the active site.

In the open conformation, the biggest conformational difference concerns the region with η6. The preceding loop uncoils and η6 is now moved toward α8, over 10 Å away from the position in the closed form. Therefore, the negatively charged active site is uncovered and ready to incorporate cofactor and substrate (**Figure 4C**). This opening of the active site is possible due to the concerted movement of the other parts of the protein as well. The beginning of α8 helix is shifted almost 6 Å away in comparison to the closed conformation. This shift has an impact not only on the conformation of the preceding loop but also on the C-terminal helix α11, which is also shifted, and it becomes more disordered. Also, the loop of the N-terminal β-hairpin slightly moves outside the pocket (**Figure 4A**) in the open state, which has a serious consequence for the substrate/inhibitor binding (see below). It is worth noting that the opened conformation of the chain H of AtSPDS2 was possible to capture only due to the crystal packing, where the residues created additional H-bonds with a symmetry-related unit in the crystal lattice.

The major transition between open and closed state of AtSPDS around the cofactor binding site (**Figure 5A**) involves Leu212/216 and Pro208/212, residues from the η6 region that are crucial for the dc-SAM stabilization. Additionally, residues in the glycine-rich region change their position to facilitate dc-SAM. Gly129/133 and Gly130/134 alter their conformation, which in consequence moves the Asp131/135 that is now poised to form a hydrogen bond with the amine group of dc-SAM. Simultaneously, the side chain of Gln107/111 rotates to complement the hydrogen-bonding network with dc-SAM.

Most likely, when the η6 is positioned in the closed conformation after the dc-SAM incorporation, the polyamine grove is adapted for substrate/inhibitor binding. Asp204/208 is rotated to form a hydrogen bond with an amine group of bound ligand inside the polyamine grove (**Figure 5B**). Also, together with the movement of α8, Glu236/240 is moved inside the active site. The movement of α8 probably has also the impact on the conformation of Trp55/59, which is pushed inside the cleft to shape the bottom wall of the polyamine grove. Also, when the ligand is bound, the side chain of Glu50/54 rotates to form a water-mediated H-bond with the ligand in the polyamine grove.

The fact that SPDS enzymes require the cofactor to be bound first in the dc-SAM binding site can be explained by the necessity of the gate region with η6 to be stabilized in the closed conformation. This helps to preserve the position of Asp204/208 and Glu236/240 in a way that they can easily create H-bonds with bound substrate or inhibitor. Most likely, in the absence of dc-SAM inside the active site the gate region is too unstable, therefore the ligand inside the polyamine grove cannot be sufficiently stabilized. The search across the PDB shows that the region with η6 in most of the APTs is disordered without the ligand bound inside the dc-SAM binding site. E. coli SPDS (PDB ID 3o4f) (Zhou et al., 2010) is another example, where some chains were captured in open conformation, similarly to chain H of AtSPDS2. On the other hand, such high instability of the η6 without ligands was not observed in MtTSPS (Sekula and Dauter, 2018). Probably, MtTSPS presents a different mechanism to open the catalytic cleft, where the active site may be opened through a relative movement of C-terminal domains with respect to the N-terminal intersubunit β-barrel.

### CONCLUSION

In this work, we have presented the crystal structures of two isoforms of SPDS from A. thaliana, AtSPDS1 and AtSPDS2, and compared the unbound and the bound conformations of these enzymes. The structures show the binding mode of dc-SAM, a universal cofactor of APTs and the donor of the aminopropyl moiety. The AtSPDS1-CHA structure gave insights into the inhibition of the plant SPDSs by CHA. This competitive inhibitor binds inside the polyamine groove of the active site creating three hydrogen bonds at the bottom of the pocket, analogical to these created by the bound substrate. Inside the polyamine grove, the inhibitor is also stabilized by the hydrophobic interactions with two perpendicularly situated Tyr residues, which also stabilize PUT. The crystallographic snapshots show in detail the structural rearrangements around the active site of AtSPDS that are required to facilitate both, the cofactor and the substrate/inhibitor. The protein undergoes concerted movement of the three major parts (i) close to the glycine-rich region where aminopropyl moiety of dc-SAM is bound, (ii) the very flexible gate region with η6, where residues interact with the adenine moiety of dc-SAM and the bound polyamine/inhibitor, and (iii) the N-terminal β-hairpin, that limits the PUT binding grove at the bottom.

### DATA AVAILABILITY

The datasets generated for this study can be found in Protein Data Bank, under the accession codes 6o63 (AtSPDS1), 6o64 (AtSPDS2), and 6o65 (AtSPDS1-CHA).

#### AUTHOR CONTRIBUTIONS

BS planned and performed the experiments, analyzed the results, and wrote the manuscript. ZD analyzed the results and supervised the work.

#### FUNDING

This project was supported in part by the Intramural Research Program of the National Cancer Institute, Center for Cancer Research.

#### ACKNOWLEDGMENTS

fpls-10-00555 May 4, 2019 Time: 16:20 # 11

The authors are grateful to Srinivas Chakravarthy, BioCAT, for the assistance during SAXS experiments and the evaluation of the data; diffraction data were collected at the Advanced Photon Source (APS), Argonne National Laboratory (ANL) at the SER-CAT beamline 22-BM [supported by the United States Department of Energy (DOE), Office of Basic

#### REFERENCES


Energy Sciences, under contract W-31-109-Eng-38]; SAXS research on the 18-ID BioCAT beamline used resources of APS, a DOE Office of Science User Facility operated by ANL (contract DE-AC02-06CH11357), a project supported by grant 9 P41 GM103622 from the National Institute of General Medical Sciences (NIGMS) and use of the PILATUS 3 1M detector was provided by grant 1S10OD018090- 01 from NIGMS.



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Sekula and Dauter. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Polyamine Putrescine Contributes to H2O<sup>2</sup> and RbohD/F-Dependent Positive Feedback Loop in Arabidopsis PAMP-Triggered Immunity

Changxin Liu, Kostadin E. Atanasov, Antonio F. Tiburcio and Rubén Alcázar\*

Department of Biology, Healthcare and Environment, Section of Plant Physiology, Faculty of Pharmacy, University of Barcelona, Barcelona, Spain

#### Edited by:

Vasileios Fotopoulos, Cyprus University of Technology, Cyprus

#### Reviewed by:

Andrés Gárriz, CONICET Institute of Biotechnological Research (IIB-INTECH), Argentina Francisco Marco, University of Valencia, Spain

> \*Correspondence: Rubén Alcázar ralcazar@ub.edu

#### Specialty section:

This article was submitted to Plant Physiology, a section of the journal Frontiers in Plant Science

Received: 10 April 2019 Accepted: 24 June 2019 Published: 16 July 2019

#### Citation:

Liu C, Atanasov KE, Tiburcio AF and Alcázar R (2019) The Polyamine Putrescine Contributes to H2O<sup>2</sup> and RbohD/F-Dependent Positive Feedback Loop in Arabidopsis PAMP-Triggered Immunity. Front. Plant Sci. 10:894. doi: 10.3389/fpls.2019.00894 Polyamines are involved in defense against pathogenic microorganisms in plants. However, the role of the polyamine putrescine (Put) during plant defense has remained elusive. In this work, we studied the implication of polyamines during pathogenassociated molecular pattern (PAMP)-triggered immunity (PTI) in the model species Arabidopsis thaliana. Our data indicate that polyamines, particularly Put, accumulate in response to non-pathogenic Pseudomonas syringae pv. tomato DC3000 hrcC and in response to the purified PAMP flagellin22. Exogenously supplied Put to Arabidopsis seedlings induces defense responses compatible with PTI activation, such as callose deposition and transcriptional up-regulation of several PTI marker genes. Consistent with this, we show that Put primes for resistance against pathogenic bacteria. Through chemical and genetic approaches, we find that PTI-related transcriptional responses induced by Put are hydrogen peroxide and NADPH oxidase (RBOHD and RBOHF) dependent, thus suggesting that apoplastic ROS mediates Put signaling. Overall, our data indicate that Put amplifies PTI responses through ROS production, leading to enhanced disease resistance against bacterial pathogens.

Keywords: polyamines, putrescine, defense, pathogen-associated molecular pattern, reactive oxygen species, PAMP-triggered immunity

### INTRODUCTION

To face against biotic stress, plants have evolved complex and effective defense systems (Dodds and Rathjen, 2010). A first barrier of plant defense is the presence of the cuticle and the cell wall, which act as physical barriers (Yeats and Rose, 2013). However, when pathogens break these preformed barriers, sophisticated mechanisms of pathogen recognition are involved (Bigeard et al., 2015). Plasma membrane pathogen or pattern recognition receptors (PRRs) recognize pathogenassociated molecular patterns (PAMPs) that lead to PAMP-triggered immunity (PTI) (Zipfel and Felix, 2005; Iakovidis et al., 2016). One of the most well-characterized PAMPs is flagellin, a structural component of the flagellum in Gram-negative bacteria. The peptide flagellin22 (flg22) is recognized by the leucine-rich repeat receptor kinase FLS2 (FLAGELLIN SENSING 2) (Felix et al., 1999; Gómez-Gómez and Boller, 2002). Known responses to PTI are the generation of

reactive oxygen species (ROS), cell wall reinforcement by callose deposition, and changes in the expression of defense-related genes (Boller and Felix, 2009; Nicaise et al., 2009; Ahuja et al., 2012). ROS production inhibits pathogen growth, stimulates cell wall cross-linking, and mediates the signal transduction for transcriptional changes (Apel and Hirt, 2004). NADPH oxidases are membrane-bound enzymes important for the generation of ROS during biotic and abiotic stresses, growth, and development. They transfer electrons from cytosolic NADPH or NADH to apoplastic oxygen, producing anion superoxide O<sup>2</sup> <sup>−</sup> in the apoplast, which can be converted to hydrogen peroxide (H2O2) by superoxide dismutase (Kadota et al., 2015). Arabidopsis thaliana (Arabidopsis) carries 10 genes encoding NADPH oxidases, which belong to the RBOH (RESPIRATORY BURST OXIDASE HOMOLOG) family. Among them, RBOHD and, to a lesser extent, RBOHF are required for the generation of apoplastic ROS during incompatible plant–pathogen interactions (Torres et al., 2002). RBOHD is required for cell death control, cell wall damage-induced lignification, and systemic signaling in response to biotic and abiotic stresses (Torres et al., 2005; Miller et al., 2009; Denness et al., 2011). RBOHD and RBOHF finetune the spatial control of ROS production and hypersensitive response (HR) in and around infection sites (Torres et al., 2002, 2005, 2006; Chaouch et al., 2012). In addition to NADPH oxidases, apoplastic ROS can also be originated from polyamine oxidation. Polyamines are small polycationic molecules bearing amino groups. Most abundant plant polyamines are putrescine (Put), spermidine (Spd), and spermine (Spm), and they can be found in free forms or conjugated to hydroxycinnamic acids. Polyamines accumulate in response to different abiotic and biotic stresses and can be oxidatively deaminated by amine oxidases generating H2O<sup>2</sup> (Tiburcio et al., 2014). Based on the cofactor involved, amine oxidases are classified in copper-containing amine oxidases (CuAOs) and FAD-dependent polyamine oxidases (PAOs). CuAOs catalyze the oxidation of Put at its primary amino group, producing 4-aminobutanal along with H2O<sup>2</sup> and NH<sup>4</sup> <sup>+</sup> (Cona et al., 2006; Angelini et al., 2010). In Arabidopsis, PAOs are involved in back-conversion reactions that convert Spm, thermospermine (tSpm), and Spd in their immediate precursors, producing 3-aminopropanal and H2O<sup>2</sup> (Moschou et al., 2012; Ono et al., 2012; Ahou et al., 2014; Kim D.W et al., 2014). Some amine oxidases are located in the apoplast and may function as a source for apoplastic H2O<sup>2</sup> during the elicitation of plant defense. For instance, inoculation of tobacco plants carrying the N resistance gene with tobacco mosaic virus (TMV) triggers HR and the accumulation of Spm in the apoplast (Yamakawa et al., 1998). In this species, Spm activates mitogen-activated protein kinases (MAPKs) SIPK (SAinduced protein kinase) and WIPK (wound-induced protein kinase) (Zhang and Klessig, 1997; Seo et al., 2007) and induces changes in the expression of Spm-responsive genes, some coding for acidic pathogenesis-related proteins (Yamakawa et al., 1998). Also in tobacco, inoculation with the hemibiotrophic bacteria Pseudomonas viridiflava and Pseudomonas syringae pv. tabaci leads to increases in Spm levels in the apoplast, which associate with disease resistance compromised by PAO inhibitors (Marina et al., 2008; Moschou et al., 2009). In Arabidopsis, Spm and its isomer tSpm trigger transcriptional changes that restrict the multiplication of cucumber mosaic virus (CMV) (Mitsuya et al., 2009; Sagor et al., 2012). Also in this species, transgenic plants that accumulate Spm by overexpression of SAMDC1 (S-ADENOSYLMETHIONINE DECARBOXYLASE 1) or SPMS (SPERMINE SYNTHASE) exhibit enhanced disease resistance against P. syringae pv. maculicola ES4326, P. syringae pv. tomato DC3000 (Pst DC3000), and P. viridiflava (Gonzalez et al., 2011; Marco et al., 2014). Overall, the polyamine Spm seems important for the establishment of HR and basal defense responses to hemibiotrophic pathogens in tobacco and Arabidopsis. Conversely, Put has not been observed to have such defense-promoting activities, although its content is remarkably increased in response to pathogens (Yoda et al., 2003; Mitsuya et al., 2009; Sagor et al., 2012; Vilas et al., 2018; Seifi and Shelp, 2019).

In this work, we studied the involvement of polyamines during PTI in Arabidopsis. We report that Put accumulates in response to inoculation with the type three secretor system (TTSS) defective P. syringae DC3000 hrcC mutant strain (hrcC), which induces a strong PTI response (Yuan and He, 1996; Tsuda et al., 2008), and this accumulation is not suppressed by Pst DC3000 type III effectors (Cunnac et al., 2009). Consistent with a potential role for Put during PTI, we show that this polyamine also accumulates in response to flg22, one of the most well-characterized PAMPs. Through the analysis of arginine decarboxylase 1 (adc1) and arginine decarboxylase 2 (adc2) loss-of-function mutants, deficient in Put biosynthesis, we find that the ADC2 isoform is the major contributor to Put biosynthesis triggered by flg22. We show that Put induces the formation of callose deposits, a typical response of PTI, when applied to Arabidopsis seedlings. In addition, we demonstrate that Put quickly induces the expression of several PTI marker genes (Huffaker and Ryan, 2007; Xiao et al., 2007; Denoux et al., 2008; Wang et al., 2009; Boudsocq et al., 2010; Cheng et al., 2013; Po-Wen et al., 2013; Shi et al., 2015), and these transcriptional changes are compromised in the presence of the H2O<sup>2</sup> scavenger dimethylthiourea (DMTU), and in atrbohD, atrbohF, and double atrbohD/F NADPH oxidase lossof-function mutants. We finally report that Put can be regarded as a priming agent that contributes to basal disease resistance against bacterial pathogens. Overall, we provide evidence that Put contributes to H2O<sup>2</sup> and RBOHD/F-dependent positive feedback loop amplification of PTI.

### MATERIALS AND METHODS

#### Plant Materials and Growth Conditions

Plants were grown on soil (peat moss:vermiculite:perlite, 40:50:10) at 20–22◦C under 12-h dark/12-h light cycles at 100– 125 µmol photons m−<sup>1</sup> s <sup>−</sup><sup>2</sup> of light intensity and 70% relative humidity. For in vitro culture, seeds were sterilized with a solution containing 30% sodium hypochlorite supplemented with 0.5% Triton X-100 for 10 min, followed by three washes

<sup>1</sup>www.arabidopsis.info

<sup>2</sup>www.anaspec.com

with sterile distilled H2O. Sterilized seeds were sown on growth media [GM, 1/2 Murashige and Skoog supplemented with vitamins, 1% sucrose, 0.6% plant agar (Duchefa Biochemie), and pH 5.7 adjusted with 1 M KOH]. Plates were kept in the dark at 4◦C for stratification for 2–3 days. Seedlings were grown under 12-h dark/12-h light cycles at 20–22◦C, 100–125 µmol photons m−<sup>2</sup> s <sup>−</sup><sup>1</sup> of light intensity. flg22 peptide was purchased from Anaspec<sup>1</sup> . The fls2 mutant was kindly provided by Jane Parker (Zipfel et al., 2004). The adc1-2 (SALK\_085350), adc2-4 (SALK\_147171), atrbohD (SALK\_109396 and SALK\_005253), atrbohF (SALK\_044584 and SALK\_077748), and double atrbohD/F (N9558) (Torres et al., 2002) mutants were obtained from the Nottingham Arabidopsis Stock Center<sup>2</sup> . The adc1-3 and adc2-3 mutants were previously reported (Cuevas et al., 2008). The gsl5 mutant was kindly provided by Christian Voigt.

### Polyamine Levels Determination

Polyamines were derivatized with dansyl chloride and analyzed by high-performance liquid chromatography (HPLC) as previously described (Marcé et al., 1995; Zarza et al., 2017). All harvested tissues were washed three times with sterile distilled H2O before processing or freezing in liquid nitrogen. Apoplastic polyamines were determined according to Yoda et al. (2009). All polyamine analyses were performed in at least three biological replicates.

### Histochemical Analyses

For aniline blue staining, seedlings were fixed and cleared in a solution of acetic acid/ethanol (1:3) overnight, followed by two washes of 30 min in 150 mM K2HPO<sup>4</sup> and staining with 0.01% aniline blue (Sigma) for 2 h in the same buffer. Observations were performed under an epifluorescence microscope and images were captured with a NIKON microscopy camera coupled to the NIS software 4.45 (NIKON). Callose intensity quantification was performed according to Daudi et al. (2012). Callose intensity was calculated with ImageJ by counting the number of callose spots and assigning a value from 1 to 10 (10, saturated signal; 9, over 250 spots; 8, between 200 and 249 spots; 7, between 150 and 199 spots; 5, between 100 and 149 spots; 3, between 50 and 99 spots; 2, between 5 and 49 spots; 1, between 0 and 5 spots). Average callose measurements were based on at least 20 leaf pictures taken from 12 different seedlings. Trypan blue staining for cell death visualization was performed as previously described (Alcázar et al., 2009).

## Real-Time qPCR Expression Analyses

Total RNA isolated from 10-day-old seedlings was extracted using TRIzol reagent (Thermo Fisher). Two micrograms of RNA was treated with DNAse I (Invitrogen) and first-strand cDNA was synthesized using Superscript IV (Invitrogen) and oligo dT. Quantitative real-time PCR using SYBR Green I dye method was performed on Roche LightCycler 480 II detector system following the PCR conditions: 95◦C for 2 min, 40 cycles (95◦C for 15 s; 60◦C for 30 s). qRT-PCR analyses were always performed on at least three biological replicates with three technical replicates each using ACTIN2 (At3g18780) as the internal control gene.

Relative expression was calculated by 2−11Ct method (Livak and Schmittgen, 2001). Primer sequences used for expression analyses are shown in **Supplementary Table 1**.

## Pseudomonas syringae pv. tomato DC3000 and hrcC Inoculation Assays

Pseudomonas syringae pv. tomato DC3000 (Pst DC3000) and P. syringae pv. tomato DC3000 hrcC (hrcC) bacteria were streaked

on solid NYGA medium (5 g/L bacto peptone, 3 g/L yeast extract, and 20 mL/L glycerol, with 15 g/L agar for solid medium) containing 25 µg/ml rifampicin. Single colonies were transferred to liquid NYGA supplemented with 25 µg/ml rifampicin and grown overnight at 28◦C. Bacterial suspensions were washed two times with water and suspended on 10 mM MgCl<sup>2</sup> to OD<sup>600</sup> = 0.2. Silwet L-77 was added to a final concentration of 0.04% (v/v) before spray inoculation of 3-week-old Arabidopsis plants. Leaves

fpls-10-00894 July 13, 2019 Time: 15:28 # 4

were harvested after 3 h and 72 h of pathogen inoculation for the determination of bacterial growth as described in Alcázar et al. (2010). At least three biological replicates were determined for each time point of analysis.

## RESULTS

### Polyamine Levels in Response to Pst DC3000 and Pst DC3000 hrcC Bacteria

The P. syringae pv. tomato DC3000 hrcC mutant (hrcC) is defective in the TTSS and mainly induces a PAMP-triggered response by failing to secrete defense-suppressing type III effectors into the plant cell (Yuan and He, 1996). In order to analyze the involvement of polyamines during PTI, we inoculated Arabidopsis wild type (Col-0) with hrcC and monitored the levels of free Put, Spd, and Spm for 3 days (**Figure 1**). Oneday post-inoculation, the levels of Put, Spd, and Spm were 2.1-, 1.4-, and 1.7-fold higher in plants inoculated with hrcC than in mock inoculated plants (**Figure 1**). These results indicated that polyamines, and particularly Put, accumulated transiently in response to non-pathogenic hrcC bacteria, thus suggesting the participation of polyamines in the metabolic reprogramming induced during PTI.

In order to determine whether type III effector proteins suppress the changes in polyamine levels observed after hrcC inoculation, we determined Put, Spd, and Spm contents in plants inoculated with P. syringae pv. tomato DC3000 (Pst DC3000), which carries a functional TTSS (**Figure 1**). Compared to mocks, the Put levels increased up to 1.6- and 2.7-fold 1 and 2 days after inoculation with Pst DC3000, respectively. Spd and Spm levels also increased up to 1.4- and 1.6-fold 1 day post-inoculation. These results indicated that type III effectors delivered by Pst DC3000 do not suppress increases in polyamine levels triggered by hrcC. Rather, Put accumulation was higher in the strain provided with a functional TTSS.

### Determination of Apoplastic Polyamines

Some polyamines have been reported to accumulate in the apoplast of Arabidopsis, tobacco, tomato, and rice during defense (Yoda et al., 2009; Vilas et al., 2018). Under basal conditions (0 h), the levels of free polyamines in the apoplastic enriched fractions were undetectable. However, apoplastic Put and Spd contents remarkably increased after 24 h of Pst DC3000 and hrcC inoculation. The levels of Put remained high in Pst DC3000 but not in hrcC inoculated plants. Apoplastic Spm was not detectable in any treatment (**Supplementary Figure S1**). We concluded that Put and Spd accumulate in the apoplast in response to Pst DC3000 and hrcC inoculation. These data suggested that polyamines could trigger some defense response from the cell surface against bacterial infection.

### Polyamine Levels in Response to flg22

To further investigate the involvement of polyamines during PTI, we analyzed polyamine levels in response to the PAMP flg22. Free Put, Spd, and Spm levels were determined in wild type and fls2 seedlings treated with 1 µM flg22 or mock (**Figure 2**). In the wild type, Put accumulated up to twofold in response to 1 µM flg22 treatment after 24 h. This increase was not evidenced in the fls2 mutant (**Figure 2**), which indicated that Put accumulation

triggered by flg22 was due to FLS2-dependent activation of PTI. The levels of Spd and Spm in seedlings treated with 1 µM flg22 did not exhibit significant changes compared with the mock control (**Figure 2**). Therefore, flg22 did not favor the synthesis or accumulation of Spd and Spm. However, increases in these polyamines were detected after 24 h of inoculation with Pst DC3000 and hrcC bacteria (**Figure 1**). We suggest that other molecules produced by P. syringae (Xin and He, 2013) and perceived by the plant might trigger the synthesis of Spd and Spm in Arabidopsis.

### Involvement of ADC Isoforms in Put Biosynthesis Triggered by flg22

Arginine decarboxylase (ADC) catalyzes the conversion of arginine into agmatine, which is a limiting step in the biosynthesis of Put. In Arabidopsis, two ADC isoforms are found (ADC1 and ADC2) that catalyze the same enzymatic reaction (Alcázar et al., 2006). To analyze the contribution of each isoform to Put synthesis in response to flg22, we treated arginine decarboxylase 1 (adc1-2, adc1-3) and arginine decarboxylase 2 (adc2-3, adc2- 4) loss-of-function mutants (Cuevas et al., 2008) with 1 µM flg22 and quantified the polyamine levels between 0 and 24 h (**Figure 2**). In adc2-3 and adc2-4, the basal level of Put was much lower than in the wild type (Cuevas et al., 2008) and Put content did not increase in response to 1 µM flg22. Conversely, in adc1-2 and adc1-3, Put content increased to a similar extent as the wild type in response to 1 µM flg22 (**Figure 2**). These results indicated that Put accumulation in response to flg22 is mainly contributed by ADC2 activity. Therefore, ADC1 and ADC2 forms do not act redundantly during PTI.

### Callose Deposition but Not Cell Death Is Induced by Put

The increases in Put triggered by flg22 perception prompted us to investigate its potential role during PTI. Deposition of the (1,3)-β-glucan callose is induced in response to flg22, and

it can be visualized by histochemical analysis based on aniline blue staining. We observed higher callose deposition in wildtype seedlings treated for 24 h with 100 µM Put or 1 µM flg22 than in seedlings treated with mock (**Figure 3**). Callose deposition induced by Put was compromised in the glucan synthase like 5 (gsl5) mutant, which is defective in inducible callose accumulation upon wounding and biotic stress (Jacobs et al., 2003) (**Figure 3**). Conversely, callose deposition in response to flg22 was not obviously compromised in adc1 or adc2 mutants (**Supplementary Figure S2**). This indicated that flg22 responses are not impaired in adc mutants. To determine whether callose deposition triggered by Put was accompanied with cell death, we performed trypan blue staining in wild-type seedlings after 24 h of infiltration with 100 µM Put or mock (**Supplementary Figure S3**). Trypan blue staining did not reveal evident symptoms of cell death related with ETI in Arabidopsis leaves treated with 100 µM Put. These data indicated that Put infiltration does not induce HR in Arabidopsis.

### Expression of PTI Marker Genes in Response to Put

Accumulation of callose by Put suggested that PTI responses were activated by this polyamine. To further investigate this hypothesis, we analyzed the expression of several PTI marker genes (PROPEP2, PROPEP3, CBP60g, WRKY22, WRKY29, WRKY53, CYP81F2, FRK1, and NHL10) (Huffaker and Ryan, 2007; Xiao et al., 2007; Denoux et al., 2008; Wang et al., 2009; Boudsocq et al., 2010; Cheng et al., 2013; Po-Wen et al., 2013; Shi et al., 2015) in wild-type seedlings treated with 100 µM Put or mock between 0 and 72 h (**Figure 4**). For most of the genes analyzed, their transcripts increased rapidly in response to 100 µM Put, with the highest expression peaks observed upon 10 min to 1 h of treatment (**Figure 4**). These results indicated that Put induces transcriptional changes consistent with activation of PTI. Because Put can be oxidized by amine oxidases, we then studied whether transcriptional responses were compromised in the presence of the H2O<sup>2</sup> scavenger dimethylthiourea (DMTU). For this, we determined the expression of WRKY29, PROPEP2, PROPEP3, and CYP81F2 (Huffaker and Ryan, 2007; Denoux et al., 2008; Cheng et al., 2013) in wild-type seedlings treated or not with 100 µM Put in the presence of 10 mM DMTU (**Figure 5**). The increase in the transcript levels of these genes triggered by Put was compromised in the presence of DMTU (**Figure 5**). We concluded that H2O<sup>2</sup> production is required for Put-triggered transcriptional up-regulation of PTI marker genes.

### Expression of PTI Marker Genes in Response to Put in atrboh D, atrboh F, and atrboh D/F Mutants

Plasma membrane RBOHD and RBOHF are important sources of ROS during plant–pathogen interactions (Kadota et al., 2015). To determine the contribution of these NADPH oxidases to changes in the expression of PTI marker genes induced by Put, we analyzed the expression of WRKY22 and CYP81F2 in atrbohD (SALK\_109396C and SALK\_005253C), atrbohF (SALK\_034674 and SALK\_077748), and double atrbohD/F loss-of-function mutants (Torres et al., 2002) treated with 100 µM Put or mock (**Figure 6**). In contrast with the wild type, up-regulation of WRKY22 and CYP81F2 expression by Put treatment was strongly compromised in atrbohD, atrbohF, and double atrbohD/F mutants (**Figure 6**). These results indicated that Put requires functional RBOHD and RBOHF NADPH oxidases for signaling.

### Disease Resistance to P. syringae pv. tomato DC3000 and hrcC in Put Treated Plants

So far, our data pointed to a role for Put contributing to amplify PTI responses. To analyze how this was translated into disease resistance, we performed pathoassays using Pst DC3000 and hrcC bacteria in wild-type plants treated with 500 µM Put, 1 µM flg22 or mock. As shown in **Figure 7**, Put treatment limited the growth of Pst DC3000 to a similar extent as 1 µM flg22, whereas no differences were detected by inoculation with the non-pathogenic hrcC strain. We concluded that Put could be regarded as a priming agent contributing to basal defense responses against some pathogenic bacteria.

## DISCUSSION

Plants are provided with an innate immune system that recognizes pathogens and activates defense responses. A first layer of the innate immunity involves the recognition of PAMPs, which are conserved signatures within a taxonomic group of pathogens. PAMPs include the flagellin peptide flg22, the elongation factor Tu (EF-Tu) peptides elf18 and elf26, lipopolysaccharides, fungal chitin, and peptidoglycan, among others (Boller and Felix, 2009). PAMPs induce the

production of ROS, which participate in defense signaling and transcriptional reprogramming (Bigeard et al., 2015). During defense, ROS are predominantly generated by NADPH oxidases RBOHD and RBOHF (Torres et al., 2002; Kadota et al., 2015). However, other sources of apoplastic ROS include the activity of apoplastic peroxidases (Daudi et al., 2012) and amine oxidases (Cona et al., 2006). The different sources of ROS might be related to the necessity of specific ROS synthesis at different stages of the defense response (Cona et al., 2006). In Arabidopsis, the copper-containing amine oxidases (CuAO) ATAO1/AtCuAOβ (At4g14940) and CuAO1/AtCuAOγ1 (At1g62810) have been localized in the apoplast (Moller and McPherson, 1998; Planas-Portell et al., 2013), whereas PAO enzymes have been found in the cytosol and peroxisomes (Tavladoraki et al., 2006; Kamada-Nobusada et al., 2008; Moschou et al., 2008; Takahashi et al., 2010; Ahou et al., 2014; Tiburcio et al., 2014). The apoplastic CuAOs preferentially catalyze the oxidation of Put (ATAO1) or Put and Spd (CuAO1) (Moller and McPherson, 1998; Planas-Portell et al., 2013), consistent with the occurrence of these polyamines in extracellular fluids (Yoda et al., 2009). Interestingly, CuAO1 expression is induced by flg22 treatment (Planas-Portell et al., 2013), which suggests its participation in PAMP-triggered ROS signaling. The involvement of CuAO activities in the defense response of incompatible plant–pathogen interactions has previously been documented. In the incompatible interaction between barley and the powdery mildew fungus B. graminis f. sp. hordei, the levels of Put, Spd, as well as diamine oxidase and PAO activities were shown to increase and to contribute to defense through H2O<sup>2</sup> production, leading to

cell wall cross-linking of polysaccharides and proteins (Cowley and Walters, 2002). In chickpea, inhibition of CuAO activity was associated with decreased defense capacity against the necrotrophic fungus Ascochyta rabiei (Rea et al., 2002). The amount of free polyamines in the apoplast seems to be a limiting factor for CuAO activity (Rea, 2004). Indeed, it has been proposed that under stress conditions, polyamine excretion is activated in plant cells (Yoda et al., 2003). Consistent with this, the levels of apoplastic Put and Spd increase in response to avirulent Pst DC3000 AvrRPM1 inoculation in Arabidopsis (Yoda et al., 2009).

Despite the growing body of evidence that shows the involvement of polyamines in defense, few studies have focused on the involvement of polyamines during PTI. In this work, we show that Put synthesis is stimulated by Pst DC3000 hrcC inoculation (**Figure 1**), a TTSS defective bacteria strain that mainly triggers a PTI response by failing to secrete effectors (Yuan and He, 1996). Consistent with this, Put level also increased by treatment with the purified PAMP flg22 (**Figure 2**). These data suggested that polyamines are part of the metabolic reprogramming response during PTI. Interestingly, inoculation with Pst DC3000, which carries a functional TTSS and can deploy effectors into the plant cell (Xin and He, 2013), did not suppress the increase in polyamine levels observed with hrcC. Rather, polyamine levels became higher (**Figure 1**). These results indicate that Pst DC3000 effectors are unlikely to suppress polyamine pathway activation. Rather, effectors might be promoting agents in polyamine biosynthesis. For example, the ADC1 isoform from Capsicum annuum is targeted by the AvrBsT effector from Xanthomonas campestris pv. vesicatoria. Their co-expression in Nicotiana benthamiana leaves promotes polyamine biosynthesis, thus leading to enhanced cell death and H2O<sup>2</sup> production (Kim N.H. et al., 2013). However, it is not known whether Arabidopsis ADC isoforms might be targets of bacterial effectors. In Arabidopsis, the ADC2 isoform is the major contributor to Put synthesis in response to flg22 (**Figure 2**). Consistent with this, Kim S.H. et al. (2013) showed that the adc2 mutant (SALK\_073977) in Arabidopsis compromises resistance to Pst DC3000, which can be rescued by infiltration with 2 µM Put.

The Put accumulation triggered by flg22 and hrcC prompted us to investigate the role of this polyamine during PTI. Interestingly, we found that exogenously supplied Put induces callose deposition in Arabidopsis seedlings (**Figure 3**). The formation of callose deposits is a typical physiological response of PTI. Callose is synthesized at the cell wall by callose synthases. The Arabidopsis genome contains 12 callose synthase (CalS) genes, also referred to as Glucan synthase-like (GSL) (Ellinger and Voigt, 2014). Among them, GSL5 (PATHOGEN MILDEW RESISTANCE 4, PMR4) is required for wound and papillary callose deposition (Jacobs et al., 2003). We found that callose deposition induced by Put supply was compromised in the gsl5 (pmr4) loss-of-function mutant (Jacobs et al., 2003) (**Figure 3**). To further investigate the involvement of Put during PTI, we selected a number of PTI marker genes based on previous reports (Huffaker and Ryan, 2007; Xiao et al., 2007; Wang et al., 2009; Boudsocq et al., 2010; Cheng et al., 2013; Po-Wen et al., 2013; Shi et al., 2015). Exogenously supplied Put rapidly led to the up-regulation of PTI marker genes tested (**Figure 4**). Interestingly, such responses were suppressed in the presence of the H2O<sup>2</sup> scavenger, DMTU (Tate et al., 1982) (**Figure 5**). Hydrogen peroxide is likely derived from amine oxidase activity, thus pointing to an important role for polyamine oxidation during the transcriptional response triggered by Put. Interestingly, up-regulation of PTI marker genes was also compromised in atrbohD, atrbohF, and double atrbohD/F loss-of-function mutants (**Figure 6**). These data indicate that plasma membrane NADPH oxidases are required for at least some transcriptional responses induced by Put. In tobacco, the NADPH oxidases RBOHD/F have been suggested to act upstream of apoplastic PAO during salt stress, contributing to cell death (Gémes et al., 2016). Our data indicate that Arabidopsis RBOHD/F are

downstream of Put or act in a concerted manner with apoplastic CuAOs during PTI. Collectively, we observed that PAMPs (flg22) induce Put biosynthesis and that Put triggers responses compatible with PTI activation, which are ROS and RBOHD/F dependent. Hence, a positive feedback loop is proposed in which Put amplifies PAMPtriggered signaling through ROS production, leading to enhanced basal disease resistance against bacterial pathogens (**Figure 7**). In this regard, apoplastic Put could act similarly to damage-associated molecular patterns (DAMPs) triggering a ROS-dependent defense response (Choi and Klessig, 2016; Versluys et al., 2017).

Collectively, our results gain insight into mechanistic processes by which polyamines contribute to disease resistance in plants. Such type of analyses should contribute to pave the road for the uses of polyamines as potential priming agents in agriculture.

#### DATA AVAILABILITY

All datasets generated for this study are included in the manuscript and/or the **Supplementary Files**.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

CL and KA performed the research. CL and RA planned the experiments. CL, KA, AT, and RA analyzed the data. RA wrote the manuscript.

#### FUNDING

This work has been supported by the BFU2017-87742-R grant of the Programa Estatal de Fomento de la Investigación Científica y Técnica de Excelencia (Ministerio de Economía y Competitividad, Spain). CL acknowledges support from the CSC (China Scholarship Council) for funding his doctoral fellowship. Project financed by the Agencia Estatal de Investigación (AEI) and the Fondo Europeo de Desarrollo Regional (FEDER).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00894/ full#supplementary-material

intracellular oxidative stress and pathogenesis responses in Arabidopsis. Plant J. 69, 613–627. doi: 10.1111/j.1365-313X.2011.04816.x



pathogenesis-related proteins and resistance against tobacco mosaic virus infection. Plant Physiol. 118, 1213–1222. doi: 10.1104/PP.118.4.1213


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Liu, Atanasov, Tiburcio and Alcázar. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

fpls-10-00894 July 13, 2019 Time: 15:28 # 12

# Spermine Differentially Refines Plant Defense Responses Against Biotic and Abiotic Stresses

Hamed Soren Seifi and Barry J. Shelp\*

Department of Plant Agriculture, University of Guelph, Guelph, ON, Canada

Roles of the major polyamines (mPA), putrescine, spermidine, and spermine (Spm), in various developmental and physiological processes in plants have been well documented. Recently, there has been increasing focus on the link between mPA metabolism and defense response during plant-stress interactions. Empirical evidence is available for a unique role of Spm, distinct from the other mPA, in eliciting an effective defense response to (a)biotic stresses. Our understanding of the precise molecular mechanism(s) by which Spm modulates these defense mechanisms is limited. Further analysis of recent studies indicates that plant Spm functions differently during biotic and abiotic interactions in the regulation of oxidative homeostasis and phytohormone signaling. Here, we summarize and integrate current knowledge about Spm-mediated modulation of plant defense responses to (a)biotic stresses, highlighting the importance of Spm as a potent plant defense activator with broad-spectrum protective effects. A model is proposed to explain how Spm refines defense mechanisms to tailor an optimal resistance response.

Edited by:

Rubén Alcázar, University of Barcelona, Spain

#### Reviewed by:

Thomas Berberich, Senckenberg Nature Research Society, Germany Andrés Gárriz, Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Argentina

> \*Correspondence: Barry J. Shelp bshelp@uoguelph.ca

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 18 December 2018 Accepted: 23 January 2019 Published: 08 February 2019

#### Citation:

Seifi HS and Shelp BJ (2019) Spermine Differentially Refines Plant Defense Responses Against Biotic and Abiotic Stresses. Front. Plant Sci. 10:117. doi: 10.3389/fpls.2019.00117 Keywords: abiotic stress, biotic stress, defense response, defense activator, signaling, spermine

#### INTRODUCTION

Polyamines are ubiquitous, small aliphatic polycations found in eukaryotic organisms. The major polyamines (mPA) in plants are the diamine putrescine (Put), the triamine spermidine (Spd) and the tetraamine spermine (Spm). They function in key developmental and physiological events such as embryogenesis, cell division, floral initiation, senescence and responses to stress (Evans and Malmberg, 1989; Galston and Sawhney, 1990). The biosynthesis and degradation of mPA are highly responsive to environmental stimuli (Liu et al., 2007). Several studies have reported that the three mPA mold plant responses to (a)biotic stresses (Bouchereau et al., 1999; Walters, 2000, 2003a,b; Urano et al., 2003; Alcázar et al., 2010; Minocha et al., 2014; Romero et al., 2018). However, there is evidence for the differential regulation of Spm/Spd and Put by stresses (see Shelp et al., 2018), and for a unique role of Spm, distinct from the other mPA, in the induction and formation of resistance responses to various types of (a)biotic stresses. For instance, Mitsuya et al. (2009) reported that Spm is the only mPA that effectively suppresses the multiplication of cucumber mosaic virus in Arabidopsis. Other research indicates that Spm strongly induces different defense-related genes in Arabidopsis seedlings, whereas similar doses of Put and Spd do not, and elevated levels of endogenous Spm are causally linked to higher tolerance to the bacterial pathogen Pseudomonas syringae and the oomycete Hyaloperonospora arabidopsidis (Marco et al., 2014). Similarly, among the mPA, only Spm strongly induces the two key defense-associated signaling

molecules, nitric oxide and hydrogen peroxide (H2O2), in Nicotiana benthamiana, ultimately leading to resistance to the bacterial pathogen Xanthomonas campestris (Kim et al., 2013). An Arabidopsis mutant deficient in Spm biosynthesis exhibits hypersensitivity to salt and drought stresses, and the phenotype is mitigated by exogenous Spm, but not Put or Spd (Yamaguchi et al., 2006; Kusano et al., 2007). Together, these findings suggest that Spm is a stress-associated signaling molecule (Yamakawa et al., 1998) due to its unique role in inducing several components of the plant defense response, including: (i) genes coding for pathogenesis related (PR) and resistance (R) proteins (Yamakawa et al., 1998; Gonzalez et al., 2011); (ii) mitogen-activated protein kinases (MAPK) (Takahashi et al., 2003; Gonzalez et al., 2011); (iii) several defense-associated transcription factors (Mitsuya et al., 2009; Gonzalez et al., 2011); (iv) phytoalexin biosynthesis (Marco et al., 2014; Mo et al., 2015); and, (v) the hypersensitive response (HR) (Takahashi et al., 2004; Sagor et al., 2009). In this review, we summarize and integrate current knowledge on Spmmediated refinement of plant defense responses to both biotic and abiotic stresses, and highlight the importance of Spm as a potent plant defense activator with broad-spectrum effects. In addition, a model is proposed to explain how Spm regulates various oxidative and hormone signaling pathways, which tailor an optimal defense response to various external stresses.

#### Spm Metabolism in Plants

Spm anabolism in plants involves two main routes (Shelp et al., 2012). The first is catalyzed by ornithine decarboxylase, which converts ornithine into Put, the main precursor for Spm biosynthesis. The second is a three-step pathway in which arginine is converted to agmatine by arginine decarboxylase, and then agmatine is converted to Put by agmatine imidohydrolase and carbamoylputrescine amidohydrolase. Put is then successively converted to Spd by Spd synthase, and then to Spm by Spm synthase. The latter reactions require the addition of aminopropyl groups, supplied from decarboxylated S-adenosylmethionine (SAM), which is a product of SAM decarboxylase (SAMDC). Spm catabolism involves flavincontaining PA oxidases (PAO), which catalyze two types of reactions, terminal oxidation and back-conversion. The terminal oxidation of Spm generates 4-N-(3-aminopropyl)-4 aminobutanal, 1,3-diaminopropane and H2O2. Alternatively, the back-conversion reaction converts Spm to Spd, and Spd to Put, resulting in the production of 3-aminopropanal and H2O2.

#### SPM METABOLISM AND BIOTIC STRESSES

#### Spm Induces Oxidative Response

The HR reaction is defined as a type of rapid programmed cell death, which is induced by the generation of reactive oxygen species (ROS, such as H2O2) at the site of pathogen entry, leading to activation of several defense mechanisms that result in cessation of growth of the pathogen, typically biotrophic, and in protection of remaining plant tissue (Govrin and Levine, 2000; Jones and Dangl, 2006). It is generally believed that the HR reaction is effective against biotrophic pathogens only, but effectiveness of HR against necrotrophic pathogens such as Botrytis cinerea has also been reported (Asselbergh et al., 2007; Azami-Sardooei et al., 2010, 2013; Seifi et al., 2013). HR induction involves two major pathways: the host HR is mediated through specific recognition of certain microbes by the surveillance system of the host, namely R proteins (Keen, 1990); and, the non-host HR is non-specific, typically induced in response to a broad spectrum of pathogens in many plants (Heath, 2000). Interestingly, Yoda et al. (2003, 2009) demonstrated that PAOmediated Spm oxidation strongly contributes to the onset of both host and non-host HRs triggered in tobacco plants by different pathogens, highlighting the importance of Spm catabolism in the regulation of the HR-dependent defense response.

Exogenous Spm induces the expression of several H2O2 dependent signaling components and transcription factors in Arabidopsis leaves, and results in HR-mediated resistance to cucumber mosaic virus (Mitsuya et al., 2009). The addition of a PAO inhibitor represses the activation of defense genes and alleviates ROS generation and HR, confirming that PAO is involved in the resistance response. Infiltration of tobacco leaf disks with Spm strongly decreases the growth of the biotrophic bacterial pathogen Pseudomonas viridiflava, but not the necrotrophic fungal pathogen, Sclerotinia sclerotiorum, and co-infiltration of Spm and a PAO inhibitor reverses this protective effect (Marina et al., 2008). Exogenous application of thermospermine, a structural isomer of Spm, induces resistance to P. viridiflava in Arabidopsis through PAO-mediated thermospermine oxidation (Marina et al., 2013). Apoplastic Spm accumulates in tobacco plants in response to infection by the (hemi)biotrophic bacterial pathogen P. syringae pv. tabaci, and PAO overexpression upregulates defense-related marker genes and cell wall-based defense responses, resulting in disease tolerance (Moschou et al., 2009). Similarly, overexpression of a cotton-derived PAO in Arabidopsis results in elevated levels of ROS and resistance to the necrotrophic vascular wilt fungus Verticillium dahlia (Mo et al., 2015). The resistance response is mainly mediated by the induction of MAPK and cytochrome P450, culminating in the accumulation of the Arabidopsisspecific phytoalexin camalexin (Mo et al., 2015). Exogenous Spm increases the disease resistance of Arabidopsis against P. viridiflava, which is compromised by the PAO inhibitor SL-11061 (Gonzalez et al., 2011). Together, these findings suggest that PAO is a key defense regulator, particularly in response to apoplastically-localized plant pathogens.

#### Mitochondrion Membrane Dysfunction

Spm induces apoptosis, a type of programmed cell death, in animal cells through the activation of a group of cell-deathinducing pathways, known as the caspase cascade, which entails the loss of mitochondrial membrane potential and leakage of electron-transfer-chain intermediates, such as cytochrome c, into the cytosol (Moffatt et al., 2000; Stefanelli et al., 2000). Similarly, plant mitochondria are known to play an important role in ROS generation and induction of HR during plantpathogen interactions (Lam et al., 2001; Hatsugai et al., 2004; Van Breusegem and Dat, 2006). Notably, exogenous Spm induces

TABLE 1 | Defense mechanisms associated with Spm-induced resistance against biotic and abiotic stresses.


Abbreviations: ABA, abscisic acid; AOS, allene oxide synthase; ASA, ascorbic acid; CAT, catalase; ERF, ethylene responsive factors; HR, hypersensitive response; GR, glutathione reductase; GSH, glutathione; GST, glutathione S-transferase; JA, jasmonic acid; LOX, lipoxigenase; MAPK, mitogen-activated protein kinase; MMD, mitochondrion membrane dysfunction; PAO, polyamine oxidase; PR, pathogenesis related; R, resistance; ROS, reactive oxygen species; SA, salicylic acid, SOD, superoxide dismutase; TF, transcription factor.

mitochondrial membrane dysfunction (Takahashi et al., 2003) and the expression of two important defense-associated MAPK, which in turn induce a subset of HR-related genes such as HSR203J (Takahashi et al., 2004). Pre-treatment with bongkrekic acid, an inhibitor of the mitochondrial permeability transition pore, suppresses the induction of HR-related genes, confirming that mitochondrial dysfunction is involved in Spm-induced HR in tobacco leaves (Takahashi et al., 2004).

#### Hormonal Regulation

fpls-10-00117 February 7, 2019 Time: 2:39 # 4

Several HR marker genes, such as HSR203J, are responsive to Spm, suggesting that it is involved in HR induction (Takahashi et al., 2004). These HR markers are also induced in NahG plants, which are highly deficient in the plant hormone salicylic acid (SA), suggesting that Spm-induced HR reaction is independent of the SA signaling pathway (Takahashi et al., 2004). This result is consistent with SA-independent, Spm-induced expression of PR proteins in tobacco (Yamakawa et al., 1998). However, several reports propose a link between JA-associated defense responses and Spm metabolism. For instance, exogenous Spm promotes JA biosynthesis in lima bean (Ozawa et al., 2009), and Spm synthaseoverexpressing plants of Arabidopsis have elevated levels of endogenous Spm (two to threefold), resistance to the bacterial pathogen P. viridiflava, and expression of components of the JA-dependent defense signaling pathway such as ERF and Myb transcription factors (Gonzalez et al., 2011). Similarly, elevated levels of endogenous Spm in SAMDC-overexpression lines of Arabidopsis are associated with resistance to Hyaloperonospora arabidopidis and P. syringae and the induction of several defenseassociated genes, such as PR and R proteins, as well as genes involved in JA biosynthesis, such as chloroplastic lipoxygenase and allene oxide synthase (Marco et al., 2014). Collectively, these findings suggest that JA signaling positively regulates Spmmediated defense response to biotic stresses.

### SPM METABOLISM AND ABIOTIC STRESSES

#### Spm Activates Antioxidant Response

Elevated levels of endogenous Spm, as well as the exogenous application of Spm, induce tolerance to various abiotic stresses (Capell et al., 2004; Yamaguchi et al., 2006; Kusano et al., 2007). Fruits of the drought-tolerant tomato cultivar Zarina have elevated levels of endogenous Spm and activities of the antioxidant enzymes superoxide dismutase (SOD) and catalase (CAT), culminating in better tolerance to dehydrationinduced oxidative stress (Sánchez-Rodríguez et al., 2016). Similarly, Spm application is associated with higher activities of SOD and CAT in pea plants, mitigating high-temperatureinduced chlorophyll degradation (Todorova et al., 2016). Also, Spm induces the tolerance of mung bean seedlings to high temperature, drought or cadmium toxicity, and this is typically associated with elevated activities of SOD, CAT, glutathione S-transferase (GST) and glutathione reductase (GR), and levels of non-enzymatic antioxidants such as ascorbic acid and glutathione (GSH), culminating in reduced ROS accumulation (Nahar et al., 2016a,b). The application of Spm to wheat leaves alleviates oxidative damage caused by cadmium and copper excess, reduces the metal-induced ROS accumulation, and restores GR activity (Groppa et al., 2007). Likewise, Spm application to soybean leaves reduces osmotic-stressinduced losses in chlorophyll, carotenoid and protein levels, and increases the activities of CAT and SOD (Radhakrishnan and Lee, 2013). Stress tolerance, elevated activities of CAT, SOD and peroxidases, and elevated expression of heat shock proteins are found in Spm-treated seedlings of trifoliate orange exposed to combined drought and heat stresses (Fu et al., 2014). Together, this body of evidence suggests that Spm induces tolerance to oxidative stress caused by abiotic stresses through the activation of both non-enzymatic and enzymatic antioxidant pathways.

### Hormonal Regulation

It has previously been shown that exogenous abscisic acid (ABA) upregulates expression of the mPA biosynthesis genes SAMDC and arginine decarboxylase (Urano et al., 2003), and

FIGURE 1 | Model for interaction of Spm with plant responses to (a)biotic stresses. Lines ending in arrowheads and closed circles, respectively, indicate positive and negative impacts. Biotic Stress: OR, oxidative response; HR, hypersensitive response; MMD, mitochondrion membrane dysfunction; PAO, polyamine oxidase; JA, jasmonic acid; LOX, lipoxygenase; AOS, allene oxide synthase; MAPK, mitogen activated protein kinase; ERF, ethylene responsive factor; PR, pathogenesis related; R: resistance. Abiotic Stress: = AR, antioxidant response; ASA, ascorbic acid; GSH, glutathione; ABA, abscisic acid; ABF, abscisic acid-binding factor; SOD, superoxide dismutase; GR, glutathione reductase; CAT, catalase; GST, glutathione S-transferase; HSP, heat shock protein.

the induction of these genes is significantly compromised in ABA-deficient mutants of Arabidopsis grown under drought stress (Alcázar et al., 2006a), suggesting a positive correlation between mPA biosynthesis and ABA-mediated response to cold, salt and drought stresses (Alcázar et al., 2010). Such a premise is supported by the existence of several abiotic stress-responsive elements (motifs), as well as, ABA-responsive elements in the promoters of mPA biosynthesis genes (Alcázar et al., 2006b). Notably, Spm treatment induces the expression of ABAresponsive element binding factors in trifoliate orange seedlings challenged by drought and heat stresses (Fu et al., 2014). Hence, crosstalk between Spm-mediated defense response to abiotic stresses and ABA-dependent signaling pathway is suggested.

#### CONTRASTING ROLES OF SPM DURING OXIDATIVE/ANTIOXIDANT RESPONSES

Many of the key reports on Spm-induced resistance discussed above are summarized in **Table 1**. Examination of the biochemical, transcriptional and molecular responses to (a)biotic stresses leads us to hypothesize dual roles for Spm in modulating the oxidative status of the plant cell. Spm seems to accumulate in response to both biotic and abiotic stresses, but this is followed by two different scenarios: (i) upon perception of biotic challenges, Spm "enhances" the oxidative response through the induction of ROS generation and HR: and (ii) upon perception of abiotic challenges, Spm "alleviates" oxidative damage through the stimulation of ROS-scavenging enzymes, leading to an antioxidant response. **Figure 1** depicts the different players involved in the two scenarios. How a plant adopts such contrasting mechanisms in order to tailor an appropriate defense response merits further consideration.

The oxidative response occurs immediately after successful recognition of the pathogen by the plant's surveillance system, following a biphasic pattern (Wojtaszek, 1997). Phase-I consists of a rapid, transient, and low-amplitude burst of ROS generation, occurs within minutes after pathogen recognition, and is known to function as an upstream trigger of several defense-related signaling cascades. Phase-II occurs after few to several hours post recognition, consists of a sustained wave of ROS generation/accumulation of much higher amplitude, and plays a key role in inducing defense-associated genes and HR (Van Camp et al., 1998; De Gara et al., 2003; Torres et al., 2006). While the oxidative response to avirulent pathogens, successfully recognized by the plant's immune system, generally exhibits a biphasic pattern of ROS accumulation, only phase-I is elicited in response to virulent pathogens that are able to avoid host recognition (Torres et al., 2006). With this in mind, it seems that PAO-mediated ROS generation (i.e., Spm oxidation) during incompatible plant-pathogen interactions exhibits the characteristics of a phase-II oxidative response, as previously proposed (Takahashi et al., 2004). Therefore, it can be posited that Spm oxidation under such conditions is not merely a metabolic feedback mechanism to maintain PA homeostasis, but beyond that, it functions as an important part of the plant immune system to provide the ROS necessary to fuel successful activation of defense genes and formation of HR.

The role of mPA as protective molecular chaperones (Jiménez-Bremont et al., 2014) might explain how Spm induces an antioxidative state in the plant tissue in response to abiotic stresses. The spatial separation of positive charges in PA at physiological pH could enable PA to bind negativelycharged molecules such as nucleic acids, phospholipids and proteins, thereby protecting the structure and function of these macromolecules from degradation and modification (Ruiz-Herrera et al., 1995; Martin-Tanguy, 2001; D'Agostino et al., 2005). This property would also enable the scavenging of free radicals and stabilization of intracellular membranes under stress conditions (Popovic et al., 1979; Groppa and Benavides, 2008; Alcázar et al., 2010; Radhakrishnan and Lee, 2013). This might also explain why mPA are abundant in green, young and actively growing tissues, whereas their titers dramatically decline in senescing organs (Galston and Sawhney, 1990; Del Duca et al., 2000). Considering that Spm contains four nitrogen groups, it could provide greater buffering capacity than Spd and Put (Shi et al., 2010). This is in agreement with previous studies that report exogenous Spm, unlike Spd and Put, has a potent anti-senescence effect on oat and lettuce leaves, as well as Jerusalem artichoketuber (Galston and Sawhney, 1990; Dondini et al., 2003; Serafini-Fracassini et al., 2010). Notably, elevated levels of Spm in an Arabidopsis mutant that lacks the PA back-conversion pathway, are associated with delayed dark-induced senescence, suggesting that Spm is a metabolic defense mechanism against senescence-induced oxidative stress and cell death (Sequera-Mutiozabal et al., 2016).

### CONCLUDING REMARKS

Many natural and synthetic compounds are known to activate defense responses against a certain type of stress only, either biotic or abiotic. Those that confer protection against a wide range of both biotic and abiotic stresses are very rare, with silicon being an important exception (Van Bockhaven et al., 2013). In light of the empirical evidence reviewed above, it seems that Spm can be considered as another exceptional molecule with broad spectrum prophylactic effects against both types of stresses. Such effects are exerted through different passive (attributed to the physical and biochemical properties of Spm) and active (attributed to molecular functions of Spm) mechanisms. Given that Spm refines the defense response according to the biotic or abiotic nature of the stress by (i) promoting appropriate hormone-mediated signaling pathways, (ii) modulating oxidative/antioxidant responses, and (iii) inducing several defense-related genes (**Figure 1**), the notion that Spm functions as a plant defense activator becomes more plausible. Nevertheless, several important questions remain regarding these mechanisms. What are the nodes of convergence between Spm-induced signaling pathway and ABA/JA-mediated defense response

during (a)biotic challenges? Which specific transcription factors or other transcription-regulating mechanisms control the Spminduced defense gene activation? What are the regulatory mechanisms that control Spm-mediated oxidative homeostasis during biotic and abiotic stress responses? Considering the immense value of environmentally-friendly methods for plant stress management in sustainable crop production systems, the application of a multidisciplinary approach benefiting from molecular, biotechnological, and breeding strategies seems to be necessary to fully unlock the potential of Spm as a natural plant defense activator with broad-spectrum protective effects.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

HS conceived and wrote the manuscript. BS supervised the writing and edited the manuscript. Both authors read and approved the final manuscript.

### FUNDING

This research was supported by funding from MITACS Canada, the Natural Sciences and Engineering Research Council (NSERC) Idea-to-Innovation Program, and NutriAg Ltd.



high salt stress in Arabidopsis thaliana. FEBS Lett. 580, 6783–6788. doi: 10.1016/j.febslet.2006.10.078


in tobacco plants. Plant Physiol. 132, 1973–1981. doi: 10.1104/PP.103. 024737

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Seifi and Shelp. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Extracellular Spermine Triggers a Rapid Intracellular Phosphatidic Acid Response in Arabidopsis, Involving PLDδ Activation and Stimulating Ion Flux

Xavier Zarza1,2, Lana Shabala<sup>3</sup> , Miki Fujita<sup>4</sup> , Sergey Shabala<sup>3</sup> , Michel A. Haring<sup>2</sup> , Antonio F. Tiburcio<sup>5</sup> and Teun Munnik1,2 \*

<sup>1</sup> Plant Cell Biology, Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, Netherlands, <sup>2</sup> Plant Physiology, Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, Netherlands, <sup>3</sup> Tasmanian Institute of Agriculture, University of Tasmania, Hobart, TAS, Australia, <sup>4</sup> Gene Discovery Research Group, RIKEN Plant Science Center, Tsukuba, Japan, <sup>5</sup> Department of Biology, Healthcare and the Environment, Faculty of Pharmacy, University of Barcelona, Barcelona, Spain

#### Edited by:

Luo Jie, Huazhong Agricultural University, China

#### Reviewed by:

Rebecca L. Roston, University of Nebraska–Lincoln, United States Thomas Berberich, Senckenberg Nature Research Society, Germany

\*Correspondence:

Teun Munnik t.munnik@uva.nl

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 05 March 2019 Accepted: 24 April 2019 Published: 21 May 2019

#### Citation:

Zarza X, Shabala L, Fujita M, Shabala S, Haring MA, Tiburcio AF and Munnik T (2019) Extracellular Spermine Triggers a Rapid Intracellular Phosphatidic Acid Response in Arabidopsis, Involving PLDδ Activation and Stimulating Ion Flux. Front. Plant Sci. 10:601. doi: 10.3389/fpls.2019.00601 Polyamines, such as putrescine (Put), spermidine (Spd), and spermine (Spm), are lowmolecular-weight polycationic molecules found in all living organisms. Despite the fact that they have been implicated in various important developmental and adaptative processes, their mode of action is still largely unclear. Here, we report that Put, Spd, and Spm trigger a rapid increase in the signaling lipid, phosphatidic acid (PA) in Arabidopsis seedlings but also mature leaves. Using time-course and dose-response experiments, Spm was found to be the most effective; promoting PA responses at physiological (low µM) concentrations. In seedlings, the increase of PA occurred mainly in the root and partly involved the plasma membrane polyamine-uptake transporter (PUT), RMV1. Using a differential <sup>32</sup>Pi-labeling strategy combined with transphosphatidylation assays and T-DNA insertion mutants, we found that phospholipase D (PLD), and in particular PLDδ was the main contributor of the increase in PA. Measuring non-invasive ion fluxes (MIFE) across the root plasma membrane of wild type and pldδ-mutant seedlings, revealed that the formation of PA is linked to a gradual- and transient efflux of K+. Potential mechanisms of how PLDδ and the increase of PA are involved in polyamine function is discussed.

Keywords: phospholipase D (PLD), phosphatidic acid (PA), lipid signaling, polyamines (putrescine, spermidine, spermine), phospholipids, MIFE

#### INTRODUCTION

Polyamines are small polycationic molecules present in all living organisms (Galston and Kaur-Sawhney, 1995). In plants, putrescine (Put), spermidine (Spd), and spermine (Spm) are the major polyamines, where they have been implicated in a broad range of cellular events, including embryogenesis, cell division, morphogenesis, senescence, and in various biotic- and abiotic stress responses (Bagni and Pistocchi, 1988; Wallace, 2009; Tiburcio et al., 2014; Michael, 2016). Despite

**152**

the fact that polyamines were discovered nearly 350 years ago, and have been intensively studied during the last decades, the molecular mechanism by which these molecules regulate such a wide range of cellular functions remains a big mystery (Bachrach, 2010; Alcázar and Tiburcio, 2014). Nonetheless, polyamines have been shown to interact with components of the nucleus and cellular membranes, including transcription factors, protein kinases and phospholipases (Miller-Fleming et al., 2015) as well as ion transporting proteins (Pottosin and Shabala, 2014; Pottosin et al., 2014). The multifaceted relationship between polyamine-mediated effects and the activation of different signaling systems adds another layer of complexity to the experimental determination of direct polyamine targets (Alcázar et al., 2010; Tiburcio et al., 2014; Miller-Fleming et al., 2015).

While most studies have focused on the interaction of endogenous polyamines with immediate subcellular targets, plants are also exposed to extracellular polyamines. In the soil, plants encounter a high degree of polyamines through decomposition of organic material by microorganisms (Young and Chen, 1997; Zandonadi et al., 2013). In addition, there are several environmental cues, such as salinity stress and abscisic acid (ABA), which trigger an efflux of polyamines into the apoplast (Moschou et al., 2008; Toumi et al., 2010). There, polyamines can be oxidized by diamineand polyamine oxidases, producing H2O<sup>2</sup> that in turn triggers downstream effects that eventually affect the plant's development and/or responses to stress (Takahashi et al., 2003; Moschou et al., 2008; Toumi et al., 2010; Pottosin and Shabala, 2014). However, not all apoplastic polyamines are oxidized, as intercellular transport and local internalization of a substantial part of these compounds also takes place (Friedman et al., 1986; Pistocchi et al., 1987; Ditomaso et al., 1992a; Yokota et al., 1994; Sood and Nagar, 2005; Pommerrenig et al., 2011).

The study of polyamine uptake and transport in plant cells remains scarce. However, with the recent characterization of several polyamine-uptake transporters (PUTs), an important new area is emerging, providing interesting genetic tools to explore its potential in plant function and signaling (Fujita et al., 2012; Mulangi et al., 2012; Li et al., 2013; Strohm et al., 2015; Martinis et al., 2016; Tong et al., 2016).

Phosphatidic acid (PA) represents a minor class of membrane lipids, constituting 1–3% of total phospholipids in most plant tissues. As a precursor of glycerolipids, PA is involved in lipid biosynthesis at the ER and plastids. Over the last decade, however, PA has also emerged as a signaling molecule, playing key roles in regulating plant development and stress responses (Munnik, 2001; Testerink and Munnik, 2005, 2011; Yao and Xue, 2018). This PA is typically formed at the plasma membrane and along the endosomal membrane system, where it recruits and modulates target proteins involved in membrane trafficking, organization of the cytoskeleton and ion transport (Testerink and Munnik, 2005, 2011; Kooijman et al., 2007; Raghu et al., 2009; McLoughlin et al., 2013; Pleskot et al., 2013; Putta et al., 2016; Yao and Xue, 2018). A local increase of PA may also induce biophysical effects, affecting membrane curvature and surface charge, which facilitate membrane fission and fusion (Kooijman et al., 2003; Wang et al., 2006; Roth, 2008), also in cooperation with other lipid signals (Testerink and Munnik, 2011).

The accumulation of PA in response to stimuli is in general relatively fast, taking place within minutes after stimulation, and is generated via two pathways, i.e., via phosphorylation of diacylglycerol (DAG) by DAG kinase (DGK) and by hydrolysis of structural phospholipids by phospholipase D (PLD). DAG itself can be produced via non-specific phospholipase C (NPC), which hydrolyses structural phospholipids, or by phosphoinositide- (PI-) specific phospholipase C (PLC), which hydrolyses inositolcontaining phospholipids (Munnik, 2014).

Both PLC- and PLD activities are known to be affected by polyamines. In vitro studies on isolated enzymes from animal cells and tissues have shown that polyamines can inhibit (Kimura et al., 1986; Smith and Snyderman, 1988; Wojcikiewicz and Fain, 1988; Sjöholm et al., 1993; Pawelczyk and Matecki, 1998) and stimulate PLC activity (Sagawa et al., 1983; Haber et al., 1991; Späth et al., 1991; Periyasamy et al., 1994; Pawelczyk and Lowenstein, 1997) and PLD activity (Jurkowska et al., 1997; Madesh and Balasubramanian, 1997). In plants, polyamines have been found to activate PLC in Catharanthus roseus roots (Echevarría-Machado et al., 2002, 2004), but to inhibit it in Coffea arabica cells, where an increase in PLD activity was observed (Echevarría-Machado et al., 2005).

Here, we show that polyamines trigger a rapid (minutes) PA response in Arabidopsis seedlings, with Spm being the most potent. Using differential <sup>32</sup>Pi-labeling techniques and a PLD-specific transphosphatidylation assay (Arisz and Munnik, 2013; Munnik and Laxalt, 2013), we provide evidence that the PLD pathway is the most important contributor. Using T-DNA-insertion PLD mutants, we identified PLDδ as the main contributor of the Spm induced-PA response. Using Microelectrode Ion Flux Estimation (MIFE), we found a differential Spm induced-K<sup>+</sup> efflux response in the pldδ KO mutant, highlighting a potential role for PA downstream of Spm signaling.

## MATERIALS AND METHODS

#### Plant Material and Growth Conditions

Arabidopsis thaliana pldα1, pldα3, pldδ, pldε, pldα1/δ, pldα1/δ/α3, pldα1/δ/ε, pldζ1, pldζ2, gapc1-1/gapc2-1, gapc1-1/gapc2-2, rmv1, spms-2 mutant null alleles and the Pro35S::RMV1 and Pro35S::SPMS-9 transgenic lines were described previously (Hong et al., 2008, 2009; Bargmann et al., 2009; Gonzalez et al., 2011; Fujita et al., 2012; Guo et al., 2012; Galvan-Ampudia et al., 2013). The lat1/2/3/5 and lat1/2/4/5 quadruple null mutant were generated by Dr. M. Fujita (unpublished; RIKEN Plant Science Center, Japan), while pldα1/δ/α3 and pldα1/δ/ε triple knock-out mutants were kindly provided by Prof. Dr. D. Bartels (University of Bonn, Germany). In most cases Arabidopsis thaliana ecotype Col-0 was used as wild type, except for rmv1 and the Pro35S::RMV1 lines, in which Ler ecotype and Col-0 empty vector, Ve-1, were used as wild type, respectively.

Seeds were surface-sterilized with chlorine gas and sown under sterile conditions on square petri dishes containing

60 µM of putrescine (Put), spermidine (Spd) or spermine (Spm), or with buffer alone (control, Ctrl), after which lipids were extracted, separated by TLC, and visualized by autoradiography. Each sample represents the extract of three seedlings. (B) <sup>32</sup>P-labeled PA fold response to 30 min Put, Spd, Spm or diaminopropane (Dap), and thermospermine (tSpm) at the indicated concentrations.

standard growth medium consisting of <sup>1</sup>/<sup>2</sup> Murashige and Skoog (MS) medium with Gamborg B5 vitamins (pH 5.7; KOH), 1% (w/v) sucrose, and 1% (w/v) agar. Plates were vernalized at 4◦C for 48 h and then placed vertically under the angle of 70◦ , in a growth chamber (16/8 light/dark cycle, 110–130 µmol m−<sup>2</sup> s −1 ) at 22◦C. Five days-old seedlings were then transferred to either 2 mL Eppendorf safe-lock tubes for <sup>32</sup>P<sup>i</sup> labeling O/N, or to treatment plates for phenotypic analyses.

### Chemicals

All chemicals were obtained from Sigma-Aldrich except <sup>32</sup>P<sup>i</sup> (orthophosphate, <sup>32</sup>PO<sup>4</sup> <sup>3</sup>−), which was purchased from PerkinElmer. Arabidopsis incubations with polyamines and chemicals were performed in incubation buffer, consisting of 2.5 mM MES buffer [2-(N-morpholino) ethanesulfuric acid], pH 5.7 (KOH), 1 mM KCl.

### <sup>32</sup>Pi-Phospholipid Labeling, Extraction and Analysis

Phospholipid levels were measured as described earlier (Munnik and Zarza, 2013). Briefly, three seedlings per sample were metabolically labeled overnight by flotation in continuous light in 2 ml safe-lock Eppendorf tubes containing 200 µl incubation buffer (2.5 mM MES-KOH, pH 5.7, 1 mM KCl) and 2.5–10 µCi <sup>32</sup>PO<sup>4</sup> <sup>3</sup><sup>−</sup> (stock <sup>32</sup>P<sup>i</sup> ; carrier-free, 2.5–10 µCi/µL). For mature plants, Arabidopsis leaf disks (∅ 5 mm) were taken from 3-week-old plants and labeled using the same conditions. Treatments were performed by adding 1:1 (v/v) of a 2× solution and incubations were stopped at indicated times by adding perchloric acid (Munnik and Zarza, 2013). Lipids were extracted and analyzed by thin-layer chromatography (TLC) using an ethyl acetate solvent system (Munnik and Laxalt, 2013). Radioactivity was visualized by autoradiography and individual spots were quantified by phosphoimaging (Typhoon FLA 7000; GE Healthcare).

For certain experiments, the protocol was slightly modified, i.e.: (1) In short-labeling experiments, <sup>32</sup>Pi was added 30 min prior treatment. (2) In transphosphatidylation assays, treatments were performed in the presence of 0.5% n-butanol (Munnik and Laxalt, 2013). (3) For tissue-dissection experiments, seedlings were labeled, treated and fixed as described above, but then carefully cut into sections with a scalpel, and every section processed separately.

For statistical analysis, letters indicate values significantly different according to Student–Newman–Keuls test at P-value <0.05, and asterisks indicate significant differences with respect to control treatments, using the Student's t-test: <sup>∗</sup>P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.005. Data represent the mean ± SD. The results obtained were confirmed by at least 2 independent experiments.

#### Detection of ROS and NO in Arabidopsis Root

ROS production in the root tip of 5-day-old seedlings was detected by DCF fluorescence as described previously (Pei et al., 2000; Zhang et al., 2009). Briefly, seedlings were treated for the indicated times and then transferred to 10 µM H2DCFDA for 10 min followed by two washes in buffer. For ROS scavenging, seedlings were pre-treated with 5 mM N,N'-Dimethylthiourea (DMTU; Lu et al., 2009) for 60 min, before the different treatments with or without Spm 120 µM for 30 min. For NO detection, seedlings were co-incubated with the corresponding treatment and 10 µM DAR-4M for 30 min, and then washed two times with buffer. For cPTIO treatment, 0.1 mM cPTIO

test at P-value < 0.05. Asterisks indicate significant differences with respect to control treatments, using the Student's t-test: <sup>∗</sup>P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.005.

was applied for 60 min prior to treatments in order to scavenge NO. All incubations were performed in dark conditions. The localization of the DCF and DAR signal was done using the AMG Evos FL digital inverted microscope equipped with transmitted light GFP (470/22 to 510/42 nm). Images were converted to 8-bit using Image-J, and data was quantified as mean pixel intensity per region of interest (ROI).

#### Ion Flux Measurement

Net K<sup>+</sup> fluxes were measured using MIFE technique (UTas Innovation, Hobart, TAS, Australia) (Newman, 2001; Shabala et al., 2006). Five days-old Arabidopsis seedlings were placed into a 30 mL measuring chamber, containing 0.5 mM KCl, 0.2 mM CaCl2, 5 mM MES, 2 mM Tris base; pH 6.0. Roots were immobilized in a horizontal position (Bose et al., 2014) and preincubated in the above buffer for at least 30 min. Electrodes were positioned near the root surface at the elongation zone (less than 2 mm from the root cap junction). First, steady-state ion fluxes were recorded over a period of 5 min, after which different concentrations of Spm were applied and net ion fluxes measured.

### Root Phenotyping Assay on Plates

Arabidopsis seedlings were grown on vertical plates containing standard growth medium for 5 days. Then, seedlings were transferred to plates supplemented with or without Spm. Plates were scanned at indicated days after transfer (DAT) using an Epson Perfection V700 Scanner at 300 dpi. For root measurements EZ-Rhizo software was used (Armengaud et al., 2009). Main root (MR) growth was expressed as growth ratio (MR length divided by MR length at 0 DAT). A paired t-test in SPSS was used for statistical analysis.

### RESULTS

#### Polyamines Trigger the Formation of PA in Arabidopsis Seedlings

To investigate whether polyamines could affect PA signaling, Arabidopsis seedlings were <sup>32</sup>Pi-prelabeled O/N and treated with physiological concentrations (60 µM) of Put, Spd, or Spm for 30 min. As shown in **Figure 1A**, in particular Spm but also Spd,

∗∗P < 0.01, ∗∗∗P < 0.005.

was found to induce an increase in PA. Put did not show any effect until millimolar concentrations were used (**Figure 1B**). Spm already induced significant PA responses at ≥15 µM, while 30– 60 µM was required for Spd. Thermospermine (tSpm), a minor structural isomer of Spm also present in Arabidopsis (Kakehi et al., 2008), was found to induce PA responses at a similar concentrations as Spm (**Figure 1B**). Diaminopropane (Dap), a diamine product of polyamine oxidation, exhibited a similar low potency like Put (**Figure 1B**). Together these results show that polyamines can trigger a PA response, with its potency depending on charge, i.e., Spm4<sup>+</sup> = tSpm4<sup>+</sup> > Spd3+>>Put2+≈Dap2+.

To further investigate the Spm-induced PA, detailed doseresponse and time-course analyses were performed. As shown in **Figure 2**, Spm induced a clear dose-dependent Michaelis– Menten PA curve when treated for 30 min; starting at low µM levels and reaching a maximum 2.5-fold increase at ∼250 µM (**Figures 2A,B**). The response was relatively fast, starting between 8 and 15 min when using 60 µM of Spm, and PA linearly increasing (**Figures 2C,D**).

### Spm Triggers PA Formation in Roots and Requires Transport Across the Plasma Membrane via RMV1

To obtain more information as to where in the seedling the PA accumulation took place, we performed the same <sup>32</sup>Pi-labeling and treatments, but now dissected root- and shoot tissues prior to lipid extraction. Interestingly, the Spm induced-PA increase was only found in the root, not in the shoot or hypocotyl (**Figure 3A**). Within the root, the PA accumulation was equally distributed along the different root sections (**Figure 3A**). A repetition of this experiment with 120 µM Spm gave similar results (data not

shown). In contrast to seedlings, we did observe a Spm induced-PA response in <sup>32</sup>Pi-labeled leaf disks of mature, 3-weeks old plants (**Supplementary Figure S1**).

The non-permeant cation transport blocker, gadolinium (Gd3+) is known to inhibit the uptake of Spm across the plasma membrane (Pistocchi et al., 1988; Ditomaso et al., 1992b; Pottosin et al., 2014). Incubation of seedlings with GdCl<sup>3</sup> prior to the application of Spm triggered a small PA response itself, but significantly reduced the Spm induced-PA response to approximately 70% of the control response (**Figure 3B**). This may indicate that most of the PA response observed is caused intracellularly.

To further characterize Spm uptake, we analyzed the Arabidopsis PUT/L-type amino acid transporter (LAT), called Resistant to Methyl Viologen 1 (RMV1, PUT3, LAT1), which is localized in the plasma membrane and responsible for the high-affinity uptake of Spm (Fujita et al., 2012). Using the knockout T-DNA insertion mutant rmv1 and two independent overexpressing Pro35S::RMV1 lines, we found a 35% decrease and ∼20–40% increase in PA, respectively (**Figures 3C,D**). These results indicate that cellular uptake of Spm is required for the PA response, and that RMV1 is one of the proteins involved in internalizing Spm across the plasma membrane.

To functionally analyze the involvement of the rest of the PUT/LAT family members, of which Arabidopsis contains five homologs (Mulangi et al., 2012), two quadruple knock-out mutants were used, i.e., lat1/2/3/5 and lat1/2/4/5, because the quintuple mutant was lethal (Fujita M., unpublished). Both lines, however, showed Spm induced-PA responses similar to wild type (**Supplementary Figure S2**). This discrepancy could be due to the fact that the single rmv1-KO allele is different from the quadruple mutants and belongs to a different ecotype (i.e., Ler vs. Col-0, respectively). While at least three LAT proteins exhibit polyamine transport activity (i.e., LAT1, LAT3, LAT4; Fujita et al., 2012; Mulangi et al., 2012), only LAT1 (RMV1, PUT3) is localized to the plasma membrane; LAT3 and LAT4 are localized to the ER and Golgi, respectively (Li et al., 2013; Fujita and Shinozaki, 2014). Hence, the results obtained may reflect distinct plasma membrane activity as well as genetic redundancy for Spm uptake, involving other members from the amino acidpolyamine-choline (APC) transporter family to which LAT/PUT transporters belong (Verrey et al., 2004; Rentsch et al., 2007).

#### Spm-Triggered PA Is Predominantly, but Not Solely Generated via PLD

A rapid PA response has traditionally been associated with increased DGK- and/or PLD activity. DGK produces PA through phosphorylation of DAG that originates from the hydrolysis of phosphoinositides or structural phospholipids by PLC or NPC, respectively (Munnik, 2014). PLD hydrolyses structural phospholipids, like PE and PC, to form PA directly. To distinguish between these two routes, a differential <sup>32</sup>P-labeling protocol was used that highlights the DGK kinase-dependent reaction (Arisz and Munnik, 2013). This method is based on the premise that the <sup>32</sup>P<sup>i</sup> added to seedlings is rapidly takenup and incorporated into ATP and subsequently into lipids that

FIGURE 4 | Analysis of DGK involvement in Spm triggered-PA response. Seedlings were pulse-labeled with <sup>32</sup>P<sup>i</sup> for 30 min and then treated with 60 µM Spm or buffer alone (Ctrl) for the indicated times. The fold-PA response of two independent experiments is shown (squares and triangles, respectively). Values were normalized to the <sup>32</sup>P-labeling of phosphatidylinositol and to Ctrl, without Spm.

are synthesized via kinase activity (e.g., DGK), which is in huge contrast to the relatively slow incorporation of <sup>32</sup>P into structural phospholipids via de novo synthesis (Munnik et al., 1994, 1998; Arisz and Munnik, 2013). Under short labeling conditions, PLD would hardly generate <sup>32</sup>P-labeled PA whereas the contribution of DGK would be augmented. As shown in **Figure 4**, Spm was still able to trigger an increase in <sup>32</sup>P-PA when seedlings were only prelabeled for 30 min rather than 16 hrs O/N, indicating that at least part of the Spm induced-PA response is generated via DGK.

To analyze the potential involvement of PLD, its unique ability to catalyze a transphosphatidylation reaction was used, which produces phosphatidylbutanol (PBut) in vivo in the presence of a low concentration of n-butanol (Munnik et al., 1995, 1998; Arisz and Munnik, 2013; Munnik and Laxalt, 2013). To get a substantial proportion (though not all) of the structural phospholipids (i.e., PLD's substrate) <sup>32</sup>P-labeled, seedlings were incubated with <sup>32</sup>P<sup>i</sup> O/N and the next day treated with Spm in the presence of 0.5% n-butanol. The subsequent formation of <sup>32</sup>P-PBut is an in vivo marker for PLD activity that can be quantified. Again dose-response and time-course experiments were performed, but this time in the presence of n-butanol. As shown in **Figures 5A,B**, Spm clearly triggered PLD activity, with the PBut following a similar pattern as PA (**Figure 2**).

Arabidopsis contains 12 PLDs, i.e., 3 PLDαs, 2 PLDβs, 3 PLDγs, 1 PLDδ, 1 PLDε, and 2 PLDζs (Zhang et al., 2005). Validating the <sup>32</sup>P-labeled PBut- and PA response in various T-DNA KO mutants, we identified PLDδ as the main contributor, with the pldδ-KO mutant alone or in combination with other KOs, showing a ∼55% reduction in PA and ∼70% reduction in PBut accumulation (**Figures 5C,D**). Interestingly, in Arabidopsis this isoform is located in the plasma membrane (Wang and Wang, 2001; Pinosa et al., 2013), while promoter-GUS analyses

suggests it is mainly expressed in roots (Katagiri et al., 2001), which is in agreement with the PA response observed here.

### H2O<sup>2</sup> or NO Are Not Involved in the Spm Induced-PA Response

Spermine is known to cause an accumulation of NO and H2O<sup>2</sup> (Cona et al., 2006; Tun, 2006; Moschou et al., 2008), which is likely mediated by polyamine oxidase (PAO) and diamine oxidase (DAO) activities (Tun, 2006; Wimalasekera et al., 2011). We confirmed that Spm was able to trigger an increase in H2O<sup>2</sup> and NO under our conditions, as evidenced by the increase in fluorescence of their reporters, i.e., 2<sup>0</sup> ,70 -dichlorofluorescein diacetate (H2DCFDA; **Figure 6A**) and diaminorhodamine-4M acetoxymethyl ester (DAR-4M AM; **Figure 6B**).

Since PLD activity can be activated by H2O<sup>2</sup> (Wang and Wang, 2001; Zhang et al., 2003, 2009) or by NO (Distéfano et al., 2008; Lanteri et al., 2008; Raho et al., 2011), this could be a potential mechanism by which the production of PA was stimulated, especially since H2O<sup>2</sup> and NO have been found to act upstream of PLDδ in response to ABA inducedstomatal closure (Distéfano et al., 2012). Similarly, H2O<sup>2</sup> has been found to promote the binding of cytosolic glyceraldehyde-3-phosphate dehydrogenase (GAPC) to PLDδ and increase its activity (Guo et al., 2012).

To investigate whether H2O<sup>2</sup> and NO were responsible for the Spm induced-PA response, the effects of ROS scavenger, DMTU and NO scavenger, carboxy-PTIO (cPTIO) were analyzed. While able to significantly reduce the accumulation of Spm-derived H2O<sup>2</sup> and NO (**Figures 6A,B**), the scavengers had no effect on the Spm induced-PA response (**Figure 6C**), suggesting that the increase in PA was independent of these secondary metabolites. Moreover, double gapc1-1 gapc2-1 or gapc1-1 gapc2-2 knock-out mutants, revealed a PA response similar to wild type (**Figure 6D**). These results indicate that the Spm-induced PA is not caused via ROS- or NO induction.

Testing polyamines on seedling growth, we observed a significant reduction in primary root growth (**Supplementary Figure S3**). Previous reports have shown that this root growth inhibition is associated to H2O<sup>2</sup> accumulation, derived from PAO activity (de Agazio et al., 1995; Couée et al., 2004; Tisi et al., 2011). However, loss-of function pldδ seedlings did not show any apparent root phenotype when transferred to agar

plates containing µM concentrations of Spm. Only a slight increase in root growth inhibition with respect to wt was observed at higher Spm concentrations, i.e., 150 µM (**Supplementary Figure S3**). Those results are consistent with previous reports indicating that pldδ is more sensitive to H2O2-induced stress (Zhang et al., 2003).

#### PLDδ Is Involved in the Spm-Induced K <sup>+</sup>-Efflux Response in the Root-Elongation Zone

Application of exogenous polyamines has been shown to trigger a K<sup>+</sup> efflux in pea- (Zepeda-Jazo et al., 2011) and maize roots (Pandolfi et al., 2010), which has consequences for the membrane potential, inducing the plasma membrane to depolarize (Ozawa et al., 2010; Pottosin et al., 2014). To study this in our context, we performed MIFE ion-flux analyses at the root elongation zone of Arabidopsis seedlings using different concentrations of Spm. As shown in **Figures 7A,B**, a clear dose-dependent efflux of K<sup>+</sup> was detected, which correlated with the response in PA (**Figure 2B**). While the Spm induced-K<sup>+</sup> efflux slowly restored to pre-treatment values after 50 min with 60–200 µM, the efflux persisted when only 10–20 µM Spm was used.

To investigate potential involvement of PLDδ, the response of wild type was compared with that of the pldδ knock-out mutant in the same root zone. Prior to Spm application, roots showed a small net K<sup>+</sup> efflux of 150–250 nmol m−<sup>2</sup> s −1 , likely due to transferring seedlings from nutrient-rich MS medium (∼20 mM K <sup>+</sup>) to poorer, basic-salt medium (BSM; 0.2 mM K+; **Figure 7**). Upon 60 µM Spm application to wild-type seedlings, the K<sup>+</sup> efflux increased gradually, reaching a peak around 15 min, and returning to basal levels after ∼50 min (**Figure 7C**). In pldδ, the response was significantly different, showing an faster K<sup>+</sup> efflux peak at 10 min, and recovery (**Figure 7C**). Overall, pldδ showed ∼60% reduction in net K<sup>+</sup> loss compared to wt (**Figure 7D**),

placing PLDδ and derived PA upstream of the Spm induced-K<sup>+</sup> efflux. Since the Spm induced-K<sup>+</sup> efflux is completely abolished by gadolinium (**Supplementary Figure S4**), these results indicate that PLDδ is likely activated by Spm from the inside of the cell, after its uptake.

### DISCUSSION

#### Polyamines Trigger a Charge-Dependent PA Response

Polyamines are naturally occurring polycationic molecules involved in a plethora of cellular events (Tiburcio et al., 2014), yet it is still largely unknown how this works at the molecular level. Here, a link to the formation of the lipid second messenger PA is reported, which is predominantly generated by PLDδ and plays a role in Spm induced-K<sup>+</sup> efflux.

In this paper, we show that low µM concentrations of Spm trigger a rapid (minutes) PA response in the roots of Arabidopsis seedlings (**Figure 2**). In older plant material, Spm also triggered a PA response in leaves (**Supplementary Figure S1**), so multiple tissues are sensitive to Spm. Other polyamines, like Spd, Put and Dap were also able to trigger an accumulation in PA but this required higher concentrations, especially the diamines. In contrast, tSpm was as effective as Spm (**Figure 1**). These results indicate that the capacity of polyamines to activate the formation of PA is charge-dependent, with Spm4<sup>+</sup> = tSpm4+>Spd3+>>Put2+, which may indicate an electrostatic interaction between the positive charges of the polyamine and a negatively charged target (Bertoluzza et al., 1988; Kurata et al., 2004; DeRouchey et al., 2010; Rudolphi-Skórska et al., 2014).

#### Triggering PA Response Requires Uptake of Polyamines

The saturation of the response at relatively low concentrations (**Figure 2**) may reflect a saturation of polyamine uptake (Pistocchi et al., 1987; Ditomaso et al., 1992a). In that regard, the Spm induced-PA responses were found to be strongly reduced in the plasma membrane localized PUT, RMV1-KO, also known as LAT1 or PUT3 (Fujita et al., 2012; Fujita and Shinozaki, 2014), while overexpression of RMV1 resulted in a much stronger PA response upon Spm treatment (**Figure 3**). The affinity of this transporter for Spm (K<sup>m</sup> = 0.6 µM), Spd (K<sup>m</sup> = 2.2 µM), and Put (K<sup>m</sup> = 56.5 µM), respectively (Fujita et al., 2012; Fujita and Shinozaki, 2014), is consistent with their potency to activate a

PA response and indicates, together with the inhibition of the PA response by gadolinium (**Figure 3**), that polyamines are taken up before triggering a PA increase.

### Link Between Polyamine Synthesis and PLDδ

Differential <sup>32</sup>P-labeling experiments combined with transphosphatidylation assays revealed that part of the PA response was generated via the PLD pathway. Using T-DNA insertion PLD-KO mutants, we found that the majority of the PA is generated through PLDδ. Arabidopsis contains 3α-, 2β-, 3γ-, 1δ-, 1ε-, and 2ζ PLDs, which differ in amino acidsequence conservation and lipid-binding domains (Bargmann and Munnik, 2006; Hong et al., 2016; Hou et al., 2016). PLDδ is typically localized in plasma membrane facing the cytosol (Wang and Wang, 2001; Pinosa et al., 2013) while all others PLDs are cytosolic though can transiently bind to various microsomal membranes (Wang, 2002). PLDδ has been implicated in drought and salinity stress (Katagiri et al., 2001; Bargmann et al., 2009; Distéfano et al., 2015) and in freezing tolerance (Li et al., 2004), which are stress responses in which polyamines have also been implicated to play a role (Alcázar et al., 2010). The specific involvement of PLDδ further implies that the polyamine induced-PA response predominantly occurs at the plasma membrane.

Both Spm and tSpm activated PLDδ equally well. In Arabidopsis, Spm is synthesized by spermine synthase (SPMS) while tSpm by ACAULIS5 (Panicot et al., 2002; Kakehi et al., 2008). Both are encoded by single genes, SPMS and ACL5, which are predominantly expressed in the phloem and xylem, respectively, in both roots and leaves (**Supplementary Figure S5**; Brady et al., 2007; Winter et al., 2007; Sagor et al., 2011; Yoshimoto et al., 2016). Interestingly, their expression strongly overlaps with that of PLDδ, especially SPMS, which are both strongly induced upon salt- and osmotic stress (**Supplementary Figures S5, S6**; Katagiri et al., 2001; Brady et al., 2007; Winter et al., 2007). So potentially, the local synthesis and/or transport of spermine could activate PLDδ to generate PA. In agreement, significantly increased PA levels were found in an SPMSoverexpressor line (35S::SPMS-9; Gonzalez et al., 2011) at control conditions (**Supplementary Figure S7A**). An spms-2 KO line exhibited normal PA levels, however, in response to salt stress, a strongly reduced PA response was found while the SPMSoverexpressor line revealed a much higher PA response than wt (**Supplementary Figures 7A,B**). These results support the idea of a novel, interesting link between PA and Spm in stress responses. Alternatively, since polyamines are rich in soil and produced by various microbes (Young and Chen, 1997; Chibucos and Morris, 2006; Zandonadi et al., 2013; Zhou et al., 2016), our observation that extracellular polyamines trigger intracellular PA responses may also reflect the action of natural, exogenous polyamines.

### PA Function

Phosphatidic acid is an important plant phospholipid. Besides its role as precursor for all glycerolipids at the ER, PA has emerged as important lipid second messenger, generated through the PLC/DGK- and/or PLD pathways in response to various (a)biotic stresses, including plant defense, wounding, salt, drought, cold, and heat stress, where it is linked to various cellular processes, like vesicular trafficking, membrane fission and -fusion, and transport (Munnik et al., 2000; Testerink and Munnik, 2005, 2011; Vergnolle et al., 2005; Zhao, 2015; Hong et al., 2016; Hou et al., 2016). A local accumulation of PA in cellular membranes may affect the enzymatic- or structural properties of protein targets in that membrane or, alternatively, recruit cytosolic protein targets via PA-binding domains. PA targets include protein kinases, phosphatases, ion transporters, PEPC, GAPDH, NADPH oxidases (Rboh) (Kim et al., 2013; McLoughlin et al., 2013; Pleskot et al., 2013; Putta et al., 2016; Ufer et al., 2017). Here, evidence is provided for a role of PA in ion transport. MIFE analyses showed that Spm triggers a rapid efflux of K<sup>+</sup> ions, which was strongly reduced in the pldδ mutant (**Figure 7**), indicating a direct or indirect role for PA in K<sup>+</sup> gating. Associated to Spm uptake, we observed a fast net H<sup>+</sup> influx (cytosolic acidification) followed by a gradual increase of H<sup>+</sup> efflux (cytosolic alkalinisation), which correlated with the K <sup>+</sup> efflux peak and its gradual recovery (**Supplementary Figure S8**). The cytosolic alkalinisation is related to the opening of voltage-gated inward-rectifying K<sup>+</sup> channels (Kin) to compensate the K<sup>+</sup> efflux (Dreyer and Uozumi, 2011; Karnik et al., 2016). In animal cells, PA has been shown to regulate voltage-gated potassium (Kv) channels and has been proposed to stabilize K+ inward channels (Kin) in its closed conformation, thus reducing K <sup>+</sup> inward currents (Hite et al., 2014). In plants, PA has been shown to inactivate Kin channels in guard cells of Vicia faba and Arabidopsis (Jacob et al., 1999; Uraji et al., 2012), in which PLDδ has been implicated (Uraji et al., 2012). These observations are in agreement with a role for PA in regulating K<sup>+</sup> fluxes of which the precise mechanism requires further investigation.

In summary, we provide molecular evidence that polyamines functionally require PLD and PA for their mode of action. This knowledge and the use of PLD- and polyamine synthesis mutants may shed new light on this phenomenon in other studies.

### DATA AVAILABILITY

All datasets for this study are included in the manuscript and the **Supplementary Files**.

## AUTHOR CONTRIBUTIONS

XZ, SS, and TM designed the experiments. LS performed the MIFE experiments while XZ performed the rest. MF, AT, and MH added materials, ideas and discussions. XZ and TM wrote the manuscript.

## FUNDING

This work was supported by the Spanish Ministerio de Ciencia e Innovación (BIO2011-29683 and CSD2007-00036), the Generalitat de Catalunya (SGR2009-1060 and BE DGR 2011), and the Netherlands Organisation for Scientific Research (NWO 867.15.020).

#### ACKNOWLEDGMENTS

fpls-10-00601 May 20, 2019 Time: 17:9 # 11

The authors would like to thank L. Tikovsky and H. Lemereis for the assistance in the greenhouse, Dr. T. Takahashi (Okayama University, Japan) for supplying tSpm and Dr. I. Pottosin

#### REFERENCES


(University of Colima, Mexico) for discussion and critical review of the MIFE experiments.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00601/ full#supplementary-material


and transcript profiling provide new insights on the role of the tetraamine spermine in Arabidopsis defense against Pseudomonas viridiflava. Plant Physiol. 156, 2266–2277. doi: 10.1104/pp.110.171413



thaliana spermine synthase gene promoter Gene Note. Plant Biotechnol. 28, 407–411. doi: 10.5511/plantbiotechnology.11.0704a



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Zarza, Shabala, Fujita, Shabala, Haring, Tiburcio and Munnik. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Spermine Confers Stress Resilience by Modulating Abscisic Acid Biosynthesis and Stress Responses in Arabidopsis Plants

*Francisco Marco1† , Enrique Busó2† , Teresa Lafuente3 and Pedro Carrasco1 \**

*1 Estructura de Recerca Interdisciplinar en Biotecnologia i Biomedicina (ERI BIOTECMED), Universitat de València, Valencia, Spain, 2 UCIM, Universitat de València, Valencia, Spain, 3Departamento de Biotecnologia de Alimentos, Instituto de Agroquímica y Tecnología de Alimentos, CSIC, Valencia, Spain*

#### *Edited by:*

*Ana Margarida Fortes, University of Lisbon, Portugal*

#### *Reviewed by:*

*Yi Shang, Yunnan Normal University, China Mingjun Li, Northwest A&F University, China Rongrong Guo, Guangxi Academy of Agricultural Science, China*

#### *\*Correspondence:*

*Pedro Carrasco pedro.carrasco@uv.es*

*† These authors have contributed equally to this work*

#### *Specialty section:*

*This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science*

*Received: 26 February 2019 Accepted: 11 July 2019 Published: 31 July 2019*

#### *Citation:*

*Marco F, Busó E, Lafuente T and Carrasco P (2019) Spermine Confers Stress Resilience by Modulating Abscisic Acid Biosynthesis and Stress Responses in Arabidopsis Plants. Front. Plant Sci. 10:972. doi: 10.3389/fpls.2019.00972*

Polyamines (PAs) constitute a group of low molecular weight aliphatic amines that have been implicated as key players in growth and development processes, as well as in the response to biotic and abiotic stresses. Transgenic plants overexpressing PA-biosynthetic genes show increased tolerance to abiotic stress. Therein, abscisic acid (ABA) is the hormone involved in plant responses to environmental stresses such as drought or high salinity. An increase in the level of free spermine (Spm) in transgenic Arabidopsis plants resulted in increased levels of endogenous ABA and promoted, in a Spm-dependent way, transcription of different ABA inducible genes. This phenotype was only partially reversed by blocking ABA biosynthesis, indicating an ABA independent response mediated by Spm. Moreover, the phenotype was reproduced by adding Spm to Col0 wild-type Arabidopsis plants. In contrast, Spm-deficient mutants showed a lower tolerance to salt stress. These results indicate that Spm plays a key role in modulating plant stress responses.

Keywords: abscisic acid, spermine, salt stress, stress response, stress tolerance

### INTRODUCTION

Polyamines (PAs) are small aliphatic amines found in all living organisms (Kusano et al., 2008). In plants, the three major PAs include putrescine (Put), spermidine (Spd), and spermine (Spm). In Arabidopsis, Put biosynthesis is done mainly by arginine decarboxylase (ADC; EC4.1.1.19) activity, which decarboxylates arginine as the first step in the PA-biosynthetic pathway. Put serves as the precursor for higher molecular weight PAs Spd and Spm. Spd synthase (SPDS;EC2.5.1.16) and Spm synthase (SPMS;EC 2.5.1.22), respectively, catalyze the addition to Put of aminopropyl groups generated from S-adenosylmethionine (SAM) by SAM decarboxylase (SAMDC, EC 4.1.1.50) (Tiburcio et al., 1990). Arabidopsis genome carries two genes that encode for ADC enzyme (ADC1 and ADC2; Urano et al., 2005), four genes for SAMDC (SAMDC1-4; Urano et al., 2003), two genes for SPDS (*SPDS1* and *SPDS2*; Hanzawa et al., 2002) and one single gene for SPMS (Hanzawa et al., 2002). *ACAULIS5* (*ACL5* gene), originally assigned as a putative Spm synthase (Hanzawa et al., 2000), encodes a thermospermine (tSpm) synthase (Knott et al., 2007). Free PAs levels of the plant are the result of a balance between biosynthesis and degradation, the former process done mainly through the activity of diamine oxidases (DAO, EC1.4.3.6) and polyamine oxidases (PAO, EC1.5.3.3), that exhibit different substrate preferences. Diamines like Put are a preferred substrate for DAOs, while higher molecular weight PAs, like Spm, tSpm, and Spd, are oxidized by PAOs (Cona et al., 2006). Some PAOs terminally oxidize PAs, while other isoforms are involved in a PA backconversion process of tSpm to Spm and Put, releasing H2O2 (Moschou et al., 2012). Ten genes encoding DAOs, as well as five genes that encode PAOs are present in the Arabidopsis genome (Takahashi et al., 2010; Planas-Portell et al., 2013).

As sessile organisms, plants respond to environmental stresses through a series of physiological, cellular, and molecular changes. Consequence of these changes, transcriptomic, proteomic, and metabolic modifications, which help to mitigate the effect of stress and lead to the adaptation of the plant, occur. Classically, PAs have been assigned to play an important role in modulating the response of plants to diverse environmental stresses (Bouchereau et al., 1999). Elevated PA levels are one of the most remarkable metabolic hallmarks in plants exposed to stresses such as drought, salinity, chilling, heat, hypoxia, ozone, UV, or heavy metals (Alcázar et al., 2010; Gill and Tuteja, 2010). In the last years, it has been elegantly summarized the relationship between the functional significance of PAs and their roles in tolerance and/or amelioration of stress responses in plants (Minocha et al., 2014). It has been suggested that PAs could act as stress messengers in plant responses interacting with different stress pathways (Takahashi et al., 2003; Tuteja and Sopory, 2008; Alcázar et al., 2010; Marco et al., 2011).

It is well known that abscisic acid (ABA) is involved in the response of plants to abiotic stresses (Vishwakarma et al., 2017). Abscisic acid (ABA) is a plant hormone that quickly accumulates in plants exposed to different abiotic stresses. ABA is a sesquiterpenoid that is synthetized from C40 carotenoids in three steps catalyzed by zeaxanthin epoxidase (Marin et al., 1996), 9-cis-epoxycarotenoid dioxygenase (NCED; Schwartz et al., 1997) and abscisic aldehyde oxidase (AAO; Seo et al., 2000). The key step in ABA biosynthesis is the cleavage reaction of epoxy carotenoids to produce xanthoxin catalyzed by NCED (Kende and Zeevaart, 1997). In fact, ectopic expression of NCED causes overproduction of ABA in tobacco and tomato (Thompson et al., 2000). Accumulation of ABA stimulates stomatal closure to limit water loss from leaves, as well as changes in the level of expression of genes involved in stress responses and in the biosynthesis of osmoprotectant species that help the cells to handle damage and restore cellular homeostasis (Fujita et al., 2011). Transcriptome comparison of Arabidopsis plants exposed to salt or drought stress shows that about half of the genes induced by these stresses are also induced by ABA (Seki et al., 2002). Moreover, the disruption of ABA synthesis by transgenic approaches in Arabidopsis leads to plants more sensitive to drought stress (Iuchi et al., 2001). These observations suggest the predominant role of this hormone as a signaling molecule in the response against drought and high salinity stresses (Sah et al., 2016).

Hierarchical clustering expression analyses in Arabidopsis using public microarray data indicate that some, but not all, PA biosynthesis gene paralogs share similar expression patterns, in agreement with their different implications in stress and development (Tiburcio et al., 2014). Moreover, transcriptome studies performed in Arabidopsis have revealed differential regulation of PA biosynthesis genes by abiotic stress (Alcazar et al., 2006b). Recent studies have been reported that Spm-overproducer plants overexpress a number of ABA related genes (Marco et al., 2011). On the other hand, the characterization of PA loss-of-function mutants has provided evidence for the involvement of PAs in resistance traits (Urano et al., 2003; Yamaguchi et al., 2007; Cuevas et al., 2008; Zarza et al., 2017). Besides that, although taken together these evidences suggest that PAs are involved in modulating plant responses to abiotic stress, the relationship between PA, and ABA remains to be fully understood.

In this work, we have observed that overexpression of *SAMDC1* gene in Arabidopsis produces Spm accumulation and leads to plants with an improved tolerance to salt stress. Furthermore, since overexpression of the *NCED3* gene and ABA accumulation were observed on these Spm-accumulating transgenic lines, the expression of several ABA-response related genes was checked in plants with different Spm levels, several Spm-deficient mutant lines, as well as WT plants treated with external Spm. A Spm-dependent expression pattern was observed for *NCDE3* and some ABA-responsive genes. These results suggest that Spm plays a key role in modulating ABA levels and that stress tolerance could be improved by manipulation of Spm biosynthesis, causing the accumulation of endogenous ABA and triggering the expression of a number of stress-response genes. On the other hand, expression of *NCDE3* and other ABA-responsive genes was increased in these Spm-overproducer plants even when ABA biosynthesis was inhibited, suggesting the existence of a Spm-dependent pathway response that would include some ABA-dependent factors.

### MATERIALS AND METHODS

### Plant Growth Conditions

Experiments were performed using several *Arabidopsis thaliana* (Arabidopsis) lines. Ecotype Col-0, obtained from the Nottingham Arabidopsis Stock Centre (University of Nottingham, Loughborough, UK) was used as the wild type (WT). On the other hand, different lines of Arabidopsis plants with altered Spm levels have been used: three transgenic lines overexpressing the SAMDC1 gene under the control of CaMV35S constitutive promoter (pBISDCs-S3', pBISDCs-S9', pBISDCs-S15, Marco et al., 2011), a T-DNA insertion mutant line of the *SPMS* gene (*spms-1*, Imai et al., 2004), an ethyl methane-sulfonatemutant line of the *ACL5* gene (*acl5-1*, Hanzawa et al., 1997) and a double mutant line obtained from crossing of the two lines mentioned above (*acl5-1/spms-1*, Imai et al., 2004). Both *acl5-1* and *acl5-1/spms-1* mutants showed stem-reduced growth phenotypes (Hanzawa et al., 1997; Imai et al., 2004), however, *spms-1* plants exhibited similar phenotypes to WT plants in terms of growth and development (Imai et al., 2004).

Plants were cultivated in growth chambers Sanyo MLR-350 (Sanyo Electric Co., Japan) under long day conditions, illumination at 23°C for 16 h, in darkness at 16° for 8 h. Previously, all the seeds were stratified for 2 days at 4°C. Adult plants were grown from seeds sown in pots with a 1:1:1 mixture of soil, vermiculite, and sand, and watered with mild nutrient solution [recipe from Arabidopsis Biological Resource Center (ABRC, The Ohio State University, USA) handling plants and seeds guide, http://www.biosci.ohio-state.edu/pcmb/Facilities/ abrc/handling.htm].

Five day-old seedlings were grown under the same light and temperature conditions. Seeds were surface sterilized by washing for 10 min in 30% (v/v) commercial bleach, 0.01% (v/v) Triton X-100 and rinsed three times with sterile distilled water. Sterile seeds were plated on 4% agar plates containing one half strength MS medium (Murashige and Skoog, 1962).

#### Plant Treatments

Adult plants were grown in pots, salt stress treatments were performed by adding NaCl (0–250 mM) to the watering solution. Visible damage of stress was estimated in 3-week old plants exposed for 10 days to 0–250 mM NaCl, and pictures were taken. Additionally, 4-week old plants were exposed for 6 h to 250 mM NaCl and rosette leaf samples were harvested to isolate RNA. Furthermore, a stress recovery assay was performed by exposing 2-week old plants to 250 mM NaCl for 2 days. After 12 days of recovery in control conditions, stem length and shoot fresh weight (FW) were measured.

Salt stress and inhibition of ABA biosynthesis experiments were also performed in plates by adding 250 mM NaCl to 5-day grown seedlings grown in ½ MS medium with or without 100 μM sodium tungstate dehydrate (Fluka). Entire seedling samples were taken after 6 h.

Seedlings were also grown in ½ MS plates supplemented with 0.1, 0.5, and 1 mM, Put, Spd, or Spm. Entire seedling samples were taken after 5 days of growth. All tissues were harvested and immediately frozen in liquid nitrogen and stored at −80°C until used.

Germination of seeds was also studied in ½ MS plates supplemented with 0, 50, 100, 150, or 250 mM NaCl. Fifty seeds were sowed per plate and the number of seeds that developed cotyledons until 8 days after stratification was counted. All experiments were done by triplicate.

#### Quantitative RT-PCR

Total RNA was extracted from plant tissue using Total Quick RNA Cells and Tissues Kit (Talent SRL, Italy), following protocol established by manufacturer. RNA was quantified by their absorbance al 260 nM, and its integrity checked by denaturing agarose gel electrophoresis.

RNA was treated with RNAse free-DNAse (Roche diagnostics, Spain), according to manufacturer's instructions. A total of 1 μg of DNA free-total RNA was reverse transcribed to Firststrand complementary DNA (cDNA) with random hexamers using SuperScript® III First-Strand Synthesis System 1st (Invitrogen, Spain) according to manufacturer's instructions. Quantitative real time PCR (qRT-PCR) was performed on Gene AmpR 5,700 Sequence Detection System (PE Applied Biosystems, Japan), using Power SYBR® Green PCR Master Mix (PE Applied Biosystems) first-strand cDNA as a template. Each 20 μl reaction contained 1 μl of cDNA, 100 nM of each pair of target primers (FW and REV), and 10 μl of SYBR Green PCR Master Mix. The PCR conditions were as follows: 95°C for 10 min, followed by 40 cycles of 95°C for 30 s and 60°C for 1 min. Three replications were performed for each sample in each experiment. Primers used for real-time PCR are described in **Table 1**. Data was analyzed according to 2−ΔΔ*<sup>C</sup>*T Method (Livak and Schmittgen, 2001). Actin-2 (AT5G09810; An et al., 1996) was used as a reference gene.

#### Abscisic Acid Quantification

ABA was measured in seedlings and 4-week old Arabidopsis plants. Plant tissue (0.2 g FW) was homogenized with 5 ml of extraction buffer (80% acetone, 100 mg L−1 butylated hydroxytoluene, 0.5 g L−1 citric acid) and centrifuged 12,000 ×*g* for 5 min. Supernatant was recovered, dried, and resuspended in 0.5 ml TBS buffer (6.05 g L−1 Tris, 0.20 mg L−1 MgCl2, and 8.8 g L−1 NaCl, pH 7.8). ABA was quantified following an indirect ELISA method (Walker-Simmons, 1987). ABA-BSA conjugates were made according to Weiler (1980) as modified by Norman et al. (1988). ABA levels are expressed as ng (g FW)−1.

#### Statistical Analyses

Data was analyzed by one-way Analysis of Variance (ANOVA) followed by *post-hoc* comparisons by Tukey's HSD *t* test. A probability level < 0.05 was considered statistically significant. Calculations were performed using IBM® SPSS® Statistics v22.0 Software.



### RESULTS

#### pBISDCs Plants Are Tolerant to Salt Stress

In a previous work, we were able to describe that overexpression of SAMDC1 gene in Arabidopsis leads to with higher Spm levels than WT plants (**Supplementary Figure S1**). Also, the transcriptome of pBISDCs transgenic lines showed an increase in the expression of a set of genes enriched in functional categories involved in defense-related processes against both biotic and abiotic stresses (Marco et al., 2011).

To assess the effects of this defense-related transcriptome expression changes in salt stress tolerance, pBISDCs and WT plants were grown in soil for 3 weeks and treated with irrigating solution with presence or absence of NaCl. Under control conditions, pBISDCs showed no phenotype in terms of growth and development. However, after 10 days of treatment, WT plants irrigated with excess NaCl showed generalized chlorosis and wilt, while pBISDCs lines appeared with a healthier aspect and diminished symptoms (**Figure 1A; Supplementary Figure S3**). A saline stress recovery assay was also performed by exposing pBISDCs and WT plants to 250 mM NaCl during 2 days and leaving them to recover for 12 days in the absence of salt. After the recovery period, salt treated plants showed a reduction in their growth compared to untreated plants, but pBISDCs lines were able to develop higher recovery percentages of stem length (46.1–48.7%) and shoot fresh weight (72.8–75.2%) than WT plants (26 and 54%, respectively) (**Supplementary Figure S4**).

Additionally, salt stress tolerance was estimated by establishing seed germination percentage after 8 days, of seeds sowed in plates with increasing concentrations of NaCl (0–250 mM). Germination percentage decreased with increasing NaCl concentration for all lines studied (transgenic and WT), dropping to 10% for the highest salt concentration (250 mM) (**Figure 1B**). However, Spm-accumulating lines maintained better germination percentages (about 15% higher than WT) at intermediate NaCl concentrations (50–150 mM) (**Figure 1B**).

#### *SAMdC1* Overexpression Alters Abscisic Acid Metabolism and Abscisic Acid-Responsive Gene Expression

Given the previous result, a more specific search of genes related to the salt stress response within the transcriptome of the pBISDCs plants was carried out, locating a set of genes that code for proteins related to the ABA synthesis and response that showed significant expression changes compared to WT plants (**Supplementary Figure S2**). The expression of some of these ABA-related genes was checked by qRT-PCR to confirm the results observed in transcriptome studies. Expression levels of ABA-induced genes *COR15A* (AT2G42540; Baker et al., 1994), *RD26* (AT4G27410; Yamaguchi-Shinozaki et al., 1992), *RD29A* (AT5G52310; Yamaguchi-Shinozaki and Shinozaki, 1993), *RD22BP1* (AT1G32640; Abe et al., 2003) and *NCED3* (AT3G14440; Tan et al., 2003) were determined in 4-week old plants (**Figure 2A**). SAMDC1-overexpressing lines showed higher expression levels of all these ABA-related genes when compared to WT plants (**Figure 2A**). ABA-biosynthesis gene *NCED3* showed the most pronounced changes (5 to 10-fold).

Plants of same age were also exposed to salt stress. After 6 h of 250 mM NaCl exposure, the level of expression of all five ABA-related genes increased in WT plants (**Figure 2B**). Regarding pBISDCs transgenic lines response to NaCl exposure,

FIGURE 1 | Salt stress tolerance of pBISDCs lines. (A) Appearance of Arabidopsis WT and pBISDCs transgenic plants overexpressing SAMDC1 (S3′, S9 and S15) after 10 days of salt treatment. 3 week-old plants were watered with solution without (control) and with the supplementation of 250 mM NaCl. Pictures were taken after 10 days. (B) Seed germination percentage in MS plates supplemented with different concentrations of NaCl. The number of seeds that developed cotyledons until 8 days after stratification was counted. Graphs show the mean ± standard deviation from MS plate triplicates. Significant changes from control treatment are highlighted (\*) (ANOVA, Tukey HSD test, *p* < 0.05).

*RD26* (all three lines) and *NCED3* (two lines) maintained expression levels similar to control conditions, while *COR15A* (two lines) raised to similar levels observed in WT plants exposed to stress. Also, *RD29A* (all lines) increased its expression in pBISDCs lines to higher levels than WT plants in salt stress conditions (**Figure 2B**).

ABA levels were also determined in both pBISDCs and WT lines in control and salt stress conditions. *SAMDC1* overexpressing lines showed also higher ABA levels (about 3-fold) than WT lines on non-stressed 4-week old plants (**Figure 2C**). After 6 h of salt exposure, ABA levels raised in WT in greater extent than transgenic lines, reaching to a slightly higher level than in Spm-accumulating transgenic lines (**Figure 2D**). Similar results in gene expression and ABA levels were found for 5 day-old plate-grown seedlings and did not changed when photoperiod was set to short day (data not shown).

#### Spm Treatment Raises ABA Levels and Induces ABA-Related Gene Expression

Results described above suggested a possible link between the observed differences in ABA and high Spm levels. To further test this possibility, PAs were exogenously supplied to plates where WT seeds were sown and let grow for 5 days. ABA levels increased in Spm-supplemented seedlings, with a positive correlation between ABA levels and external Spm concentration (**Figure 3A**). This positive correlation affected also to expression of the chosen set of ABA-related genes, with *NCED3* showing the most important induction in response to external Spm (**Figure 3B**). However, the growth of seedlings in the presence of external Put or Spd did not produce significant changes in ABA levels or in the expression of ABA-related genes (**Figure 3**).

#### Spm Deficiency Affects Expression of Abscisic Acid-Induced Genes

Moreover, the possible relationship between ABA-related stress response and Spm levels was also tested in different Spm-deficiency scenarios. Therefore, ABA levels were determined in 5-day old plate-grown seedlings of tSpm synthase mutants (*acl5-1*), that have similar free Spm levels than WT plants (Hanzawa et al., 2000; Imai et al., 2004), as well as in *spms-1* mutants and double mutant *acl5-1/spms-1*, with no detectable levels of Spm (**Figure 4A**; Imai et al., 2004). In control conditions, all mutants showed similar ABA levels to those of WT seedlings. On the other hand, when seedlings were exposed to 250 mM NaCl for 6 h (salt stress conditions), ABA on *acl5-1* mutant rose to levels similar to those observed stressed WT seedlings. On the contrary, *spm-1* and *acl5-1/spm-1* mutants showed slightly lower ABA levels than WT stressed seedlings (**Figure 4A**). In the same conditions, Spm-accumulating line pBISDCs-S15 had the highest levels of ABA for both control and salt stress conditions (**Figure 4A**).

Expression of the ABA-responsive genes previously analyzed was also determined by qRT-PCR in Spm-deficient mutants (**Figures 4B–E**). Under control conditions, expression levels in the mutants were similar to those found in WT plants for all genes, but their expression pattern changed in salt exposed seedlings, depending on the mutant considered. Thus, *spms-1* and *acl5-1/spms-1* mutants were not able to raise *NCED3* and *RD26* expression when exposed to salt to the levels observed

in salt-stressed WT and *acl5-1* seedlings (**Figures 4C,E**, respectively). Similar observations were made for *COR15A*, although the expression levels were more close to WT stressed seedlings (**Figure 4B**). On the other side, *RD22BP1* gene showed expression fold-change levels closer to the observed in WT salt-stressed seedlings for both *spms-1* mutants assayed (**Figure 4D**).

#### Spm Is Able to Modulate Plant Responses to Salt Stress in an Abscisic Acid-Independent Way

ABA synthesis was blocked by addition of 0.1 mM tungstate in plates. As expected, ABA levels dropped to levels <1 ng ABA g FW−1 in 5 days seedlings of WT, Spm-treated WT or pBISDCs plants (**Figure 5A**). Expression levels of *COR15A*

and *RD22BP1* were higher in Spm-treated WT and pBISDCs lines than in WT seedlings (**Figures 5B,D**). In the presence of tungstate, expression of those genes was similar to the levels observed in the WT seedlings. (Fold changes ≤1) (**Figures 5B,D**), suggesting that their induction by Spm is determined by ABA-dependent pathways. On the contrary, *NCED3* expression levels in presence of tungstate remained higher in Spm-treated plants and pBISDCs lines than in WT plants (**Figure 5C**). *RD26* also showed maintenance of their expression after tungstate treatment combined with Spm or in pBISDCs seedlings (**Figure 5E**) but with fold change differences relative to WT control less pronounced than those observed for *NCDE3* (**Figure 5E**).

Once determined that higher Spm levels affect expression of ABA-related genes, we also tested the effect of blocking ABA biosynthesis in the different Spm-deficient situations

determine the expression levels of salt stress-related response genes in all conditions assayed (B–G). For each gene, data are expressed as fold change relative to the level measured in WT plants in control conditions (2−ΔΔCT). Graph show the mean of three biological replicates ± standard deviation. Significant differences between treatments are indicated with letters (ANOVA, Tukey HSD test, *p* < 0.05).

described previously. Expression of the ABA dependent genes *COR15A*, *RD22BP1, NCED3*, and *RD26* was analyzed under control and salt stress conditio*ns* in the presence or absence of tungstate. Expression levels of *COR15A* and *RD22BP1* did not respond to salt stress in the presence of tungstate either in WT type, Spm-deficient mutants or the pBISDCs line S15 (**Figures 4B,D**). However, after salt stress, *NCDE3* rises to levels higher than control conditions in WT, the pBISDCs line S15 and *acl5-1* seedlings, and only dropped to similar levels observed in WT in Spm-deficient *spm-1* and *acl5-1/ spm-1* mutants (**Figure 4C**). *RD26* also showed maintenance of their expression with tungstate treatment combined with salt stress, but with fold change differences relative to WT control less pronounced than those observed for *NCDE3* (**Figure 4E**). Similar results were obtained when ABA biosynthesis was blocked with Fluoridone (data not shown).

The specificity of Spm in relation with ABA-dependent stress responses was checked by analyzing the expression levels

of *SOS1* (AT2G01980) and *SOS3* (AT5G24270), two members of the ABA-independent SOS (Salt Overly Sensitive) saltsignaling response pathway (Ji et al., 2013). *SOS1* and *SOS3*·expression was similar in all seedlings of the lines assayed, and salt stress induced their expression at similar levels independently of tungstate presence or the level of Spm in the seedlings (**Figures 4F,G**).

deviation. Significant differences between treatments are indicated with letters (ANOVA, Tukey HSD test, *p* < 0.05).

#### DISCUSSION

During the last years, the manipulation of polyamine levels by transgenic approaches or use of loss or gain mutants has proved a useful tool to gain knowledge about possible polyamine roles in plant processes (Alcazar et al., 2006b). Transgenic approaches include heterologous overexpression studies of *ODC*, *ADC*, *SAMDC*, and *SPDS* from different animal and plant sources in rice, tobacco, tomato, and Arabidopsis (Alcázar et al., 2010).

Likewise, Spm accumulation has been previously observed in transgenic rice plants constitutively overexpressing the *Datura stramonium* SAMDC cDNA in transgenic rice (Thu-Hang et al., 2002) or the yeast SAMDC in tobacco (Cheng et al., 2009), as well as using ABA inducible expression of Tritordeum SAMDC in rice (Roy and Wu, 2002) or ripening-induced expression of yeast SAMDC in tomato (Mehta et al., 2002), where accumulation of Spd and Spm was observed. High levels of Put, Spd, and Spm were also observed in tobacco plants that overexpress constitutively carnation SAMDC (Wi et al., 2006). In a previous work, we were able to describe that overexpression of *SAMDC1* gene in Arabidopsis leads to plants with higher Spm content than WT plants. Also, the transcriptome

of pBISDCs transgenic lines showed an increase in the expression of a set of genes enriched in functional categories involved in defense-related processes against both biotic and abiotic stresses (Marco et al., 2011). In line with these results, Arabidopsis Spm-accumulating lines obtained by overexpression of SPMS share a common set of 234 genes that includes genes related with the response to water deprivation or cold acclimation (Gonzalez et al., 2011; Marco et al., 2011). In addition, it was previously observed that external Spm treatment modulated the expression of a large number of defense-related genes (Mitsuya et al., 2009). In fact, 28 genes induced from that study are also overexpressed in the pBISDCs transgenic lines, including the transcription factor AtbZIP60 (Iwata and Koizumi, 2005) and the mitogen-activated protein kinase AtMAPK3 (Takahashi et al., 2003). Moreover, Arabidopsis AtPO4-deficient plants show increased Spm levels in the roots and up-regulation of several genes encoding drought stress response proteins (Kamada-Nobusada et al., 2008).

The activation of this set of stress response genes due to Spm accumulation could be the factor that would explain the tolerance of those plants to salt stress (**Figure 1**). Absence of Spm on mutant line *acl5-1/spm-1* causes a defect of Ca2+ homeostasis and resulted in hypersensitivity to salt stress, being this phenotype mitigated only by exogenously applied Spm, but not by Spd or Put (Yamaguchi et al., 2006). On the other hand, increased tolerance to abiotic stress by increasing Spm levels had been previously observed in other transgenic plant systems. Rice plants over-expressing *Tritordeum* SAMDC under the control of an ABA inducible promoter accumulate Spd and Spm and are less sensitive to salt stress (Roy and Wu, 2002). Also, rice plants with constitutive expression of *D. stramonium SAMDC* showed increased levels of Spm and an improved recovery after exposure to drought conditions (Peremarti et al., 2009). In the same trend, over-expression of carnation SAMDC produced accumulation of total PAs in tobacco, and generated a broad-spectrum tolerance to abiotic stresses (Wi et al., 2006). More recently, is has been described that constitutive overexpression of *Capsicum annuum S*AMDC in Arabidopsis increases Spd and Spm levels and leads to an increased drought tolerance of the transgenic plants compared to WT (Wi et al., 2014). Raise of Spm and Spd levels was also obtained in Arabidopsis plants by overexpression of cucurbita *SPDS* gene, leading to an enhanced tolerance to multiple environmental stresses (Kasukabe et al., 2004). Conversely, alterations of Spm levels have also been reported by downregulation of SAMDC gene by RNA interference strategies in tobacco (Moschou et al., 2008) and rice (Chen et al., 2014), leading to plants with reduced Spd and Spm levels and an enhanced salinity-induced programmed cell death in tobacco (Moschou et al., 2008) or with a reduced tolerance to stress by drought, salinity or chilling in rice (Chen et al., 2014).

The relationship among PA metabolism, abiotic stress, and ABA has been previously reported in several studies. Expression of *ADC2*, *SPMS*, and *SAMDC2* genes is induced by the exogenous application of ABA in Arabidopsis (Urano et al., 2003). Also, induction of *ADC2*, *SPDS1*, and *SPMS* by drought stress is Arabidopsis is an ABA-dependent response, since up-regulation is not observed in ABA deficient (*aba2*) and insensitive (*abi1*) mutants (Alcazar et al., 2006a). Moreover, maize *ADC2*, *ZmSPDS1* and *ZmSPDS2* genes are also induced by NaCl and ABA treatments (Jiménez-Bremont et al., 2007).

Based on these results, the effect of Spm accumulation on the expression of ABA-related genes in pBISDCs plants was examined more in depth with the MAPMAN tool (**Supplementary Figure S2**; Usadel et al., 2005). We found three upregulated genes (*NCED3*, *NCED4*, and *ABA2*) that code for ABA biosynthesis enzymes (González-Guzmán et al., 2002; Tan et al., 2003). In addition, other genes coding for ABA-induced proteins like ATHVA22A, ATHVA22B (Chen et al., 2002), KIN1 (Wang et al., 1995) or the gram-domain containing protein GER5 (AT5G13200; Baron et al., 2014) were also up-regulated (**Supplementary Figure S2**, overexpressed). The set of under-expressed genes also includes some ABA-responsive genes as AAO2 (Seo et al., 2000), ABF4 (Kang et al., 2002) and a couple more of ABA-responsive proteins. (**Supplementary Figure S2**, under-expressed). These observations pointed us to carry out a more detailed study of the ABA levels and ABA response in pBISDCs plants. In order to confirm the expression data observed in the transcriptome study, expression levels of several ABA-related genes were checked by qRT-PCR. pBISDCs lines have higher levels of expression of genes that code for ABA biosynthesis genes (**Supplementary Figure S2**), including *NCED3* (**Figure 2A**), and consequently higher ABA levels than WT lines in control conditions (**Figure 2C**).

Expression levels of a set of ABA-responsive genes were also checked by qRT-PCR. Cold-Regulated 15A gene (*COR15A*) codes for a member of the Late Embryogenesis Abundant protein family with a role in the protection of the chloroplast structures during freeze-induced dehydration (Steponkus et al., 1998). Responsive to Desiccation 26 gene (*RD26*) is induced in response to desiccation, and encodes a transcriptional activator that acts in ABA-mediated dehydration response (Yamaguchi-Shinozaki et al., 1992; Song et al., 2016). *RD22BP1* gene also encodes a MYC-related transcriptional activator that is induced by dehydration stress and ABA treatment (Abe et al., 2003; Liu et al., 2018). *RD29A* encodes a hydrophilic protein of unknown function that is induced by salt and drought stresses (Yamaguchi-Shinozaki and Shinozaki, 1993). Expression levels of all these ABA-dependent genes were increased in the pBISDCs lines (**Figure 2A**). Induction of *RD26* and *COR15A* has been previously observed in Arabidopsis plants overexpressing cucurbita *SPDS*, with increased levels of Spd and Spm (Kasukabe et al., 2004). Higher ABA levels on pBISDCs lines could also explain the activation of at least part of the stress-related genes observed in these Spm-accumulating lines (**Supplementary Figure S2**).

Additionally, external application of Spm to WT plants lead to similar changes in *NCDE3* expression, ABA levels, and expression levels of the other ABA-responsive genes (**Figure 5**). Moreover, Spm-deficient mutants (*spm-1*, *acl5-1/ spm-1*) were not able to raise ABA to similar levels than WT plants when exposed to salt stress (**Figure 4A**). Same observation was done for the expression levels of genes *NCED3*

(**Figure 4C**) and *RD26* (**Figure 4E**). An increase in ABA levels has been also reported in soybean seeds treated with Spm (Radhakrishnan and Lee, 2013).

Finally, when ABA synthesis was blocked with tungstate, *NCED3* maintained a certain level of induction in plants with normal (WT, *acl5-1*) or high (Spm-treated WT and pBISDCs lines) levels of Spm, whereas this was not observed in Spm-deficient mutants *spm-1* and *acl5-1/spm-1* (**Figure 5**, tungstate and **Figure 4**, saline + tungstate). The existence of an ABA-independent pathway responsible for part of the induction of ABA biosynthesis genes by salt stress was proposed previously (Barrero et al., 2006). Those authors pointed out that severe ABA-deficient mutants still showed a NaCl-dependent induction of *NCED3*, *AAO3*, and *ABA1*, being *NCED3* the gene that showed a stronger induction with NaCl in ABA absence. Taken together, our results suggest that Spm could have a possible role in the induction of the expression of *NCED3* by this ABA-independent pathway salt stress response.

On the other hand, expression levels of two members of the SOS signaling pathway, involved in the maintenance of ion homeostasis during salt stress (Ji et al., 2013) are not altered in Spm-accumulating lines in response to NaCl stress (**Figures 4F,G**). This observation was previously reported in Spm-deficient mutants (Yamaguchi et al., 2006).

In summary, the results obtained in this study add more evidences to the involvement of Spm in plant stress responses, for which various protective roles have been proposed. Our results suggest that one of these mechanisms could involve the modulation of ABA levels in salt stress response through modulation of ABA biosynthesis by affecting *NCED3* gene expression. At the same time, Spm could be involved in an ABA-independent stress response pathway, as suggested by the results observed when ABA biosynthesis is blocked with tungstate.

#### REFERENCES


It remains to be determined which other changes in gene expression observed in Spm-accumulating plants results from the direct action of Spm or which gene expression changes are the consequence of cross-talking between Spm and other stress-response signaling pathways, including ABA.

### DATA AVAILABILITY

All datasets generated for this study are included in the manuscript and/or the **Supplementary Files**.

### AUTHOR CONTRIBUTIONS

FM, EB and PC conceived the experimental design. FM and EB conducted the experiments. EB and TL conducted ABA measurements FM, EB and PC conducted gene selection and primer design. FM and PC wrote the manuscript.

#### ACKNOWLEDGMENTS

We are grateful to Drs Oliver Laule and Andreas Fürholz (RIP) for their assessment in the Arabidopsis genomic experiments carried out in ETH Zürich.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00972/ full#supplementary-material

development and inflorescence architecture. *Plant Sci.* 223, 153–166. doi: 10.1016/j.plantsci.2014.03.017


cold acclimation by regulating abscisic acid levels in response to low temperature. *Plant Physiol.* 148, 1094–1105. doi: 10.1104/pp.108.122945


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Marco, Busó, Lafuente and Carrasco. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Increasing Polyamine Contents Enhances the Stress Tolerance *via*  Reinforcement of Antioxidative Properties

*So Yeon Seo, Yu Jung Kim and Ky Young Park\**

*Department of Biology, Sunchon National University, Suncheon, South Korea*

The diamine putrescine and the polyamines (PAs), spermidine (Spd) and spermine (Spm), are ubiquitously occurring polycations associated with several important cellular functions, especially antisenescence. Numerous studies have reported increased levels of PA in plant cells under conditions of abiotic and biotic stress such as drought, high salt concentrations, and pathogen attack. However, the physiological mechanism of elevated PA levels in response to abiotic and biotic stresses remains undetermined. Transgenic plants having overexpression of *SAMDC* complementary DNA and increased levels of putrescine (1.4-fold), Spd (2.3-fold), and Spm (1.8-fold) under unstressed conditions were compared to wild-type (WT) plants in the current study. The most abundant PA in transgenic plants was Spd. Under salt stress conditions, enhancement of endogenous PAs due to overexpression of the *SAMDC* gene and exogenous treatment with Spd considerably reduces the reactive oxygen species (ROS) accumulation in intra- and extracellular compartments. Conversely, as compared to the WT, PA oxidase transcription rapidly increases in the *S16-S-4* transgenic strain subsequent to salt stress. Furthermore, transcription levels of ROS detoxifying enzymes are elevated in transgenic plants as compared to the WT. Our findings with OxyBlot analysis indicate that upregulated amounts of endogenous PAs in transgenic tobacco plants show antioxidative effects for protein homeostasis against stress-induced protein oxidation. These results imply that the increased PAs induce transcription of PA oxidases, which oxidize PAs, which in turn trigger signal antioxidative responses resulting to lower the ROS load. Furthermore, total proteins from leaves with exogenously supplemented Spd and Spm upregulate the chaperone activity. These effects of PAs for antioxidative properties and antiaggregation of proteins contribute towards maintaining the physiological cellular functions against abiotic stresses. It is suggested that these functions of PAs are beneficial for protein homeostasis during abiotic stresses. Taken together, these results indicate that PA molecules function as antisenescence regulators through inducing ROS detoxification, antioxidative properties, and molecular chaperone activity under stress conditions, thereby providing broad-spectrum tolerance against a variety of stresses.

Keywords: polyamines, spermidine, reactive oxygen species, chaperone activity, S-adenosylmethionine decarboxylase

#### *Edited by:*

*Ana Margarida Fortes, University of Lisbon, Portugal*

#### *Reviewed by:*

*Autar Krishen Mattoo, Agricultural Research Service (USDA), United States Shaohua Zeng, Chinese Academy of Sciences, China*

> *\*Correspondence: Ky Young Park plpm@sunchon.ac.kr*

#### *Specialty section:*

*This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science*

*Received: 12 March 2019 Accepted: 25 September 2019 Published: 31 October 2019*

#### *Citation:*

*Seo SY, Kim YJ and Park KY (2019) Increasing Polyamine Contents Enhances the Stress Tolerance via Reinforcement of Antioxidative Properties. Front. Plant Sci. 10:1331. doi: 10.3389/fpls.2019.01331*

#### INTRODUCTION

Polyamines (PAs) are naturally occurring polycations ubiquitous to all living cells, which are essential for development, growth, and survival (Tabor and Tabor, 1985; Tiburcio et al., 2014). The diamine putrescine (Put) and the higher PAs [spermidine (Spd; triamine) and spermine (Spm; tetraamine)] contribute to biochemical and physiological cellular processes such as ion channel regulation and the maintenance of chromatin structure and cell membranes. They also modulate enzyme functions and are required for the regulation of DNA replication, transcription, and translation (Madeo et al., 2018), many of which are related to the positive charge at physiological pH for strong binding capacity to the negatively charged ions of DNA, RNA, and protein molecules (Murray-Stewart et al., 2018). Especially in plants, PAs and their metabolic products act as cell signaling molecules in enhancing tolerance to pathogen attacks and numerous abiotic stresses, including antisenescence in fruits (Wi et al., 2006; Mattoo et al., 2007; Moschou et al., 2008; Pathak et al., 2014; Jiménez-Bremont et al., 2014; Tiburcio et al., 2014; Handa et al., 2018). Attenuation of whole-plant senescence in transgenic plants with overexpression of yeast Spd synthase provides evidence for the role of PAs, particularly Spd, in increasing fruit shelf life, probably by reducing the postharvest senescence (Nambeesan et al., 2010).

The biosynthesis and degradation pathway of PAs in plants is well studied (Bagni and Tassoni, 2001; Moschou et al., 2008). The starting points for PA biosynthesis are the basic amino acids ornithine and arginine, which are decarboxylated by ornithine decarboxylase and arginine decarboxylase, respectively, to yield Put, which serves as the substrate for biosynthesis of Spd and Spm *via* the activities of S-adenosylmethionine decarboxylase (SAMDC) and Spd synthase and Spm synthase (Walters, 2003; Tiburcio et al., 2014). The oxidation of PAs is catalyzed by amine oxidases (AOs) including diamine oxidases (DAOs) and PA oxidases (PAOs), localized either intercellularly (i.e., apoplast) or intracellularly (i.e., cytoplasm and peroxisomes) (Tiburcio et al., 2014; Gémes et al., 2016). The activities of these two enzymes produce hydrogen peroxide (H2O2), which acts as a signal molecule or an antimicrobial compound involved in the resistance to pathogen attack (Walters, 2003; Moschou et al., 2008).

PAs have been linked to ROS homeostasis, in which PAs act as scavengers of reactive oxygen species (ROS) and activate the antioxidant enzyme machinery (Pottosin et al., 2014). An important rate-limiting step in PA biosynthesis is catalyzed by SAMDC. Cellular accumulation of ROS significantly reduces under drought stress in transgenic *Arabidopsis SAMDC* overexpressor plants exhibiting higher endogenous PAs (Wi et al., 2014). On the other hand, PAs exhibit an inverse relationship with PAOs, which correlate with developmental and stress responses (Paschalidis and Roubelakis-Angelakis, 2005). Furthermore, the respiratory burst oxidase homologs [nicotinamide adenine dinucleotide phosphate (NADPH) oxidase] and the apoplastic PAO form a feedforward ROS amplification loop, which impinges on oxidative state and culminates in the execution of cell damages. This loop is a central hub in the plethora of responses controlling

salt stress tolerance, with potential functions extending beyond stress tolerance (Gémes et al., 2016). Therefore, both functions of PAs are proposed to augment antioxidants for protection against oxygen-radical-mediated damages and are substrates for oxidation reactions that produce H2O2 (Murray-Stewart et al., 2018).

Under physiological or stress conditions, superoxide anions (O2 •−) are generated mainly by NADPH oxidase. Superoxide dismutation by superoxide dismutase is considered one of the major routes for subsequent H2O2 production (Gémes et al., 2016). At low/moderate concentrations, ROS are implicated as second messengers in intracellular signaling cascades that mediate several plant responses in plant cells, including stomatal closure, programmed cell death (PCD), gravitropism, and acquisition of tolerance to both biotic and abiotic stresses such as systemic acquired resistance (Sharma et al., 2012). However, it remains unknown whether the major ROS generators, namely, PAOs and NADPH oxidase, are functionally inked or interplayed. In addition, it remains ambiguous as to which enzyme is more effective in generating ROS under abiotic and biotic stress (Tiburcio et al., 2014).

An imbalance between ROS generation and scavenging often results in oxidative stress, which is a common phenomenon for stress-induced detrimental effects in abiotic stress, which sometimes induces a hypersensitive response against incompatible pathogens (Pottosin et al., 2014). In plants, the relative abundance of the PAs depends on the species, the developmental stages, and environmental conditions (Tiburcio et al., 2014). PAs are detected at relatively high concentrations in actively growing tissues and under conditions of biotic or abiotic stress (Jiménez-Bremont et al., 2014).

Theoretically, under stressed conditions, both phenomena of increased accumulations of PAs and ROS can occur independently in plants. Even if PAs are increased in response to a variety of stresses, the stress-induced ROS generation is more rapid and effectively increased than the antioxidative function of PAs. Although the exact mechanism of action of PAs remains elusive in plants under stressed conditions, their indispensable roles are getting recognized as beneficial SAMDC activity in plants and as differential roles in human health and disease (Handa et al., 2018). It has recently been proposed that high doses of PAs are detrimental to disease conditions involving higher cellular proliferation, although PAs are involved in increased longevity and reducing some age-related diseases. Despite the increasing trend in our understanding regarding effects on health and diagnostic use of PAs in humans, the physiological functions of PAs remain ambiguous, since the functions are often opposed to each other depending on the conditions. In fact, this phenomenon is associated to the strict homeostasis of PAs in living organisms.

Recently, there is an increased focus on the link between metabolism and defense response of major PAs during plant– stress interactions (Seifi and Shelp, 2019). We therefore undertook to investigate the physiological mechanism of PAs on cellular homeostasis in response to salt stress, using transgenic tobacco plants with overexpression of *SAMDC* complementary DNA (cDNA) from the carnation (*Dianthus caryophyllus* L.) flower. Our results provide evidence for the physiological function of PAs required for stress tolerance under high salt condition, *via*  signaling through the intracellular and intercellular ROS levels.

### MATERIALS AND METHODS

#### Chemicals and Reagents

All chemicals procured were either analytical or laboratory grade and were used as received without further purification. Reagent and stock/standard solutions were prepared in deionized water (pH 7.0) purified by Barnstead Water Purification Systems (Thermo Scientific, Waltham, MA, USA). NaCl and sodium carbonate were purchased from Junsei (Japan). Standard PAs, Put, Spd, and Spm were obtained from Sigma-Aldrich, Saint Louis, MO, USA. 4-(2-Hydroxyethyl)-1-piperazineethanesulfonic acid, phenol, chloroform, trimethylamine, perchloric acid, and most other chemicals were obtained from Sigma-Aldrich, Saint Louis, MO, USA. Glycerol and lactic acid were purchased from Biobasic (Canada).

### Plant Materials and Growth Conditions

In our previous study, transgenic tobacco (*Nicotiana tabacum* Wisconsin 38) plants were generated with overexpression construct of the *CaMV 35S* promoter-driven *SAMDC* gene *via Agrobacterium tumefaciens*-mediated gene transfer (Wi et al., 2006). A DNA fragment corresponding to nucleotides 39-1794 of the full-length *SAMDC* cDNA (GenBank accession no. U38527) from carnation (*D. caryophyllus* L.) flowers was included in the 465 bp of 5′-untranslated leader sequence, which included 52 amino acids of the upstream open reading frame (uORF) and 377 amino acids of SAMDC ORF. Although several independent transgenic tobacco lines were obtained from the transgenic experiment, we used the *S16-S-4* transgenic plant (T3 homo line) in this study since the expression trait of the SAMDC gene was characterized to the maximum in our previous studies (Wi et al., 2006). Surface-sterilized seeds were cultured on solid Murashige and Skoog medium (pH 5.8) (Duchefa Biochemi, Netherlands) under light (16L/8D, 100 μmol photons m−2 s−1) at room temperature (25°C). Wild-type (WT) tobacco plants were used as the controls. Fully expanded, green, healthy leaves were plucked, and whole leaves were treated with 200 mM NaCl solution for inducing salt stress. For mock treatment, whole tobacco leaves were floated on 2-(N-morpholino)ethanesulfonic acid buffer (pH 6.1) devoid of any other chemicals.

### RNA Isolation and Real-Time qPCR

Using a High-Fidelity PrimeScript RT-PCR kit (Takara, Japan), total RNA was extracted from whole leaves that had been subjected to salt stress (Wi et al., 2006). Gene-specific PCR primers, sequence information of which was obtained from the GenBank database, were designed using a stringent set of criteria (**Supplemental Table S1**). Real-time quantitative PCR (qPCR) was performed in optical 96-well plates using a Chromo 4 continuous fluorescence detector (Bio-Rad, USA). Reaction mixtures (20 ml) comprised of 10 ml 23 SYBR Green master mix, 0.5 mM of each primer, and 10 ng cDNA. PCR conditions were as follows: 95°C for 15 min; 45 cycles of 95°C for 30 s, 57°C for 30 s, and 72°C for 30 s; followed by 72°C for 10 min. Fluorescence threshold data were analyzed using the MJ Opticon monitor software version 3.1 (Bio-Rad, USA) and subsequently exported to Microsoft Excel for further analysis. Relative expression levels in each cDNA sample were normalized to the reference gene β-actin. PCR efficiencies (90–95%) for all primers were determined by serial dilution of cDNA from RNA samples.

### Determination of PA Contents by High-Performance Thin-Layer Chromatography

PA levels were assessed as described by Pedrol and Tiburcio (2001). Leaves (0.3 g) were homogenized in 1 ml of 5% (*v*/*v*) perchloric acid and centrifuged at 12,000×*g* for 15 min. To 0.2 ml of the isolated supernatant, we added 0.2 ml of saturated sodium carbonate and 0.4 ml of dansyl chloride (5-dimethylaminonaphthalene-1-sulfonyl chloride; Sigma-Aldrich, Saint Louis, MO, USA) (1 mg ml−1 stock solution prepared in acetone), and incubated the mixture overnight at room temperature*.* The dansylated product was extracted with toluene and separated on thin layer chromatography in chloroform/trimethylamine (4:1, *v*/*v*) using a silica gel 60 Å high-performance thin-layer chromatography plate (Merck, USA). The separated PAs were scraped off and quantified against commercial standards using a spectrophotofluorometer (RF-1501, Shimadzu, Japan), where emission was recorded at 495 nm after excitation at 350 nm.

#### Trypan Blue Staining

To monitor plant cell death, tobacco leaves were stained as described previously (Wi et al., 2012). Whole leaves were immersed for 1 min in a boiling solution comprising 10 ml lactic acid, 10 ml glycerol, 10 g phenol, and 0.4% (*w*/*v*) trypan blue (Sigma, USA). After cooling to room temperature for 1 h, the solution was replaced with 70% (*w*/*v*) chloral hydrate. Stained plants were decolorized overnight and photographed using a digital camera.

#### Histochemical Detection of Hydrogen Peroxide and Superoxide Anion

The accumulation of superoxide (O2 •**<sup>−</sup>**) anion and hydrogen peroxide (H2O2) was histochemically assessed by nitroblue tetrazolium (NBT) (Biobasic, Canada) and 3,3′-diaminobenzidine (DAB) (Sigma-Aldrich, Saint Louis, MO, USA) staining (Kumar et al., 2013). For detection of superoxide, the leaves were floated in 50 mM potassium phosphate (pH 7.8) containing 0.2% NBT for 2 h at 25°C, leading to the formation of the dark blue insoluble formazan compound. For the detection of hydrogen peroxide, the leaves were immersed for 2 h in a solution of DAB (1 mg ml−1, pH3.8) at 25°C. Thereafter, chlorophyll was removed by boiling in 96% (*v*/*v*) ethanol for 10 min. After complete removal of chlorophyll, the stained leaves were photographed using a digital camera. H2O2 was visualized as a reddish-brown stain formed by the reaction of DAB with endogenous H2O2. The O2 •**<sup>−</sup>** content was detected as dark blue stain of formazan compound formed as a result of NBT reacting with the endogenous O2 •**−**.

#### Measurement of ROS With 2**′**,7**′**-Dichlorodihydrofluorescein Diacetate

For histochemical staining of total ROS level, leaf epidermal strips were peeled from tobacco leaves subjected to stress treatment for the indicated time. Leaf epidermal strips were floated for 10 min on a solution of 50 μM 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) (Invitrogen, USA) prepared in 20 mM potassium phosphate buffer (pH 6.0). Mechanistically, DCFH-DA can be transported across the cell membrane and deacetylated by esterase to form nonfluorescent DCFH. This compound is then trapped inside the cells and is subsequently converted into the highly fluorescent compound (DCF) through the action of ROS and peroxidase, which can be detected and quantified based on fluorescence intensity. ROS was observed by fluorescence microscopy (excitation: 450 ± 490 nm; barrier 520 ± 560 nm) equipped with a cooled charge-coupled device camera (Olympus, FV300, Japan). Fluorescence intensity of histochemical staining from each photograph was quantified by densitometry in ImageJ.

#### Detection of Intra- and Extracellular ROS in Guard Cells

For fluorescent detection of ROS, we used the lower leaf epidermal strips. For detecting the intra- and extracellular superoxide anions, benzene sulfonyl (BES)-So-Am and BES-So (WAKO Chemicals, Japan), respectively, were used at a concentration of 20 mM, prepared in potassium phosphate buffer (pH 6). For detecting intra- and extracellular hydrogen peroxide, BES-H2O2-Ac and BES-H2O2 (WAKO Chemicals, Japan), respectively, were used at a concentration of 50 mM in 20 mM potassium phosphate buffer (pH 6). After incubation of epidermal tissues with ROS/superoxide detection solution at room temperature in the dark for 1 h, fluorescence was observed using the confocal laser scanning microscope FluoView 300 (FV 300; Olympus, Japan). Fluorescence intensity for superoxide detection was observed in the dark with excitation at 485 nm and emission at 530 nm. Fluorescence intensity for hydrogen peroxide was observed with excitation at 505 nm and emission at 544 nm.

#### Oxidized Protein Analysis

Oxidized proteins were detected using an OxyBlot protein oxidation detection kit (Merck Millipore, USA), according to the manufacturer's instructions. Dinitrophenyl hydrazine was added to crude total proteins (10 μg) extracted from tobacco leaves post-NaCl treatment to derive carbonyl groups from the protein side chains. Carbonylated proteins were resolved on sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and Western blot analysis was performed using the provided 2,4-DNP antibody (1:150). The DNP signals in integrated intensity of each fraction were quantified by densitometry in ImageJ and normalized to the total protein value of the WT 0 h control, which was set as 1.

#### Analysis of Chaperone Activity

The chaperone activity of Spd was assayed by measuring its capacity to suppress thermal aggregation of malate dehydrogenase (MDH) (Sigma-Aldrich, Saint Louis, MO, USA) from malic dehydrogenase, using porcine heart (Sigma-Aldrich, Saint Louis, MO, USA) as a model substrate (Jung et al., 2015). MDH (1 μM) was incubated in 50 mM 4-(2-hydroxyethyl)-1 piperazineethanesulfonic acid-KOH (pH 8.0) buffer with total proteins extracted from salt-stressed plants after addition of 0.2 mM Spd. Aggregation of the substrate was monitored under heat denaturation at 45°C for 30min, by measuring the turbidity at 340nm using a Shimadzu UV-1601 spectrophotometer (Shimadzu, Japan).

#### Quantitation and Statistical Analysis

All experiments were repeated at least three times with three independent biological replicates. The photographs represented are from one representative experiment in more than three independent biological experiments with more than three replicates after verifying the reproducibility of the results (**Figures 2A**, **3A**–**C**, **4A**, **B**, **5A**, **B**, **7A**, and **8A**, **B**). Statistically significant differences using Student's *t* test (Microsoft Excel) between transgenic lines and respective controls at each time point are indicated with one asterisk (\*) (*P* < 0.05) or two asterisks (\*\*) (*P* < 0.01). A two-way analysis of variance (ANOVA) is performed to determine significant differences between NaCl and PAs treatment (*P* < 0.001) using GraphPad Prism (GraphPad, San Diego, CA, USA). Evaluation of differences between two groups are evaluated using Student's *t* test of GraphPad Prism.

#### RESULTS

#### Enhancement of PA Biosynthesis Is Induced in Transgenic Tobacco Plants

The full-length of tobacco *SAMDC16* cDNA clone (GenBank accession no. U38527) was overexpressed with the *35S CaMV* promoter-driven construct. We originally isolated two *SAMDC* cDNA clones, *SAMDC9* and *SAMDC16*, from the petals of carnation flowers (Lee et al., 1997). Since the *SAMDC16* clone is more effectively expressed in the leaves of carnation plants, we used *SAMDC16* cDNA for making the transgenic plants (Wi et al., 2006).

Transgenic plants overexpress the full-length *SAMDC16* cDNA fragment that contains an uORF of 52 amino acids located at 5′-untranslated region. It is previously reported that the *SAMDC uORF* sequence or SAMDC uORF protein is a translational inhibitor of its own downstream ORF, which is responsible for the homeostatic regulation of PAs even under stressed conditions (Hanfrey et al., 2003; Hu et al., 2005; Choi and Park, 2011). Therefore, we used the *SAMDC* gene with *uORF* region for developing the transgenic plant in our attempt to overexpress the *SAMDC* gene without it being affected by the removal of *SAMDC* uORF.

After confirming *SAMDC* gene integration in kanamycinresistant transgenic lines by Southern blot analysis and Northern blot analysis, four transgenic lines were finally selected. SAMDC activity in three transgenic lines (*S16-S-1*, *S16-S-2*, and *S16- S-4*) was significantly increased as compared to the WT (**Supplementary Material Figure S1**); the *S16-S-4* transgenic line had the highest SAMDC activity (increased by 70%), compared to the WT control.

Within 3–4 months, the four transgenic plants produced flowers and seeds from self-fertilization. The seeds produced from each flower of transgenic plants were counted. Seed mass was increased by 1.6 times in all transgenic lines, as compared to the nontransformed WT plants (Wi et al., 2006). However, the transgenic plants with empty vector control (*pBI121*) produced almost the same number of seeds as the WT tobacco. Based on these results, the *S16-S-4* transgenic line was selected for further studies.

In our experiments, the major PAs such as Put, Spd, and Spm were significantly increased in the leaves of transgenic *S16-S-4* under unstressed condition, having the most effective increase in Spd levels (2.3-fold), as compared to the WT (**Figure 1A**). In addition, the most abundant PA in transgenic plants was Spd. After being subjected to salt stress, we observed a rapid increase in the amount of *SAMDC* transcript, which in turn may result in increased PA biosynthesis. Transcript levels of *SAMDC16* were

significantly increased in WT and transgenic tobacco leaves in response to salt stress with 200 mM NaCl, resulting in elevated *SAMDC* transcription. This was more significant in stressed tobacco T3 transgenic homozygous lines (*S16-S-4*) as compared to stressed WT plants (**Figure 1B**). Although *SAMDC16* was driven by constitutive *35S CaMV* promoter in these transgenic plants, the transcripts of *SAMDC16* increased by only ~1.5 fold transcripts when compared to WT plants during the entire period of high salinity.

In addition, induction of transcription of *PAO* rapidly increased from 1 h after salt stress (**Figure 1C**), which contributed to the degradation of the PAs. The induction of *PAO* transcripts was much higher in transgenic plants as compared to WT. Stress-induced increase in transcriptions of PA biosynthetic gene and catalytic gene were transient in both the WT and transgenic plants, which peaked at 12 and 1 h, respectively, after salt stress (**Figures 1B**, **C**).

Therefore, we next examined for alterations of endogenous PA levels of tobacco leaves in transgenic plants under salt stress. Following salt stress, levels of the major PAs (Put, Spd, and Spm) were found to be considerably higher in *SAMDC*overexpressing transgenic plants as compared to WT plants (**Figure 1D**). These increases in Spd and Spm levels were mainly due to overexpression of the *SAMDC* transgene. Furthermore, the levels of Put, a precursor of these PAs, were also increased and were found to be higher in transgenic plants than in WT during the entire period of salt stress. Increase in the amount of PAs due to high salinity stress was highest at 12 h after salt stress but decreased to almost initial levels after 24 h. These phenomena indicate the tight regulation of PA biosynthesis for homeostasis, even under abiotic stress conditions.

#### Enhanced Expression of *SAMDC16* Induces a Tolerance in Response to Salt Stress

To elucidate the physiological functions of PAs in response to salt stress, we first compared stress-induced damages between WT and PA-overproducing transgenic plants. We observed that leaves of *S16-S-4* transgenic plants showed greater tolerance to salt stress when compared to WT and vector control transgenic plants, as determined by trypan blue staining for cell death (**Figure 2A**) and attenuation in stress-induced chlorophyll degradation (**Figure 2B**). Furthermore, the maximal protective effects of *SAMDC* overexpression appeared after 2 days of salt stress, as determined by chlorophyll degradation.

The effects of increased PAs in the *S16-S-4* line were examined on the expression of metacaspase type II gene (*MCP2*), which plays a positive role in biotic and abiotic-induced PCD (Watanabe and Lam, 2011). Upon salt stress, the transcription levels of *MCP2* rapidly increased in both WT and *S16-S-4* transgenic plants, showing lower levels in *S16-S-4* as compared to WT (**Figure 2C**). Our findings show that upregulation of *MCP2* expression under salt stress is significantly attenuated in transgenic plants. Taken together, we believe that enhancement of endogenous PAs by overexpression of *SAMDC* gene induces stress tolerance based on cell damages under conditions of high salinity.

#### Downregulation of ROS Accumulation by PAs Under High Salinity

To investigate ROS generation in whole leaves of WT and transgenic tobacco plants following induction of high salinity with 200 mM NaCl, we histochemically monitored two important ROS, superoxide anion and hydrogen peroxide, using NBT and DAB as chromogenic substrates, respectively. NBT reacts with O2 •**<sup>−</sup>** to form a dark blue insoluble formazan compound, whereas DAB is oxidized by hydrogen peroxide to produce a reddishbrown precipitate (Kumar et al., 2013). Both ROS were detected in the leaves of WT from 1 h after salt stress, following which the levels were significantly increased until 24 h after salt stress (**Figures 3A**, **B**). However, both the stress-induced ROS were significantly downregulated in transgenic plants as compared to WT, suggesting that the upregulated PAs in transgenic tobacco plants inhibit ROS accumulation under salt stress. These results indicate that PAs have significant roles as antioxidative agents rather than acting as a source of hydrogen peroxide through PA degradation.

Next, by applying the DCFHDA histochemical assay, we analyzed ROS accumulation at the cellular level in response to salt stress. Since guard cells have been used as a well-developed single-cell model system and are particularly useful for the study of ROS signaling (Qi et al., 2017), we examined the histochemical pattern of ROS produced by the guard cells. Our data reveal that ROS are rapidly and transiently generated in guard cells in response to salt stress and significantly downregulated in transgenic plants of *S16-S-4* as compared to WT (**Figure 3C**). No differences were observed in rapid ROS accumulation in the WT and vector control transgenic plants under salt stress (**Supplementary Material Figure S2**). Surprisingly, only very low levels of ROS were produced in guard cells of *S16-S-4*, compared to WT. These results imply that stress-induced ROS generation is effectively inhibited by upregulated PA levels by overexpression of the *SAMDC* gene.

In aerobic organisms, ROS are produced during normal cellular metabolism as by-products of metabolic pathways and electron flows in both mitochondria and chloroplasts. RboH, called neutrophil NADPH oxidase, is a transmembrane protein that generates superoxide radicals in plant cells. Its isoforms, NtRbohD and NtRbohF, are expressed in all tobacco plants and produce the superoxide anion, which is unable to permeate cell membranes under ambient pH conditions due to the presence of a negative charge. The highly reactive superoxide anion is catalyzed to O2 and H2O2 either spontaneously or by superoxide dismutase (SOD) (Libik-Konieczny et al., 2015). H2O2 is less reactive than O2 •**<sup>−</sup>** but is more stable and can diffuse through membranes *via*  aquaporins (Bienert et al., 2007). Thus, H2O2 is recognized as the most potent signaling ROS in plants. We therefore determined gene expression profiles of tobacco plants in the context of ROS, including H2O2, following salt stress.

Transcription levels of the two *NtRboh* genes were evaluated in tobacco leaves using real-time qPCR analysis. Transcription of *NtRbohD* and *NtRbohF* occurred in a time-dependent manner, which were significantly downregulated in leaves of transgenic plants during the entire period of stress treatment, as compared

were generated from one representative experiment with three independent biological replicates after verifying the reproducibility of the results in three experiments. An asterisk indicates a significant difference between WT and transgenic plants (\*\**P* < 0.01).

to WT (**Figures 3E**, **F**). These results suggest that PAs inhibit ROS generation by downregulating the expression of both *NtRbohD* and *NtRbohF* genes during stress responses. Therefore, our findings confirm that increased PAs inhibit the generation and accumulation of total ROS.

We further determined the accumulation of each superoxide anion and hydrogen peroxide in the subcellular regions using specific fluorescent dyes. Intracellular generation of O2 •− was detected using BES-So-AM, a highly specific fluorescent probe (Park and Roubelakis-Angelakis, 2018). Under control conditions, no significant accumulation of O2 •− was detected in the guard cells of WT and *S16-S-4* (**Figure 4A**, 0 h). After 1 and 6 h of salt stress, the levels of intracellular O2 •− were increased significantly in guard cells of WT, with ROS accumulation in the chloroplasts of guard cells, and a very large amount of ROS accumulation especially in the nucleus. At 6 h after stress treatment, the accumulated ROS in the chloroplast were considerably small but still very large in the guard cell nuclei of WT tobacco leaves. However, the accumulation of intracellular O2 •− was dramatically reduced in the chloroplasts and nucleus of guard cells of the *S16-S-4* transgenic plants (**Figure 4A**).

Next, extracellular O2 •− levels were detected using BES-So. Similar to the intracellular O2 •−, no significant accumulation of fluorescent BES-So was detected under control conditions in the guard cells of WT and *S16-S-4* transgenic plants (**Figures 4A, C**, 0h). Although extracellular ROS levels were observed somewhat after 1 h, increased levels were observed after 6 h on the exterior of guard cells in WT under salt stress condition (**Figures 4B**, **D**, 1 and 6 h). In addition, a higher amount of BES-So fluorescence was observed in the stomatal pores of WT tobacco leaves under salt stress, indicating that stress-induced superoxide anions were localized outside the guard cells, since its anionic property prevented entry

FIGURE 3 | Kinetics of ROS production in response to salt stress. (A) Superoxide radical (sporadic O2 •− accumulation in leaves was detected after NBT staining in WT and transgenic plants with or without 200 mM NaCl. (B) Sporadic accumulation of H2O2 was determined by 3,3′-diaminobenzidine (DAB) staining in leaves from WT and transgenic plants after administering 200 mM NaCl. (C, D) Histochemical analysis of cellular reactive oxygen species (ROS) accumulation in response to salt stress. ROS accumulation was determined by incubation with 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) for 10 min. Staining images of leaves were obtained by confocal microscopy (C) and quantified by ImageJ software (D). (E, F) Relative mRNA levels of *NtRbohD* and *NtRbohF* genes after salt stress. Transcription levels of *NtRbohD* or *NtRbohF* are expressed as means ± SD. Transcription levels are expressed relative to the reference gene β*-actin* after qPCR. The photographs represented (A–C) are from one representative experiment after verifying the reproducibility of the results in three independent experiments with three leaves. Data were generated from one representative experiment with ten guard cells (D) and three independent biological replicates (E, F). An asterisk indicates a significant difference between WT and transgenic plants (\**P* < 0.05; \*\**P* < 0.01).

•− (C) and extracellular O2 by ImageJ. The photographs represented (A, B) are from one representative experiment with three leaves after verifying the reproducibility of the results in three experiments. Data (C, D) were generated from 10 cells in one representative experiment after verifying the reproducibility of the results in three experiments. An asterisk indicates a significant difference between WT and transgenic plants (\*\**P* < 0.01).

into the cells. These results suggest that increased levels of PAs are responsible for significant reduction in intra- and extracellular accumulation of superoxide anion in response to salt stress. Differences between WT and transgenic plants were significantly greater in the late phase under salt stress, implying that endogenous PAs might have more important effects on the later stage in response to salt stress.

We then determined the intra- and extracellular H2O2 accumulations after salt stress. Since DCF and DAB are highly sensitive to peroxidase (Noctor et al., 2016), we employed a peroxidase-independent method for the estimation of H2O2 levels. We used the highly specific BES-H2O2 and BES-H2O2-Ac probes to estimate intra- and extracellular H2O2 levels, respectively (**Figure 5**). Higher levels of intra- and extracellular H2O2 accumulation were observed in the guard cells of WT than in corresponding *S16-S-4* plant cells under salt stress (**Figure 5A**). Intra- and extracellular H2O2 remained at significant levels after 6 h of salt stress in leaf cells of WT. However, H2O2 accumulation was only marginally detected by BES-H2O2 and BES-H2O2-Ac in the *S16-S-4* transgenic plants under salt stress (**Figures 5A**, **B**). Our results additionally imply that BES-H2O2 and BES-H2O2-Ac enter the guard cells under salt stress conditions.

#### Downregulation in the Expression of ROS-Detoxifying Enzymes by PAs Under Salt Stress

We next investigated whether PAs contribute to the expression of ROS-detoxifying enzymes in stress-induced ROS accumulation. An enzymatic dismutation reaction converts superoxide into a more stable, membrane-permeable H2O2 derivative, which is required for cell-to-cell signaling. ROS-scavenging enzymes such as SOD, ascorbate peroxide (APX), and catalase (CAT) provide the cells with a highly efficient machinery for detoxifying superoxide and H2O2 (Foyer and Noctor, 2005).

The level of ROS is tightly regulated by enzymes involved in the ROS-detoxifying pathways, including mitochondrial manganese-SOD (MnSODmi), cytosolic copper/zinc SOD (CuZnSODc), cytosolic APX (APXc), CAT (CAT1 and CAT2), and phi glutathione-S-transferase (GSTF). In this study, *CAT1*

determined using confocal scanning microscopy followed by incubation with BES-H2O2-Ac. Images of H2O2 stained with BES-So-Am (green) and chlorophyll autofluorescence (red). (B) Extracellular hydrogen peroxide was determined using confocal scanning microscopy by incubation with BES-H2O2. Images of the H2O2 stained with BES-H2O2 (green) and chlorophyll autofluorescence (red). (C, D) Green fluorescence signals for intracellular H2O2 (C) and extracellular H2O2 (D) were quantified by ImageJ. The photographs represented (A, B) are from one representative experiment with three leaves after verifying the reproducibility of the results in three experiments. Data (C, D) were generated from 10 cells in one representative experiment after verifying the reproducibility of the results in three experiments. An asterisk indicates a significant difference between WT and transgenic plants (\*\**P* < 0.01).

and *CAT2* expressions were induced biphasically, which peaked at 6 h, after which they decreased, but again showed a high level at 24 h (**Figure 6**, *CAT1* and *CAT2*). On the other hand, expressions of other ROS-detoxifying enzymes in response to salt stress were significantly induced monophasically. Salt stress upregulated the transcription of *SODs*, *MnSOMmi*, and *CuZnSODc*, reaching maximum peaks at 24 and 12 h, respectively, with accompanying elevation of *APXc* transcription peaking at 6 h for detoxification of H2O2. Plant GSTs are known to induce tolerance to abiotic stresses due to their ability to regulate specific redox signaling pathways responsible for activating defense gene transcription (Cummins et al., 2011). Expression of GSTF, a plant-specific phi class of stress-induced GST, reached a peaked at 12 h under salt stress. Taken together, these results indicate that PAs activate the ROS detoxification pathway, thereby lowering ROS accumulation and resulting in the prevention of severe stressinduced cell damages.

### Downregulation of Oxidized Proteins by PAs Under Salt Stress

Oxidized proteins were detected using an OxyBlot protein oxidation detection kit (Merck Millipore, USA), according to the manufacturer's instructions. After dinitrophenyl hydrazine was added to crude total proteins to derive carbonyl groups from the protein side chains, Western blot analysis was performed using the provided 2,4-DNP antibody (1:150). DNP signals in integrated intensity of each fraction were quantified by densitometry in ImageJ and normalized to the total protein value of the WT 0 h control, which was set as 1.

We observed that, under unstressed condition, the amount of oxidized proteins was much higher in WT than in transgenic plants. In addition, salt stress caused a rapid increase in the amount of oxidized protein, which was higher in leaves of WT as compared to transgenic plants (**Figure 7**). These results indicate that the upregulation of endogenous PAs in transgenic tobacco

the results in three experiments. An asterisk indicates a significant difference between WT and transgenic plants (\**P* < 0.05; \*\**P* < 0.01).

plants might show antioxidative effects for protein homeostasis against stress-induced protein oxidation.

#### Exogenous Addition of PA Downregulates ROS Accumulation and Upregulates the Chaperone Activity Induced by Salt Stress

Abiotic stresses such as salt stress usually cause the accumulation of not only oxidized proteins but also protein aggregation (Luo et al., 2017). Molecular chaperones are key components contributing to the homeostasis and the quality control of proteins under stress conditions (Wang et al., 2004). In the current study, MDH protein alone as the target protein showed substantial aggregation after 6 min at 45°C, whereas proteins purified from the leaves after cotreatment with salt stress and PAs for 6 h induced a marked reduction in light scattering, indicating that PAs prevent the heat-induced aggregation of cellular target proteins and enhance the chaperone activity of cellular proteins under salt stress (**Figure 9**). Two-way ANOVA found statistically significant effects of PAs for chaperone activity [*F*(60, 156) = 2.13, *P* < 0.0001] compared to mock- and NaCl-treated conditions. Spd and Spm show similar induction of chaperone activity, and their effect was significantly higher than that exerted by Put.

#### DISCUSSION

PAs are small ubiquitous polycations involved in numerous processes of plant growth and development and are well known for their antisenescence and antistress effects due to their acid neutralizing and antioxidant properties, as well as for their membrane and cell wall stabilizing abilities (Gill and Tuteja, 2010; Tiburcio et al., 2014). Engineered PA accumulation by the expression of the yeast SAMdc enhances phytonutrient content, juice quality, and vine life in tomato, which supplied the direct evidence for a physiological role of PAs and demonstrates an approach to improving nutritional quality (Mehta et al., 2002). Mattoo et al. (2007) suggested that higher levels of Spd and Spm in the tomato fruit causes a marked shift for antisenescence in gene expression, especially for genes involved in metabolism, signal transduction, and defense/stress responses. It has been suggested that Put plays more active roles against pathogen attacks (Rodrıguez-Kessler et al., 2008), whereas Spd and Spm modulate the defense response of plants to numerous environmental stresses including metal toxicity, oxidative stress, drought, salinity, and chilling stress (Liu et al., 2000). Handa and Mattoo (2010) proposed that due to the chemical attributes of PAs and their derivatives/conjugates, they interact differentially with cellular components including chromatin, transcriptional

expressed as means ± SD. The photographs represented (A) are from one representative experiment in four independent biological experiments after verifying the reproducibility of the results. Data (B) were generated from four independent biological replicates. An asterisk indicates a significant difference between WT and transgenic plants (\*\**P* < 0.01).

machinery, translational machinery, and macromolecules, and result in modified metabolic profiles. However, little is understood regarding the detailed physiological mechanism of elevated PA levels in response to abiotic and biotic stresses.

The PA-derived H2O2 by PAO triggers signal transduction pathways causing defense gene expression and stress tolerance in plants, particularly due to the potential of H2O2 to act as a secondary messenger and a signaling molecule (Tavladoraki et al., 2012; Handa et al., 2018). However, it is possible that PAO is involved in the back-conversion reactions during PA catabolism, in which Spm is oxidized to Spd, and Spd is further oxidized to Put (Cona et al., 2006; Tavladoraki et al., 2012).

As previously reported, there exists a feedforward amplification loop between the apoplastic PAO and the plasma membrane NADPH oxidase for controlling ROS accumulation in response to abiotic stress (Gémes et al., 2016). This model suggests that the

(D) Oxidation index (integrated density/area) was quantified with Oxyblot by ImageJ. The photographs represented (A) and data (B) were generated from 10 cells in one representative experiment after verifying the reproducibility of the results at three experiments. The photograph (C) are from one representative experiment after verifying the reproducibility of the results in four experiments. Data (D) were expressed as means ± SD from four independent experiments. An asterisk indicates a significant difference between WT and transgenic plants (\**P* < 0.05; \*\**P* < 0.01).

apoplastic PAO feeds a stress-inducible ROS amplification loop that can lead to tolerance responses during stresses. Although it is hypothesized that PAO-related ROS triggers further cascade of ROS signaling for stress tolerance, apoplastic ROS was not detected in our histochemical experiments in *S16-S-4* transgenic plants. The temporal differences between transcript levels of *PAO* and *SAMDC16* indicate that the early increase in *PAO* is not induced due to the catabolism of stress-induced PAs. We therefore hypothesize that the catabolism of PA occurs independently from stress-induced PA accumulation.

The proposed role of PAs is for mediating stress responses through redox homeostasis as antioxidative molecules (Saha

treatment under salt stress. Light scattering of total proteins was determined using a model substrate MDH (1 μM) under thermal denaturing conditions (45°C) for 30 min. Data were expressed as means ± SD from three independent biological replicates. The statistical analysis was performed with two-way ANOVA using GraphPad Prism, comparing NaCl vs. NaCl + Put, NaCl vs. NaCl vs. NaCl + Spd, or NaCl vs. NaCl + Spm. Significant differences are displayed (\**P* < 0.05; \*\**P* < 0.01; \*\*\**P* < 0.001).

et al., 2015). One possible conjecture is that the PA molecules themselves induce stress-induced ROS detoxification and/or encompass an antioxidative property. The remnant extracellular superoxide anion is thought to cause serious damage by salt stress. However, the accumulation of extracellular superoxide anion is dramatically inhibited in *S16-S-4* under conditions of salt stress. These data indicate that under salt stress, PAs are bestowed with tolerance by the removal of extracellular superoxide anions.

Although NADPH oxidase plays a role in the production of extracellular superoxide anions, various mechanisms are involved in the production of different ROS, including H2O2. Transcript accumulation of *NtRbohD* and *NtRbohF* was significantly inhibited in *S16-S-4* transgenic plants and by exogenously added PAs, indicating that ROS production is specifically inhibited by the increase in cellular PAs in transgenic plants. Even 1 h after salt stress treatment, in which *PAO* transcripts exhibited a maximum value, only a low level of apoplastic H2O2 was observed in *S16-S-4*. Therefore, stress-induced *PAO* transcripts do not seem to contribute significantly to induce extracellular H2O2 accumulation in transgenic plants.

We also observed that superoxide anions and hydrogen peroxide prominently accumulate in the nucleus after salt stress. Intracellular ROS are mainly produced in chloroplasts, endoplasmic reticulum, and peroxisomes and, to a lesser extent, in mitochondria in plants (Sharma et al., 2012; Qi et al., 2017). It is yet unknown how ROS accumulates in the nucleus of plant cells. Recently, however, the isoform of NADPH oxidase, which produces ROS in the nucleus, has been reported in animal cells including human vascular epithelial cells (Kuroda et al., 2005; Weyemi et al., 2012). In plant cells, it is suggested that ROS produced in the chloroplast or peroxisome may be transferred into the nucleus (Shapiguzov et al., 2012). On the other hand, it is impossible to rule out the possibility that ROS are produced directly in the nucleus. The fact that nuclear ROS is significantly downregulated by PAs implies that PAs are effective on the nuclear gene expression for stress tolerance in a redoxdependent manner.

Stress-induced ROS burst and oxidative stress drive the protein oxidation and result in cellular toxicity that culminates in cell death (Sharma et al., 2012). The effects of PAs imparting antioxidative properties and antiaggregation of proteins contribute towards maintaining physiological cellular functions against abiotic stresses. It is suggested that these functions of PA are advantageous for protein homeostasis during abiotic stresses. Transcriptome analysis in transgenic fruit that accumulates higher PAs revealed upregulation of chaperones transcripts as compared to WT control (Mattoo et al., 2007). Interestingly, recent reports indicate that high concentrations of Spm (2.5 mM) and Spd (5 mM) exhibit chaperone-like activity against thermal and oxidative stress when incubated with bovine seminal plasma proteins (Singh et al., 2017).

The cytotoxicity of PAs through their oxidation resulted in PCD, which triggers the caspase-3 activity in leukemia cells (Stefanelli et al., 1999). In some cases, analogues of Spm are known to induce PCD in animal systems by activation of PA degradation and subsequent generation of H2O2 (Walters, 2003). Only transient enhancement of PA levels was observed in both WT and transgenic plants, which induce tolerant responses under salt stress condition. Taken together, these results indicate that the amounts of increased PA under salt stress were below the threshold levels of toxicity.

Increased resistance to stress correlates to extended leaf longevity and the overall life span (Sharabi-Schwager et al., 2009). They reported that older leaves had lower levels of antioxidants and persistently lower activities of the antioxidative enzymes in tobacco plants. These changes resulted in lower stress tolerance, with consequent acceleration of leaf senescence. Our current research supports the inference that increased resistance to salt stress results in attenuation of leaf senescence in transgenic plants. Taken together, our findings are consistent with the known role of Spd and Spm as antisenescence effectors (Mattoo et al., 2007). The next study will be towards understanding the roles of PAs and ROS signals with ROS generation and detoxification,

### REFERENCES


antioxidative machinery, and molecular chaperone activity for the fine orchestration of stress tolerance.

### DATA AVAILABILITY STATEMENT

The datasets generated for this study are available on request to the corresponding author.

### AUTHOR CONTRIBUTIONS

KYP designed the experiments, analyzed the data, and wrote the manuscript. SYS and YJK executed all experiments and data analysis.

### FUNDING

This research was supported by funding from National Research Foundation of Korea (Project No. NRF-2017R1D1A3B03034134) to KYP.

### ACKNOWLEDGMENTS

The authors thank Prof. K.A. Roubelakis-Angelakis for giving helpful advice on interpreting the data.

### SUPPLEMENTARY MATERIALS

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.01331/ full#supplementary-material


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Seo, Kim and Park. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Differential Nitrogen Nutrition Modifies Polyamines and the Amino-Acid Profile of Sweet Pepper Under Salinity Stress

M. C. Piñero<sup>1</sup> , Manuel E. Porras<sup>2</sup> , Josefa López-Marín<sup>1</sup> , Mari C. Sánchez-Guerrero<sup>2</sup> , Evangelina Medrano<sup>2</sup> , Pilar Lorenzo<sup>2</sup> and Francisco M. del Amor<sup>1</sup> \*

<sup>1</sup> Department of Crop Production and Agri-Tecnology, Murcia Institute of Agri-Food Research and Development, Murcia, Spain, <sup>2</sup> Agricultural Research and Development Centre of Almería (IFAPA), Almería, Spain

#### Edited by:

Ana Margarida Fortes, Universidade de Lisboa, Portugal

#### Reviewed by:

Christine Becker, Hochschule Geisenheim University, Germany Oscar A. Ruiz, CONICET Institute of Biotechnological Research (IIB-INTECH), Argentina

#### \*Correspondence:

Francisco M. del Amor franciscom.delamor@carm.es

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 29 October 2018 Accepted: 25 February 2019 Published: 02 April 2019

Citation:

Piñero MC, Porras ME, López-Marín J, Sánchez-Guerrero MC, Medrano E, Lorenzo P and del Amor FM (2019) Differential Nitrogen Nutrition Modifies Polyamines and the Amino-Acid Profile of Sweet Pepper Under Salinity Stress. Front. Plant Sci. 10:301. doi: 10.3389/fpls.2019.00301 The horticultural industry demands high-quality resources to achieve excellence in yield and optimal revenues. Nitrogen is a pivotal nutrient to accomplish these goals for plant growth and product quality. However, competition for water in semi-arid regions can force the use of brackish waters, which can impair N uptake. The lower N uptake can be due to several reasons, such as an antagonism between ions, an absence of ATP, and/or alteration of N metabolism. The effect of supplying N as NO<sup>−</sup> 3 alone or in combination with NH<sup>+</sup> 4 , coupled with low or high salinity (8 or 20 mM NaCl), has been studied in sweet pepper fruits (Capsicum annuum L. cv. Melchor). The application of NH<sup>+</sup> 4 at high salinity affected chromatic parameters (a<sup>∗</sup> , b<sup>∗</sup> , and C<sup>∗</sup> ), while chlorophyll a and b levels declined and β-carotene increased. The concentrations of P, K, Ca, Mg, and Cu were reduced in the fruits of plants irrigated with NH<sup>+</sup> 4 . The concentration of Na was only reduced when NH<sup>+</sup> <sup>4</sup> was supplied. Likewise, the concentration of total phenolics was also reduced at high salinity. However, total protein was unaffected. The amino acid profile was altered by the supply of NH<sup>+</sup> 4 , which reduced the concentrations of histidine and phenylalanine. Moreover, the concentrations of putrescine and cadaverine were increased by NH<sup>+</sup> 4 at high salinity, whereas that of cadaverine was reduced by NH<sup>+</sup> 4 at low salinity. The observed changes in fruit quality triggered by salinity, under the conditions of this study, should be borne in mind for this crop with regard to the envisaged palliative effect of the supply of N-NH<sup>+</sup> 4 .

Keywords: nitrogen, salinity, Capsicum annuum L., nutrients, free amino acids, polyamines

## INTRODUCTION

Consumers are increasingly demanding high-quality products with benefits for human health (Bortolin et al., 2016). However, the agricultural industry usually prioritizes the maximization of yield and quality without investing in the improvement of plant tolerance of potentially harmful abiotic stress factors, which frequently impair crop yields. The current climate drift that presumably will increase the frequency and severity of dry heatwaves is now a source of concern to many growers worldwide, but especially in Mediterranean-climate areas (Surasinghe, 2011).

Therefore, new, efficient, and effective ameliorative tools are necessary, to reduce the impacts and to maintain ecosystem biodiversity, sustainability, economic growth, food security, and social equity (Dhankher and Foyer, 2018). The competition for good-quality water, which is forcing farmers to use brackish waters for irrigation, could be alleviated by the use of salt-tolerant crops or appropriate management (fertirrigation) practices (Piñero et al., 2018). Salinity is one of the main stressors limiting plant development and crop productivity. Munns and Tester (2008) observed that under salinity stress the plant N content declines while the leaf Cl<sup>−</sup> content increases. However, the negative effects of salinity could, in part, be compensated by an appropriate, balanced N supply. Some studies indicated that the supply of NO<sup>−</sup> 3 as the sole N source may be detrimental under salt stress, since salinity reduces its uptake rate (Kant et al., 2007) and decreases the plant N content (Piñero et al., 2016). Several reasons have been put forward for the decrease in leaf N content under salinity: (i) an antagonism between Cl<sup>−</sup> and NO<sup>−</sup> 3 transport and/or inactivation of NO<sup>−</sup> 3 transporters by the toxic effects of salinity; (ii) an absence of ATP, which is required for active NO<sup>−</sup> 3 transport (Piñero et al., 2016); and (iii) an alteration in the activities of enzymes involved in N metabolism induced by salt (Srivastava and Mishra, 2014). The supply of NH<sup>+</sup> 4 has an advantage, related to the lower energy cost of its uptake, compared with NO<sup>−</sup> 3 (Fernández-Crespo et al., 2012), but it can induce cell acidification, nutrient deficiencies, and inhibition of root growth (Sarasketa et al., 2016). Moreover, NH<sup>+</sup> 4 has also been associated with amino acids that can act as signaling molecules to trigger metabolic pathways that could limit oxidative damage (Hessini et al., 2013), such as that provoked by salinity stress. Consequently, the growth response of plants to N fertilization under salinity stress varies depending on whether N is supplied as NO<sup>−</sup> 3 or NH<sup>+</sup> 4 , as well as on the species considered (Misra and Dwivedi, 1990).

Sweet pepper (Capsicum annuum L.) is one of the most valuable crops in the Mediterranean area and is usually cultivated in greenhouses, which allow higher yield and exceptional fruit quality in comparison with open field (conventional cultivation) conditions (Serret et al., 2018). The fruit contains a large number of health-promoting compounds such as vitamins, carotenoids, lycopene, capsaicinoids, phenolic compounds, and amino acids, all of which have antioxidant properties and provide protection against cancer (Bramley, 2000; Navarro et al., 2006; González-Chavira et al., 2018). Moreover, the synthesis of phenolic compounds, vitamin C, and carotenoids in pepper and other vegetables depends on several factors, such as the cultivar, agricultural practices, maturity, and storage conditions. It has been observed that the availability of N has the ability to modify the synthesis of phenolic compounds and soluble solids (Doll et al., 1994; González-Chavira et al., 2018).

Polyamines, also known as biogenic amines, are low molecular weight compounds involved in plant responses to abiotic and biotic stresses (Hussain et al., 2011), as well as in the processes of cell growth and differentiation (Regla-Márquez et al., 2016). They are also linked to molecular plant defense mechanisms involved in the scavenging and/or generation of free radicals, regulation of gene expression, and formation of toxic defense products (Cai et al., 2015). Moreover, polyamines are also related to the hormonal balance (Houdusse et al., 2008) and the direct effect of climate change, as influenced by CO<sup>2</sup> concentrations (Piñero et al., 2017).

In this study, we hypothesized that the additional supply of NH<sup>+</sup> 4 to the nutrient solution (instead of using NO<sup>−</sup> 3 as the sole N source) might partially or totally overcome the predicted deleterious effect of salinity on the fruit quality of sweet pepper, as this crop is considered salt sensitive (del Amor and Cuadra-Crespo, 2011; Piñero et al., 2014). Most plants show a strong preference for NO<sup>−</sup> 3 over NH<sup>+</sup> 4 ions, and others grow best if they have access to both NO<sup>−</sup> 3 and NH<sup>+</sup> 4 (Zhou et al., 2011). However, the optimum NO<sup>−</sup> 3 /NH<sup>+</sup> 4 ratio changes depending on the plant species, the stage of development, and the environmental conditions (Hu et al., 2015). Since the reported studies concerning the effect of different N forms and salinity on sweet pepper quality are very few, this study offers a new insight into the impact of the interaction between the NO<sup>−</sup> 3 /NH<sup>+</sup> 4 ratio and salinity, in this important crop. Therefore, our objective was to determine the influence of the NO<sup>−</sup> 3 /NH<sup>+</sup> 4 nutritional regime, under salt stress, on the fruit colorimetric properties, pigments (chlorophylls, lycopene, and β-carotene), mineral composition, total protein and total phenolics contents, polyamines, and amino-acids profile.

### MATERIALS AND METHODS

#### Plant Material and Growth Conditions

The experiments took place in a greenhouse located in the IFAPA center "La Mojonera" (Almería, Spain, latitude 36◦ 48<sup>0</sup> N, longitude 2◦ 41<sup>0</sup> W). It was of the multi-tunnel type (3 spans), oriented east–west, and had an area of 720 m<sup>2</sup> . The greenhouse height was 4.7 m to the ridge and 3 m to the gutter of each span. The greenhouse cover was thermal polyethylene (PE) (0.2 mm thick). Ventilation was provided by two roof vents (opening area: 1 m × 30 m) and two side vents (opening area: 1.5 m × 26 m) set in both the south and north arches. The greenhouse was equipped with a commercial climate control system (CDC, INTA S.A.). A continuous register of the temperature and humidity (HMP45C sensors, Campbell Sci.) was maintained. In the greenhouse, pepper seedlings (cv. Melchor) were transplanted on 19 August 2013, two plants into each 27-L container filled with perlite, with a density of 2.5 plants per m<sup>2</sup> . The cultivation was carried out with a type of Dutch pruning; two stalks were left on each plant.

Water and fertilizer were delivered by an automated drip irrigation system (CDN, INTA S.A.). The nutrient solution was applied with one dripper (3 L h−<sup>1</sup> ) per container. Two levels of nutrient solution salinity were used: 8 mM NaCl (C) for half of the plants and 25 mM NaCl (S) for the other half. Also, the plants in each salinity treatment were supplied with different N sources: half received NO<sup>−</sup> 3 alone and the other half a NO<sup>−</sup> 3 /NH<sup>+</sup> 4 mixture. To adjust the N input to the crop demand, two phases were established. In the first phase (until October 31) the N inputs


TABLE 1 | Effect of two different N forms (NF) (NO<sup>−</sup> 3 and NO<sup>−</sup> 3 /NH<sup>+</sup> 4 ) and salinity (8 and 25 mM NaCl) on sweet pepper fruits: CIEL∗a ∗b <sup>∗</sup> color coordinates.

Different letters within a column indicate significant (P ≤ 0.05) differences between treatments. <sup>a</sup>Analysis of variance: ns, not significant; <sup>∗</sup>P ≤ 0.05; ∗∗P ≤ 0.001. <sup>b</sup>Nitrogen form.

were: NO<sup>−</sup> 3 12 mM NO<sup>−</sup> 3 and NO<sup>−</sup> 3 /NH<sup>+</sup> 4 10 mM NO<sup>−</sup> <sup>3</sup> + 2 mM NH<sup>+</sup> 4 ; in the second phase (October 31 to the end of the cycle) the contributions were: NO<sup>−</sup> 3 10 mM NO<sup>−</sup> 3 and NO<sup>−</sup> 3 /NH<sup>+</sup> 4 8 mM NO<sup>−</sup> <sup>3</sup> <sup>+</sup> 2 mM NH<sup>+</sup> 4 . The volume of nutrient solution supplied by each irrigation event was 500 mL per container. The irrigation frequency fluctuated between one and five times per day depending on the needs of the plants, maintaining approximately 40% drainage. The harvest period was between 28 October 2013 and 24 February 2014, the fruits being harvested once they had reached commercial maturity (red color). At the second harvest (29 January 2014) 12 fruits per treatment were collected from different plants, for the measurement of color, chlorophylls, lycopene, β-carotene, mineral content, total proteins, total phenolic compounds, amino acids, and polyamines. Two fruits were considered a sample; therefore, the analyses were carried out using six replicates per treatment.

#### Skin Color

Pepper fruit color was measured with a Konica-Minolta CR-300 colorimeter (Konica-Minolta, Kyoto, Japan) having a D65 illuminant, making three measurements along the equatorial perimeter. The color data are provided as CIEL∗a ∗b ∗ coordinates, which define the color in a three-dimensional space. C ∗ is chroma [C <sup>∗</sup> = √ (a ∗2 ) + (b ∗2 )], 0 being at the center of a color sphere, and the value increases according to the distance from the center. Finally, hab is the hue angle [hab = arc tg( b ∗a ∗ )], which is defined


as starting at the +a ∗ axis and is expressed in degrees; 0◦ would be +a ∗ (red), 90◦ would be +b ∗ (yellow), 180◦ would be −a ∗ (green), and 270◦ would be −b ∗ (blue).

#### Fruit Chlorophylls, Lycopene, and β-Carotene

The β-carotene, lycopene, and chlorophylls were extracted from 1-g samples of frozen pepper fruits (−80◦C) with 25 mL of acetone–hexane (2:3) solvent. The samples were homogenized using a polytron and centrifuged at 3,500 rpm for 6 min, at 4 ◦C. Subsequently, the optical density of the supernatant was measured spectrophotometrically at wavelengths of 663, 645, 505, and 453 nm. The concentrations of chlorophylls a and b, lycopene, and β-carotene were calculated using Nagata and Yamashita (1992) equations:

Chlorophyll a (mg100 mL−<sup>1</sup> ) = 0.999∗A<sup>663</sup> − −0.0989∗A<sup>645</sup> .

Chlorophyll b (mg100 mL−<sup>1</sup> ) = − 0.328∗A<sup>663</sup> + 1.77∗A<sup>645</sup> .

Lycopene (mg100 mL−<sup>1</sup> ) =

− 0.0458∗A<sup>663</sup> + 0.204∗A<sup>645</sup> + 0.372∗A<sup>505</sup> − 0.0806<sup>∗</sup> A453.

β-Carotene (mg100 mL−<sup>1</sup> ) =

0.216∗A<sup>663</sup> − 1.22∗A<sup>645</sup> − 0.304∗A<sup>505</sup> + 0.452<sup>∗</sup> A453.

### Mineral Content

The Ca, K, Mg, B, Cu, Fe, Mn, P, and Na concentrations in the dry matter of fruits were determined with an inductivelycoupled plasma (ICP) spectrometer (Varian Vista MPX, Palo Alto, CA, United States). An ETHOS ONE microwave digestion system (Milestone, Inc., Shelton, CT, United States) was applied for fruit sample preparation. This digestion procedure has many advantages due to the speed of the digestion process, the lower acid consumption and lower possibility of contamination, and the high extraction efficiencies (Soylak et al., 2004).

#### Total Protein

The fruit dry weight was determined after at least 72 h at 70◦C and the total nitrogen was measured in the dry matter, using a LECO FP-528 (Leco Corporation, St. Joseph, MI, United States). We use the conversion factor 6.25 to convert total nitrogen to total protein (Jones et al., 1942).

#### Total Phenolic Compounds

The total phenolic compounds were extracted from 0.5 g of frozen pepper fruit (−80◦C) with 5 mL of 80% acetone. The homogenate was centrifuged at 10,000 rpm for 10 min, at 4◦C. For the determination, Folin–Ciocalteu reagent was used, diluted with Milli-Q water (1:10). The diluted reagent (1 mL) was mixed with 100 µL of supernatant and 2 mL of Milli-Q water, and

fpls-10-00301 March 29, 2019 Time: 18:51 # 4

5 mL of sodium carbonate (20%) were then added. The mixture was kept for 30 min in the dark and then the absorbance was measured at 765 nm, according to the methodology of Kähkönen et al. (1999). The total phenolic content was expressed as gallic acid equivalents, in mg g−<sup>1</sup> fresh weight.

### Free Amino Acids

The free amino acids were extracted from fruits frozen at −80◦C: the homogenate was extracted, after vortexing at 5,000 rpm (10 min, 4◦C), and analyzed by the AccQ·Tag-ultra ultraperformance liquid chromatography (UPLC) method (Waters, UPLC Amino Acid Analysis Solution, 2006). For derivatization, 70 µL of borate buffer were added to the hydrolyzed sample or to 10 µL of the fruit homogenate. Next, 20 µL of reagent solution were added. The reaction mixture was mixed immediately and heated at 55◦C for 10 min. After cooling, an aliquot of the reaction mixture was used for UPLC injection. The UPLC was performed with an Acquity system (Waters, Milford, MA, United States) equipped with a fluorescence detection (FLR) system. A BEH C18 100 mm × 2.1 mm, 1.7 µm column (Waters) was used. The flow rate was 0.7 mL min−<sup>1</sup> and the column temperature was kept at 55◦C. The injection volume was 1 µL. The excitation (λex) and emission (λem) wavelengths were set at 266 and 473 nm, respectively. The solvent system consisted of two eluents: (A) AccQ·Tag-ultra eluent A concentrate (5%, v/v) and water (95%, v/v); (B) AccQ·Tag-ultra eluent B. The following elution gradient was used: 0–0.54 min, 99.9% A–0.1% B; 5.74 min, 90.9% A–9.1% B; 7.74 min, 78.8% A–21.2% B; 8.04 min, 40.4% A–59.6% B; 8.05–8.64 min, 10% A–90% B; 8.73–10 min, 99.9% A–0.1% B. Empower 2 (Waters) software was used for system control and data acquisition. External standards (Thermo Scientific) were used for quantification of (Ala) alanine; (Arg) arginine; (Asp) aspartic acid; (Cys) cysteine; (Glu) glutamic acid; (Gly) glycine; (His) histidine; (Ile) isoleucine; (Leu) leucine; (Lys) lysine; (Met) methionine; (Phe) phenylalanine; (Pro) proline; (Ser) serine; (Thr) threonine; (Tyr) tyrosine; and (Val) valine.

## Polyamine Analysis

Free polyamines were extracted by homogenizing 1.0 g of tissue in 10 mL of 5% perchloric acid, using a Polytron (Kinematica, Bohemia, NY, United States) homogenizer, and were analyzed by the benzoylation method (Serrano et al., 1995), using highperformance liquid chromatography (HPLC) (Hewlett-Packard). As an internal standard, 1,6-hexanediamine ([100 nmol (g fresh weight)−<sup>1</sup> of tissue] was used, and standard curves of putrescine, cadaverine, and histamine were prepared. The results are expressed as nmol (g fresh weight)−<sup>1</sup> .

### Statistical Analysis

The data were tested first for homogeneity of variance and normality of distribution. The significance was determined by analysis of variance (ANOVA) and the significance (P ≤ 0.05) of differences between mean values was tested by Duncan's New Multiple Range Test, using Statgraphics Centurion <sup>R</sup> XVI (StatPoint Technologies, Inc.). Four combinations of treatments were used, involving two N forms (NO<sup>−</sup> 3 or NO<sup>−</sup> 3 /NH<sup>+</sup> 4 ) and two levels of nutrient solution salinity (8 and 25 mM NaCl), with six replications per combination.

## RESULTS AND DISCUSSION

### Color, Chlorophylls, Lycopene, and β-Carotene

The fruit skin lightness (L<sup>∗</sup> ) values were only affected by salinity. However, the fruit of plants irrigated with NH<sup>+</sup> 4 -containing nutrient solution, averaged over the two salinity levels, had lower values of the parameters a<sup>∗</sup> , b<sup>∗</sup> , and C<sup>∗</sup> , compared with the supply of NO<sup>−</sup> 3 alone. Fruits of plants grown with NH<sup>+</sup> <sup>4</sup> were less red in color (a <sup>∗</sup> = 24.01 and b <sup>∗</sup> = 11.56) and had a less intense and less vivid color (C <sup>∗</sup> = 26.73), although such differences would hardly be perceived by consumers (**Table 1**). Additionally, the lower a<sup>∗</sup> values coincided with a higher chlorophyll a concentration and

lower contents of lycopene and β-carotene (**Figure 1**). Thus, this reduction in the intensity of the red color appears to be due to reduced chlorophyll degradation and a considerable reduction in the contents of lycopene and β-carotene. On the other hand, the salt treatment had no significant general effect on the CIELab color coordinates (with the exception of L<sup>∗</sup> ). This could be because the chlorophyll a increased in the same proportion (from 0.37 to 0.65 mg kg−<sup>1</sup> FW) as the degradation of chlorophyll b (from 0.74 to 0.32) (**Figures 1A,B**). Salinity significantly modified the lycopene and β-carotene contents in fruits (**Figures 1C,D**). The concentrations of lycopene and β-carotene were reduced at high salinity in fruits grown with NO<sup>−</sup> 3 alone. But, this effect was not observed when NH<sup>+</sup> <sup>4</sup> was added to the nutrient solution, especially for β-carotene – whose concentration was significantly increased. These responses highlight the interaction between the N form and salinity concerning chlorophyll a, lycopene, and β-carotene, while chlorophyll b was only clearly affected by salinity.

#### Mineral Content

The P, K, Ca, Mg, and Cu concentrations in pepper fruits were influenced significantly by the salinity and N form, decreasing with the presence of NH<sup>+</sup> 4 in the nutrient solution (**Table 2**). Thus, the N form used for fertilization can be considered an important factor in the uptake of mineral nutrients by plants (Kowalska and Sady, 2012). In plants, NH<sup>+</sup> 4 can act through different physiological and biochemical mechanisms – like acidification of the growth medium and NH<sup>+</sup> 4 -toxicity per se, leading to antagonism in cation uptake, and/or alterations in the osmotic balance (Horchani et al., 2010). Borgognone et al. (2013) studied the mineral composition of tomato and showed that the uptake of K, Ca, and Mg was lower when the proportion of NH<sup>+</sup> 4 in the nutrient solution was increased. These authors considered that this was due to the mechanism of charge balance in ion uptake, since N is a dominant macronutrient and its ionic form controls cation and anion uptake. Horchani et al. (2010) observed that NH<sup>+</sup> 4 toxicity led to antagonism in cation uptake and/or alterations in the osmotic balance, which lowered the uptake of cations. However, in the current work, the rise in the external salinity only had a notable effect on the concentration of Na, which increased from 64 to 322 mg kg−<sup>1</sup> DW and from 21 to 145 mg kg−<sup>1</sup> DW in fruits of plants irrigated with NO<sup>−</sup> 3 or NO<sup>−</sup> 3 /NH<sup>+</sup> 4 , respectively. This agrees with de Souza Miranda et al. (2016), who found that, under salinity, Na<sup>+</sup> accumulation was severely limited in the presence of NH<sup>+</sup> 4 . Likewise, Houdusse et al. (2008) pointed out that K and Ca levels declined in plants grown under salinity when they were supplied with NH<sup>+</sup> 4 . This agrees with our data for fruits and again emphasizes the effect of adding NH<sup>+</sup> 4 under saline stress conditions. Moreover, the concentrations of microelements (B, Mn, Fe, and Cu) were not affected by salinity when NH<sup>+</sup> <sup>4</sup> was included in the nutrient solution, but they were increased when salinity was imposed without it. Especially notable was the effect of the N form on Cu, since NH<sup>+</sup> 4 dramatically reduced the Cu concentration in fruits.

### Total Protein

When NH<sup>+</sup> <sup>4</sup> was supplied in the nutrient solution, the total protein content was reduced slightly, but did not differ

TABLE 3 | Effect of two different N forms (NF) (NO<sup>−</sup> 3 and NO<sup>−</sup> 3 /NH<sup>+</sup> 4 ) and salinity (8 and 25 mM NaCl) on sweet pepper fruits: amino acid profiles.


Different letters within a column indicate significant (P ≤ 0.05) differences between treatments. <sup>a</sup>Analysis of variance: ns, not significant; <sup>∗</sup>P ≤ 0.05; ∗∗P ≤ 0.01; ∗∗∗P ≤ 0.001. <sup>b</sup>Nitrogen form.

significantly from the levels of the control plants (**Figure 2A**). Likewise, salinity had no effect on the total protein content.

#### Total Phenolics

fpls-10-00301 March 29, 2019 Time: 18:51 # 7

The content of total phenolics was highest in fruits of plants irrigated with NH<sup>+</sup> 4 in the absence of salinity stress (**Figure 2B**). The total phenolic concentration was increased by 20% in the fruits of sweet pepper by the supply of NH<sup>+</sup> 4 in the nutrient solution, compared to the control (**Figure 2B**), in accordance with Leja et al. (2008). Abu-Zahra (2011) attributed differences in the concentrations of phenolics to nutrient availability. The importance of phenolic compounds lies in the nutritional, organoleptic, and commercial properties of agricultural food products, since they contribute to sensory properties such as color and flavor (Pérez-López et al., 2007). Phenolic compounds in our diet provide health benefits associated with a reduced risk of chronic diseases. It has been found that they have the ability to protect against cardiovascular disease and have anticarcinogenic properties due to their antioxidant activity and their role as free radical scavengers (Gani et al., 2012). Moreover, it is wellknown that environmental stresses stimulate the biosynthesis of phenolic compounds (Mitchell et al., 2007). Authors like Wahid and Ghazanfar (2006) found that greater synthesis of phenolics compounds is directly correlated with salt tolerance. However, our study clearly shows that the response to salinity differed according to the N form(s) supplied, being reduced when NH<sup>+</sup> 4 was added to the nutrient solution. Therefore, the beneficial effect for human well-being triggered by salinity (referred to earlier) is diminished under such a plant nutrition strategy.

### Free Amino Acids

Amino acids are essential for human health. They are required for the growth, development, regeneration, and reconstruction of the body and are responsible for the production of antibodies, blood cells, hormones, and enzymes (Zou et al., 2015). The concentration of total amino acids was highest in the fruits of the salinity-stressed plants that received NH<sup>+</sup> 4 (**Table 3**). In a more detailed analysis considering each amino acid individually, Gly was the most abundant free amino acid in the fruits of all treatments, and its relative content was not significantly affected by the N form or salinity. Moreover, Pro, Asp, and Ser, which are non-essential amino acids, followed Gly in abundance, in red fruits. The Ser concentration was significantly increased by the interaction between salinity and the supply of NH<sup>+</sup> 4 . By contrast, the supply of NH<sup>+</sup> 4 caused the greatest reduction in the concentrations of Tyr and His (64 and 52%, respectively), while the levels of Tyr (38%) and Met (36%) were also reduced by salinity. Thus, an interaction occurred, since the presence of NH<sup>+</sup> 4 in the nutrient solution only resulted in decreased Tyr concentrations under low salinity. Furthermore, salinity did not reduce the Met concentrations when NH<sup>+</sup> <sup>4</sup> was added to the nutrient solution.

#### Polyamines

Putrescine, cadaverine, and histamine were present in pepper fruit. However, histamine was scarcely detectable; therefore, only the data for putrescine and cadaverine are presented (**Figures 3A,B**). The putrescine levels in pepper fruit were not affected by salinity; however, the interactive effect of salinity and NH<sup>+</sup> 4 caused a sharp increase (of 76%, compared to the control; **Figure 3A**). The cadaverine levels were significantly reduced by the supply of NH<sup>+</sup> 4 , from 3.14 to 1.34 nmol g−<sup>1</sup> FW (**Figure 3B**), while the combined effect increased the cadaverine levels by 16%, compared to the control. The effects of salinity on polyamine biosynthesis have been studied in several plant species and the response seems to be dependent on the species, the plant system used, and/or the duration of exposure to salinity (Mutlu and Bozcuk, 2007). Houdusse et al. (2008), studying leaves of pepper plants, found that the effect of salinity depended on the N level. This agrees with Tiburcio et al. (1986), who indicated that dicotyledons submitted to osmotic stress had increased putrescine accumulation, but this was more dependent on the plant species considered (tomato) than on differences in N nutrition. In fruits, one of the main postharvest effects attributed to polyamines is the preservation of their flesh firmness, which has been shown to be enhanced

by salinity (Botella et al., 2000). Although NH<sup>+</sup> 4 did not lead to increased concentrations of putrescine or cadaverine at the lower salinity level, at the higher level both polyamines were increased by the supply of NH<sup>+</sup> 4 . Beyond their contributions to stress tolerance in plants, polyamines (aliphatic amines) are involved in the regulation of differentiation of immune cells, inflammatory reactions, intestinal immunoallergic responses, diabetes, and food allergy prevention in children (Dandrifosse et al., 2000; Moinard et al., 2005). Therefore, their increased accumulation in plants in response to stress could enable benefits to be derived from the consumption of the edible parts of plants grown with the appropriate crop management, as the level of polyamines decreases with age in vital animal organs (Das and Kanungo, 1982).

#### CONCLUSION

Irrigation with a nutrient solution in which NO<sup>−</sup> <sup>3</sup> was partly replaced by NH<sup>+</sup> 4 had important effects on the fruit quality of sweet pepper. There were changes in the color parameters and the mineral content was impaired, but at the same time the specific effect of a dramatic reduction in the Na concentration could be of paramount importance for specific dietary prescriptions. Additionally, the amino-acid profile was altered whilst the polyamines levels increased notably under salinity. Such responses should be taken into account in further studies that could include post-harvest quality and high salinity stress, to elucidate whether increasing polyamines levels can effectively counteract the deleterious effects of salinity on this crop. Moreover, crop management effects on polyamines

#### REFERENCES


accumulation should be taken into account with regard to dietary treatments of many ailments that affect the elderly.

#### DATA AVAILABILITY

The datasets generated for this study are available on request to the corresponding author.

#### AUTHOR CONTRIBUTIONS

FdA, PL, MS-G, and EM conceived and supervised the whole study. FdA and MCP wrote the manuscript with inputs from all authors. MEP carried out the field experiment. MCP analyzed the plant material. MCP and JL-M performed the statistical analysis. All authors discussed the results and provided critical feedback on the manuscript.

### FUNDING

This work has been supported by the Instituto Nacional de Investigaciones Agrarias (INIA), through project RTA2011- 00026-C02-01. Part of this work was also funded by the European Social Fund.

#### ACKNOWLEDGMENTS

We thank Dr. David J. Walker for assistance with the correction of the English. MCP and MEP are recipients of pre-doctoral fellowships from the INIA-CCAA.


stress: evidence for a priming effect of ammonium? Plant Soil 370, 163–173. doi: 10.1007/s11104-013-1616-1


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Piñero, Porras, López-Marín, Sánchez-Guerrero, Medrano, Lorenzo and del Amor. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Contribution of Maize Polyamine and Amino Acid Metabolism Toward Resistance Against Aspergillus flavus Infection and Aflatoxin Production

Rajtilak Majumdar<sup>1</sup> , Rakesh Minocha<sup>2</sup> , Matthew D. Lebar<sup>1</sup> , Kanniah Rajasekaran<sup>1</sup> , Stephanie Long<sup>2</sup> , Carol Carter-Wientjes<sup>1</sup> , Subhash Minocha<sup>3</sup> and Jeffrey W. Cary<sup>1</sup> \*

<sup>1</sup> Food and Feed Safety Research Unit, Southern Regional Research Center, United States Department of Agriculture, Agricultural Research Service, New Orleans, LA, United States, <sup>2</sup> United States Department of Agriculture Forest Service, Northern Research Station, Durham, NH, United States, <sup>3</sup> Department of Biological Sciences, University of New Hampshire, Durham, NH, United States

#### Edited by:

Rubén Alcázar, University of Barcelona, Spain

#### Reviewed by:

Manchikatla Venkat Rajam, University of Delhi, India Tibor Janda, Centre for Agricultural Research (MTA), Hungary

> \*Correspondence: Jeffrey W. Cary jeff.cary@ars.usda.gov

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 12 February 2019 Accepted: 08 May 2019 Published: 24 May 2019

#### Citation:

Majumdar R, Minocha R, Lebar MD, Rajasekaran K, Long S, Carter-Wientjes C, Minocha S and Cary JW (2019) Contribution of Maize Polyamine and Amino Acid Metabolism Toward Resistance Against Aspergillus flavus Infection and Aflatoxin Production. Front. Plant Sci. 10:692. doi: 10.3389/fpls.2019.00692 Polyamines (PAs) are ubiquitous polycations found in plants and other organisms that are essential for growth, development, and resistance against abiotic and biotic stresses. The role of PAs in plant disease resistance depends on the relative abundance of higher PAs [spermidine (Spd), spermine (Spm)] vs. the diamine putrescine (Put) and PA catabolism. With respect to the pathogen, PAs are required to achieve successful pathogenesis of the host. Maize is an important food and feed crop, which is highly susceptible to Aspergillus flavus infection. Upon infection, the fungus produces carcinogenic aflatoxins and numerous other toxic secondary metabolites that adversely affect human health and crop value worldwide. To evaluate the role of PAs in aflatoxin resistance in maize, in vitro kernel infection assays were performed using maize lines that are susceptible (SC212) or resistant (TZAR102, MI82) to aflatoxin production. Results indicated significant induction of both PA biosynthetic and catabolic genes upon A. flavus infection. As compared to the susceptible line, the resistant maize lines showed higher basal expression of PA metabolism genes in mock-inoculated kernels that increased upon fungal infection. In general, increased biosynthesis and conversion of Put to Spd and Spm along with their increased catabolism was evident in the resistant lines vs. the susceptible line SC212. There were higher concentrations of amino acids such as glutamate (Glu), glutamine (Gln) and γ-aminobutyric acid (GABA) in SC212. The resistant lines were significantly lower in fungal load and aflatoxin production as compared to the susceptible line. The data presented here demonstrate an important role of PA metabolism in the resistance of maize to A. flavus colonization and aflatoxin contamination. These results provide future direction for the manipulation of PA metabolism in susceptible maize genotypes to improve aflatoxin resistance and overall stress tolerance.

Keywords: Aspergillus flavus, s-adenosylmethionine decarboxylase, polyamine oxidase, mycotoxin, polyamine uptake, amino acids

### INTRODUCTION

fpls-10-00692 May 22, 2019 Time: 18:40 # 2

Mycotoxin contamination of food and feed crops is a global threat. The major fungal genera that are the primary contributors of mycotoxin contamination in crop plants are Aspergillus, Fusarium, Penicillium, and Alternaria. Among these fungi, Aspergillus flavus has the most adverse impact on crop loss and human/animal health (Ismaiel and Papenbrock, 2015; Mitchell et al., 2016; Umesha et al., 2016). The fungus infects oilseed crops such as maize and peanut where it produces carcinogenic aflatoxins and other toxic secondary metabolites (SMs). Maize is a major food and feed crop grown worldwide. In 2013, economic losses resulting from A. flavus contamination in maize grown in the U.S. were estimated to be \$686.6 million (Mitchell et al., 2016). The health impact of aflatoxin contamination is severe. Aflatoxin causes liver cancer, stunted growth, and A. flavus causes aspergillosis in immune compromised individuals (reviewed in Ojiambo et al., 2018). Abiotic stressors such as drought increase aflatoxin production in maize (Kebede et al., 2012; Fountain et al., 2014). The role of polyamines (PAs) in drought tolerance and tolerance to other abiotic stresses is well-established (reviewed in Minocha et al., 2014; Tiburcio and Alcázar, 2018). Therefore, drought tolerance and aflatoxin resistance could be useful traits in maize. A number of approaches to enhance aflatoxin resistance in maize are being examined that include conventional and marker-assisted breeding, transgenic expression of resistance-associated proteins, RNAinterference-based host induced gene silencing, and biocontrol (reviewed in Ojiambo et al., 2018).

Polyamines are ubiquitous aliphatic amines found throughout all life forms. Their relative amounts vary depending on tissue type, developmental stage, or exposure to abiotic and biotic stressors (reviewed in Jiménez-Bremont et al., 2014; Minocha et al., 2014; Majumdar et al., 2017; Masson et al., 2017). The PA pathway (**Figure 1**) starts with the production of the diamine putrescine (Put) by the enzymes ornithine decarboxylase (ODC; E.C. 4.1.1.17) or arginine decarboxylase (ADC; E.C. 4.1.1.19) from the substrates ornithine (Orn) and arginine (Arg) respectively. Putrescine is converted into higher PAs, spermidine (Spd; a tri-amine) and Spm (Spm; a tetraamine), by the enzymes Spd synthase (SPDS; E.C. 2.5.1.16) and Spm synthase (SPMS; E.C. 2.5.1.22) both of which require decarboxylated S-adenosylmethionine (dcSAM) produced by S-adenosylmethionine decarboxylase (SAMDC; E.C. 4.1.1.50). In animals, Spm and Spd are back-converted to Spd and Put, respectively, by Spm/Spd N<sup>1</sup> -acetyltransferase (SSAT; E.C. 2.3.1.57) and PA oxidase (PAO; E.C. 1.5.3.11); in plants, the SSAT is not well-characterized.

The PA pathway plays a critical role during normal development, stress response, and is biochemically linked to AA metabolism in plants (Majumdar et al., 2013, 2016; Minocha et al., 2014); especially the AAs Glu, Arg, Pro, Orn, and GABA. As Orn is synthesized from Glu, which serves both as an entry point for N, and a major precursor for several other AAs in plants, alterations in PA metabolism can impact AA levels in living cells (Mohapatra et al., 2010; Majumdar et al., 2013, 2016). The role of PAs in plant disease resistance is highly evident from several studies that involved alteration of tissue PA concentrations through genetic manipulations or exogenous application of PAs (reviewed in Jiménez-Bremont et al., 2014; Takahashi, 2016; Pal and Janda, 2017). Over-expression of a human SAMDC gene in tobacco resulted in increased accumulation of free and conjugated PAs and conferred tolerance to Verticillium dahliae and Fusarium oxysporum (Waie and Rajam, 2003). Transgenic eggplants over-expressing an oat ADC gene showed increased resistance against Fusarium oxysporum, the causal organism of wilt disease (Prabhavathi and Rajam, 2007). Transgenic plants accumulated both free and conjugated Put and Spd, and exhibited higher DAO activity than the control plants. Over-expression of other PA biosynthetic genes, SPDS and SPMS in sweet orange (Citrus sinensis) and Arabidopsis thaliana, respectively, increased resistance against bacterial pathogens (Fu et al., 2011; Gonzalez et al., 2011; Fu and Liu, 2013). Polyamine-associated disease resistance in plants is affected by an increase in free and conjugated PAs, catabolism of PAs, production of H2O<sup>2</sup> and activation of defense signaling pathways. Other PA functions include up-regulation of genes involved in the production of pathogenesis-related (PR) proteins (especially by Spm; reviewed in Seifi and Shelp, 2019), transcription factors (e.g., basic leucine zipper protein family), and plant defense hormones, e.g., methyl-jasmonate (reviewed in Hussain et al., 2011; Jiménez-Bremont et al., 2014). Polyamines have also been implicated to indirectly lower the plant immune response against pathogenic bacteria (Erwinia amylovora – Oh et al., 2005, and Pseudomonas syringae – O'Neill et al., 2018) by the production of novel compounds like Phevamine A (a conjugate of L-Phe, L-Val, and a amidino-Spd).

Polyamines produced by the invading fungus are of importance with respect to growth, development, pathogenesis, and production of SMs (reviewed in Valdes-Santiago et al., 2012; Valdes-Santiago and Ruiz-Herrera, 2013). An A. flavus 1spds mutant failed to grow in vitro in the absence of exogenously supplied Spd in the growth medium and showed reduced growth and aflatoxin production during infection of maize kernels (Majumdar et al., 2018). In the WT A. flavus strain grown in vitro, an exogenous supply of Spd and Spm in the growth medium significantly increased fungal growth, sporulation, and production of aflatoxin and other toxic SMs, namely aflatrem, aflavinine, and cyclopiazonic (CPA) acid. Up-regulation of PAs in the host plant in response to a pathogen can be favorable to both host and pathogen. The outcome depends in large part on the relative abundance of the diamine Put versus higher PAs, Spd and Spm, accompanied by their catabolism, and also on the type of the pathogen and host species.

**Abbreviations:** ADC, arginine decarboxylase; ADI, agmatine deiminase; AL, argininosuccinate lyase; Ala, alanine; AA, amino acid; Arg, arginine; AS, argininosuccinate synthase; Asp, aspartate; CPA, N-carbamoylputrescine amidohydrolase; Cys, cysteine; DAO, diamine oxidase; dcSAM, decarboxylated S-adenosylmethioninie; GABA, γ-aminobutyric acid; Glu, glutamate; Gly, glycine; His, histidine; Ile, isoleucine; Leu, leucine; Lys, lysine; CS, N 0 ,N00-di-coumaroyl Spd; FP, N 0 ,N00-di-feruloyl-Put; Orn, ornithine; ODC, ornithine decarboxylase; PA, polyamine; OTC, ornithine transcarbamoylase; P5C, 1<sup>1</sup> -pyrroline-5-carboxylate; PAO, polyamine oxidase; PCA, perchloric acid; Pro, proline; Put, putrescine; SAMDC, s-adenosylmethionine decarboxylase; Ser, serine; SM, secondary metabolite; Spd, spermidine; Spm, spermine; SPDS, spermidine synthase; SPMS, spermine synthase; SSAT, Spd and Spm N 1 -acetyl transferase; Thr, threonine; Trp, tryptophan; tSPMS, thermospermine synthase; Val, valine; WT, wild-type.

Knowing the diverse roles of PAs in plant disease resistance, the current study was undertaken to investigate the role of host PA metabolism during maize-A. flavus pathogenic interaction. The work described here, used maize lines that were previously characterized as either resistant or susceptible to A. flavus infection and aflatoxin contamination. The genotypes used in this current study were selected based on the aflatoxin data obtained from multiyear field studies (reviewed in Brown et al., 2016). The data presented here demonstrate a significant induction of Put biosynthetic genes accompanied by greater conversion of Put into higher PAs (Spd and Spm) in resistant lines in comparison with a susceptible maize line. In mock-inoculated kernels, higher basal expression of PA metabolism genes in resistant, as compared to susceptible lines, may form the basis for future breeding or transgenic approaches to improve aflatoxin resistance and overall stress tolerance in maize. In addition, higher Spd and/or Spm content could possibly be used to screen maize genotypes for potential aflatoxin resistance.

### MATERIALS AND METHODS

#### Maize Kernel Inoculation and Incubation

Undamaged and uniformly sized kernels of one susceptible genotype, SC212, and two resistant genotypes (TZAR102 – highly resistant, and MI82 – moderately resistant; Brown et al., 2016) of maize were processed for the kernel screening assay (KSA) as described in Rajasekaran et al. (2013). Briefly, kernels were surface sterilized with 70% ethanol, air dried, and kept sterile until experiments were started. The AF13 strain of A. flavus [SRRC 1532, a highly pathogenic and high aflatoxin-producing isolate (Cotty, 1989)] was grown on V8 agar medium for 7 days under illumination at 30◦C prior to the collection of spores for kernel inoculation. Sterile kernels were inoculated by placing in a sterile 300 ml beaker containing 100 ml of AF13 spore suspension (4 × 10<sup>6</sup> spores/ml) for 3 min with continuous stirring. Excess inoculum was removed and the kernels were placed in plastic caps arranged in trays containing filter paper on the bottom. The filter paper was moistened by addition of sterile ddH20 and kept moist during the course of the experiment to maintain high relative humidity. Aspergillus flavus inoculated or water-inoculated (control, "mock-inoculated") kernels were placed inside trays (with lids on top) and kept in an incubator at 31◦C in the dark. Kernels were collected at 8 h, 3 and 7 days post-inoculation.

### Quantification of Polyamines and Amino Acids

Infected or control (water inoculated) ground maize kernels previously stored at −80◦C were subjected to repeated (3X)

freeze (−20◦C) and thaw (room temperature) cycles in 5% PCA. The samples were then vortexed for 2 min after final thaw and centrifuged at 14,000 × g for 8 min. Dansylation and quantification of PAs and AAs was performed simultaneously using the method as described in Minocha and Long (2004) with minor modifications (Majumdar et al., 2018). The samples were incubated for 30 min at 60◦C followed by cooling for 3 min and centrifuged for 30 s at 14,000 × g. The reaction was terminated by adding glacial acetic acid. Microfuge tubes containing samples were kept open under a flow hood for 3 min for evaporation of CO2. Acetone used to solubilize dansyl chloride was evaporated in a SpeedVac Evaporator (Savant, Farmingdale, NY, United States) for 5 min. Filtered HPLC grade methanol was added to each sample for a final volume of 2 ml.

The HPLC system was comprised of a Series 200 pump, autosampler, and fluorescence detector (Perkin-Elmer Corporation, Waltham, MA, United States) fitted with a 200 ml injection loop (20 ml injection volume). A column heater (Bio-Rad Laboratories, Hercules, CA, United States) was set at 40◦C. The other components of the HPLC system included a Perkin-Elmer-Brownlee Pecosphere scavenger cartridge column (CRC18, 3 mm, 33 mm × 4.6 mm I.D.), a Phenomenex SecurityguardTM guard column (C18, 5 mm, 3 mm × 4 mm I.D.; Phenomenex, Torrance, CA, United States), and a Phenomenex SynergiTM Hydro-RP analytical column (C18, 80 Å, 4 mm, 150 mm × 4.6 mm I.D.). Excitation and emission wavelengths were set at 340 and 515 nm respectively, and the data were processed using Perkin Elmer TotalChrom software (version 6.2.1).

#### Aflatoxin and Polyamine Conjugates Analysis

Ground maize kernels (∼20–70 mg) inoculated with either AF13 strain or water (control) were extracted in 1 ml of methanol for 24 h with shaking (175 rpm) at room temperature. The extracts were filtered using cotton plugs and the filtrates dried under a stream of nitrogen. Extracts were reconstituted in 250 µl of methanol, centrifuged to remove particulates, and analyzed in a Waters Acquity UPLC system (Waters Corporation, Milford, MA, United States) (isocratic separation with 40% methanol in water, and a BEH C18 1.7 µm, 2.1 mm × 50 mm I.D. column) using fluorescence detection (excitation at 365 nm, and emission at 440 nm). Samples were diluted 10-fold if the aflatoxin signal saturated the detector. Analytical standards (Sigma-Aldrich, St. Louis, MO, United States) were used to identify and quantify aflatoxins [retention time of aflatoxin B1 (AFB1), 4.60 min; retention time of aflatoxin B2 (AFB2), 3.55 min]. Aflatoxin content was expressed as ng/mg fresh weight (FW) of homogenized kernels. Putrescine and Spd conjugates were analyzed on a Waters Acquity UPLC system equipped with a PDA UV detector and an Acquity QDa mass detector using the following conditions: solvent A = 0.1% formic acid (FA) in water; solvent B = 0.1% FA in acetonitrile; flow rate: 0.5 ml/min; solvent gradient: 5% B (0–1.25 min), gradient to 25% B (1.25–1.5 min), gradient to 100% B (1.5–5.0 min), 100% B (5.0–7.5 min), then re-equilibration to 5% B (7.6–10.1 min). Putative Put and Spd conjugates were identified by their molecular ion and corresponding UV spectrum [N 0 ,N00-di-feruloyl-Put: 2.85 min (M+H)<sup>+</sup> = 441.2 m/z, λmax = 218.5, 235.0, 293.3, 317.9; N 0 ,N<sup>00</sup> di-coumaroyl Spd: 2.50 min (M+H)<sup>+</sup> = 438.2 m/z, λmax = 212.7, 225.8, 297.6, 307.4]. Quantification of N 0 ,N00-di-feruloyl-Put and N 0 ,N00-di-coumaroyl Spd was achieved by peak integration of the extracted ion chromatograms and normalized by sample weight.

### RNA Isolation, cDNA Synthesis, and Gene Expression

Total RNA from infected or control (water-inoculated) ground maize kernels was isolated using a 'SpectrumTM Plant Total RNA kit' (Sigma-Aldrich). cDNA was synthesized using an iScriptTM cDNA synthesis kit (Bio-Rad). Manufacturer's protocols were followed for both RNA isolation and cDNA synthesis. Quantitative RT-PCR (qRT-PCR) was performed in an iCycler iQ5 Multicolor real-time PCR detection system (Bio-Rad) using SYBR green I chemistry. The thermocycler conditions comprised of a pre-incubation step at 95◦C for 3 min, dye activation at 95◦C for 10 s, primer annealing at 55◦C for 30 s, elongation at 55◦C for 50 s, and a dissociation curve between 65 and 95◦C for 30 min (with 0.5◦C increments). The primers used for the qRT-PCR analyses are listed in **Supplementary Table S1**. Gene expression was normalized by 11C<sup>T</sup> method (Livak and Schmittgen, 2001) to Zea mays ribosomal structural gene GRMZM2G024838 or A. flavus β-tubulin gene (AFLA\_068620) expression (Shu et al., 2015) using the gene expression analysis software package of the Bio-Rad iQ5.

Fungal loads in the infected maize kernels were estimated at 8 h, 3 and 7 days post A. flavus infection. Quantification of fungal load was performed according to Thakare et al. (2017), and calculated as relative expression of A. flavus β-tubulin gene (AFLA\_068620) to the expression of maize ribosomal structural gene GRMZM2G024838 (Shu et al., 2015).

### Statistical Analysis

Student's t-test was performed to determine statistical significance between the A. flavus susceptible line (SC212) and the resistant maize genotypes (TZAR102 and MI82), and between mock-inoculated and Af-inoculated within each line at a specific time point. The level of significance was determined at P ≤ 0.05 and is depicted in the figure legends and graphics as ∗ and # respectively.

### RESULTS

#### Polyamine Content

Polyamine content varied with incubation period, and with or without A. flavus infection between the resistant and susceptible maize lines. At 8 h mock-inoculated samples of resistant TZAR102 and MI82 lines had 35–43% higher Put content than the susceptible line (**Figure 2A**). No significant change in Put content was observed among Af-inoculated kernels of resistant and susceptible lines at this time point. At 3 days post-inoculation (dpi), Put content was significantly lower in the inoculated TZAR102 line (152 ± 10 nmol/g FW) as compared to the

(

(Af-inoc) kernels of susceptible (SC212) and resistant (TZAR102, MI82) maize genotypes. Data are Mean ± SE of 4 replicates, each replicate consists of six seeds

<sup>∗</sup>P ≤ 0.05, between the susceptible line SC212 and other lines; #P ≤ 0.05, between mock and +Af treatments within each line at different times after inoculation).

susceptible line SC212 (223 ± 30 nmol/g FW). Put content was significantly higher in the Af-inoculated TZAR102 line in comparison to the mock-inoculated kernels at this time point. At 7 dpi, Put content increased in all lines in both Af-inoculated and mock-inoculated kernels. At 7 dpi, Put content was highest in the Af-inoculated kernels of SC212 (1092 ± 129 nmol/g FW); 33–49% higher than the resistant lines, and significantly higher than the SC212 mock-inoculants. A 160% increase in Put content was observed in the Af-inoculated kernels of SC212 line in comparison to the mock-inoculated kernels at this time point.

Cellular content of Spd was significantly higher (102–117%) at 8 h (**Figure 2B**) in both mock and Af-inoculated TZAR102 and MI82 lines in comparison with SC212. There were no major differences in Spd between the lines at 3 or 7 dpi.

With respect to Spm, its content was significantly higher (∼45%) in TZAR102 (vs. SC212) at 8 h, in both the Af- and mock-inoculated samples (**Figure 2C**). No major change in Spm content was observed at 3 and 7 dpi in the susceptible and resistant lines, except for relatively lower levels in mockinoculated TZAR102 and MI82 samples in comparison with SC212 at 3 dpi. In the TZAR102 line at 3 dpi, there was a significant increase in all three of the polyamines in the Af- vs. mock-inoculated samples.

In comparison with the SC212 susceptible line, the ratio of Spd/Put was significantly higher (33–79%) in the mockinoculated kernels of MI82 and in the mock- and Af-inoculated kernels of TZAR102 (both resistant lines) at 8 h (**Figure 2D**). The TZAR102 line maintained higher Spd/Put ratio in both groups of inoculants (vs. SC212) at 3 days. At 7 days the Spd/Put ratio was significantly higher (102–104%) in the Af-inoculated kernels of TZAR102 and MI82 lines than in those from the SC212 line, and the Af-inoculated kernels of SC212 had a Spd/Put ratio that was significantly less than mock-inoculated kernels. For all lines the ratio of Spd/Put decreased over time.

replicate consists of six seeds (∗P ≤ 0.05, between the susceptible line

SC212 and other lines at different times after inoculation).

### Changes in Put and Spd Conjugates in Response to A. flavus Infection

Polyamine conjugates are known to have antimicrobial properties; we therefore investigated PA conjugates in the susceptible and resistant maize genotypes during A. flavus infection. The two maize PA conjugates that were detected in the current study were N 0 ,N00-di-feruloyl-Put (FP) and N 0 ,N00-dicoumaroyl Spd (CS) (**Figure 3**). The FP content was ∼60–230% higher in the resistant lines as compared to SC212 at 8 h. Between the two resistant lines, TZAR102 maintained a higher level (by 100–200%) of FP content throughout the infection period as compared to the other lines. The CS content on the other hand was similar at 8 hpi in both resistant and susceptible lines except for TZAR102, which was significantly lower than the other lines. At 3 and 7 dpi, CS content was decreased by 38–59% in the resistant lines in comparison to SC212.

#### Amino Acids Analyses

As cellular Put is produced from the substrates Orn and Arg, both of which are derived from Glu (a key precursor of AAs), any change in Orn and Arg utilization will affect related AAs (**Figure 1**). There were no major changes in cellular levels of Glu in the resistant and susceptible lines at 8 hpi except for TZAR102, which had significantly lower Glu in the mockinoculated samples as compared to SC212 mock-inoculants (**Figure 4A**). At 3 dpi Glu content increased by 100% in both the mock and Af-inoculated SC212 line with a further increase in both at 7 dpi. At days 3 and 7 Glu content was significantly higher in both SC212 inoculants than all other inoculants of the susceptible and resistant lines.

The differences in cellular Gln content between the susceptible and the resistant lines were relatively small at 8 dpi, however Gln in both of the resistant lines was significantly lower in the mock-inoculated and the Af-inoculated MI82 samples than in the susceptible SC212 line. At 3 dpi, cellular Gln content was significantly higher in SC212 relative to the other two lines for both Af-inoculated and mock-inoculated samples (**Figure 4B**). The trend remained the same for inoculated samples at 7 days.

Cellular Orn content in the Af-inoculated MI82 and Af- and mock-inoculated TZAR102 lines was significantly lower than the susceptible line SC212 at 8 h and 3 dpi (**Figure 4C**). At 7 dpi, Orn in the mock inoculants of both resistant lines was significantly lower than the mock-inoculated susceptible line and the Af-inoculated samples of the same lines. Mock-inoculated TZAR102 samples had significantly lower Orn relative to SC212 at all times tested.

As Pro is derived from Orn, any change in cellular Orn content will affect Pro content. In TZAR102, cellular content of Pro was significantly higher (100–144%) in mock- and Af-inoculated samples at 8 hpi, and 7 dpi Af-inoculated samples when compared to SC212 (**Figure 4D**). At 3 dpi, Pro was significantly lower in the TZAR102 mock-inoculated kernels, and significantly higher at 7 dpi in Af-inoculated TZAR102 vs. SC212.

Cellular Arg content increased by 600% from 8 hpi to 3 dpi in both inoculants of SC212, and was significantly higher than all other susceptible and resistant lines at this time point (**Figure 4E**).

(E) arginine, and (F) γ-aminobutyric acid (GABA) at 8 h, 3 and 7 days in the mock-inoculated (mock-inoc) and A. flavus inoculated (Af-inoc) kernels of susceptible (SC212) and resistant (TZAR102, MI82) maize genotypes. Data are Mean ± SE of 4 replicates, each replicate consists of six seeds <sup>∗</sup>P ≤ 0.05, between the susceptible line SC212 and other lines; #P ≤ 0.05, between mock and +Af treatments within each line at different times after inoculation).

At 7 dpi, in comparison with mock-inoculated samples, Arg content was significantly lower in the Af-inoculated samples of both resistant lines.

The non-protein AA GABA is produced in two ways, via catabolism of Put and directly from Glu by Glu decarboxylase (**Figure 1**). At 8 hpi, cellular contents of GABA were significantly lower in the mock-inoculated samples of TZAR102, and both inoculants of the MI82 kernels in comparison to SC212 (**Figure 4F**). At 3 dpi this trend was seen only for mockinoculated TZAR102 line. Cellular GABA content increased by >200% at 7 dpi (vs. 8 hpi and 3 dpi time points) in SC212 and was significantly higher than the Af-inoculated MI82 line. For the inoculated samples, there were no significant differences in GABA between the susceptible line SC212 and resistant line TZAR102 at any time tested (**Figure 4F**).

The data on AAs that are not directly related to the Glu-Orn-Arg-Pro-Put pathway, are presented in the **Supplementary Data** (**Supplementary Figure S1**). In general, the AAs that were decreased by >63–92% both in mock-inoculated and Af-inoculated samples of TZAR102 and MI82 resistant lines were: His, Ser, Leu, Ala, Val, Ile, Gly, Met, and Thr. The majority of these decreases were seen at 3 dpi.

#### Expression of Polyamine Genes in Maize

Biosynthesis of Put takes place via ODC and ADC pathways. Among the three ZmODC genes, expression of ZmODC3 was

highest in the TZAR102 resistant line. With one exception, mock and Af-inoculated kernels of both resistant lines had significantly higher expression of ZmODC3 than the susceptible SC212 line at 3 and 7 dpi (**Figures 5A,B**). Expression of ZmODC3 was highly up-regulated upon fungal inoculation at 3 and 7 dpi in SC212 and MI82 lines. Expression of ZmODC2 was lower than ZmODC3 in all lines at both time points with no differences between lines or inoculants. No expression of ZmODC1 was observed in any of the samples.

The expression of four putative ZmADC genes varied between the susceptible and resistant lines at 3 and 7 dpi (**Figures 5C,D**). Expression of ZmADC1 was significantly higher (5 to 24-fold) in both inoculants of the TZAR102 line compared to the SC212 line at 3 and 7 dpi. At 3 dpi in the MI82 line and 7 dpi in the SC212 line, ZmADC1 expression was relatively higher in the Af-inoculants than the mock-inoculants. Expression of ZmADC2 expression was 9 to 12-fold higher in the mock-inoculated TZAR102 kernels at 3 dpi, and 3 to 110-fold higher at 7 dpi in comparison to the SC212 line. At both 3 and 7 dpi mock inoculants of both resistant lines had significantly higher levels of ZmADC2 expression than the susceptible line. Expression of ZmADC4 was significantly higher in both mock and Afinoculated kernels of TZAR102 as compared to SC212 at both 3 and 7 dpi. In comparison with mock-inoculants, Af-inoculants of SC212 had significantly higher levels of expression of ZmADC4 at 7 dpi. No expression of ZmADC3 was observed in any of the lines.

Among the four putative ZmSAMDC genes, ZmSAMDC1 expression was 3 to 25-fold higher in the mock-inoculated kernels of TZAR102 line at 3 and 7 days, and MI82 at 7 days in comparison to the susceptible line (**Figures 5E,F**). Expression of ZmSAMDC1 and ZmSAMDC2 increased by ≥20-fold in the Afinoculated samples at 7 dpi in comparison to their corresponding values at 3 dpi. At 3 days, ZmSAMDC3 showed higher induction in the inoculated samples than their corresponding mockinoculated controls in the susceptible SC212 and resistant MI82 lines. At 7 days, Af-inoculated SC212 kernels showed >50-fold upregulation of ZmSAMDC1 and ZmSAMDC2 expression vs. their corresponding mock-inoculated samples.

Overall, relative induction of ZmSPDS and ZmSPMS genes (upon Af-inoculation) was substantially lower than all other PA biosynthetic genes (**Figures 5G–J**). Expression of ZmSPDS1 and ZmSPDS3 were significantly higher in the mock-inoculated kernels of TZAR102, and MI82 lines (vs. SC212) and was upregulated upon fungal infection at 3 dpi (**Figure 5G**). At 7 dpi, the trend of ZmSPDS3 expression in the resistant lines was similar to 3 dpi and was up-regulated in the TZAR102 line upon fungal inoculation (**Figure 5H**). No expression of the ZmSPDS2 gene was observed in any of the samples. Expression of ZmSPMS2 was up-regulated in the susceptible SC212 line both at 3 and 7 days upon fungal infection (**Figures 5I,J**). Expression of both ZmSPMS1 and ZmSPMS2 was highest in the mock-inoculated TZAR102 kernels both at 3 and 7 days and maintained similar level of expression during fungal infection.

Increased expression of Spd and Spm biosynthetic genes affected the expression of ZmPAO genes (**Figure 6**). Among the six ZmPAO genes, the pattern of relative change in ZmPAO1-3 expression (mock-inoculated vs. Af-inoculated) was similar in all lines. Expression of ZmPAO5 at 3 days was 50 to 100-fold higher in the TZAR102 and MI82 resistant lines vs. SC212 susceptible lines in both inoculants (**Figure 6A**). At 3 and 7 dpi, expression of ZmPAO4-6 was higher in the mock-inoculated kernels of resistant lines in comparison with the susceptible line SC212 (**Figure 6B**). There was a sevenfold increase in ZmPAO6 expression in Af-inoculated SC212 than the mockinoculated samples.

#### Fungal Load and Aflatoxin Production

The susceptible line (SC212) showed higher A. flavus colonization on the kernels at 3 dpi and increased by 7 dpi as compared to the TZAR102 and MI82 resistant lines (**Figure 7A**). At both 3 and 7 dpi, the susceptible line had higher fungal load than the resistant lines (**Figure 7B**). Eight hpi data were inconsistent and are not presented.

In general, aflatoxin B1 (AFB1) content was several fold higher than aflatoxin B2 (AFB2; **Figure 8**) in all lines. No aflatoxin was detected in any samples at 8 hpi. The susceptible SC212 line accumulated the highest amount of aflatoxins at both 3 and 7 dpi. At 3 dpi AFB1 content in SC212 kernels was significantly higher than in both inoculants of the resistant lines TZAR102 and MI82. Aflatoxin content was significantly higher at 7 dpi in the resistant lines vs. 3 dpi, but it was still significantly lower than the susceptible line. The trend was similar with AFB2 content in the infected kernels of susceptible and resistant maize lines.

### DISCUSSION

Polyamines are present in all life forms and are involved in a plethora of cellular processes including growth, development, stress response, and pathogenesis (reviewed in Valdes-Santiago et al., 2012; Minocha et al., 2014; Miller-Fleming et al., 2015; Takahashi, 2016; Pal and Janda, 2017). Consequently, the PA metabolic pathway in the host plant as well as in the pathogen has been the target of a number of studies to improve disease resistance in plants.

#### Polyamine Biosynthesis, Conversion, and Catabolism Play an Important Role in Resistance to A. flavus

Polyamines are widely implicated in plant defense or susceptibility to disease (depending on the type of PAs and their relative abundance) during interaction with pathogens and pests (Subramanyam et al., 2015; Takahashi, 2016; Pal and Janda, 2017). Often, transgenic expression of the PA biosynthetic genes using ADC, SAMDC, SPDS, or SPMS leads to increased accumulation of free or conjugated PAs, and improves host plant resistance against a wide variety of pathogens including fungi and bacteria. Increase in Put biosynthesis in the host plant without proportionate conversion of Put into Spd and Spm in some cases, increased susceptibility to fungal pathogen. During compatible and incompatible interactions between oat (Avena sativa L.) and powdery mildew (Blumeria graminis f.sp. avenae) pathogen, the susceptible oat cultivar accumulated higher amount of Put vs. the resistant cultivar at early (24 h post-inoculation)

infection stage (Montilla-Bascón et al., 2014). Whereas, Spd content was higher in the resistant cultivar in comparison to the susceptible one at the same time point. The results indicate that increase in host Spd production may contribute to resistance against the powdery mildew pathogen. Cotton (Gossypium hirsutum) cultivars tolerant to the necrotrophic fungal pathogen, Verticillium dahliae, showed greater increase in Spd and Spm content post-fungal inoculation vs. susceptible cultivars, indicating the contribution of higher PAs in resistance against the fungus (Mo et al., 2015). In the present study,

treatments within each line at different times after inoculation).

TZAR102 and MI82 maize lines previously shown to have resistance to A. flavus infection and aflatoxin production (Brown et al., 2016), accumulated higher amounts of Spd and Spm as compared to the susceptible line at the earliest time tested after inoculation (**Figures 2B,C**). High basal accumulation of PAs in mock-inoculated kernels of the resistant maize lines compared to the susceptible line might suggest the presence of a host defense priming mechanism by polyamines as reported earlier in plants (Hussain et al., 2011). The high Spd to Put ratio in the TZAR102 and MI82 resistant lines compared to the SC212 susceptible line

(**Figure 2D**) indicates a possible role of the host Spd in resistance to aflatoxin production by A. flavus. Higher Spd content will contribute to its greater availability for catabolism, conjugation, and other regulatory roles. The mode of action of free Spd and Spm in plant resistance against fungal pathogens has been described through their interactions with PAOs, antioxidant systems, and defense-signaling pathways as reported in several studies (Koç, 2015; Takahashi, 2016; Pal and Janda, 2017; Seifi and Shelp, 2019). Higher Spd/Put ratio in the host plant could also potentially affect uptake of PAs by the fungus especially, Put uptake by A. flavus from the maize kernels during pathogenesis. Using radiolabeled PAs in A. nidulans, it was shown that the rate of uptake of Put was 2- to 3-fold more rapid than Spd and higher concentrations of Put inhibited Spd uptake (Spathas et al., 1982). Significant increase in the expression of putative Put uptake transporters in A. flavus during kernel infection of a susceptible maize variety indicates that uptake of diamine by the fungus may contribute to increased pathogenicity (Majumdar et al., 2018). Besides the role of free PAs in plant pathogen resistance, specific conjugates of Put and Spd have been demonstrated to possess antimicrobial properties in addition to their aid in reinforcing plant cell walls during pathogen infection (Takahashi, 2016; Pal and Janda, 2017). The most resistant line TZAR102 showed up to 250% higher accumulation of the PA conjugate FP, at early and later stages of infection as compared to the SC212 susceptible and MI82 resistant lines used in the present study (**Figure 3**). Our results on FP in A. flavus resistance are in line with an earlier report by Mellon and Moreau (2004), where a concentration dependent inhibition of growth of the AF13 strain was demonstrated in vitro.

The role of PAs in plant resistance or susceptibility toward fungal pathogens depends on the type of pathogen (biotroph vs. necrotroph), host species, severity of infection, relative abundance and conversion of the different PAs (reviewed in Pal and Janda, 2017). High basal expression of PA biosynthetic and catabolic genes (in mock-inoculated kernels) in the TZAR102 and MI82 resistance lines also suggests a possible defense priming against A. flavus infection. Future RNA-seq studies comparing the transcriptome profiles of inoculated and mockinoculated resistant and susceptible maize lines will add to our understanding of the regulation of other defense-related signaling pathways in relation to the regulation of PA genes. Catabolism

of Spd and Spm by PAOs produces H2O2, which activates mitogen-activated protein kinases (MAPKs), and woundinduced protein kinases (WIPKs) associated with defense-related pathways in plants (reviewed in Hussain et al., 2011; Moschou et al., 2012). Maize has six PAO genes (Jasso-Robles et al., 2016). Based on the predicted intracellular localization of the proteins, ZmPAO1 is extracellular, ZmPAO2, 3, and 4 are peroxisomal/endoplasmic reticulum, ZmPAO5 is cytoplasmic, and ZmPAO6 is cytoplasmic/peroxisomal (Jasso-Robles et al., 2016). Among the different ZmPAO genes investigated in the current study, the expression of the ZmPAO5 gene was highest at 3 dpi and ZmPAO6 at 7 dpi in the resistant vs. the susceptible lines (**Figures 6A,B**) at the basal level (mock-inoculated kernels). This might indicate a possible role of ZmPAO5 and ZmPAO6 in resistance against A. flavus during early and late infection stages in the TZAR108 and MI82 lines. Higher basal expression of cytoplasm and peroxisome specific ZmPAOs (2, 3, 6) at early and later development stages respectively (**Figure 6B**) may have implications on back conversion of Spm or tSpm to Spd and production of H2O<sup>2</sup> involved in plant defense responses (Kamada-Nobusada et al., 2008; Alcázar et al., 2010; Liu et al., 2014; Tiburcio et al., 2014; Pál et al., 2015). Polyamine back conversion and its role against a necrotrophic plant pathogen has been reported (Mo et al., 2015). Heterologous over-expression of a cotton PAO gene in Arabidopsis significantly increased Spd content (through back-conversion of Spm to Spd) in the transgenic plants and increased resistance against the fungal necrotroph Verticillium dahliae (Mo et al., 2015). In another study, the role of PAOs/DAOs in defense response against the necrotrophic fungal pathogen Botrytis (B.) cinerea, was studied in grapevine (Hatmi et al., 2018). Increases in free PAs in the berries followed by osmotic stress and B. cinerea infection without increase in PA catabolism (by oxidases) led to increased berry susceptibility. The results presented here along with previous studies indicate that increase in free PAs accompanied by increased PA catabolism improves maize resistance against A. flavus.

### Amino Acid Pools Are Significantly Altered in Susceptible vs. Resistant Maize Lines

It can be expected that any major change in cellular PA content would affect the cellular pool of several AAs as the pathways share common substrates such as Glu, Orn, and Arg (Mohapatra et al., 2010; Majumdar et al., 2013, p. 16). Although storage proteins are less favored as carbon sources by A. flavus to produce aflatoxins (Mellon et al., 2000, 2002), Liu et al. (2016) showed that specific AAs such as Glu, Asp, and Asn could significantly increase AFB1 production. Given the observation that cellular contents of Glu and Gln were significantly higher in the susceptible line SC212 than the other lines, it is likely that these two AAs may be responsible for higher aflatoxin accumulation (**Figures 4A,B**).

Several other AAs including Ala, Pro, and GABA have also been positively correlated with aflatoxin production both in planta and in vitro studies (Gupta et al., 1977; Payne and Hagler, 1983; Falade et al., 2018). Among these, Pro and GABA are widely associated with abiotic and biotic stress responses and tolerance in plants (reviewed in Hayat et al., 2012; Shelp et al., 2017); both these AAs are closely associated with the PA biosynthetic pathway. The resistant line TZAR102 showed relatively higher Pro content than SC212 at early and late infection stages (**Figure 4D**). The TZAR102 line is of African origin and associated with drought tolerance (Brown et al., 2016) which might account for its relatively higher Pro content than the SC212 susceptible line. Cellular content of non-protein AA, i.e., GABA, significantly increases in response to diverse stresses and contributes to stress tolerance in plants (reviewed in Shelp et al., 2017).

Proline and GABA are commonly co-induced in many plants in response to various forms of abiotic stress, and both use Glu as the primary substrate. However, the relative proportion of the two is quite different in most cases (Templer et al., 2017; Lawas et al., 2018; Kumar et al., 2019). Stress-induced GABA production in plants has been reported to stimulate fungal pathogenicity (reviewed in Oliver and Solomon, 2004). Consistent with the previous reports, the SC212 line produced >40% higher amount of GABA than the resistant lines during A. flavus infection both at early and late infection stages (**Figure 4F**). In fact, increase in cellular content of GABA was proportionate to the increase in Put content (**Figure 2A**) in response to Af-inoculation. A concurrent increase in cellular content of GABA and increase in aflatoxin production (**Figures 4F**, **8**) in the current study is in line with an earlier report where GABA accumulation was high during infection of maize seeds at different developmental stages (Falade et al., 2018). This may indicate that the production of Pro may be

more important for lowering aflatoxin production than GABA. This is consistent with the observation that Orn (substrate for Pro and Put) content in the Af-resistant lines (**Figure 4C**) was lower than that in the susceptible line. Lower Orn content in both the resistant lines (**Figure 4C**) indicates increased utilization of Orn (either directly through ODC pathway or via ADC pathway) to produce PAs. The observation that Af-inoculated TZAR102 and M182 resistant lines had significantly higher content of Orn as compared to their mock-inoculated counterparts on 7 dpi, provides further support to this argument.

Among other AAs, a decrease in Ile and Leu catabolism was associated with a reduction in aflatoxin production in A. flavus in vitro studies (Chang et al., 2015). In the present study, with the exception of His, Ser, and Cys at 7 dpi, the content of most AAs were lower in the TZAR102 and MI82 resistant lines compared to SC212 line at 3 and 7 dpi (**Figure 4** and **Supplementary Figure S1**). This could be because large quantities of Glu are being driven toward the synthesis of PAs.

Higher cellular content of specific AAs in the plant (such as Glu, and GABA) coupled with observed increase in aflatoxin levels during infection, noted in the present and previous studies, indicate a possible role for these AAs in susceptibility of the plant to aflatoxin contamination. However, the exact role of specific host plant AAs in aflatoxin production by the fungus can only be delineated through down-regulation of key genes involved in AAs biosynthesis in the plants. Even if the beneficial effects of specific AAs (e.g., GABA) mentioned above are known to improve stress tolerance in plants, their absolute amounts and regulatory roles may vary depending upon the type of environmental stress, abiotic or biotic. There might also be a threshold beyond which excess AAs produced by plants during infection might favor the pathogen through uptake of AAs by the pathogen from the host (Struck, 2015). From a nutritional perspective, several of the AAs discussed above are highly desired in maize seeds. Therefore, future strategies of selection or metabolic engineering of these AAs must be aimed toward balancing the two aspects (need for higher quantities vs. contribution to pathogen susceptibility) of their metabolism in developing Af-resistant maize varieties, without compromising the nutritional value of the product in important food and feed crop.

#### Polyamines Modulate Aflatoxin Production During Host–Pathogen Interaction

A relationship between PA metabolism and aflatoxin production has previously been reported (Khurana et al., 1996; Jin et al., 2002; Khatri and Rajam, 2007; Majumdar et al., 2018). An association with Put accumulation (along with GABA accumulation; which is a direct product of Put catabolism) and aflatoxin production is evident in the susceptible line SC212 compared to the resistant lines, TZAR102 and MI82. High Put accumulation at a later stage of infection (7 dpi) in SC212 (**Figure 2**) was highly correlated with increased aflatoxin accumulation (**Figure 8**). Polyamines are common to both plants and their pathogens, and an increase in PA biosynthesis by the host plant during pathogen infection can be advantageous to the pathogen as they can take up PAs (an excellent source of N as well as promoters of growth) that are

produced by the host. Using radiolabeled substrates it was shown that Put uptake was more efficient than Spd and Spm uptake in A. nidulans (Spathas et al., 1982). As Put levels increased several-fold in the susceptible line at later infection stages (**Figure 2A**) in the present study, it can be argued that the fungus predominantly takes up Put from the host plant. Upregulation of Put transporters in A. flavus during maize kernel infection has been recently reported (Majumdar et al., 2018). Uptake of plant PAs by the invading pathogen was supported by the observation that application of a fungal PA uptake/transport inhibitor reduced Fusarium graminearum infection and decreased DON production in wheat by >100-fold (Crespo-Sempere et al., 2015). The results presented here along with the observations of Valdes-Santiago et al. (2012) suggest that Put is an inducer of aflatoxin production in A. flavus. Gardiner et al. (2009, 2010) found a similar role of Put in mycotoxin production in wheat-Fusarium graminearum pathogenic interaction. The levels of PA pathway intermediates were strongly correlated with the production of deoxynivalenol (DON). Putrescine increased DON production in vitro by up-regulating the expression of the biosynthetic genes involved in DON production (Gardiner et al., 2009). The reduction in aflatoxin content in the resistant lines is possibly due to the reduction in fungal load (**Figures 7**, **8**). Similar correlation between fungal load and aflatoxin production in response to infection of maize kernels infection has also been reported in several recent studies (Gilbert et al., 2018; Lebar et al., 2018; Majumdar et al., 2018). The reduction in fungal load in the resistant lines could be a combinatorial result of an increase in free PAs, PA-conjugates, reduction in Orn (**Figure 9**) along with other genetic factors such as kernel pericarp wax (Maupin et al., 2003). The data presented here suggest that the observed resistance in the TZAR102 and MI82 maize lines to A. flavus in maize lines can in part be ascribed to the higher amounts of Spd and Spm rather than increased Put accumulation during A. flavus infection.

### CONCLUSION

Biosynthesis of PAs is critical to both plants and their pathogens, for promoting stress tolerance and pathogenicity, respectively. Equally important is the role of PA catabolism in host plant resistance against pathogens. The current work shows the role of the diamine Put and the higher PAs, Spd and Spm, in susceptibility and resistance of maize to A. flavus infection and aflatoxin accumulation, respectively. In general, high basal expression of genes involved in PA biosynthesis and catabolism in absence of the pathogen and their induction upon fungal infection was observed in the resistant lines in comparison to the susceptible line. The data presented here indicate that higher Spd and/or Spm content in maize genotypes may have implication for higher resistance to A. flavus and aflatoxin contamination. It should be noted, as mentioned above, that PA metabolism might not be the only factor contributing to resistance against A. flavus infection and aflatoxin accumulation in the TZAR102 and MI82 maize lines. Involvement of other metabolites, differences in the molecular genetics of defense responses and physical attributes of kernels, such as waxy seed coat in MI82 (Maupin et al., 2003), are additional characteristics that are likely to contribute to the overall Af-resistance in these lines. Future studies focused on analysis of global gene expression along with targeted metabolomics approaches using these resistant and susceptible maize lines will allow for a greater understanding of the mechanisms of PAs in host plant defense responses and A. flavus pathogenicity.

### DATA AVAILABILITY

All datasets generated for this study are included in the manuscript and/or the **Supplementary Files**.

### AUTHOR CONTRIBUTIONS

RMa, KR, and JC conceived and designed the experiments. RMa performed the experiments. RMa, RMi, SM, ML, SL, and CC-W analyzed the data. RMa and ML wrote the manuscript. RMi, KR, SL, SM, and JC edited the draft manuscript. All authors reviewed and approved the final manuscript.

## FUNDING

This research was funded by the United States Department of Agriculture (USDA), Agricultural Research Service (CRIS No. 6054-42000-025-00D). Partial funding for this research was also provided by the USDA Forest Service, the New Hampshire Agricultural Experiment Station (NHAES), and USDA National Institute of Food and Agriculture (McIntire-Stennis) Project (NH00076-M).

### ACKNOWLEDGMENTS

The authors greatly acknowledge the help from Mr. David Ambrogio in setting up the kernel screening assays and Ms. Mary Lovisa with sample processing.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00692/ full#supplementary-material

FIGURE S1 | Amino acids are differentially regulated in the A. flavus susceptible vs. resistant lines. Cellular contents of (A) histidine, (B) serine, (C) cysteine, (D) phenylalanine, (E) tryptophan, (F) leucine, (G) alanine, (H) valine, (I) isoleucine, (J) lysine, (K) glycine, (L) methionine, (M) threonine, (N) aspartic acid at 8 h, 3 d, and 7 d in the mock-inoculated (mock-inoc) and A. flavus inoculated (Af-inoc) kernels of susceptible (SC212) and resistant (TZAR102, MI82) maize genotypes. Data are Mean ± SE of 4 replicates, each replicate consists of 6 seeds (∗P ≤ 0.05, between the susceptible line SC212 and other lines; #P ≤ 0.05, between mock and +Af treatments within each line at different times after inoculation).

TABLE S1 | Oligonucleotide primers used for the qRT-PCR work.

#### REFERENCES

fpls-10-00692 May 22, 2019 Time: 18:40 # 15



genotypes using a GFP-expressing Aspergillus flavus strain. World Mycotoxin J. 6, 151–158.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Majumdar, Minocha, Lebar, Rajasekaran, Long, Carter-Wientjes, Minocha and Cary. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Four FAD-Dependent Histone Demethylases of Arabidopsis Are Differently Involved in the Control of Flowering Time

Damiano Martignago1,2, Benedetta Bernardini<sup>1</sup> , Fabio Polticelli1,3, Daniele Salvi<sup>4</sup> , Alessandra Cona<sup>1</sup> , Riccardo Angelini<sup>1</sup> and Paraskevi Tavladoraki<sup>1</sup> \*

<sup>1</sup> Department of Science, Roma Tre University, Rome, Italy, <sup>2</sup> Centre for Research in Agricultural Genomics, Spanish National Research Council–Institute for Food and Agricultural Research and Technology–Autonomous University of Barcelona–University of Barcelona, Barcelona, Spain, <sup>3</sup> 'Roma Tre' Section, National Institute of Nuclear Physics, Rome, Italy, <sup>4</sup> Department of Life, Health and Environmental Sciences, University of L'Aquila, L'Aquila, Italy

#### Edited by:

Ana Margarida Fortes, University of Lisbon, Portugal

### Reviewed by:

Sureshkumar Balasubramanian, Monash University, Australia Xuncheng Liu, South China Botanical Garden (CAS), China

#### \*Correspondence:

Paraskevi Tavladoraki paraskevi.tavladoraki@uniroma3.it

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 28 January 2019 Accepted: 02 May 2019 Published: 04 June 2019

#### Citation:

Martignago D, Bernardini B, Polticelli F, Salvi D, Cona A, Angelini R and Tavladoraki P (2019) The Four FAD-Dependent Histone Demethylases of Arabidopsis Are Differently Involved in the Control of Flowering Time. Front. Plant Sci. 10:669. doi: 10.3389/fpls.2019.00669 In Arabidopsis thaliana, four FAD-dependent lysine-specific histone demethylases (LDL1, LDL2, LDL3, and FLD) are present, bearing both a SWIRM and an amine oxidase domain. In this study, a comparative analysis of gene structure, evolutionary relationships, tissue- and organ-specific expression patterns, physiological roles and target genes for the four Arabidopsis LDL/FLDs is reported. Phylogenetic analysis evidences a different evolutionary history for the four LDL/FLDs, while promoter activity data show that LDL/FLDs are strongly expressed during plant development and embryogenesis, with some gene-specific expression patterns. Furthermore, phenotypical analysis of loss-of-function mutants indicates a role of all four Arabidopsis LDL/FLD genes in the control of flowering time, though for some of them with opposing effects. This study contributes toward a better understanding of the LDL/FLD physiological roles and may provide biotechnological strategies for crop improvement.

#### Keywords: flowering time, histone demethylases, FLD, FLC, FWA, LDL, LSD1

## INTRODUCTION

Histone methylation is involved in a wide range of biological processes (Pfluger and Wagner, 2007). It decorates both transcriptionally silenced and active chromatin domains, depending on which residues are methylated and the degree of methylation. One of the most relevant and studied histone marks in plants is the methylation of lysine 3 on histone 4 (H3K4me). H3K4 can be mono-, di-, and tri-methylated (respectively me1, me2, me3) by different classes of SET domain-containing methyltransferases and this process is reversed by histone demethylases in a dynamic fashion. Two types of lysine-specific histone demethylases are present in both animals and plants, the Jumonji C (JmjC) domain-containing histone demethylases and the FAD-dependent histone demethylases (JHDMs and LSDs, respectively; Shi et al., 2004; Tsukada et al., 2005; Xu et al., 2015; Gu et al., 2016). In animals, two LSDs are found, LSD1 and LSD2 (Shi et al., 2004; Karytinos et al., 2009), which contain a FAD-dependent amine oxidase (AO; Polticelli et al., 2005) domain and a SWIRM domain (Stavropoulos et al., 2006). LSD1 has also an ∼100 amino acid protruding domain, known as 'Tower' domain, which interacts with the corepressor CoREST, among other proteins, and is

required for LSD1 catalytic activity on nucleosomes (Shi et al., 2005; Chen et al., 2006; Stavropoulos et al., 2006; Yang et al., 2006; Burg et al., 2015). Unlike LSD1, LSD2 does not have the 'Tower' domain and does not interact with CoREST, but possesses both a CW-type zinc finger motif and a C4H2C2-type zinc finger motif joined by a linker domain composed of two α-helices. This suggests that LSD2 may interact with different targets or co-regulatory molecules and may be involved in transcriptional programs distinct from those of LSD1 (Burg et al., 2015).

Arabidopsis thaliana has four homologs of the human LSD1 (HsLSD1) gene: At1g62830 (LSD1-LIKE1; LDL1), At3g13682 (LDL2), At4g16310 (LDL3), and At3g10390 (FLD), all bearing both a flavin AO domain and a SWIRM domain (Shi et al., 2004; Jiang et al., 2007; Spedaletti et al., 2008). Like HsLSD1, Arabidopsis LDL1 is able to specifically demethylate H3K4me2 and H3K4me1 peptides and to discriminate between different epigenetic marks (Forneris et al., 2005; Spedaletti et al., 2008). Furthermore, LDL1 interacts with a SET-domain histone methyltransferase and a histone deubiquitinase to form co-repressor complexes (Krichevsky et al., 2007, 2011). However, plant LDL/FLDs are probably directed to their substrates by mechanisms different from those of their animal counterparts (Sadiq et al., 2016). Indeed, plants do not encode CoREST homologs. In addition, LDL/FLDs do not interact with plant homologs of SFMBT1, which functions as part of the LSD1 based repressor complex and is known to bind different forms of methylated histones (Tang et al., 2013; Zhang J. et al., 2013; Sadiq et al., 2016).

Most of the physiological studies on the Arabidopsis LDL/FLDs focus on their role in the control of flowering time. The developmental transition from the vegetative to the reproductive stage is a critical event in the plant life cycle. In A. thaliana, a complex regulatory network controls the timing of floral transition, a key component of which is FLOWERING LOCUS C (FLC), a MADS-box transcriptional regulator that inhibits floral transition largely by reducing the expression of flowering-time integrators, such as SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1) and FLOWERING LOCUS T (FT) (He et al., 2003; He, 2009). This regulatory network integrates the endogenous developmental state of the plant (autonomous pathway and gibberellin-dependent pathway) with environmental cues (Amasino and Michaels, 2010). FLD is involved in the autonomous pathway by constitutively repressing FLC. Indeed, Arabidopsis fld loss-of-function mutants are known to be late-flowering or non-flowering due to increased FLC expression levels (Sanda and Amasino, 1996; Chou and Yang, 1998; He et al., 2003; Jiang et al., 2007). FLD is also required in chromatin silencing of FLC mediated by the RNAbinding protein FCA (Liu et al., 2007). Furthermore, the physical interaction between FLD and the histone deacetylases HDA5 and HDA6 plays an important role in the control of both H3 acetylation and H3K4 trimethylation at FLC and its homologs MADS AFFECTING FLOWERING 1 (MAF1), MAF4 and MAF5 (Yu et al., 2011; Luo et al., 2015). Indeed, fld mutants display altered H3 and H4 acetylation levels at FLC (He et al., 2003; Zhang Y. et al., 2013; Hu et al., 2014). FLC is down-regulated also by LDL1 and LDL2, which act in partial redundancy with FLD, the latter playing a more prominent role (Jiang et al., 2007). Consistently, ldl1ldl2 mutants display increased H3K4me3 levels at FLC as compared to wild-type plants, but to a lesser degree than ldl1fld mutants. LDL1 and LDL2, but not FLD, are additionally involved in the control of H3K4 methylation state at FWA, a homeodomain-containing transcription factor which interferes with floral transition (Jiang et al., 2007). Altogether, these data suggest that the Arabidopsis LDL/FLD gene family plays a critical role in the histone methylation pattern of flowering genes. A similar function was also suggested for LDL/FLD homologs in other plant species (Hu et al., 2014; Gu et al., 2016; Shibaya et al., 2016).

Recent studies have evidenced the involvement of the LDL/FLD gene family also in several developmental and stress defense processes (Yu et al., 2016). In fact, LDL1 is involved in root elongation and lateral root initiation (Krichevsky et al., 2009; Singh et al., 2012). In addition, LDL1 and LDL2 repress the expression of seed dormancy-related genes and act redundantly in repressing seed dormancy (Zhao et al., 2015). Furthermore, FLD is required for activation of systemic acquired resistance, through a FLC-independent pathway, and for up-regulation of important modulators of plant immune responses (Singh et al., 2013, 2014; Banday and Nandi, 2018). In wheat, a LDL1 homolog is up-regulated in heat-primed plants suggesting a role of this gene family in the epigenetic mechanisms regulating stress memory (Wang et al., 2016).

The increasing evidence for the involvement of the LDL/FLD gene family in different physiological processes raises the need for a comparative analysis of this gene family. To this end, in the present study the gene and protein structure, as well as the evolutionary history of all four LDL/FLDs have been dissected. Furthermore, the tissue- and organ-specific expression patterns of the four LDL/FLDs were analyzed. Phenotypical analyses of loss-of-function mutants for all four LDL/FLD genes were also performed, with particular attention to the flowering time, revealing functional differences among them.

## MATERIALS AND METHODS

### Protein Sequence Homology Search and Retrieval

The amino acid sequence of LSD1-like proteins from various plant and animal organisms were retrieved by sequence similarity searches in BLASTP (Altschul et al., 1997) using the amino acid sequence of HsLSD1 and HsLSD2, as well as of the A. thaliana LDL1, LDL2, FLD, and LDL3 as query sequences. The amino acid sequence of additional LSD1-like proteins was retrieved from the National Center for Biotechnology Information (NCBI) database based on sequence annotation. Abbreviations and accession numbers are listed in **Supplementary Table 1**. To determine SWIRM and AO domains, multiple amino acid sequence alignments were performed using Clustal Omega (Sievers et al., 2011). For genomic exon–intron structure comparisons, manual alignment between genomic and cDNA sequences was performed. Information on intron number was additionally obtained from the NCBI database.

### Molecular Modeling

fpls-10-00669 June 1, 2019 Time: 10:29 # 3

Molecular models of A. thaliana LDL3, and LDL3 homologs from Physcomitrella patens (PpLDL3) and Selaginella moellendorffii (SmLDL3) have been built using the ab initio/threading protocol implemented in the I-TASSER pipeline (Yang et al., 2015). No query/template alignment has been provided in input as I-TASSER uses LOMETS (Local Meta-Threading Server) to thread the query sequence through a representative library of PDB structures and select the folds compatible with the sequence of the query protein (Wu and Zhang, 2007). Best models have been selected on the basis of the I-TASSER quality score (C-score) whose values range from −5 to 2, higher values indicating higher quality models (Yang et al., 2015). C-score values of the selected models for Arabidopsis LDL3, SmLDL3 and PpLDL3 are −0.27, 0.86, and 0.6, respectively.

### Phylogenetic Analyses

Amino acid sequences were aligned with MAFFT v.7 (Katoh and Standley, 2013) using the E-INS-i iterative refinement algorithm. Two alignments were built, one with the entire protein sequence, and another one including only amino acids of the AO domain. For each alignment, the optimal model of protein evolution was selected by ModelTest-NG v0.1.5<sup>1</sup> under the corrected Akaike Information Criterion. The JTT model (Jones et al., 1992) with gamma distributed rates across site (+G) was selected for both alignments. Phylogenetic analyses were performed with the Maximum Likelihood method using RAXML v.8.2.10 (Stamatakis, 2014) with the PROTGAMMAJTT substitution model. Node support was evaluated with 1,000 rapid bootstrap inferences. The sequence of the polyamine oxidase 1 of A. thaliana (AtPAO1; At5g13700; **Supplementary Table 1**) was used as outgroup. Phylogenetic analyses were computed in the CIPRES Science Gateway V. 3.3<sup>2</sup> (Miller et al., 2010).

### Plant Material

All experiments were performed with Arabidopsis ecotype Columbia-0 plants grown under long-day (16 h day/8 h night) photoperiod conditions. To determine the flowering time (expressed as the number of rosette leaves at bolting), seeds were sown in a 3:1 soil:perlite mixture and plants were grown to mature stage. For RT-PCR and qRT-PCR analyses, seedlings were grown for 7 days on plates containing half-strength Murashige and Skoog basal medium supplemented with Gamborg's vitamins and 0.5% (w/v) sucrose (1/2MS) and solidified with 0.7% agar. Then, seedlings were transferred in 6-well plates containing <sup>1</sup>/2MS liquid medium and were left to grow for 7 more days.

### Characterization of Loss-of-Function LDL/FLD Mutants

Arabidopsis ldl1, ldl2, and fld loss-of-function mutants were obtained from the SALK collection (SALK\_142477.31.30.x, SALK\_146346.52.50.x, and SALK\_015053.35.80.x, respectively; Alonso et al., 2003), while ldl3 mutant was obtained from

<sup>1</sup>https://github.com/ddarriba/modeltest

<sup>2</sup>http://www.phylo.org/

the SAIL library (SAIL\_640\_B10.v1; Sessions et al., 2002). The presence of T-DNA insertion was confirmed by PCR, and homozygous mutant plants were selected. RT-PCR analysis using primers upstream and downstream from the T-DNA insertion confirmed the absence of correct mRNA for the corresponding genes, whereas qRT-PCR analysis confirmed reduced genespecific expression levels (**Supplementary Figure 2**). Primer sequences are listed in **Supplementary Table 2**.

### Construction and Characterization of Arabidopsis Transgenic Plants

To construct LDL/FLD::GFP-GUS transgenic Arabidopsis plants, 2- to 3-kb promoter regions including the 50UTR were amplified from Arabidopsis genomic DNA by PCR and cloned into the pDONR207 vector (Invitrogen) via Gateway Technology (Invitrogen). Sequences of oligonucleotides used for the amplification of promoter regions are shown in **Supplementary Table 2**. Following sequencing, promoter regions were inserted into the Gateway binary vector pKGWFS7 vector (Karimi et al., 2002) in-frame with the downstream green fluorescent protein (GFP) and β-glucuronidase (GUS) reporter genes. The resulting constructs were used to transform A. thaliana wild-type plants by the Agrobacterium tumefaciensmediated floral dip transformation method (Bent, 2006). Independently transformed plant lines were tested by PCR.

#### Histochemical GUS Assay

GUS staining of Arabidopsis LDL/FLD::GFP-GUS transgenic plants was performed essentially as previously described (Fincato et al., 2012). Briefly, samples were gently soaked in 90% (v/v) cold acetone for 1 h at −20◦C, rinsed with 50 mM sodium phosphate buffer pH 7.0, vacuum infiltrated in staining solution (1 mM 5-bromo-4-chloro-3-indolyl-β-D-glucuronide, 2.5 mM potassium ferrocyanide, 2.5 mM potassium ferricyanide, 0.2% Triton X-100, 10 mM EDTA, 50 mM sodium phosphate buffer, pH 7.0) and incubated at 37◦C for 18 h. Chlorophyll was extracted with ethanol:acetic acid (3:1). Samples were kept in 70% ethanol. To improve destaining, reproductive organs were washed with Hoyer's light medium (Stangeland and Salehian, 2002).

#### Quantitative RT-PCR Analysis

Total RNA was isolated from whole Arabidopsis seedlings using the RNeasy Plant Mini kit (QIAGEN) and treated with RNase-free DNase during RNA purification (RNase-Free DNase Set; QIAGEN) according to the manufacturers' protocol. RNA concentration was measured with a NanoDrop ND-1000 UV-Vis spectrophotometer (NanoDrop Technologies). Synthesis of cDNA and PCR amplification were carried out using GoTaq <sup>R</sup> 2-Step RT-qPCR System (Promega). The qPCR reactions were performed in a Corbett RG6000 (Corbett Life Science, QIAGEN) following the program: 95◦C for 2 min then 40 cycles of 95◦C for 3 s and 60◦C for 30 s. Primers were designed using Primer3 software (Untergasser et al., 2007) and tested for specificity using Primer-BLAST. UBIQUITIN-CONJUGATING ENZYME 21 (UBC21, At5g25760) was chosen as a reference gene (Czechowski et al., 2005). Primer sequences are listed in

**Supplementary Table 2**. Relative expression levels are expressed as fold-changes (2−11Ct). Reactions were performed in triplicate and mean values ± SE were calculated. At least three independent biological replicates were performed for each experiment, and mean values of relative expression levels from the different biological replicates are shown.

#### RESULTS

#### The Arabidopsis LDL/FLD Gene Family

The Arabidopsis LDL1, LDL2, and FLD display a high amino acid sequence identity with each other (48–52%; Spedaletti et al., 2008) and a shared gene structure, although with a different number of introns (**Figure 1**). In particular, LDL1 gene has no intron, LDL2 has one and FLD four, one of the FLD introns at the same position as the single intron in LDL2 (**Figure 1**, red diamonds). These similarities suggest that LDL1, LDL2, and FLD are recent derivatives of a common ancestor. In contrast, LDL3 gene structure is different from that of the other three LDL/FLD genes displaying seven introns, all of them at different position with respect to the FLD introns (**Figure 1**). Furthermore, the amino acid sequence identity of LDL3 with the other three LDL/FLDs is low (25–30%). LDL3 amino acid sequence (1,628 amino acids) is also significantly longer than that of the other LDL/FLDs (746–884 amino acids). In particular, LDL3 displays longer N-terminal and C-terminal extensions, as well as a larger region linking SWIRM and AO domains (SWIRM/AO distance), in respect to those of LDL1, LDL2, and FLD. Interestingly, at the C-terminal extension of LDL3, a putative structured domain with some similarity to transcription factor IIS was identified, which may have a regulatory role (**Figure 1**). DNA-binding domains are also present in the fungal (SWIRM1 and SWIRM2) and HsLSD2 homologs, which display a HMG box and a zinc finger domain, respectively (Nicolas et al., 2006; Zhang Q. et al., 2013).

LDL1, LDL2, and FLD do not have the HsLSD1 protruding 'Tower' domain. However, despite the absence of the 'Tower'

LDL3 also presents an insertion in the amine oxidase domain and a putative structured domain with similarity to the transcription factor IIS. Intron positions are shown. Numbers indicate the length of the amino acid primary sequence. The schematic representations are in scale.

domain, demethylase activity has been shown for Arabidopsis LDL1 (Spedaletti et al., 2008), as shown for the mouse LSD2 which also lacks the 'Tower' domain (Karytinos et al., 2009). Conversely to LDL1, LDL2, and FLD, a small insertion (about 33 amino acids) is present inside the AO domain of LDL3 (**Figure 1**), which, however, has low sequence similarity to the HsLSD1 'Tower' domain. Molecular modeling of LDL3 indicates that this region is probably unstructured (**Figure 2**). Nonetheless, it cannot be excluded that this region becomes structured upon interaction with, yet unknown, binding partners. Furthermore, comparative analysis of the LDL3 model with respect to the threedimensional structure of HsLDS1 in complex with a substratemimic peptide (Forneris et al., 2007) and to the molecular model of LDL1 (Spedaletti et al., 2008) indicates that almost all of the residues involved in substrate binding in HsLSD1 are conserved in both LDL1, whose demethylase activity has been demonstrated experimentally (Spedaletti et al., 2008), and LDL3 (**Table 1**). This analysis suggests that LDL3 is a lysine demethylase with a substrate specificity similar to that of LDL1.

#### Evolutionary History of Plant LDL/FLDs

To investigate the evolutionary history of the plant LDL/FLDs, a phylogenetic analysis of the amino acid sequence of 159 LDL/FLD homologs from 57 different representative animal, plant and algal species (**Supplementary Table 1**) was carried out. Maximum likelihood phylogenetic trees of both the fulllength (**Figure 3**) and AO domain (**Supplementary Figure 1**) amino acid sequences show four main clades: clades AI and AII grouping, respectively, HsLSD1 and HsLSD2 homologs from animals, and clades PI and PII grouping LDL/FLD homologs from plants and algae. Relationships between these animal and plant clades are unclear since the most basal nodes of the tree lack strong bootstrap support, in agreement with a recent phylogenetic analysis of the AO domains based on a limited number of plant and animal LDL/FLDs (Zhou and Ma, 2008). Within both clade PI and clade PII, green algae LDL/FLDs form a subclade that is sister to the clade formed by land plant LDL/FLDs (**Figure 3** and **Supplementary Figure 1**), in agreement with organism relationships. Phylogenetic relationships between plant LDL/FLDs indicate that plant LDL1, LDL2, and FLD homologs share a recent common ancestor (**Figure 3**, node c; bootstrap support, DBS = 100), whereas plant LDL3 homologs belong to a different evolutionary lineage (node b; BS = 98), consistently to what has been suggested for the Arabidopsis LDL/FLDs based on gene structure analysis. Moreover, within clade PI, plant LDL1, LDL2, and FLD homologs form three well supported clades (clades PIa, PIb1, and PIb2, respectively), with LDL2 homologs sister to FLD homologs (node e; BS = 80). The LDL/FLD homologs of PIa, PIb1, and PIb2 clades are well distributed among the various flowering plant species, being present both in dicotyledonous and monocotyledonous plants, as well as in Amborella trichopoda, which represents the sister lineage to all other extant flowering plants (Amborella Genome Project, 2013). In these clades, phylogenetic relationships between LDL/FLD amino acid sequences closely reflect evolutionary relationships between plant families to which they belong (**Figure 3**). This phylogenetic pattern suggests that LDL1, LDL2, and FLD genes

respectively) were obtained through molecular modeling approaches based on the three-dimensional structure of HsLDS1 and the molecular model of Arabidopsis LDL1 (Spedaletti et al., 2008). The FAD cofactor is shown in ball-and-stick representation and colored in blue. Shaded ellipses highlight the 'Tower' domain of HsLSD1 and PpLDL3, and the corresponding regions in SmLDL3 and AtLDL3.

have evolved through gene duplications. A first duplication would have occurred in correspondence of node c and a second one at node e. The fact that Amborella trichopoda shows one copy of each of LDL1, LDL2, and FLD genes with sister relationships to the corresponding clades formed by flowering plant LDL/FLD homologs indicates that such duplication events occurred before the split between eudicots and monocots. Most likely duplications took place early during the diversification of land plants, as suggested by the occurrence of a supported subclade of FLD (**Figure 3**, node h; BS = 98) clustering homologs found in the moss Physcomitrella patens, the liverwort Marchantia polymorpha and the seedless ancient vascular plant Selaginella moellendorffii. According to this scenario, gene duplications would have been followed by LDL1 and LDL2 gene loss in these ancient plants. Phylogenetic relationships between plant LDL1, LDL2, and FLD homologs also account for their shared gene structure. Indeed, with some exceptions, in most flowering plants LDL1 homologs lack introns, LDL2 homologs display one intron, whereas FLD homologs have four to five introns, all at conserved positions (**Table 2** and **Supplementary Table 2**). This suggests that sequential insertions have occurred first in the common ancestor of LDL2 and FLD genes (node e) and later in the ancestor of FLD genes (node g).

The phylogenetic pattern observed in lineage PII provides no evidence for old duplication events behind the diversification of LDL3 homologs, whereas species-specific duplication events might account for the occurrence of multiple LDL3 genes in the rosids Gossypium hirsutum, Populus trichocarpa, and Glycine max. Furthermore, the LDL/FLD homologs of group PII bear a higher number of introns (5 to 19 introns, **Table 2** and **Supplementary Table 2**) than those of clade PI. Whether such



Binding residues have been selected on the basis of the three-dimensional structure of human LSD1 (HsLSD1) in complex with a substrate-mimic peptide (PDB code 2V1D; Forneris et al., 2007). Orthologous residues in Arabidopsis LDL1 and LDL3 have been identified on the basis of structure-based sequence alignment using the molecular models of Arabidopsis LDL1 (Spedaletti et al., 2008) and LDL3 (present work). <sup>∗</sup>Conserved or conservatively substituted residues are shown in black, non-conserved residues in red.

as outgroup.

performed with the Maximum Likelihood method using RAXML v.8.2.10 (Stamatakis, 2014) with the PROTGAMMAJTT substitution model. Node support was evaluated with 1,000 rapid bootstrap inferences. The sequence of the polyamine oxidase 1 of A. thaliana (AtPAO1; At5g13700; Supplementary Table 1) was used


The number of introns in the corresponding genes and the number of the amino acid residues constituting the 'Tower' domain, the N- and C-terminal extensions and the distance between the SWIRM and amine oxidase (AO) domains are indicated. <sup>∗</sup>For p-LDL1-, p-LDL2-, p-FLD-, p-LDL3-, a-LSD1-, and a-LSD2-like proteins, mean values from all the flowering plant (p-) and animal (a-) species considered in the present study (Supplementary Figure 2) are indicated. Complete data are reported in Supplementary Figure 2. Pp, Physcomitrella patens; Sm, Selaginella moellendorffii; Mp, Marchantia polymorpha; Ol, Ostreococcus lucimarinus; Ot, Ostreococcus tauri; f-Sw1 and f-Sw2, LSD1-like proteins from Schizosaccharomyces pombe. NA, genomic sequence not available.

difference among LDL/FLDs of clade PI and clade PII has a physiological significance is still unknown. Also animal LSD1 and LSD2 homologs have a large number of introns (larger than the plant FLD and LDL3 homologs; **Table 2**), not at conserved positions when the genes of the two animal clades are compared. These data suggest a different evolutionary history for the two animal clades.

Phylogenetic results can also provide important insight into the structural evolution of LDL/FLDs and LSDs. Several structural differences can be pointed out among the different plant and animal clades. In particular, all LDL/FLD homologs of group PII have long N- and C-terminal extensions, as well as long SWIRM/AO regions, as compared the LDL/FLD homologs of group PI, with the exception of the LDL3 homologs of the two green algal species and S. moellendorffii which display short N-terminal extensions (**Table 2**). Also animal LSD homologs of both clade AI and AII have large N-terminal extensions, but small or null C-terminal extensions (**Table 2**). In particular, LSD2 homologs display longer N-extensions with respect to the LSD1 homologs, probably through acquisition of DNA- or protein-interaction domains. Indeed, HsLSD2 possesses both CW-type and C4H2C2-type zinc finger motifs (Burg et al., 2015). Furthermore, similarly to the Arabidopsis LDL3, all LDL/FLD homologs of group PII, except the Ostreococcus tauri one, are characterized by the presence of a small insertion inside the AO domain (**Table 2**), at the same position of the 'Tower' domain in the animal LSD1 homologs. This insertion is smaller than the HsLSD1 'Tower' domain, being of 47 amino acids in S. moellendorffii, 50 amino acids in A. trichopoda, 39 to 43 amino acids in monocots, 33 to 39 amino acids in dicots, and 24 amino acids in Ostreococcus lucimarinus. Only the LDL3 homolog of P. patens (PpLDL3) displays an insertion of a size (94 amino acids) similar to that of the animal 'Tower' domain (**Table 2**). Molecular modeling analyses indicate that this insertion may adopt a fold similar to that of the HsLSD1 'Tower' domain (**Figure 2**). In contrast, the S. moellendorffii insertion appears unstructured (**Figure 2**). The functional significance of these structural differences among the different plant and animal LDL/FLDs and LSDs are not clear so far.

### Expression Pattern of the Four Arabidopsis LDL/FLD Genes During Seedling Development

Since information concerning the tissue- and organ-specific gene expression pattern may be useful to determine physiological roles, promoter regions of the four Arabidopsis LDL/FLD genes were cloned upstream of a GFP-GUS fusion gene, and LDL/FLD::GFP-GUS transgenic Arabidopsis plants were obtained. Histochemical GUS staining of developing seedlings showed that LDL1 is expressed in the shoot apical meristem (SAM), in the newly emerging leaves (**Figures 4A,B**), and in the root tip (**Figure 4C**). Cotyledons also appeared stained mainly at the tips and along the vascular system (**Figure 4A**). Strong LDL1-specific GUS staining was also observed in trichomes (**Figure 4D**). LDL2-specific GUS staining was observed in the root elongation and differentiation zones up to the meristematic region and in the columella of primary and secondary roots (**Figures 4E,G,H**). LDL2 expression was also observed in the SAM and the newly emerging leaves (**Figures 4E,F**). FLD is expressed in the root apex of primary (**Figures 4I,K**) and secondary roots (**Figure 4L**), in the SAM and in the newly emerging leaves (**Figures 4I,J**). It is also expressed in the vascular system of cotyledons, roots (**Figures 4I,L**) and fully developed leaves. LDL3 is expressed in newly emerging leaves (**Figures 4M,N**), in the columella and in the root vascular system (**Figures 4M,O**). LDL3-related GUS staining was also observed in the vascular system of leaves (**Figure 4P**), in guard cells (**Figure 4Q**), and in trichomes (**Figure 4N**).

### Expression of Arabidopsis LDL/FLD Genes During Flower Development and Embryogenesis

LDL1-related GUS signal was observed in young, completely closed floral buds (**Figure 5A**). Staining in developing anthers and in particular in both anther tapetum and filaments was also observed (**Figure 5B**). In later steps of flower development, strong GUS staining was evident in mature pollen grains (**Figure 5C**), while in non-fertilized ovules only faint staining was observed. During embryo development, LDL1-related GUS staining was present in the funiculus of the fertilized ovule, mainly at the ovule proximal region (**Figure 5D**). Furthermore,

developing and mature embryos presented staining at the central part of cotyledons (**Figures 6A–C**) and this pattern was maintained in fully developed embryos, as observed by GUS histochemical analysis of imbibed seeds (**Figure 6D**). The LDL1 expression in imbibed seeds is in agreement with the public Arabidopsis microarray database and the reported essential role of LDL1 and LDL2 in seed dormancy (Zhao et al., 2015).

LDL2-specific GUS staining was observed in developing pistils and anthers of floral buds (**Figure 5E**). Later during development, mature pollen grains (**Figure 5F**) and embryo sacs (**Figures 5G,H**) were stained too, embryo sacs presenting a strong signal at the micropylar end (**Figure 5H**). Following fertilization, strong LDL2-specific staining was observed in developing embryos at the heart and torpedo stages and in mature embryos (**Figures 6E–G**). This staining was extended in the entire embryo, excluding only the embryonal root tip. In embryos within the imbibed seeds, the expression pattern resembled the one of the young seedlings, with strong promoter activity in SAM, cotyledons, and in the root elongation zone (**Figure 6H**).

FLD-related GUS staining was observed in the anther– filament junction and in the tapetum (**Figures 5I,J**), but not in mature pollen grains. Ovules also appeared stained (**Figures 5K,L**). Following fertilization, strong FLD-specific staining was observed in developing embryos at the heart and torpedo stages and in mature embryos (**Figures 6I–L**). Mature embryos in imbibed seeds also presented staining (**Figure 6M**). Interestingly, FLD expression was evident in the provascular tissues of embryonic roots and cotyledons (**Figure 6L**, arrows),

similarly to FLD expression in root, cotyledon and leaf vascular system of young seedlings (**Figures 4I,L**).

LDL3-specific staining was observed in peduncles, sepals (**Figure 5M**), stamen filaments (**Figure 5O**) and mature pollen grains (**Figure 5N**). Pistils were also stained (**Figures 5M,O**), in particular ovules (**Figures 5O,P**). Following fertilization, LDL3-specific staining was observed both in developing and in mature embryos (**Figures 6N–Q**). Funiculus of fertilized ovules presented staining as well (**Figure 6O**). In mature embryos, staining of the SAM and root tip was evident (**Figure 6Q**). The same staining pattern was observed in mature embryos inside imbibed seeds (**Figure 6R**).

### LDL3 Mutant Plants Show Early-Flowering Phenotype

To elucidate the physiological roles of the four Arabidopsis LDL/FLD genes, loss-of-function mutants for each of the four

Arabidopsis LDL/FLD genes were identified and characterized to confirm disruption of gene expression (**Supplementary Figure 2**). The ldl/fld mutants were initially examined for flowering time by measuring the number of rosette leaves upon bolting. In agreement with previously published data (He et al., 2003), the fld mutant presented an extremely late-flowering phenotype. Indeed, floral transition was not obtained under our experimental conditions unless the fld mutant plants were treated with gibberellins or grown under low-temperature conditions for prolonged periods, further confirming that FLD is involved in the autonomous pathway controlling flowering time. Under the same conditions, the ldl1 and ldl2 mutants displayed only a very short, not statistically significant, delay in flowering (**Figure 7**) as previously reported (Jiang et al., 2007). Conversely, data

presented here evidence that ldl3 mutant display early-flowering phenotype (**Figure 7**).

### FLC Is Up-Regulated in ldl1, ldl2, and fld Mutants, but Down-Regulated in ldl3 Mutants

To verify whether the early-flowering phenotype of the ldl3 mutant depends on FLC, as the late-flowering phenotype of ldl1, ldl2, and fld mutants does (He et al., 2003; Jiang et al., 2007), a comparative analysis of the FLC expression levels in the four lsd/fld mutants was performed. The qRT-PCR analysis evidenced twofold and fourfold increase in FLC expression levels in the ldl1 and ldl2 mutants, respectively, comparing to the wild-type plants, as opposed to a 100-fold increase in the fld mutant (**Figure 8**), consistently with the flowering phenotypes. Conversely, a twofold decrease in FLC expression levels was observed in the ldl3 mutant as compared to the wild-type plants (**Figure 8**), which is also consistent with the early-lowering phenotype of this mutant. These results suggest that the various members of LDL/FLD gene family contribute in a different way to the control of FLC expression and thus to the flowering time, LDL3 having an opposing effect with respect to the others.

MAF1 to MAF5, MADS-containing transcription factors homologs to FLC, also contribute to the control of the flowering time (Ratcliffe et al., 2003; Rosloski et al., 2013). Data from qRT-PCR analysis evidenced no statistically significant difference in the expression levels of MAF1 to MAF4 in all four ldl/fld mutants as compared to the wild-type plants (**Supplementary Figure 3**). These data indicate that Arabidopsis LDL/FLDs are not involved in the regulation of the MAF gene family despite the fact

that fld mutants were previously shown to display altered H3K4 trimethylation levels at MAF4 and MAF5 (Yu et al., 2011).

indicate statistically significant differences from WT plants (one-way ANOVA

FWA is a transcription factor which participates in the control of floral transition and which has been demonstrated to be under epigenetic control. In particular, FWA is silenced during plant vegetative development and in the sporophytes by repressive DNA methylation in its 5<sup>0</sup> region, its expression being confined to the central cell of the female gametophytes and to the endosperm. Moreover, fwa epi-alleles cause a late-flowering phenotype due to ectopic FWA expression in sporophytic tissues (Soppe et al., 2000; Kinoshita et al., 2004). In addition, FWA was shown

test, p < 0.05).

to be ectopically activated in rosette leaves of ldl1 and ldl2 mutants, but not of fld mutants, suggesting that LDL1 and LDL2 contribute to the repression of FWA expression during vegetative development (Jiang et al., 2007). Here, to determine the specific contribution of the four LDL/FLDs to FWA regulation, qRT-PCR analysis was performed. Results showed a strong increase in FWA expression levels in the ldl2 mutant and a smaller one in the ldl1 and fld mutants (**Figure 8**). Instead, no difference in FWA expression levels was observed in the ldl3 mutant as compared to the wild type plants. These data suggest that it is mainly LDL2, among the four Arabidopsis LDL/FLDs, that is involved in FWA repression during vegetative growth of Arabidopsis plants. This well correlates with the LDL2 expression in the female gametophyte presenting a pattern similar to that of FWA (Kinoshita et al., 2004). Indeed, both LDL2 and FWA are expressed in embryo sacs (**Figure 5N**) (Kinoshita et al., 2004).

### DISCUSSION

In the present work, a comparative study on gene structure, phylogenetic relationships, spatio-temporal expression patterns, physiological roles and target genes for the four LDL/FLD genes was performed, which evidenced several similarities, but also important differences among them.

Data from exon/intron structure analyses and phylogenetic studies suggest that the LDL1, LDL2, and FLD homologs of the various plant species derive from a single copy of the LDL/FLD gene present in the early ancestor of land plants through two duplication events. Furthermore, the different number of introns observed in LDL/FLD genes is likely the result of two sequential insertion events occurred first in the common ancestor of LDL2 and FLD genes and later in the ancestor of FLD genes. This process might have brought about differences in expression levels and function among LDL/FLD homologs of clade PI. Indeed, it has been shown that although all three LDL1, LDL2, and FLD act redundantly in the control of the flowering time, FLD plays a major role in this process (Jiang et al., 2007). Intron acquisition occurred also during evolution of the LDL3 homologs which might have contributed to increase gene expression levels. Differently from LDL1, LDL2, and FLD, LDL3 homologs are characterized by the presence of a small insertion in the same position as that of the 'Tower' domain in HsLSD1. This insertion, which in P. patens is long enough to allow the presence of a structural domain similar to the HsLSD1 'Tower' domain, became shorter during the transition from the moss P. patens (94 amino acids), to the ancient vascular plant S. moellendorffii (47 amino acids), to the basal angiosperm A. trichopoda (50 amino acids), to dicots and monocots (33–47 amino acids). The evolutionary pressure leading to such changes is not known yet. It is possible that these changes have been accompanied by evolution of new protein/protein interaction motifs. Altogether, these data indicate a different evolutionary history of the two main plant LDL/FLD clades, similarly to the two animal LSD clades.

In the present study, an analysis of the main LDL/FLD target genes showed that all four LDL/FLDs are involved in the control of FLC expression. In particular, in agreement with previously published data (Jiang et al., 2007), LDL1, LDL2, and FLD were shown to have a repressive effect on FLC expression levels, with the effect of FLD being much more pronounced than that of LDL1 and LDL2. Instead, LDL3 has an enhancing effect on FLC expression. Indeed, ldl3 mutant plants display decreased FLC transcript levels, as compared to the wild-type plants, while ldl1, ldl2, and fld mutants display increased FLC levels (**Figure 8**). These differences in FLC expression levels reflect the differences in flowering time, fld displaying a non-flowering phenotype, while ldl3 an early-flowering phenotype (**Figure 7**). LDL1, LDL2, and FLD repress also FWA expression LDL2 having a more pronounced effect than LDL1 and FLD (**Figure 8**). The lack of differences in flowering time among ldl1 and ldl2 mutants and wild-type plants, despite the altered FLC and FWA expression levels, suggests a quantitative effect of FLC and FWA on floral transition. Differences in tissue- and temporal-specific expression pattern, as well as in protein/protein interactions among the four LDL/FLDs may also explain the different flowering phenotypes of the four ldl/fld mutants.

Previous studies have shown that LDL1, LDL2, and FLD repress FLC transcription by reducing H3K4 methylation levels at specific regions of the FLC chromatin (Jiang et al., 2007). This raises the question of which are the underlying mechanisms determining the opposing effects of the different LDL/FLDs on FLC expression levels considering the similarity of the catalytic sites (**Table 1**) (Spedaletti et al., 2008). To get through these mechanisms, the LDL3 substrate specificity, the FLC chromatin regions with which LDL3 specifically interacts and the LDL/FLD specific partners have to be determined. On the other hand, the physiological significance of the different/opposing effects of the four LDL/FLDs on floral transition is not clear so far. They may contribute to a fine-tune regulation and optimization of the flowering time.

FLC expression is promoted by FRIGIDA (FRI) and is repressed by sets of genes in the autonomous and vernalization pathways (Amasino and Michaels, 2010; Yang et al., 2017). FLC is expressed in shoot and root apical regions, as well as in leaf vasculature, in pollen mother cells, in the tapetum surrounding these cells, and in the anther connective tissue, but not in mature pollen grains (Bastow et al., 2004; Michaels et al., 2005; Sheldon et al., 2008; Choi et al., 2009). It is also expressed in the ovule integuments before and after pollination, but not in the female gametophytes. FLC is additionally expressed in the developing embryo during all stages of embryogenesis reaching a maximum when the seed has been fully formed (Sheldon et al., 2008; Choi et al., 2009; Berry and Dean, 2015). In old embryos, FLC is expressed in the provascular tissue of both the embryonic roots and cotyledons. Thus, FLC expression is repressed in mature male and female gametophytes to be reactivated after fertilization, in reprogramming processes that are considered important for plant reproduction, mainly ensuring a vernalization requirement in each generation (Berry and Dean, 2015). Little is known so far about how the several FLC regulators control FLC transcription in the various developmental stages (Choi et al., 2009).

Martignago et al. Histone Demethylases and Flowering Time

The promoter activity studies presented here evidence that all four Arabidopsis LDL/FLD genes are expressed in SAM and/or newly emerging leaves. The Arabidopsis LDL/FLD genes are also expressed in roots, though with a gene-specific pattern. Furthermore, all LDL/FLD genes, except LDL2, are expressed in the vascular system of the roots and/or leaves. LDL3, differently from the other LDL/FLD genes, is also expressed in guard cells. LDL/FLDs are also expressed during reproductive development, though with some differences from each other. In particular, prior to fertilization all four LDL/FLDs are expressed in ovules. However, while LDL1, FLD and LDL3 are expressed in the entire ovule, probably mainly involving the ovule integuments, LDL2 is specifically expressed in the embryo sacs. Furthermore, LDL1, LDL2, and LDL3 are expressed in mature pollen grains, as opposed to FLD that is not expressed in male gametophytes. Following fertilization, all four LDL/FLDs are expressed in developing embryos. LDL1 and LDL3 are additionally expressed in the funiculus of developing embryos. Altogether, the promoter activity studies presented here show that the four LDL/FLDs, both the FLC repressors and the FLC activator, display overlapping and complementary expression patterns with respect to each other and to FLC, thus not allowing to assign a specific role to each of them in FLC regulation at certain developmental stages. What appears to be an important difference among the different LDL/FLDs is the lack of FLD-specific expression in pollen grains. These data exclude the possibility that FLD, which among the four LDL/FLDs is the best FLC repressor, is responsible for FLC repression in pollen grains, in agreement with previous data showing that FLC expression pattern during gametogenesis and embryogenesis is not altered in an fld genotype (Choi et al., 2009). LDL1 and LDL2 may have a role in this process, although it is again difficult to explain the presence of LDL3, which acts as an FLC activator, in pollen. Further detailed analyses of FLC expression pattern in single and multiple ldl/fld mutants may give useful information on the specific contribution of the different LDL/FLDs to FLC regulation in a tissue- and organ-specific way and in the reprogramming processes. It is also likely that a balanced activity of different FLC regulators is necessary for proper FLC levels to be established at the various developmental stages. On the other hand, the high expression levels of LDL/FLDs during gametogenesis and embryogenesis suggests a function for this gene family in the transgenerational reset of epigenetic memory, known to affect not only DNA methylation level, but also histone methylation (Zheng et al., 2016).

FWA is specifically expressed in the female gametophytes, mainly in the central cell and in the developing endosperm, for 48 h after pollination (Kinoshita et al., 2004). However, in the present study, a similar expression pattern has been evidenced for LDL2 (**Figure 5H**), which, among the three LDL/FLDs, appears to have a major role in the control FWA expression during vegetative growth of Arabidopsis plants, repressing it. LDL2 may be necessary together with FWA activators in multi-protein complexes for optimal FWA levels in embryo sacs. It is also possible that LDL2 is responsible for repression of FWA expression in ovules 48 h after pollination (Kinoshita et al., 2004).

Altogether, data presented here suggest functional differences among the four Arabidopsis LDL/FLD genes, even among the LDL1, LDL2, and FLD, which are recent derivatives of a common ancestor gene. It is possible that following gene duplication, LDL1, LDL2, and FLD genes have undergone subfunctionalization or neo-functionalization which might have helped in the optimization of the regulatory network controlling floral transition and defense responses (Zhou and Ma, 2008).

Several studies have evidenced a relevant role of the different epigenetic mechanisms in the control of plant developmental and defense/adaptation processes and their impact on agronomical traits other than flowering time, such as yield and fruit ripening (Gallusci et al., 2017; Giovannoni et al., 2017; Annacondia et al., 2018). In this context, the contribution of the LDL/FLD gene family in these processes should be analyzed and involved target genes should be identified. These pieces of information may provide novel biotechnological strategies for crop improvement.

#### AUTHOR CONTRIBUTIONS

PT conceived the research plan. DM and PT designed and performed the experiments. BB, DS, and PT contributed to the phylogenetic analyses. FP performed the molecular modeling analyses. DM, DS, FP, and PT wrote the manuscript. AC and RA provided advice and comments for the manuscript.

#### FUNDING

This work was supported by the Italian Ministry of Education, University and Research (Grant to Department of Science, University 'Roma Tre'-'Dipartimenti di Eccellenza', ARTICOLO 1, COMMI 314–337, LEGGE 423 232/2016), and University 'Roma Tre'. DS is currently supported by the program 'Rita Levi Montalcini' (MIUR, Ministero dell'Istruzione dell'Università e della Ricerca) for the recruitment of young researchers at the University of L'Aquila.

### ACKNOWLEDGMENTS

We are grateful to Plant Systems Biology (University of Gent) for the kind gift of the pKGWFS7 and pK2GW7 binary vectors, the Nottingham Arabidopsis Stock Center for the SAIL\_640\_B10.v1 mutant, and Arabidopsis Biological Resource Center for the SALK lines.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00669/ full#supplementary-material

Altschul, S. F., Madden, T. L., Schäffer, A. A., Zhang, J., Zhang, Z., Miller, W., et al. (1997). Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25, 3389–3402.

Alonso, J. M., Stepanova, A. N., Leisse, T. J., Kim, C. J., Chen, H., Shinn, P., et al. (2003). Genome-wide insertional mutagenesis of Arabidopsis thaliana. Science


REFERENCES

fpls-10-00669 June 1, 2019 Time: 10:29 # 14

interact with plant Tudor/PWWP/MBT-domain proteins. Biochem. Biophys. Res. Commun. 470, 913–916. doi: 10.1016/j.bbrc.2016.01.151


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Martignago, Bernardini, Polticelli, Salvi, Cona, Angelini and Tavladoraki. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Nexus Between Spermidine and Floral Organ Identity and Fruit/Seed Set in Tomato

#### *Savithri U. Nambeesan1, Autar K. Mattoo2 and Avtar K. Handa3\**

*1 Department of Horticulture, University of Georgia, Athens, GA, United States, 2 Sustainable Agricultural Systems Laboratory, USDA-ARS, Beltsville Agricultural Research Center, Beltsville, MD, United States, 3 Center of Plant Biology, Department of Horticulture and Landscape Architecture, Purdue University, West Lafayette, IN, United States*

Polyamines (PAs) constituting putrescine (Put), spermidine (Spd), and spermine (Spm) are ubiquitous in all organisms and play essential roles in the growth and developmental processes in living organisms, including plants. Evidences obtained through genetic, biochemical, and transgenic approaches suggest a tight homeostasis for cellular PA levels. Altered cellular PA homeostasis is associated with abnormal phenotypes. However, the mechanisms involved for these abnormalities are not yet fully understood, nor is it known whether cellular ratios of different polyamines play any role(s) in specific plant processes. We expressed a yeast spermidine synthase gene (*ySpdSyn*) under a constitutive promoter *CaMV35S* in tomato and studied the different phenotypes that developed. The constitutive expression of *ySpdSyn* resulted in variable flower phenotypes in independent transgenic lines, some of which lacked fruit and seed set. Quantification of PA levels in the developing flowers showed that the transgenic plants without fruit and seed set had significantly reduced Spd levels as well as low Spd/Put ratio compared to the transgenic lines with normal fruit and seed set. Transcript levels of *SlDELLA*, *GA-20oxidase-1*, and *GA-3oxidase-2*, which impact gibberellin (GA) metabolism and signaling, were significantly reduced in bud tissue of transgenic lines that lacked fruit and seed set. These findings indicate that PAs, particularly Spd, impact floral organ identity and fruit set in tomato involving GA metabolism and signaling. Furthermore, we suggest that a nexus exists between PA ratios and developmental programs in plants.

#### *University of Valencia, Spain*

*Edited by: Ana Margarida Fortes, University of Lisbon, Portugal Reviewed by: Pedro Carrasco,* 

*Daqi Fu, China Agricultural University (CAU), China*

> *\*Correspondence: Avtar K. Handa ahanda@purdue.edu*

#### *Specialty section:*

*This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science*

*Received: 14 March 2019 Accepted: 24 July 2019 Published: 25 September 2019*

#### *Citation:*

*Nambeesan SU, Mattoo AK and Handa AK (2019) Nexus Between Spermidine and Floral Organ Identity and Fruit/Seed Set in Tomato. Front. Plant Sci. 10:1033. doi: 10.3389/fpls.2019.01033*

Keywords: polyamines, seed set, parthenocarpy, gibberellins, transgenic tomatoes

## INTRODUCTION

Polyamines (PAs), putrescine (Put), spermidine (Spd), spermine (Spm), and thermospermine (Tspm) are hormone-like biogenic amines ubiquitously present in all living organisms including plants. They have been implicated in regulating growth and developmental processes in plants (Kaur-Sawhney et al., 1988; Martin-Tanguy, 2001; Alcazar et al., 2005; Mattoo and Handa, 2008; Nambeesan et al., 2008; Igarashi and Kashiwagi, 2010; Nambeesan et al., 2010; Pegg, 2016; Handa et al., 2018; Tiburcio and Alcázar, 2018). PAs impact primary, lateral and adventitious root development (Flores and Galston, 1982), plant architecture (Cui et al., 2010), *in vitro* plant regeneration by somatic embryogenesis (Bastola and Minocha, 1995) and organogenesis (Ikeuchi et al., 2016), flowering (Malmberg and McIndoo, 1983; Tiburcio et al., 1988), fruit ripening (Mehta et al., 2002), and leaf and flower senescence (Kaur-Sawhney et al., 1980; Nambeesan et al., 2010; Mattoo and Sobieszczuk-Nowicka, 2019). Genetic, biochemical, and transgenic approaches have demonstrated that the metabolism of various PAs is tightly regulated, and imbalance in their homeostasis results in abnormal phenotypes (Kaur-Sawhney et al., 1980; Flores and Galston, 1982; Malmberg and McIndoo, 1983; Kaur-Sawhney et al., 1988; Tiburcio et al., 1989; Bastola and Minocha, 1995; Mehta et al., 2002; Ge et al., 2006; Cui et al., 2010; Nambeesan et al., 2010; Ikeuchi et al., 2016; Murray-Stewart et al., 2018). Furthermore, each PA may have a different effect on metabolism, growth, and development in plants (Handa and Mattoo, 2010; Mattoo et al., 2010; Tiburcio and Alcázar, 2018). Mutations in PA biosynthetic pathway or inhibitors that reduce Put and Spd levels cause plant lethality (Kusano et al., 2008). In addition, it is now apparent that Spm and Tspm play significant role(s) in abiotic responses and xylem differentiation, respectively (Takahashi and Kakehi, 2009; Yoshimoto et al., 2016; Seiff and Shelp, 2019). However, mechanisms that impart PA-associated phenotypes are not yet fully understood.

A close association of PAs with floral development in plants is known (Galston and Sawhney, 1990; Walden et al., 1997). Specifically, PA levels have been correlated with flower development and fruit bearing in olive (Pritsa and Voyiatzis, 2004; Pritsa and Voyiatzis, 2005), apricot (Alburquerque et al., 2006), and plum (DeDios et al., 2006). Early studies also reported higher Put, Spd, and Spm in floral organs of a male-sterile *stamenless-2* (*sl-2/sl-2*) tomato mutant (Rastogi and Sawhney, 1990a; Rastogi and Sawhney, 1990b). In addition, regeneration of tobacco plant from a cell line impaired in PA metabolism produced flowers with a second row of petals in place of anthers (Malmberg, 1980), while a MGBG-[methylglyoxalbis (guanylhydrazone)]-resistant tobacco cell line with elevated PA levels also exhibited abnormal floral phenotype producing flowers with anthers in place of ovules (Malmberg and McIndoo, 1983). MGBG is a known inhibitor of S-adenosyl methionine decarboxylase (SAMdc), a critical enzyme in PA biosynthesis pathway. Inhibition of SAMdc by MGBG led to impaired flowering, which was reversed by Spd in *Spirodela punctata* (De Cantu and Kanderler, 1989). Similarly, inhibition of other PA-biosynthesis enzymes using alpha-difluoromethylarginine (DFMA), an arginine decarboxylase inhibitor, and alpha-difluoromethylornithine (DMFO), an ornithine decarboxylase inhibitor, prevented floral bud initiation and subsequent development of floral bud in tobacco cell culture explants (Tiburcio et al., 1988). Tobacco and petunia mutants impaired in PA production also exhibited abnormal flower phenotype (Malmberg and McIndoo, 1983, Gerats et al., 1988; Malmberg and McIndoo, 1988), whereas high levels of PA in a tomato mutant resulted in abnormal stamen development (Rastogi and Sawhney, 1990a). Although these studies provided indirect evidence in favor of a role of PAs in flower development, the differential role(s) of specific PAs in these processes has remained to be discerned.

To determine the role of Spd during plant development including flower initiation, maturation, fruit set, and seed set, we introduced a yeast spermidine synthase gene (*ySpdSyn*) under the control of either a fruit-ripening-specific promoter (*SlE8*) or a constitutive *CaMV35S* promoter in tomato (Nambeesan et al., 2010). The *ySpdSyn* transgenic line with fruit-specific promoter *SlE8* developed normal fruit and seed set, and the fruits had longer shelf life (Nambeesan et al., 2010). However, the expression of *ySpdSyn* under *CaMV35S* promoter resulted in two types of transgenic plant populations: one that normally set fruit and seed and the other that did not set fruit/seed and exhibited a range of flower and fruit phenotypes. Thus, these lines with constitutive expression of *ySpdSyn* were employed in our studies presented here to address the question of PA role in floral organ identity and fruit/seed set in tomato.

Our results demonstrate that flowers of the transgenic plants that set fruit and seeds exhibit significantly higher Spd/Put and Spd/Spm ratios compared to flowers from the transgenic plants that failed to set fruit and seeds. The lack of fruit and seed set was found associated with a reduction in the transcript levels of GA biosynthesis genes and *DELLA*, a negative regulator of GA signaling. We interpret these results to indicate that PAs influence floral organ identity and fruit set in tomato by modifying the genetic program that controls expression of GA biosynthesis and signaling genes.

### MATERIALS AND METHODS

#### Generation of Transgenic Plants and Phenotypic Measurements

Transgenic plants overexpressing yeast spermidine synthase (*ySpdSyn*) were generated as described previously (Nambeesan et al., 2010). The *ySpdSyn* gene (ScSpe3, systematic name: YPR069C) was amplified from a yeast genomic library and cloned in the sense orientation between a *CaMV35S* promoter and the 3' end of a pea rbcS-E9 gene in pKYLX71 (Nambeesan et al., 2010). This construct was introduced into disarmed *Agrobacterium tumefaciens* LBA4404, and *Agrobacterium* strains harboring the chimeric constructs were used for transformation of tomato cv. Ohio 8245 cotyledons (Tieman et al., 1992). Successful plantlets obtained were transplanted into potted soil and grown in the greenhouse.

Fifteen independent transgenic plants expressing *CaMV35SySpdSyn* were generated. Two independent transgenic lines C13, C14 that exhibited floral morphological defects, impaired fruit and seed set, and sterility were propagated as vegetative cuttings and characterized alongside the isogenic wild-type (WT) Ohio 8245 control.

#### Phenotypic Analysis of Floral Morphology

For phenotypic measurements, 27–36 flowers were harvested from WT and transgenic plants, and the number of sepals, petals, and stamens were counted. Any abnormality in the stamen morphology was noted. The flower size was measured diagonally across the flower. The flowers were dissected to observe the morphology of the gynoecium. Pollen morphology from dissected stamens was observed under a light microscope (Olympus, Center Valley, PA).

#### Pollen Germination Assay

For pollen germination assay, open flowers were hydrated for 1 h in a Petri dish with wet Kim wipes to maintain a relative humidity close to 100%. The hydrated pollen from the flowers was spread on a glass slide with 30 µl germination medium [20 mM MES, 2% sucrose, 15% PEG 4000, 1 mM KNO3, 3 mM Ca(NO3)2, 0.8 mM MgSO4, and 1.6 mM H3BO3, pH 6.0]. The pollen grains were incubated for 3 h in the dark at 25°C under high relative humidity by placing the glass slide on a moist filter paper in a Petri dish. Germination was terminated by the addition of 5 µl of phenol solution. Pollen germination was scored using a light microscope (Olympus, Center Valley, PA). Pollen tubes that were twice the diameter of the pollen grain were scored as positive for successful pollination. Pollen grains were examined from at least five flowers with 100–200 pollen counted per flower. Digital images were captured using the SPOT software (Draper, Utah).

#### Polyamine Measurement

Mature leaf and tissue from various stages of flower development, including early buds (0.1–0.7 mm in length), late buds (0.8– 1.5 cm in length), and fully open flowers were harvested. Three independent biological replicates at each developmental stage were collected and were frozen immediately in liquid nitrogen and stored at −80°C until analyzed. All samples were processed and quantified in triplicates for polyamine measurement as described previously (Nambeesan et al., 2010).

#### Semiquantitative and Quantitative RT-PCR

Mature leaf and flowers were collected at two stages of development, ~135 early buds (0.1–0.7 mm in length) and 75 late buds (0.8–1.5 cm in length) from WT, C13, and C14 lines. Total RNA was extracted, and cDNA synthesis was performed following the manufacture's recommendation (Promega, Madison, WI). Briefly, total RNA (1 μg) was treated with DNase and ImPromII reverse transcriptase to perform reverse transcription. The cDNA was diluted five times, and 1 μl of diluted cDNA was used for semiquantitative reverse transcription PCR (RT-PCR) in a 10-μl total reaction volume. For each primer pair, cycle number was determined for exponential amplification, and all subsequent reactions were performed for the predetermined number of cycles. Each PCR reaction typically included the following cycles: 95°C (3 min); [95°C (30 s); 55/56°C (30 s); 70°C (1 min)] 25 cycles; 72°C, 10 min. Tomato *ACTIN* gene was used for normalization. After PCR amplification, the products were separated on a 1.2% agarose gel, stained with 0.5 μg/ml ethidium bromide, and images were captured and processed using Image J software (National Institutes of Health, Bethesda, Maryland). Primers used for PCR amplification are listed in **Supplemental Table 1**.

Quantitative RT-PCR analysis was performed using realtime PCR (Stratagene Mx3005P, San Diego, CA). All primers (**Supplemental Table 1**) along with control *ACTIN* (*SlACTIN*) were validated for relative gene expression analysis. One microliter of the cDNA was utilized in a 15-μl reaction using the SYBR green-PCR master mix (Applied Biosystems, Foster City, CA). Melting curve analysis was performed after the PCR reaction to determine specificity of the PCR products. At least three biological replicates were analyzed for gene expression analyses. Statistical analyses were performed using one-way analysis of variance to compare across transgenic C13 and C14 lines and WT separately for a given developmental timepoint or treatment, using JMP Pro 12 (SAS Institute, Cary, NC, USA). Means were separated using Tukey's honestly significant difference (HSD) test (α = 0.05).

### RESULTS

#### Fruit and Seed Set in *ySpdSyn* Transgenic Tomato Lines

Eleven independent events in transgenic tomato lines positive for the expression of *CaMV35S*‐*ySpdSyn* were evaluated in greenhouse. Five of them—C5, C11, C12, C13, and C14—were found severely impaired in fruit set. They varied from having no seeds to very few seeds and exhibited partial parthenocarpy. Other four independent transgenic lines—C1, C3, C4, and C15—exhibited normal fruit set, but their seed number/fruit varied being 18, 15, and 30% lower in C1, C3, and C4, respectively, except for C15 in which seed numbers were similar to the WT control (**Table 1**). Variable phenotypes, including impaired fruit and seed set and seed germination among the independent transgenic plants, may have resulted from the influence of transgene integration site(s) (De Buck et al., 2013).

#### Expression of *ySpdSyn* and PA Quantification in C13 and C14 Transgenic Lines

Expression of *ySpdSyn* and endogenous *SlSpdSyn* in tomato leaves of C13 and C14 T0 lines is shown in **Figure 1A**. Both transgenic lines expressed *ySpdSyn* and *SlSpdSyn*, while, as expected, WT leaves expressed only the endogenous *SlSpdSyn* gene. The *SlSpdSyn* transcript levels remained unaltered in C13 and C14 lines compared to WT leaves, which indicated lack of endogenous gene silencing (**Figure 1A**). Quantification of different polyamines in these leaves indicated that Spd levels increased by 1.3- and 1.7-fold in C13 and C14 lines, respectively, as compared to the WT (**Figure 1B**). However, the transgenic event caused a 2.1- and 1.5-fold decrease in Spm in C13 and C14, respectively, compared to WT plants. Leaves from C14 lines exhibited significant 1.4-fold increase in Put compared to WT leaves, whereas C13 leaves did not show any change in Put levels compared to WT leaves (**Figure 1B**).

#### Altered Floral Morphology of *ySpdSyn* Under *CaMV35S* Transgenic Plants

The floral and fruit morphology in transgenic and WT plants was evaluated to determine the basis of impaired fruit and seed set in


*Seeds from 10 fruits were extracted for each genotype and counted. Shown are the average 10 independent extraction of seeds from each genotype.*

FIGURE 1 | Generation and characterization of transgenic plants overexpressing yeast spermidine synthase (*ySpdSyn*). (A) The *ySpdSyn* construct was cloned under the *CaMV35S* promoter in pKYLX71 vector. Shown are levels of endogenous *SlSpdSyn* and the control *SlACT* using semiquantitative RT-PCR analysis indicating the expression of *ySpdSyn* and endogenous tomato *SlSpdSyn* in WT, C13, and C14 leaves. The tomato *Actin* gene was used as an internal control (*SlACT*). (B) Polyamine levels in WT, C13, and C14 leaves measured using HPLC. Shown are mean values and standard from three independent biological leaf replicates. Means followed by the same letter are not significantly different from each other based on one-way analysis of variance (α = 0.05) for a given polyamine (PA). Put, putrescine, Spd, spermidine; and Spm, spermine.

C13 and C14 lines. Flower size or floral organ number did not differ significantly among these lines compared to WT plants. All genotypes had five sepals per flower. However, the average petal number was significantly different in line C13 (6.6 petal/flower) as compared to WT and C14 which had 5.8 and 5.3 petals/flower, respectively (**Figure 2D**). Similarly, the average number of stamens in C13 (7 per flower) was significantly higher than the WT and C14 lines that had 5.8 and 5.4 stamens per flower, respectively (**Figure 2D**). However, C13 line had severe morphological defects, with ~17% flowers having 8–10 stamens in comparison to 5.4 in the WT. Most flowers from C14 and C13, however, did not form a normal staminal cone (**Figure 2A–C**). Dissection of flowers from transgenic lines revealed that carpels from C13 (**Figure 2F**) and C14 flowers (**Figure 2G**) were significantly enlarged as compared to the WT flowers (**Figure 2E**). Occasionally, fusion of multiple gynoecia at the base was observed in C13 flowers. The length of the style was reduced in C13 and C14 flowers (**Figures 2E**–**G**) with the tip of the style close to the stigmatic surface curved in transgenic flowers (**Figures 2E**–**G**). Other phenotypes in transgenic flowers included delayed abscission of petals during fruit growth (**Figure 2K**), multiple fruit set in the same flower by fusion of multiple gynoecia at the base (**Figures 2L**, **M**), and altered fruit shape with multiple locules (**Figure 2N**).

We next investigated pollen morphology and germination to determine the basis of differences in fruit set in transgenic lines. WT pollen appeared spherical and intact, and 96% of the pollen displayed a normal morphology (**Figure 2H**). However, ~68% of pollen from C13 and 37% from C14 flowers exhibited a collapsed morphology (**Figures 2I**, **J**). *In vitro* pollen germination assays indicated reduced pollen germination in C13 and C14 lines compared with WT. Among the intact-appearing pollen, 79% of WT and only 17% of C13 and 40% of C14 pollen grains germinated (data not shown). None of the abnormal looking pollen grains germinated (data not shown). The reciprocal crosses using C13 and C14 pollen on WT stigma did not result in fruit set, but pollination of C13 and C14 using WT pollen resulted in 100 and 80% parthenocarpic fruit set, respectively (data not shown), Among the fruit that set seed in C14, only one seed was present. These results indicated that fruit development occurred without fertilization. Reduced seed set in C14 indicated partial female sterility.

#### PA Homeostasis and Changes in Spd/Put and Spd/Spm Ratios

We determined the levels of free polyamines (Put, Spd, and Spm) in fully opened flowers from WT and four *CaMV-ySpdSyn* lines to evaluate if altered PAs homeostasis was associated with the observed phenotypes (**Figure 3**). The steady-state Spd level in the flowers of other transgenic lines, viz., C4, C15, and WT plants was similar; these genotypes had normal fruit and seed set (**Figure 3**). The flowers of C13 and C14 lines exhibited parthenocarpic phenotype and had significantly lower Spd levels than the WT flowers (**Figures 2** and **3**). Notably, the Spd/Spm ratio in the flowers of C4 and C15 plants was 8.8 and 7.0, respectively, and only 2.9 and 3.1 in C13 and C14 flowers, respectively (**Figure 3**). Similarly, the Spd/Put ratio was significantly higher in C4 and C15 flowers compared to the WT flowers but significantly lower in C13 to C14 flowers. This pattern was also reflected in the Spd/ TPA (**Figure 3**). Notably, the ratio of Put/TPA was significantly lower in C4 and C15 flowers and significantly higher in C13 and C14 flowers compared to WT flowers, respectively. Consistent patterns for Spm/TPA ratio were not observed for transgenic flowers from the four independent lines examined (**Figure 3**). Taken together, these results suggest that reduced Spd and Spd/ Put ratios are associated with the flower phenotype and may therefore affect fruit and seed set in tomato.

#### Establishment of Polyamine Homeostasis During the Early Flower Development

Expression patterns of *ySpdSyn* and endogenous *SlSpdSyn* were compared at various stages of flower development including early

and late bud stages and fully open flowers. The *ySpdSyn* transcripts were present at all the stages of flower development in the C13 and C14 lines and, as expected, not in the WT (**Figure 4A**). As anticipated, *SlSpdSyn* transcript was present at all the stages of flower development, indicating the lack of silencing of the endogenous or *ySpdSyn* transgene in the transgenic lines (**Figure 4A**).

We quantified the levels of Put, Spd, and Spm in early and late flower buds to determine if the observed changes in various PA levels described in fully opened WT flower were also present during the flower development in C13 and C14 genotypes that had abnormal fruit and seed set (**Figure 4B**). The Spd level was not significantly different at early and late bud stages in the C13 line but was significantly lower in the C14 line at both stages of bud development. Furthermore, Spm levels in the C13 and C14 genotypes were significantly lower than the WT at early and late bud stages of flower development, whereas TPA and Put levels were lower at the latter stage.

The Spd/Put ratio was significantly higher at early bud stage in the C13 and C14 lines but not at late bud stage as compared with the WT. Spd/Spm ratio was significantly higher in the early and late bud stage in C13, whereas it decreased at the late bud stage in C14 line (**Figure 4C**). A significant decrease in Spm/Put ratio in both the C13 and C14 genotypes compared to WT was apparent only at the early bud development stage (**Figure 4**). Although the ratio of Spd/TPA was higher at the early bud stage, however, in general,

FIGURE 3 | Changes in the free total PAs (TPA), Put, Spd, and Spm (A), ratios of Spd/Put, Spd/Spm, and Spm/Put (B), and ratios of Put, Spd, and Spm to TPA (C) in fully open flowers of WT, C4, C15, C13, and C14 plants. WT, Wild-type Ohio 8245; C4, C15, C13, and C14 transgenic plants harboring *CaMV-SpdSyn* transgene. C4 and C15 plants displayed fruit and seed set, while the C13 and C14 plants exhibited abnormal flower phenotype and impaired seed set. Shown are the mean values and standard errors from the three independent biological replicates. Means followed by the different letters are significantly different from each other based on one-way analysis of variance (α = 0.05) for a given polyamine (PA).

fruit formation from a single flower in C13. (M, N). Altered fruit shape and

multilocular fruits in C13 in comparison to wild-type fruit.

Put/TPA and Spm/TPA ratios in both genotypes at the early and

#### Steady-State Levels of *SlDELLA* and GA Biosynthesis Genes Are Lower in the Flowers of *ySpdSyn* Transgenic Plants With Impaired Fruit and Seed Set

late bud stages of development were not different from the WT.

Flower initiation and development are known to be regulated by the hormone gibberellins (GAs) (Davis, 2009). Therefore, we surmised that PAs and GAs may interact to regulate floral programs. To understand such an association between GAs and PAs, we analyzed the expression of tomato *DELLA* (*SlDELLA*) and GA-biosynthesis genes (**Figure 5**). *DELLA* expression decreased significantly at early stages of bud development in C13 and C14 lines being 1.8- and 1.4-fold lower in C13 and C14, respectively, as compared to WT (**Figure 5A**).

The expression of *GA20-oxidase1* (*GA20ox1*) decreased by 2.2-fold in the C13 line at the early bud stage, decreasing further by 5.6-fold in the late bud stage (**Figure 5B**). A similar pattern was obtained for flower bud development in C14 line. Expression of *GA20-oxidase2* (*GA20ox2*) was not significantly different at the early bud stage of both the transgenic lines; however, at the late bud stage, *GA20ox2* expression was elevated ~2- and ~2.5-fold in the C13 and C14 lines, respectively (**Figure 5C**). In contrast, *GA3ox2* expression was significantly

Means followed by the different letters are significantly different from each other based on one-way analysis of variance (α = 0.05) for a given polyamine (PA).

FIGURE 5 | Transcript levels of *SlDELLA* and GA-biosynthesis genes during flower development of *ySpdSyn and WT* tomato lines. Flower development stages, early buds (0.1–0.7 cm) and late buds (0.8–1.5 cm), were harvested from plants grown in a green house. Expression of the tomato homologue of (A) *DELLA* (*SlDELLA*); (B) GA-20oxidase-1 (*GA20ox1*); (C) GA-20oxidase-2 (*GA20ox2*); and (D) GA-3oxidase-2 (*GA3ox2*), during flower development in wild type, C13 and C14 flower buds was analyzed using quantitative RT-PCR (Supplemental File 1). The relative gene expression analysis was performed by the ∆∆Ct method using tomato *ACTIN* as a reference gene. Data shown are mean + standard error based on ting three independent biological replicates of WT, C13, and C14 at early and late bud stage of flower development. Means followed by the same letter are not significantly different from each other based on one-way analysis of variance (α = 0.05) for a given tissue type of a gene.

downregulated ~4-fold in both C13 and C14 lines at the early bud stage with no significant differences at the late bud stage between the WT and two transgenic lines (**Figure 5D**).

## DISCUSSION

Put, Spd, and Spm have two, three, and four positive charges, respectively, and thus can compete differentially to bind various biomolecules (Igarashi and Kashiwagi, 2010). Changes in the levels of any specific PAs (Put, Spd, or Spm) may affect their binding to biomolecules resulting in the perturbation of various biological processes, including transcription and metabolome (Mattoo et al., 2006; Srivastava et al., 2007). We demonstrate here that the expression of *CaMV35S-ySpdSyn* plants results in significant changes in the ratios of Spd/Put, Put/TPA, and Spd/TPA in transgenic flowers. Transgenic plants expressing *ySpdSyn* gene under the control of *CaMV35S* promoter gave rise to two distinct populations of transgenic plants. About one-half of the independent transgenic plants exhibited normal flower development with fruit and seed set, while the other half of the transgenic population showed floral abnormalities and parthenocarpic fruit set (**Figure 2**). The Spd level in the transgenic flower with normal fruit and seed set phenotype was similar to WT flowers but lower by threefold in transgenic flowers with abnormal parthenocarpic fruit than WT flowers. We propose that such alterations of individual PAs and their ratios affect the phenotypes observed in transgenic flowers. Altered PA homeostasis with increased PA ratios such as Spd to Spm ratio has been recently reported in Snyder–Robinson syndrome (Murray-Stewart et al., 2018). These findings provide further support and add to the role(s) of PAs in flower development and parthenocarpy (Rastogi and Sawhney, 1990b; Antognoni et al., 2002; Fos et al., 2003; Sinha and Rajam, 2013; Chen et al., 2014)

Furthermore, these data corroborate earlier envisaged reports that each polyamine Put, Spd, and Spm differentially influence plant phenotypes, metabolism, and gene expression (Handa and Mattoo, 2010; Mattoo et al., 2010; Tiburcio and Alcázar, 2018). Thus, Spd and Spm were positively correlated with tomato fruitquality attributes and specific metabolites in poplar cell culture and tomato pericarp in contrast to Put, which exhibited a negative correlation with these same traits (Handa and Mattoo, 2010; Mattoo et al., 2010). The importance of PA ratios in regulating various biological systems including plants and mammalian systems has been previously suggested, but their physiological significance has remained elusive (Huang et al., 2004; Murray-Stewart et al., 2018). Changes in PA ratios were also found in Gy mice and human patients with Snyder–Robinson syndrome, a genetic condition, resulting from the lack of SpmSyn activity. Both these phenotypes were found associated with increased Spd/Spm ratio due to a decrease in Spm (Pegg, 2016). Since high levels of Spd restored the normal growth of Gy mice lacking SpdSyn in cultured fibroblast, it has also been suggested that the Spm/Spd ratio plays an important role in myocyte differentiation (Luchessi et al., 2009). Interestingly, the Spd/Spm ratio has also been implicated in altering activity of Kir channels affecting the resting membrane potential, cardiac and neuronal electrical activity, and electrolyte balance (Stanfield and Sutcliffe, 2003). In addition, our findings are in sync with the contention that cellular levels of individual PAs, their homeostasis, particularly ratio among different PAs, regulate various physiological responses (Murray-Stewart et al., 2018).

Molecular basis of reduced levels of Put, Spm, and total PAs in transgenic flowers from all the transgenic lines tested here and reduction in Spd levels in only about half the independent transgenic lines tested here is not clear. Variable expression of the transgenes under the *CaMV35S* promoter has been previously observed in several investigations (Tieman et al., 1992; Somssich, 2019). Nonetheless, that the ratio of Spd/Spm and Spd/Put in transgenic flower with normal fruit and seed set were higher compared to transgenic abnormal transgenic flowers (**Figure 3**) may indicate that it is Spd ratio with Put and Spm that is crucial for the observed flower phenotypes. This warrants further investigation.

PA and GA metabolism and signaling were linked to floral and early fruit development in tomato, but the nature of these interactions was not clear (Anwar et al., 2015). High GA levels in *pat-2* ovaries were associated with increased PA biosynthesis and higher Spm levels (Fos et al., 2003). In *Arabidopsis*, GA promotes flower development by suppressing expression of DELLA group of proteins and, in turn, promotes the expression of floral homeotic genes (Okamuro et al., 1997; Cheng et al., 2004; Tyler et al., 2004; Yu et al., 2004; Cao et al., 2006). The tomato *procera* mutant in which *SlDELLA* gene is mutated has a constitutively induced GA phenotype with facultative parthenocarpy and meristematic alterations (Carrera et al., 2012). In addition, overexpression of *GA20oxidase* resulted in some flowers having protruding stigma and partial parthenocarpy in tomato (García-Hurtado et al., 2012). In the present study, the expression of *SlDELLA* and GA-biosynthesis-related genes, *GA20ox1* and *GA3ox2*, was significantly downregulated in those transgenic lines that had abnormal flower phenotype. Furthermore, *DELLA*silenced lines are downregulated in *GA20ox* and *GA3ox* during early fruit development, suggesting that silencing of *DELLA* may result in higher GA levels which in turn reduce the expression of GA-biosynthetic genes through a feedback regulation (Martí et al., 2007). For instance, GA-deficient *Arabidopsis* mutants have a higher expression of GA biosynthesis genes, *AtGA20ox1* and *AtGA3ox1*, which could be reduced by GA application (Yamaguchi et al., 1998; Xu et al.,

### REFERENCES

Alburquerque, N., Egea, J., Burgos, L., Martinez-Romero, D., Valero, D., and Serrano, M. (2006). The influence of polyamines on apricot ovary development and fruit set. *Ann. Appl. Biol.* 149, 27–33. doi: 10.1111/j.1744-7348.2006.00067.x

1999; Matsushita et al., 2007). In light of these observations, our data on decreased expression of *GA20ox1* and *GA3ox2* in the C13 and C14 transgenic lines may be a result of a similar feedback regulation and suggest that PAs play an important role in flower development and fruit set likely by interacting with GA signal transduction and biosynthetic pathways in tomato. The contention that Spd levels may have an important role in flower development has been previously proposed (Malmberg, 1980; Malmberg and McIndoo, 1983; Gerats et al., 1988; Malmberg and McIndoo, 1988; Tiburcio et al., 1988; Galston and Sawhney, 1990; Rastogi and Sawhney, 1990a, Rastogi and Sawhney, 1990b; Pritsa and Voyiatzis, 2004, Pritsa and Voyiatzis, 2005; Alburquerque et al., 2006; DeDios et al., 2006). However, much more definitive evidence is required to conclude that altered PA ratios regulate floral developmental process(es). In conclusion, our studies present a compelling suggestion for a nexus between PA ratios and regulation of developmental programs in plants.

### DATA AVAILABILITY STATEMENT

All datasets for this study are included in the manuscript/the **Supplementary files**.

### AUTHOR CONTRIBUTIONS

All authors had equal roles in conceptualization, data analyses, and manuscript preparation.

### FUNDING

Studies were partly supported by a US–Israel Binational Agricultural Research and Development Fund to AH and AM (grant no. IS-3441-03) and USDA-IFAFS program grant (award no. 741740) and USDA/NIFA Hatch IND011872 to AH. AM is supported through USDA-ARS intramural Project No. 8042-21000- 143-00D.

## ACKNOWLEDGMENTS

We thank Purdue University Library and University of Georgia for providing the publication cost of this manuscript.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.01033/ full#supplementary-material

Alcazar, R., Garcia-Martinez, J. L., Cuevas, J. C., Tiburcio, A. F., and Altabella, T. (2005). Overexpression of ADC2 in *Arabidopsis* induces dwarfism and late-flowering through GA deficiency. *Plant J.* 43, 425–436. doi: 10.1111/j.1365-313X.2005.02465.x Anwar, R., Mattoo, A. K., and Handa, A. K. (2015). "Polyamine interactions with plant hormones: crosstalk at several levels," in *Polyamines universal molecular*  *nexus for growth, survival and specialized metabolism*. Eds. T. Kusano and H. Suzuki (Japan: Springer), 267–302. doi: 10.1007/978-4-431-55212-3\_22


and development in *Polianthes tuberosa. J. Plant Physiol.* 161, 709–713. doi: 10.1078/0176-1617-01256


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2019 Nambeesan, Mattoo and Handa. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Conservation of Thermospermine Synthase Activity in Vascular and Non-vascular Plants

Anna Solé-Gil, Jorge Hernández-García, María Pilar López-Gresa, Miguel A. Blázquez and Javier Agustí\*

Instituto de Biología Molecular y Celular de Plantas, Consejo Superior de Investigaciones Científicas – Universidad Politécnica de Valencia, Valencia, Spain

In plants, the only confirmed function for thermospermine is regulating xylem cells maturation. However, genes putatively encoding thermospermine synthases have been identified in the genomes of both vascular and non-vascular plants. Here, we verify the activity of the thermospermine synthase genes and the presence of thermospermine in vascular and non-vascular land plants as well as in the aquatic plant Chlamydomonas reinhardtii. In addition, we provide information about differential content of thermospermine in diverse organs at different developmental stages in some vascular species that suggest that, although the major role of thermospermine in vascular plants is likely to be xylem development, other potential roles in development and/or responses to stress conditions could be associated to such polyamine. In summary, our results in vascular and non-vascular species indicate that the capacity to synthesize thermospermine is conserved throughout the entire plant kingdom.

#### Edited by:

Rubén Alcázar, University of Barcelona, Spain

#### Reviewed by:

Thomas Berberich, Senckenberg Nature Research Society, Germany Taku Takahashi, Okayama University, Japan

> \*Correspondence: Javier Agustí jagusti@ibmcp.upv.es

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 08 March 2019 Accepted: 02 May 2019 Published: 11 June 2019

#### Citation:

Solé-Gil A, Hernández-García J, López-Gresa MP, Blázquez MA and Agustí J (2019) Conservation of Thermospermine Synthase Activity in Vascular and Non-vascular Plants. Front. Plant Sci. 10:663. doi: 10.3389/fpls.2019.00663

Keywords: plants, polyamines, thermospermine, evolution, development

#### INTRODUCTION

Polyamines are positively charged aliphatic compounds with a widespread presence in all living organisms (Tabor and Tabor, 1984). The diamine putrescine (Put) and the triamine spermidine (Spd) are the most commonly found polyamines, while the tetra-amine spermine (Spm) is found in yeasts, most animals, seed plants and some bacteria (Pegg and Michael, 2010). Another tetra-amine, thermospermine (Tspm), has been detected in archaea, diatoms and plants, but not in animals or bacteria (Michael, 2016).

The ability to synthesize specific polyamines is clade-specific and is mainly the result of evolutionary adjustments in the polyamine biosynthesis pathway, reflected in the presence or absence of specific polyamine biosynthetic enzymes in given clades. Triamines and tetra-amines are mainly synthesized by a set of aminopropyl transferases, which are evolutionarily related to each other. It has been proposed that while Spm synthase (SPMS) genes in fungi and plants have likely emerged independently after duplication and neofunctionalization of Spd synthase (SPDS) genes (Minguet et al., 2008), the Tspm synthase (TSPMS) gene in plants was probably acquired through endosymbiosis of a cyanobacterium. However, it is arguable whether such gene originally encoded a TSPMS or an agmatine aminopropyl transferase (Pegg and Michael, 2010; Michael, 2016). All in all, the capacity to synthesize different polyamines in a given species is defined by (i) the presence or absence of specific polyamine biosynthetic enzymes and (ii) the degree of specificity

of the polyamine biosynthesis enzymes toward their substrates. For instance, it has been reported that an aminopropyl transferase from the archaea Pyrobaculum aerophilum displays its highest specificity in vitro toward norspermidine, resulting in norspermidine biosynthesis, but it is also able to synthesize Tspm from Spd (Knott et al., 2007). Similarly, the gymnosperm Pinus sylvestris lacks a specific SPMS gene, but the aminopropyl transferase encoded by PsSPDS efficiently converts Put to Spd, as well as Spd to Spm (Vuosku et al., 2018). Both this flexibility in substrate recognition and the repeated independent generation of new aminopropyl transferases along evolution can be explained by the alteration of a few key residues in the active site of aminopropyl transferases that determine their characteristic substrate discrimination, as proposed through structural modeling and crystal structure comparisons of active sites (Wu et al., 2007; Minguet et al., 2008; Sekula and Dauter, 2018).

In plants, polyamines have been implicated in the response to stress and in the modulation of developmental processes (Chen et al., 2018). Correlations between specific endogenous levels of polyamines in plants under different stress conditions or during the progression of specific developmental processes have been extensively reported. In addition, the effects provoked by exogenous polyamines application have been largely documented. However, beyond such physiological reports, solid evidences for polyamines roles come from the analysis of loss of function mutants in polyamine metabolism genes. In this way, Put and Spd have been proved to be essential for life (Imai et al., 2004), while the tetra-amine Spm has been shown to be required for proper acclimation to salt, drought and heat stress (Yamaguchi et al., 2006, 2007; Sagor et al., 2013). Furthermore, both Spm and Tspm promote protection against bacterial pathogens (Gonzalez et al., 2011; Marina et al., 2013). However, the participation of Tspm could be a secondary effect of its primary function in vascular development (Vera-Sirera et al., 2010).

The most extensively studied role for a polyamine is that of Tspm in the regulation of xylem differentiation. The Arabidopsis acaulis5 (acl5) mutant, impaired in TSPMS activity, displays stunted growth (Hanzawa et al., 1997, 2000) which has been associated to an increase in vascular cell proliferation, premature xylem cell death and miss-regulated lignin deposition (Clay and Nelson, 2005; Muniz et al., 2008). In Arabidopsis, the ACL5 gene is specifically expressed in developing xylem cells (Clay and Nelson, 2005; Muniz et al., 2008), and the HD-ZIPIII transcription factor AtHB8 -which directs xylem differentiation and displays the same expression domain as ACL5- mediates its induction by auxin (Baima et al., 2014). Auxin has been shown to promote cell proliferation by ensuring the formation of complexes between Target of Monopteros5 (TMO5) and Lonesome Highway (LHW), leading to the induction of cytokinin biosynthesis (De Rybel et al., 2013, 2014). Recent evidence suggests that Tspm is part of a negative feed-forward loop triggered by auxin that maintains proliferation of vascular cells within the correct range (Katayama et al., 2015; Vera-Sirera et al., 2015). The mechanism involves the promotion of translation of a small family of Suppressor of ACL5 (SACL) transcriptional regulators (Imai et al., 2006; Kakehi et al., 2008) which compete with TMO5 for the formation of inactive SACL-LHW complexes (Katayama et al., 2015; Vera-Sirera et al., 2015). This is not the only case in which polyamines have been implicated in translational regulation (Hanfrey et al., 2002, 2005).

Systematic sequencing of plant transcriptomes and genomes has revealed the presence of ACL5 putative homologs in all plant lineages of vascular and non-vascular plants (including algae), raising the following questions: Do all putative ACL5 genes across plant lineages encode enzymes with bona-fide TSPMS activity, even those present in non-vascular species? Is the expression of these genes associated to xylem development in all vascular plants? To start answering these questions, we have (i) examined the enzymatic activity of the ACL5 homologs of several species in key plant lineages, (ii) searched for Tspm presence in vascular and non-vascular plants and (iii) studied the ACL5 expression pattern in vegetative and reproductive organs of several seedplant species at different developmental stages.

### MATERIALS AND METHODS

#### Plant Material and Growth Conditions

Arabidopsis thaliana Columbia-0 (Col-0) plants were grown on MS media with sucrose (1%) under long-day conditions (16 h light, 8 h dark) at 22◦C in a growth room. Samples for polyamine analysis and RNA extraction from Col-0 were taken after 15 days of growth. P. abies plant material was collected from adult trees that were identified in Catalonia (Spain) and frozen at –80◦C until further analysis. S. lepidophylla plants were obtained from an external vendor, and hydrated before freezing the material at –80◦C for further analysis. P. patens plant material was obtained from Jesús Vicente Carbajosa's Lab (CBGP-Madrid, Spain) and grown in BCD medium + 1 mM Ca2<sup>+</sup> under continuous light conditions in a growth room (Ashton and Cove, 1977). M. polymorpha Tak-1 plants were grown on <sup>1</sup>/<sup>2</sup> strength Gamborg's B5 medium for 15 days before tissue recollection under continuous light conditions in a growth room. C. reinhardttii plant material was obtained from Federico Valverde's Lab (IBVF-Sevilla, Spain).

### Phylogenetic Analysis

Sequences used in this study were obtained by extensive BLAST analysis (tblastn) using the Arabidopsis thermospermine synthase (ACL5-AT5G19530), spermine synthase (SPMS-AT5G53120), and spermidine synthase 1 and 2 (SPDS1- AT1G23820, SPDS2-AT1G70310) genes as baits in the NCBI<sup>1</sup> , Phytozome<sup>2</sup> and OneKP<sup>3</sup> databases. All the sequences were managed using the Benchling tool<sup>4</sup> . Alignment of the sequences was done with using the MUSCLE algorithm (Siebers et al., 2017) included in the SeaView 4.6.4 GUI (Gouy et al., 2010), with 16 iterations, default clustering methods, gap open score

<sup>1</sup>http://www.ncbi.nlm.nih.gov

<sup>2</sup>http://www.phytozome.net

<sup>3</sup>http://db.cngb.org/onekp/

<sup>4</sup>https://benchling.com

of –2.7, and hydrophobicity multiplier of 1.2, followed by manual curation. To select the best-fit model of amino acid substitution, the ProtTest v3.4.2 (Darriba et al., 2011) was used on final multiple sequence alignment, together with AIC model for ranking. Maximum likelihood tree was produced with PhyML v3.1 (Guindon et al., 2010), using the best-scored model of amino acid substitution. Statistical significance was evaluated by bootstrap analysis of 1000 replicates. Phylogenetic tree graphical representations were initially generated using FigTree (version 1.4.3) software<sup>5</sup> , and final cartoons edited manually.

#### RNA Extraction and PCR Analysis

Total RNA was isolated from 200 mg of frozen powdered tissues using RNeasy plant mini kit (Qiagen) for A. thaliana and M. polymorpha tissues following the manufacturer's instructions. RNA from C. reinhardtii, P. patens, S. lepidophylla and P. abies was extracted by Trizol-Chloroform method as described (Siebers et al., 2017). cDNA synthesis was performed on 1 µg DNasetreated RNA using PrimeScriptTM 1st strand cDNA synthesis kit (Takara). The resulting cDNA was used for PCR analysis with target gene-specific primers (**Supplementary Table S2**).

#### Cloning and Expression in Yeast

Coding sequence regions of thermospermine synthase genes from the selected species were obtained either from cDNA (M. polymorpha, S. lepidophylla and A. thaliana) or synthesized (Integrated DNA Technologies, IDT). The primers used for amplification are summarized in **Supplementary Table S2**. For plasmid construction, the CDS of the genes was cloned into pDONR207 through Gateway recombination (Invitrogen) and eventually into the destination vector pAG426GPD-ccdb-HA (a gift from Susan Lindquist; Addgene plasmid number 14252) for expression in yeast. The A. thaliana thermospermine synthase CDS was cloned into pCR8 through Golden Braid technology (Sarrion-Perdigones et al., 2011) and then shifted to Gateway technology to clone the CDS into the pDEST like the other species.

Yeast strain BY4741 (MATa his3 leu2 met15 ura3), kindly provided by Ramón Serrano's lab (IBMCP-Valencia, Spain), was grown on Synthetic Defined (SD) medium and transformed with the pAG426GPD containing the CDS of interest by LiAc/ssDNA/PEG. Selected transformants were grown on 50 mL liquid SD medium lacking Trp and Ura for 2 days, then harvested by centrifugation (about 200 mg of pellet) and frozen for further polyamine quantification.

### Polyamine Quantification

Polyamine extraction from tissues and standards was done with a modification of a previously described method (Naka et al., 2010) as follows. About 1 g of plant samples in 3 replicates were frozen in liquid nitrogen and kept at –80◦C until use. The tissue was ground with mortar and pestle (in the case of yeast pellets, the cells were broken with glass balls) and resuspended in 2.5 mL of 5% perchloric acid (PCA) in a 15-mL tube. At this point 500 µL of the internal standard (diethylamine 1 mM, DEA) was added to the 5% PCA. The samples were kept on ice for 1 h and centrifuged at 4◦C for 20 min at 15000×g. The whole supernatant was collected (between 2 and 3.5 mL) and transferred to a fresh 15 mL falcon tube. For each mL of supernatant, 0.66 mL of 2 M NaOH was added. Then, benzoylation started with the addition of 10 µL of benzoyl chloride. After 1 min vortex, plant extracts were left at room temperature for 20 min. Then, for each mL of initial supernatant used, 1.33 mL of saturated NaCl solution was added. Two mL of diethyl ether was added, vortexed, and centrifuged at 3000×g for 1 min for the separation of the phases. The supernatant was transferred to a new pyrex vial and dried completely using N2. The remaining polyamines were resuspended in 130 µL methanol and filtered with a filter syringe (pore size 0.2 µm). Then, the filtrate was transferred to a plastic vial for HPLC analysis. Briefly, 30-µL aliquots were injected through a Waters 717plus autosampler into a 1525 Waters Binary HPLC pump equipped with a 996 Waters PDA detector and using a Luna C18(2) (Phenomenex) column (250 × 4.6 mm, i.d. 5 µm). The column was equilibrated with 58% solvent A (acidic H2O containing 10 mL acetic acid for each liter of distilled water) and 42% solvent B (acetonitrile). Elution was carried out at room temperature and for polyamine separation a 1 mL min−<sup>1</sup> flow rate was used the isocratic gradient of 42% acetonitrile for 25 min. Then, the column was washed with 42–100% acetonitrile within 3 min and kept at 100% acetonitrile for 10 min. Eventually, the column was equilibrated with 42% acetonitrile for 17 min before the next injection. Detection of polyamines was performed at 254 nm.

#### Expression Analysis

Data for TSPMS gene expression in A. thaliana, P. trichocarpa, M. truncatula, S. lycopersicum, Symphytum tuberosum, O. sativa, G. max, V. vinifera, B. distachyon, E. grandis, P. abies, Ananas comosus and P. patens in different tissues was gathered from the Bio-Analytic Resource for Plant Biology (BAR<sup>6</sup> ). Expression data was classified according to the organ tissue (stem, root, flower and leaf) and tissue age (mature or young). The total expression data per gene and species in the target organs was used to calculate the percentage of expression of the TSPMS gene in each organ. Heatmaps were drawn using the Matrix2png tool<sup>7</sup> .

#### RESULTS

### Identification of Thermospermine Synthase Genes Across Plant Lineages

Polyamine biosynthesis enzymes display high degree of similarity (Hashimoto et al., 1998; Panicot et al., 2002; Teuber et al., 2007). Therefore, to guarantee accurate identification of putative ACL5 orthologs, we screened the Phytozome and OneKP databases using as baits the Arabidopsis ACL5 sequence, and also the

<sup>5</sup>http://tree.bio.ed.ac.uk/software/figtree/

<sup>6</sup>http://bar.toronto.ca

<sup>7</sup>https://matrix2png.msl.ubc.ca/

FIGURE 1 | Phylogenetic analysis of polyamine aminopropyl transferases. (A) Phylogenetic analysis of polyamine aminopropyl transferases protein sequences with special focus in plant sequences. Support values associated with branches are maximum likelihood bootstrap values from 1000 replicates depicted as a color range (gray scale). Black branches indicate a bootstrap of 1 (100% support). Blue background indicates TSPMS clade sequences and light brown SPDS-related sequences. The remaining sequences belong to metazoan SPMS used as outgroups to root the tree. Purple and dark brown indicate SPMS sequences from Spermatophyta and PMT sequences from Solanaceae/Convolvulaceae, respectively. Green ribbons mark Viridiplantae sequences. Gray shaded sequences with an asterisk indicate basal branches of the TSPMS clade. (B) Phylogenetic tree of TSPMS basal branches extracted from panel (A), marked with an asterisk. Support values associated with branches and displayed as bar thickness are maximum likelihood bootstrap values from 1000 replicates. Values under 0.75 have been merged to ease visualization. Names in brackets indicate experimentally tested forms of aminopropyl transferases: TSPMS, thermospermine synthase; TAAPT, triamine/agmatine aminopropyl transferase; PAAPT, polyamine aminopropyl transferase. Viridiplantae sequences have been collapsed. (C) Phylogenetic analysis of Viridiplantae TSPMS from tree (A). Support values associated with branches and displayed as bar thickness are maximum likelihood bootstrap values from 1000 replicates. Background colors indicate sequences from major lineages of plants. Red colored branches indicate chosen sequences for experimental analysis. Scale bars in panels (A–C) indicate substitution per residue distances.

sequences of the two Arabidopsis genes encoding SPDS, and the single gene for SPMS (see Section "Materials and Methods" for details). We obtained 542 sequences covering every major plant lineages and included members of representative bacteria, archaea and other non-plant eukaryotic groups, which were aligned and used for the construction of a phylogenetic tree (**Figure 1A** and **Supplementary Files S1–S3**). This is the most extensive phylogenetic study of plant polyamine aminopropyl transferases done to date, and it not only confirms some of the previous evolutionary models (Hashimoto et al., 1998; Panicot et al., 2002; Teuber et al., 2007; Minguet et al., 2008), but also provides new details on certain key events.

First, our analysis expands the presence of ACL5 homologs and SPDS genes to all plant lineages not previously studied, like charophytes and chlorophytes (**Figure 1A**). Second, the exclusive origin of SPMS genes in angiosperms, previously proposed based on only a few sequences (Minguet et al., 2008), is now set in the common ancestor of all Spermatophytes with more sequences from gymnosperms and their absence in multiple genomes of other land plants. Third, we find indications for a possible transfer of an additional SPDS of Holomycota origin to Setaphyta, based on the presence of extra copies in the genomes of the bryophyte Sphagnum and of several liverworts that unequivocally cluster with some sequences of that class

(**Supplementary Files S1–S3**). Fourth, as proposed in previous analyses (Minguet et al., 2008; Takano et al., 2012), our more extensive search still identifies ACL5 orthologs only in plants, red algae, diatoms, archaea and bacteria (**Figures 1A,B**). The new data are compatible with the proposed model that TSPMS activity in Archaeplastida has a prokaryotic origin, while the common root with the Archaea Sulfolobus and Pyrobaculum, of two additional copies of ACL5 orthologs in Thalassiosira pseudonana suggests a possible second event of horizontal gene transfer to Chromista (**Figure 1B**). And fourth, there are indications of an early ACL5 duplication in Spermatophyta, followed by several species-specific losses of one of the copies within the angiosperms (some Brassicaceae like Arabidopsis thaliana, some Poaceae) (**Figure 1C**).

### Thermospermine Is Synthesized Across All Plant Lineages

Our phylogenetic analyses identified at least one putative ACL5 ortholog in all studied non-vascular plant species (**Figure 1C**). Since the only confirmed function for Tspm in plants is the regulation of xylem maturation dynamics (which, by definition, does not occur in non-vascular plants), we wondered whether the identified sequences encode enzymes displaying actual TSPMS activity. Therefore, we expressed the ACL5 homologs of representative plant lineages in yeast, an organism that is unable to produce Tspm. Among the non-vascular plants, we tested ACL5 homologs of a chlorophyte (Chlamydomonas reinhardtii), a liverwort (Marchantia polymorpha) and a moss (Physcomitrella patens (**Supplementary Table S1**). As representative vascular plants we chose one lycophyte (Selaginella lepidophylla), and one gymnosperm (Picea abies), with the angiosperm A. thaliana as a control (**Supplementary Table S1**). As previously shown, HPLC analysis showed the presence of Put, Spd and Spm, but not Tspm, in the extracts of a wild-type yeast strain transformed with an empty plasmid (**Figure 2A**). In contrast, all the tested ACL5 homologs allowed the production of Tspm in yeast (**Figures 2B–G**) demonstrating that these genes encode enzymes with TSPMS activity. This is in accordance with the previously reported partial complementation of the Arabidopsis acl5 mutant by one of the PpACL5 orthologs (Takahashi and Kakehi, 2010). The observed differences in Tspm production between the different ACL5 genes suggest either species-specific variation in TSPMS activity kinetic parameters or in the variable capacity of yeast to express the different heterologous genes.

To check whether the presence of ACL5 orthologs with TSPMS activity correlated with the ability of these species to synthesize Tspm in vivo, we examined the polyamine levels in samples of these plants grown in standard conditions (see Section "Materials and Methods"). As expected, all vascular and non-vascular species accumulated Put and Spd to different levels (**Table 1**). For instance, Put levels in the chlorophyte C. reinhardtii were between one and three orders of magnitude higher than in the land plants examined, while Spd levels were generally higher in the land plants than in C. reinhardtii, except for P. patens. On the other hand, Spm was detectable in A. thaliana and P. abies, in agreement with previous reports for the occurrence of this tetraamine in seed plants (Gonzalez et al., 2011; Vuosku et al., 2018) and the presence of SPMS orthologs in this clade (**Figure 1A**). The detection of Spm in S. lepidophylla despite the absence of SPMS orthologs in lycophytes indicates that this tetra-amine might be synthesized by a less strict SPDS, as already suggested for P. sylvestris (Vuosku et al., 2018). With respect to Tspm, all the species tested were able to synthesize this polyamine at different levels, varying between 1.13 nM/g (fresh weight) in C. reinhardtii and 154 nM/g in P. abies. It is worth noting that, while previous reports show that Tspm tends to accumulate less than Spm in Arabidopsis (Naka et al., 2010; Rambla et al., 2010; Cai et al., 2016; Yoshimoto et al., 2016), in the species, tissues and conditions analyzed here, the two polyamines showed comparable levels. Although this result does not demonstrate that the ACL5 orthologs identified in all the selected species are responsible for the Tspm synthesis shown here, this possibility is further supported by the observation that these genes are actually expressed in the same growth conditions as the ones used for Tspm quantification (**Figure 3**).

### Differential ACL5 Expression Dynamics in Vegetative and Reproductive Organs Throughout Development in Seed Plants

Tspm synthase activity has been associated to xylem development in A. thaliana (Muniz et al., 2008; Vera-Sirera et al., 2015), Populus trichocarpa (Milhinhos et al., 2013) and P. sylvestris (Vuosku et al., 2018). To investigate whether this association can be extended to other vascular plants and to point to additional potential roles in these species, we took advantage of the vast amount of publicly available transcriptomic data and examined the expression pattern of the ACL5 orthologs in vegetative and reproductive organs of several vascular species at different developmental stages.

Although for some species we did not find data for stem or root, the available transcriptomic data shows that, for all studied species, ACL5 transcripts tend to accumulate more in vegetative than in reproductive organs (**Figure 4A**). In the species for which transcriptomic data exists for both stem and root, ACL5 expression is mostly higher in stems than in roots (7 out of 13 of the ACL5 homologs), while only in two cases is the opposite behavior observed. In Medicago truncatula, where two ACL5 paralogs exist, we found that one of them is more prominently accumulated in the stem, while the other one accumulates more in the root. Since our analyses are based on percentage of transcript accumulation across organs, in the species for which no transcriptomic data is available for stem tissue (i.e., Solanum lycopersicum), the proportion of transcript accumulation appears to be highly enriched in roots. For the same reason, in general, we regard the cases in which we found strong ACL5 accumulation in flowers to the non-availability of expression data for stem or root (i.e., Oryza sativa or Glycine max ACL5#4). Only in the case of Vitis vinifera (where transcriptomic data was available for all tested organs) we found that one of the paralogs showed a (slight) preferential expression in flowers than in vegetative organs. In P. trichocarpa, where three paralogs exist, we found

FIGURE 2 | Functional analysis of ACL5 orthologs. Saccharomyces cerevisiae BY4741 strain, unable to produce Tspm (A), was transformed with the putative ACL5 from P. abies (B), A. thaliana (C), P. patens (D), S. lepidophylla ACL5.1 (E), C. reinhardtii (F) and M. polymorpha (G), and the ability to produce Tspm was tested with an HPLC analysis. Graphs show the polyamine profile of yeast extracts after benzoylation and fluorescence detection with the help of a fluorimeter. Tspm, thermospermine; AU, absorbance units.

TABLE 1 | Polyamine quantification in plant tissues.


Polyamine quantification from A. thaliana, P. abies, S. lepidophylla, P. patens, M. polymorpha, and C. reinhardttii tissues. Polyamines were extracted and benzoylated from fresh tissues and measured with an HPLC analysis. Polyamine measurements were done in triplicates. FW, Fresh Weight; SE, Standard Error; n.d., not detected.

FIGURE 3 | Detection of the expression of ACL5 homologs in several plant species by PCR. (A) Schemes of the genes selected for this analysis. Exons are represented as boxes, and arrowheads indicate the primer pairs used for PCR detection, which are located on different exons, except for MpACL5, in which one of them spans an exon-exon junction. (B) PCR products amplified from cDNA. Plant material for RNA extraction and later cDNA synthesis was grown in the same standard conditions than the plants used for polyamine quantification by HPLC analysis.

that none of them display preferential expression for any of the analyzed organs.

We also observed a marked preference of ACL5 expression for young vs. mature organs (**Figure 4B**). In A. thaliana, P. trichocarpa, O. sativa and V. vinifera, this was visible in both flowers and leaves. There was no preferential expression between leaves and flowers except for the case of V. vinifera, in which both paralogs displayed higher preference for flowers than for leaves. For S. lycopersicum, we only found transcriptomes for young and mature flowers, while in Brachypodium distachyon and Eucalyptus grandis, we only found transcriptomes for young and mature leaves, and in all of them we found enriched ACL5 expression in young organs. These results indicate a fairly widespread expression of ACL5 homologs across organs, and mostly in young tissues with active vasculature development.

### DISCUSSION

Our current knowledge about thermospermine is that its only confirmed function in plants is the coordination of xylem maturation. However, previous phylogenetic work reported the existence in the genomes of non-vascular plants of sequences clustering with TSPMS (Minguet et al., 2008;

gene per species. (A) Expression data of TSPMS in A. thaliana, P. trichocarpa, M. truncatula, S. lycopersicum, S. tuberosum, G. max, V. vinifera, E. grandis, A. comosus, O. sativa, B. distachyon and P. abies in stem, root, flower and leaf. (B) Expression data of TSPMS in from A. thaliana, P. trichocarpa, S. lycopersicum, O. sativa, V. vinifera, B. distachyon and E. grandis in different developmental stages of leaf and flower.

lineage, followed by an additional loss in Glaucophyta. Putative origin of TSPMS genes is marked with an asterisk, and predicted loss of TSPMS genes is marked with a 1.

Pegg and Michael, 2010), implying that Tspm might exist and, perhaps, play developmental and/or stress responserelated roles also in such species. On one hand, as mentioned above, sequence similarity between all polyamine biosynthesis enzymes – especially those using putrescine or spermidine as substrates- has been extensively reported (Hashimoto et al., 1998; Panicot et al., 2002; Teuber et al., 2007). On the other hand, the sequencing of large numbers of genomes and transcriptomes of non-vascular plant species is relatively recent, and therefore previously published comparative analyses missed key clades in plant phylogeny.

It has been proposed that TSPMS had appeared in plants by endosymbiosis of a cyanobacterium (Minguet et al., 2008). Our data are compatible with three alternative hypotheses in this respect (**Figure 5**). The most parsimonious one is that TSPMS was already present before Eubacteria and Archaea and was lost in the major eukaryotic branch (**Figure 5C**). A second loss would have occurred during the divergence of Glaucophyta, explaining the absence in such branch and the presence in the Chromista, Rhodophyta and Viridiplantae branches. A less parsimonious, but plausible possibility is that TSPMS was lost in the last eukaryotic common ancestor (LECA) and that it then was transferred either from Eubacteria or from Archaea (i) to Archaeplastida after Glaucophyta had diverged and (ii) to Chromista (**Figure 5B**). A third, less parsimonious hypothesis would postulate that TSPMS was lost in LECA and transferred either from Eubacteria or from Archaea to the early lineage of Archaeplastida and to Chromista (**Figure 5A**). TSPMS would have been lost again in Glaucophyta. However, phylogenetic positioning of TSPMS points to the horizontal transfer of TSPMS genes from Rhodophyta to diatoms (and possibly other algal groups originated during secondary chloroplast acquisition), thus excluding this latter hypothesis, which is also based in a very low sampling among Chromista representatives. Given that it is almost practically impossible to differentiate between the two remaining models, the hypothesis that contemplates only one loss and two reported horizontal transfers seems the most likely one (**Figure 5B**).

The identification of TSPMS activity in non-vascular plants clearly argues for possible functions of Tspm other than in xylem development. It has been proposed that Tspm might play a role in defense against stress conditions (Gonzalez et al., 2011; Marina et al., 2013), so it could also be the case that such activity is conserved across vascular and nonvascular plants and that developmental functions – either related with vascular development or with other, yet to be identified, processes – were only acquired in vascular plant lineages. In any case, further experimentation using mutant, overexpression and marker lines in emerging non-vascular model plant species such as M. polymorpha or P. patens might provide further information about what developmental and/or stress related processes might be controlled by this polyamine.

The in silico analyses of relative abundance of ACL5 transcripts in different organs in representative seed-plant species presented here aimed at carrying out a first approach toward understanding whether, apart from coordinating xylem maturation, TSPM plays any other role in vascular plants. The higher prevalence of ACL5 transcripts in vegetative than in reproductive organs and in young (presumably developing) than in mature organs argues for higher association of ACL5 expression in organs developing more proportion of vascular tissues (vegetative) at early developmental stages, when vasculature is probably developing. However, the presence of ACL5 transcripts in reproductive organs and in mature vegetative organs also argues for potential other (maybe less prominent) activities of TSPM in plants.

In summary, our work shows that Tspm exists in non-vascular plants and that there is a correlation between Tspm presence, TSPMS activity and ACL5 expression throughout vascular and non-vascular plant lineages. Our results indicate that Tspm might play developmental and/or stress-related roles (apart from the described role in xylem development) not only in non-vascular but also in vascular plants. Future experimentation will shed new light on such intriguing topic.

#### DATA AVAILABILITY

fpls-10-00663 June 7, 2019 Time: 18:29 # 9

All datasets generated for this study are included in the manuscript and/or the **Supplementary Files**.

### AUTHOR CONTRIBUTIONS

AS-G, JH-G, MB, and JA conceived and designed the work. AS-G, JH-G, and ML-G performed all in silico and experimental analyses. JA and MB wrote the first draft of the manuscript, to which all authors contributed.

#### FUNDING

This work in the laboratories was funded by grants BFU2016- 80621-P and BIO2016-79147-R of the Spanish Ministry of Economy, Industry and Competitiveness. AS-G and JH-G are recipients of Fellowships of the Spanish Ministry of Science, Innovation and Universities BES-2017-080387 and

#### REFERENCES


of the Spanish Ministry of Education, Culture and Sport FPU15/01756, respectively. JA holds a Ramón y Cajal Contract RYC-2014-15752.

#### ACKNOWLEDGMENTS

We thank the members of the Hormone Signaling and Plasticity Lab at IBMCP (http://plasticity.ibmcp.csic.es/) for useful discussions and suggestions. We acknowledge Federico Valverde (IBVF-Sevilla, Spain), Jesús Vicente Carbajosa (CBGP-Madrid, Spain) and Ramón Serrano (IBMCP-Valencia, Spain) for sharing C. reinhardttii, P. patens and yeast materials, respectively.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00663/ full#supplementary-material



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Solé-Gil, Hernández-García, López-Gresa, Blázquez and Agustí. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Complexity and Conservation of Thermospermine-Responsive uORFs of SAC51 Family Genes in Angiosperms

Soichi Ishitsuka, Mai Yamamoto, Minaho Miyamoto, Yoshitaka Kuwashiro, Akihiro Imai† , Hiroyasu Motose and Taku Takahashi\*

Division of Earth, Life and Molecular Sciences, Graduate School of Natural Science and Technology, Okayama University, Okayama, Japan

#### Edited by:

Rubén Alcázar, University of Barcelona, Spain

#### Reviewed by:

Francisco Vera-Sirera, Spanish National Research Council (CSIC), Spain Jaana Marketta Vuosku, University of Oulu, Finland

> \*Correspondence: Taku Takahashi perfect@cc.okayama-u.ac.jp

†Present address: Akihiro Imai, Faculty of Life Sciences, Hiroshima Institute of Technology, Hiroshima, Japan

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 05 February 2019 Accepted: 15 April 2019 Published: 01 May 2019

#### Citation:

Ishitsuka S, Yamamoto M, Miyamoto M, Kuwashiro Y, Imai A, Motose H and Takahashi T (2019) Complexity and Conservation of Thermospermine-Responsive uORFs of SAC51 Family Genes in Angiosperms. Front. Plant Sci. 10:564. doi: 10.3389/fpls.2019.00564 ACAULIS5 (ACL5) encodes thermospermine synthase in Arabidopsis and its loss-offunction mutant acl5 shows excess xylem differentiation and severe dwarfism. SAC51 encodes a basic helix-loop-helix (bHLH) protein and was identified from sac51-d, a dominant suppressor mutant of acl5, which restores the wild-type phenotype without thermospermine. The 5<sup>0</sup> leader of the SAC51 mRNA contains multiple upstream openreading frames (uORFs) and sac51-d has a premature stop codon in the fourth uORF. This uORF is conserved among SAC51 family genes in vascular plants. According to the GUS reporter assay, the SAC51 promoter was not responsive to thermospermine but the SAC51 5 0 leader fused to the constitutive 35S promoter enhanced the GUS activity in response to thermospermine. Disruption experiments of each start codon of the SAC51 uORFs revealed that uORF4 and uORF6 whose start codon corresponds to the second methionine codon of uORF4 had an inhibitory effect on the main ORF translation while the other four uORFs rather had a stimulatory effect. The response of the 5<sup>0</sup> leader to thermospermine was retained after disruption of each one of six start codons of these uORFs but abolished by mutating both uORF4 and uORF6 start codons, suggesting the importance of the C-terminal sequence shared by these uORFs in the action of thermospermine. We introduced GUS fusions with 5<sup>0</sup> leaders of SAC51 family genes from other angiosperm species into Arabidopsis and found that all 5<sup>0</sup> leaders responsive to thermospermine, so far examined, contained these two conserved, and overlapping uORFs.

Keywords: thermospermine, translational regulation, uORF, SAC51, ACL5, Arabidopsis thaliana

### INTRODUCTION

Thermospermine is a structural isomer of spermine and present in some bacteria and ubiquitously in plants (Minguet et al., 2008; Fuell et al., 2010; Takano et al., 2012). In Arabidopsis thaliana, the ACAULIS5 (ACL5) gene encodes thermospermine synthase (Knott et al., 2007; Kakehi et al., 2008) and is expressed exclusively in differentiating xylem vessels (Clay and Nelson, 2005). Its loss-of-function mutants show over-proliferation of xylem vessels and severe dwarfism

**254**

(Hanzawa et al., 1997). Expression of the ACL5 gene is upregulated by auxin but down-regulated by thermospermine while the excess xylem phenotype of the acl5 mutant is enhanced by synthetic and persistent auxin analogs but suppressed by thermospermine, indicating that both thermospermine synthesis and auxin-induced xylem differentiation are under negative feedback control by thermospermine (Yoshimoto et al., 2012). Except in Arabidopsis, however, the function of thermospermine in xylem development has only been investigated in few species including poplar (Milhinhos et al., 2013) and cotton (Mo et al., 2015). On the other hand, we have developed an artificial inhibitor of thermospermine synthase, named xylemin, which showed a potent inducing effect of ectopic xylem formation in tobacco leaves in the presence of a synthetic auxin analog (Yoshimoto et al., 2016). Thus, in view of biotechnological applications, the combinatorial use of these plant growth regulators might have a potential for the control of woody biomass.

To clarify the mode of action of thermospermine, previous studies have identified suppressor mutants named suppressor of acl5 (sac) that rescue the acl5 phenotype without thermospermine. The results revealed that SAC51 encodes a basic helix-loop-helix (bHLH) transcription factor (Imai et al., 2006) while SAC52, SAC53, and SAC56 encode ribosomal proteins, RPL10, RACK1, and RPL4, respectively (Imai et al., 2008; Kakehi et al., 2015). SAC51 contains six upstream open reading frames (uORFs) in the 5<sup>0</sup> leader region of the mRNA. A dominant allele, sac51-d, has a premature stop codon in the fourth uORF. uORFs are highly abundant in the genomes of angiosperms and the fourth uORF of SAC51 is one of the highly conserved uORFs among different plant species (Hayden and Jorgensen, 2007). Conserved uORFs are generally present in regulatory genes and have an inhibitory effect on the main ORF translation, which is often attenuated by ribosomal translation reinitiation (Tran et al., 2008; Jorgensen and Dorantes-Acosta, 2012; von Arnim et al., 2014). Thermospermine may specifically counteract this effect on the SAC51 mRNA and lead to translation of the main ORF. Dominant or semi-dominant alleles of SAC52, SAC53, and SAC56, have been suggested to enhance translation of the SAC51 main ORF instead of thermospermine (Kakehi et al., 2015). SAC51 in turn may be involved in repressing the expression of ACL5 and a subset of genes required for xylem differentiation. Another study identified point mutations of the conserved uORF in SACL1 and SACL3 of the same SAC51 family as suppressors of acl5 (Vera-Sirera et al., 2015). We also found that sac57-d has a point mutation in this conserved uORF of SACL3 (Cai et al., 2016).

upstream open-reading frames-dependent translation regulated by polyamines is well known for S-adenosylmethionine decarboxylase (AdoMetDC). Mammalian AdoMetDC mRNAs contain a conserved uORF encoding the hexapeptide MAGDIS (Ruan et al., 1996). Ribosomes synthesizing this peptide are stalled by high concentrations of polyamines and blocked to access to the main ORF encoding the enzyme that catalyzes the production of decarboxylated AdoMet, an aminopropyl group donor for the synthesis of polyamines. In Arabidopsis, the AdoMetDC1 mRNA contains two conserved uORFs that are overlapping in different frames and have also been shown to regulate the translation of the main ORF according to polyamine levels (Franceschetti et al., 2001; Hanfrey et al., 2005). In mammals and yeasts, high levels of polyamines cause +1 and −2 ribosomal frameshifting, respectively, during the translation of the ornithine decarboxylase (ODC) antizyme mRNA, which contains an extra nucleotide in the coding sequence, and lead to the synthesis of a functional protein that mediates degradation of ODC, a rate-limiting enzyme in polyamine biosynthesis (Matsufuji et al., 1996). This regulatory mechanism may not be conserved in plants (Ivanov and Atkins, 2007). On the other hand, the translational enhancement by thermospermine has been reported so far only for Arabidopsis SAC51 family members (Takahashi, 2018). Our studies have shown that, among the four members of the SAC51 family, SAC51 and SACL1 are responsive to thermospermine but SACL2, and SACL3 are not (Cai et al., 2016; Yamamoto and Takahashi, 2017). We were thus interested in determining how much the response mechanism to thermospermine is conserved across plant species. Here we carried out a more detailed study of the responsiveness to thermospermine of the 5<sup>0</sup> leader region of the SAC51 mRNA and also extended our analysis to that of SAC51 family genes in different angiosperm species.

### MATERIALS AND METHODS

#### Plant Material and Growth Conditions

Arabidopsis thaliana Col-0 accession was used as the wild-type plant. Seeds were surface-sterilized in bleach solution containing 0.01% (v/v) Triton X-100 for 3 min, washed 3 times with sterile water, and sown onto MS medium (Murashige and Skoog, 1962) containing 3% sucrose and 0.8% agar. Plants were grown under 16 h light and 8 h dark at 22◦C. For thermospermine treatment, seedlings were incubated for 24 h in liquid MS medium with 100 µM thermospermine-4HCl, which was purchased from Santa Cruz Biotechnology.

#### Genomic DNA Preparation

Arabidopsis genomic DNA was prepared as described previously (Imai et al., 2006). Genomic DNAs of broccoli (Brassica oleracea), soybean (Glycine max), poplar (Populus tremula × alba), and rice (Oryza sativa) were prepared from each seedling by using NucleoSpin Plant II kit (Macherey-Nagel) following the manufacturer's instruction.

### T-DNA Construction and Plant Transformation

For the SAC51 promoter-driven expression of the GUS reporter gene, a 990-bp genomic fragment upstream from the first exon of SAC51 was amplified by PCR with primers, SAC51 proFCl, and SAC51-proRBg (**Supplementary Table S1**), cloned into a pGEM-T easy vector (Promega), and then transferred as a ClaI-BglII fragment to ClaI-BamlHI sites of Ti-plasmid vector pBI101 (Clontech). The GUS gene construct fused to the SAC51 promoter with the first untranslated exon

and intron was similarly made using primers, SAC51-proFCl, and SAC51-ex2RBg (**Supplementary Table S1**). The GUS construct containing the SAC51 promoter and the whole 5<sup>0</sup> leader region has been described previously (Imai et al., 2006). The mutant versions were generated by PCR-based site-directed mutagenesis as follows. First-round PCR was performed in 10 cycles of two separate reactions using the wild-type construct as a template with a primer pair of SAC51-proFCl and SAC51-mXR or with a primer pair of SAC51-mXF and SAC51-5RBg (**Supplementary Table S1**). The PCR products were mixed and subsequently amplified by 10 cycles of PCR with a primer pair of SAC51-proFCl and SAC51-5RBg. The product from a second round of PCR was cloned into pGEM-T Easy and transferred as a ClaI-BglII fragment to pBI101. The no-uORF version was generated sequentially by introducing a point mutation in the start codon of each uORF.

To generate the CaMV 35S promoter-GUS construct containing a genomic or cDNA fragment of the 5<sup>0</sup> leader region of SAC51, the fragment was amplified from genomic DNA or reverse-transcribed cDNA by PCR with primers, SAC51-5FSp and SAC51-5RBg (**Supplementary Table S1**), cloned into pGEM-T easy, and then inserted as a SpeI- BglII fragment between the 35S promoter and the GUS coding region of pBI121 (Clontech). Point mutations were sequentially introduced into the start codon of each uORF as described above. The other 35S-5<sup>0</sup> -GUS constructs containing a 5<sup>0</sup> leader region of SAC51 homologs from different plant species were made in a similar way. All PCR was performed using Ex Taq DNA polymerase (Takara) according to the manufacturers protocol. PCR conditions were 50 cycles of 94◦C for 30 s, 55◦C for 30 s, and 72◦C for 90 s, unless otherwise stated. The primers used were shown in **Supplementary Table S1**.

Ti plasmid constructs were introduced into Agrobacterium tumefaciens C58C1 by electroporation (Mersereau et al., 1990). Arabidopsis plants were transformed using the floral dip method (Clough and Bent, 1998). Transgenic lines were selected on kanamycin and confirmed by PCR using PBI-Cl and GUS primers (**Supplementary Table S1**). At least five independent homozygous lines carrying single copy of the transgene were further selected based on the segregation ratio of kanamycinresistant plants in subsequent generations.

#### GUS Assays

Fluorometric assay of GUS activity was performed as described previously (Jefferson et al., 1987). The fluorescence was measured with an RF-5300PC spectrofluorophotometer (Shimadzu, Japan). Total protein content was measured by using the Bradford assay (BioRad). For histochemical staining of GUS activity, samples were prefixed for 20 min in ice-cold 90% (v/v) acetone under vacuum, rinsed three times with water, and incubated in GUS staining buffer containing 50 mM Na2HPO4/NaH2PO<sup>4</sup> (pH7.0), 2 mM K3Fe(CN)6, 2 mM K4Fe(CN)6, 0.1% Triton-X100, and 1 mM 5-bromo-4-chloro-3-indolyl glucuronide, at 37◦C overnight. Samples were then treated with 70% ethanol to remove chlorophyll.

### Quantitative RT-PCR Analysis

Total RNA was extracted from Arabidopsis seedlings by the SDSphenol method (Hanzawa et al., 1997) and reverse transcribed by using the PrimeScript II first strand cDNA Synthesis Kit (Takara). Quantitative RT-PCR was performed on the Thermal Cycler Dice TP760 (Takara) using the KAPA SYBR FAST Universal (KAPA Biosystems). ACTIN8 (ACT8) was used as a reference gene for normalization. Means of expression levels were calculated from three technical replicates. Primers used were ACT8-F (5<sup>0</sup> -GTGAG CCAGA TCTTC ATTCG TC-3<sup>0</sup> ) and ACT8-R (5<sup>0</sup> -TCTCT TGCTC GTAGT CGACA G-3<sup>0</sup> ) for ACT8 and SAC51-FF (5<sup>0</sup> -CATTC CTTTC TAAGA TACTA AAG-3<sup>0</sup> ) and GUS (**Supplementary Table S1**) for the SAC51 5 0 -fused GUS, respectively.

### In vitro Transcription and Translation

The GFP reporter gene was amplified by PCR from pEGFP (Clontech) using primers, GFP-ATG and GFP-3 (**Supplementary Table S1**), and cloned as a BamHI-XhoI fragment into pT7 Blue-2 (Promega) to generate a control plasmid, pSI020. For the SAC51 5 0 leader-GFP transcriptional fusion construct, an 860-bp cDNA fragment of the SAC51 5 0 leader region was amplified by PCR with primers, SAC51-5FBal, and SAC51-5RBal (**Supplementary Table S1**), and inserted into the BalI restriction site of pSI020 just upstream of the GFP coding sequence. The no-uORF version of the SAC51 5 0 leader was generated by PCR from that of the 35S-SAC51 5 0 -GUS T-DNA and similarly cloned into pSI020.

FIGURE 1 | The response of the SAC51 promoter and 5' leader regions to thermospermine. (A) Gene structure of SAC51 and the regions used for GUS fusion constructs. Bars are promoter regions and introns of SAC51. Exons are shown in white boxes in which colored and black areas represent uORFs and a main ORF, respectively. sac51-d contains a premature stop codon in uORF4. (B) GUS staining patterns in cotyledons carrying the GUS fusion with only the SAC51 promoter (upper panel) and that with the SAC51 promoter-5<sup>0</sup> leader (lower panel). (C) Relative GUS activity of GUS fusion constructs. Ten-day-old seedlings were treated with or without 100 µM thermospermine for 24 h before GUS assays. Results are from single representative transgenic lines for each construct and the GUS activity of the SAC51 promoter-5<sup>0</sup> leader-GUS fusion without thermospermine is set as 1. Error bars represent SD (n = 4). Different letters indicate statistically significant differences between means by two-way ANOVA with Tukey–Kramer multiple comparison test (P < 0.05).

These plasmids were digested with XbaI to generate linear DNA templates for transcription and transcribed using T7 RNA polymerase in the presence of Ribo m7G cap analog (Promega). The resulting capped RNAs were translated in wheat germ extract with Transcend biotinylated lysine-tRNA (Promega) in the presence or absence of thermospermine according to the manufacturer's instructions.

In vitro-translated proteins were subjected to SDS– polyacrylamide gel electrophoresis, transferred to PVDF membrane, and detected using Transcend non-radioactive translation detection systems (Promega). The gel images were visualized using a LAS-4000 mini luminescent imaging analyzer (Fujifilm).

#### Statistical Analysis

All statistical analyses were performed using the EZR software (Kanda, 2013), which is a graphical user interface for R (The R Foundation for Statistical Computing). Significance of differences between groups was estimated by one-way or two-way ANOVA with Tukey-Kramer multiple comparisons test.

### RESULTS

#### The SAC51 Promoter Is Not Responsive to Thermospermine

There are six uORFs in the 5<sup>0</sup> leader region of the SAC51 mRNA, including the sixth uORF, which presumptively starts from an in-frame start codon, namely, the second methionine codon of the fourth uORF (Yamamoto and Takahashi, 2017). These uORFs are encoded in the second and third exons of SAC51. To confirm whether more upstream regions are responsive to thermospermine or not, we generated transgenic lines carrying the GUS reporter gene under the control of the SAC51 promoter

FIGURE 2 | Effect of disruption of each start codon of SAC51 uORFs on the GUS activity. (A) The regions used for GUS fusion constructs and relative GUS activity of each construct. Assays were performed as shown in Figure 1C. Error bars represent SD (n = 4). Different letters indicate statistically significant differences between means by two-way ANOVA with Tukey–Kramer multiple comparison test (P < 0.05). (B) GUS staining patterns in cotyledons carrying the GUS fusion with the SAC51 promoter-5<sup>0</sup> leader of each mutant uORF version. (C) Relative ratio of the GUS activity to its mRNA level. The GUS activity and the GUS mRNA level were measured using 10-day-old seedlings of each transgenic line. Error bars represent SD (n = 4). Different letters indicate statistically significant differences between means by one-way ANOVA with Tukey–Kramer multiple comparison test (P < 0.05).

or that followed by the first exon and intron with its splice acceptor site (**Figure 1A**) and examined the GUS activity. The results showed that the SAC51 promoter directed the GUS expression sharply to vascular cells (**Figure 1B**) and neither only the promoter nor the promoter followed by the first exon and intron was responsive to thermospermine (**Figure 1C**). However, when the SAC51 upstream regions containing the whole 5<sup>0</sup> leader sequence was used, weak GUS expression was detected in additional tissues to the vasculature (**Figure 1B**) and the GUS activity was increased by 24-h treatment with thermospermine (**Figure 1C**), as described previously (Kakehi et al., 2008). The sac51-d mutant construct in which the fourth uORF contains a premature stop codon (**Figure 1A**) shows much higher GUS expression than the wild-type construct (Imai et al., 2006) but still retained the responsiveness to thermospermine (**Figure 1C**).

### Both uORF4 and uORF6 of SAC51 Are Responsive to Thermospermine

To address which uORFs of the SAC51 mRNA are involved in the response to thermospermine, we generated transgenic lines that carry the GUS fusion constructs with the SAC51 promoter and 5<sup>0</sup> leader containing a base substitution in the start codon of each uORF (**Figure 2A**) and examined the effect of thermospermine on the GUS activity. All constructs in which one of the uORF start codons is mutated were shown to retain the response to thermospermine. As also shown by GUS staining (**Figure 2B**), the results revealed that the basal GUS activity was rather reduced in uORF1, uORF2, uORF3, and uORF5-disruption constructs and increased in uORF4 and uORF6-disruption constructs compared with that in the wild-type construct, suggesting that uORF4 and uORF6 are inhibitory but uORF1, uORF2, uORF3, and uORF5 are stimulatory to the main ORF translation. We confirmed by RT-PCR experiments that the relative ratio of the GUS activity to the GUS transcript level was reduced in transgenic lines with uORF1, uORF2, uORF3, and uORF5-disruption constructs (**Figure 2C**). Disruption of uORF4 or uORF6 appeared to reduce the responsiveness to thermospermine. Our previous study suggests the importance of this uORF6 because the 5 0 leaders of both SAC51 and SACL1 contain this in-frame uORF and are responsive to thermospermine but those of SACL2 and SACL3 don't contain it and show no response to thermospermine (Yamamoto and Takahashi, 2017). We thus disrupted start codons of both uORF4 and uORF6 and found that this mutant construct resulted in no response to thermospermine (**Figure 2A**). Disruption of all of six start codons of the uORFs also caused no response to thermospermine but the construct with the loss of all but the uORF6 start codon showed the response (**Figure 2A**). These results collectively suggest the requirement for at least one of uORF4 and uORF6 in the response to thermospermine.

We also examined the response of the SAC51 5 0 leader region to thermospermine under the control of the constitutive cauliflower mosaic virus (CaMV) 35S promoter. Although the 35S promoter is not responsive to thermospermine, insertion of a genomic or cDNA fragment of the SAC51 5 0 leader region between the promoter and the GUS reporter gene conferred the response to thermospermine in the GUS activity (**Figure 3**). Furthermore, the mutant cDNA construct in which all but uORF6 is disrupted was responsive to thermospermine whereas the 5<sup>0</sup> fragment containing the minimum uORF6 alone resulted in no response (**Figure 3**).

#### The Response to Thermospermine Is Conserved in Dicots and Monocots

Because the uORF4 of SAC51 is widely conserved in SAC51 family genes of different plant species, we examined whether 5<sup>0</sup> leader regions of these mRNAs are responsive to thermospermine or not. The 5<sup>0</sup> regions were cloned from broccoli, soybean, poplar, and rice genomic DNA, inserted between the 35S promoter and the GUS gene, and introduced into Arabidopsis. The deduced amino acid sequences of the conserved uORFs, some of which contain in-frame ATG codons, namely, in-frame uORFs, are aligned in **Figure 4A**, and their phylogenetic relationships are shown in **Figure 4B**. We detected the response to thermospermine in some constructs including

FIGURE 4 | Comparison of the conserved uORFs of SAC51 family genes. (A) Alignment of amino acid sequences deduced from the conserved uORFs of SAC51 family genes from different plant species. The amino acids encoded by the internal uORF corresponding to uORF6 of AtSAC51 are colored in red and those encoded by another uORF in SACL3 homologs are shaded. Dashes indicate gaps. At, Arabidopsis thaliana; Bo, Brassica oleracea; Os, Oryza sativa; Pt, Populus trichocarpa; Gm, Glycine max. Gene ID or GenBank accession numbers are given in parentheses. (B) Phylogenetic relationship of the conserved uORFs shown in (A). The tree based on the deduced amino acid sequences was constructed using the neighbor-joining method of the MEGA7 software (Kumar et al., 2016). The scale bar indicates the number of amino acid substitutions per site. (C) The response of each 5<sup>0</sup> leader region fused with the CaMV 35S promoter to thermospermine. GUS assays were performed as shown in Figure 1C. Error bars represent SD (n = 4). Asterisks indicate a significant increase as compared with control (t-test, <sup>∗</sup>P < 0.05, ∗∗P < 0.01).

FIGURE 5 | In vitro translation of the SAC51 5 0 leader-GFP fusion transcript using a wheat-germ extract. (A) Structure of the three constructs used. The constructs of GFP alone (I), GFP fused with the SAC51 5 0 leader (II), and GFP fused with the SAC51 5 0 leader in which all start codons of uORFs are mutated (III), were transcribed in vitro using T7 RNA polymerase. (B) Effect of the amount of substrate transcripts on GFP synthesis. (C) Effect of thermospermine on GFP synthesis. 0.4 µg each of the in vitro transcript from the construct II was used in the translation reaction. (D) Effect of pretreatment by heat of the transcripts on GFP synthesis. 0.4 µg each of the in-vitro transcripts was treated at 25◦C (Ct) or 65◦C (Hs) for 10 min before translation.

BoSACL1, OsSACL3A, OsSACL3C, and GmSACL3, but not in others including OsSACL2, PtSACL2, GmSACL2, and BoSACL3 (**Figure 4C**).

### The Inhibitory Effect of the 5<sup>0</sup> Leader on Translation Is Repressed in vitro by Heat

We finally tested whether or not the response of the SAC51 transcript to thermospermine can be reproduced in vitro. The cDNA fragment of the SAC51 5 0 leader region was fused to the GFP reporter gene (**Figure 5A**), transcribed in vitro, and then translated by using a wheat germ extract translation system. Detection of the chemiluminescent-labeled GFP protein showed that the efficiency of GFP translation was reduced by adding the SAC51 5 0 leader sequence of both wild-type and no-uORF versions to the GFP transcript and further reduced by increasing the amount of the transcript in the translation reaction (**Figure 5B**). Addition of thermospermine to the translation reaction mixture had no effect on the GFP production (**Figure 5C**). We found, however, that pretreatment of the 5<sup>0</sup> - GFP fusion transcript of both wild-type and no-uORF versions with heat at 65◦C for 10 min effectively increased the translation efficiency (**Figure 5D**).

### DISCUSSION

The results of GUS expression experiments, first of all, confirmed that the SAC51 promoter, the first exon, which contains no coding sequence, and the first intron are not

responsive to thermospermine. On the other hand, the fulllength 5<sup>0</sup> leader region solely conferred the responsiveness to thermospermine on its fused mRNA under a certain promoter. We have previously shown that the mRNA levels of SAC51 and SACL1 are increased by 24-h treatment with thermospermine (Kakehi et al., 2010; Cai et al., 2016). Because these mRNAs have also been identified as nonsense-mediated mRNA decay (NMD) targets (Kurihara et al., 2009), these increases may be caused by enhancement of the main ORF translation followed by the avoidance of NMD. We thus conclude that the response of SAC51 to thermospermine is predominantly regulated at the level of translation. A previous study of the SAC51 expression in root tissue has shown that it is broadly expressed within the vascular cylinder (Vera-Sirera et al., 2015). ACL5 expression is more restricted to differentiating xylem vessels (Katayama et al., 2015). Given that thermospermine is a mobile signal, its concentration gradient could potentially control spatiotemporal distribution of the SAC51 protein through enhancing its mRNA translation. The mechanism of cellto-cell transport of thermospermine remains, and needs to be addressed.

Disruption of the conserved uORF4 or its internal uORF6 increased basal levels of the GUS activity, suggesting their inhibitory role in the main ORF translation. In contrast, disruption of each single uORF of uORF1, uORF2, uORF3, and uORF5, which are not conserved among plant species, rather reduced basal levels of the GUS reporter activity. It is possible that these uORFs serve to lead the scanning ribosomes to bypass the conserved uORF4 and its internal uORF6 and reinitiate translation from the downstream main ORF to some extent. Disruption of both uORF4 and uORF6 completely abolished the response to thermospermine. Thus, these two uORFs may play a cooperative or redundant role in the regulation of the SAC51 main ORF translation, although at least uORF6 alone within the context of the full-length 5<sup>0</sup> leader region is sufficient for the response to thermospermine. Simultaneous disruption of these uORFs also reduced the basal level of the GUS activity. It is possible that point mutations in these uORFs alter the secondary structure of the 5<sup>0</sup> leader region of the transcript, thereby affecting the translation efficiency of the main ORF. The internal uORF6 within uORF4 is also present in SACL1 but not in SACL2 and SACL3. We have previously suggested the importance of this short uORF because SACL2 and SACL3 show no clear response to thermospermine (Yamamoto and Takahashi, 2017). However, it should be noted that the response to thermospermine was still detected in the case of the uORF6 disruption construct. More detailed studies such as the construction of synonymous substitutions of the conserved uORF6 will be necessary.

The response of SAC51 family genes to thermospermine has not been investigated before in other plant species than Arabidopsis. Our results showed that 5<sup>0</sup> leader regions of Brassica SACL1, soybean SACL3, rice SACL3A, and SACL3C, were responsive to thermospermine in transgenic Arabidopsis plants. All of these contain at least the two uORFs corresponding to the conserved uORF4 and uORF6 of the Arabidopsis SAC51, suggesting again the relevance of these two uORFs to the response to thermospermine. Although SACL3-homologous mRNAs tested contain another uORF in the conserved uORF (**Figure 4A**), it may not always relate to the response to thermospermine. Further analysis of SACL3-homologous mRNAs in different plant species will give a clue as to the arrangement of uORFs responsive to thermospermine. In Arabidopsis, TMO5-LHW and T5L1-LHW heterodimeric transcription factors have been shown to commonly regulate expression of ACL5 and SACL3 in xylem precursor cells in the root (Katayama et al., 2015). Thus, translational response to thermospermine might no longer be critical for SACL3 expression. On the other hand, SACL2 homologous mRNAs tested contain no additional uORF in the conserved uORF and showed no response to thermospermine. These results pose a question about the significance of the conserved uORF in the response to thermospermine.

Unfortunately, the response of the 5<sup>0</sup> leader region of SAC51 to thermospermine was not reproduced in the wheat germ in vitro translation system. It is possible that additional mediators are required for the function of thermospermine as is the case for other plant hormone signals, which are mediated by hormone receptor proteins. However, given the mode of action of polyamines in mRNA translation so far identified and their strong affinity to nucleic acids (Igarashi and Kashiwagi, 2015), it is most likely that thermospermine directly interacts with RNA molecules. Thermospermine might be required to integrate with ribosomal RNA during the formation of small and large ribosomal subunits although the possibility cannot be excluded that thermospermine interacts with a specific mRNA sequence such as the SAC51 5 0 leader region and alters its secondary structure to work as a probable thermospermine-responsive riboswitch. The in vitro transcribed SAC51 5 0 leader was not responsive to thermospermine but to heat treatment, suggesting the importance of the secondary structure of the 5<sup>0</sup> leader sequence in translation. Whether or not all the uORFs are intact, the long 5<sup>0</sup> leader of SAC51 reduced GFP translation as the amount of the transcript added to the reaction increased, suggesting the need for more ribosomes in the reaction. Further investigation of optimal reaction conditions may be required. It would also be interesting and worth to examine whether or not the translational enhancement by thermospermine can be reproduced and applied as a biotechnological tool in animal and fungal systems.

#### AUTHOR CONTRIBUTIONS

SI, YK, AI, HM, and TT designed the experiments. SI, MY, MM, and YK carried out the experiments and analyzed the data. SI and TT wrote the manuscript.

### FUNDING

This work was supported in part by the Japan Society for the Promotion of Science (JSPS) Grants-in-Aid for Scientific Research (Nos. 26113516, 16H01245, and 19K06724) to TT.

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00564/ full#supplementary-material



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Ishitsuka, Yamamoto, Miyamoto, Kuwashiro, Imai, Motose and Takahashi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Thermospermine Synthase (ACL5) and Diamine Oxidase (DAO) Expression Is Needed for Zygotic Embryogenesis and Vascular Development in Scots Pine

Jaana Vuosku1\*, Riina Muilu-Mäkelä<sup>2</sup> , Komlan Avia1† , Marko Suokas <sup>1</sup> , Johanna Kestilä<sup>1</sup> , Esa Läärä<sup>3</sup> , Hely Häggman<sup>1</sup> , Outi Savolainen<sup>1</sup> and Tytti Sarjala<sup>2</sup>

#### Edited by:

Ana Margarida Fortes, University of Lisbon, Portugal

#### Reviewed by:

Ana Milhinhos, New University of Lisbon, Portugal Fiammetta Alagna, ENEA - Centro Ricerche Trisaia, Italy

> \*Correspondence: Jaana Vuosku jaana.vuosku@oulu.fi

† Present address: Komlan Avia, Université de Strasbourg, INRA, SVQV UMR-A 1131, Colmar, France

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 15 March 2019 Accepted: 14 November 2019 Published: 20 December 2019

#### Citation:

Vuosku J, Muilu-Mäkelä R, Avia K, Suokas M, Kestilä J, Läärä E, Häggman H, Savolainen O and Sarjala T (2019) Thermospermine Synthase (ACL5) and Diamine Oxidase (DAO) Expression Is Needed for Zygotic Embryogenesis and Vascular Development in Scots Pine. Front. Plant Sci. 10:1600. doi: 10.3389/fpls.2019.01600 <sup>1</sup> Ecology and Genetics Research Unit, University of Oulu, Oulu, Finland, <sup>2</sup> Production Systems, Natural Resources Institute Finland, Espoo, Finland, <sup>3</sup> Research Unit of Mathematical Sciences, University of Oulu, Oulu, Finland

Unlike in flowering plants, the detailed roles of the enzymes in the polyamine (PA) pathway in conifers are poorly known. We explored the sequence conservation of the PA biosynthetic genes and diamine oxidase (DAO) in conifers and flowering plants to reveal the potential functional diversification of the enzymes between the plant lineages. The expression of the genes showing different selective constraints was studied in Scots pine zygotic embryogenesis and early seedling development. We found that the arginine decarboxylase pathway is strongly preferred in putrescine production in the Scots pine as well as generally in conifers and that the reduced use of ornithine decarboxylase (ODC) has led to relaxed purifying selection in ODC genes. Thermospermine synthase (ACL5) genes evolve under strong purifying selection in conifers and the DAO gene is also highly conserved in pines. In developing Scots pine seeds, the expression of both ACL5 and DAO increased as embryogenesis proceeded. Strong ACL5 expression was present in the procambial cells of the embryo and in the megagametophyte cells destined to die via morphologically necrotic cell death. Thus, the high sequence conservation of ACL5 genes in conifers may indicate the necessity of ACL5 for both embryogenesis and vascular development. Moreover, the result suggests the involvement of ACL5 in morphologically necrotic cell death and supports the view of the genetic regulation of necrosis in Scots pine embryogenesis and in plant development. DAO transcripts were located close to the cell walls and between the walls of adjacent cells in Scots pine zygotic embryos and in the roots of young seedlings. We propose that DAO, in addition to the role in Put oxidation for providing H2O2 during the cell-wall structural processes, may also participate in cell-to-cell communication at the mRNA level. To conclude, our findings indicate that the PA pathway of Scots pines possesses several special functional characteristics which differ from those of flowering plants.

Keywords: arginine decarboxylase, developmental regulation, diamine oxidase, enzyme pathway evolution, polyamine, Scots pine, thermospermine synthase, zygotic embryogenesis

## INTRODUCTION

Polyamines (PAs), ancient small polycations, are found in all living organisms (Michael, 2016). The most common PAs in eukaryotic cells are putrescine (Put), spermidine (Spd), and spermine (Spm) (Tiburcio et al., 1997; Pegg, 2009). Other PAs include cadaverine (Cad) (Bagni and Tassoni, 2001) and thermospermine (T-Spm), a structural isomer of Spm, which is widely found in the plant kingdom (Knott et al., 2007; Takano et al., 2012). PAs show high affinity for polyanionic macromolecules, such as DNA, RNA, proteins, and phospholipids, and function in various fundamental cellular processes, such as DNA and protein syntheses, gene expression, cell division and elongation, differentiation, free radical scavenging, and programmed cell death (Ha et al., 1998; Childs et al., 2003; Seiler and Raul, 2005; Igarashi and Kashiwagi, 2010; Schuster and Bernhardt, 2011). In plants PAs are involved in numerous physiological events as well as different abiotic and biotic stress responses (Galston and Sawhney, 1990; Kumar et al., 1997; Bouchereau et al., 1999; Alcázar et al., 2006; Kusano et al., 2008) during which PA homeostasis is achieved by modulating PA biosynthesis, catabolism, conjugation, and transport (Tiburcio et al., 2014).

In plants PAs are biosynthesized and catabolized via branched enzymatic pathways (Supplementary Figure S1). Ornithine decarboxylase (ODC) produces Put directly from ornithine. In addition, plants generally possess an additional route for Put formation from arginine consisting of the enzymes arginine decarboxylase (ADC), agmatine iminohydrolase (AIH), and N-carbamoylputrescine amidohydrolase (CPA) (Tiburcio et al., 1997). Diamine Put is the immediate precursor of the tri- and tetra-amines Spd, Spm, and T-Spm which are synthesized by combined actions of S-adenosylmethionine decarboxylase (SAMDC) and the aminopropyltransferases spermidine synthase (SPDS), spermine synthase (SPMS), and thermospermine synthase (TSPMS) (Shao et al., 2012). Both SPDS and SPMS activities can also occur in the same bifunctional enzyme (Vuosku et al., 2018). Put is catabolized by the action of diamine oxidases (DAOs) belonging to the group of copper-containing diamine oxidases (CuAOs) and the higher PAs by the flavoprotein-containing PA oxidases (PAOs) (Bagni and Tassoni, 2001; Bais and Ravishankar, 2002; Marina et al., 2013) (Supplementary Figure S1).

Because Put and Spd are the only PAs produced in all PAsynthesizing eukaryotes, the extant core of the eukaryotic PA biosynthetic pathway in the last eukaryotic common ancestor might have consisted of the ODC and SPDS enzymes (Michael, 2016). In plants the complex evolutionary history of the PA biosynthesis pathway includes the transfers of the ADC pathway from the cyanobacterial ancestor of the chloroplast (Illingworth et al., 2003) and TSPMS encoding ACL5 genes from archaea or bacteria (Minguet et al., 2008). The duplications of SPDS genes have led to the evolution of separate SPDS and SPMS enzymes in flowering plants (Minguet et al., 2008), whereas a bifunctional progenitor enzyme possessing both SPDS and SPMS activity was preserved in an evolutionary old conifer (Vuosku et al., 2018). All plant species, including Scots pine (Pinus sylvestris L.) and the moss Physcomitrella patens (Hedw.) Bruch & Schimp possess sequences identified as SPDS or ACL5, but no genomic sequence like the one described for SPMS has been reported from gymnosperms or Physcomitrella so far (Minguet et al., 2008; Rodríguez-Kessler et al., 2010; Vuosku et al., 2018). Although the synthesis of T-Spm likely resembles the formation of Spm (Knott et al., 2007), the TSPMS and SPMS enzymes have different evolutionary origin—TSPMS being more ancient than SPMS in plants.

The importance of PAs in plant embryogenesis has been documented in both angiosperms and gymnosperms (Baron and Stasolla, 2008). PA biosynthetic knock-out mutants have indicated that ADC (Urano et al., 2005), SPDS (Imai et al., 2004) and SAMDC (Ge et al., 2006) are essential for embryo development in Arabidopsis (Arabidopsis thaliana L.). Both ADC and SPDS expressed in the mitotic cells of Scots pine zygotic embryos, which supports the essential roles of Put and Spd in basic cell functions (Vuosku et al., 2006; Vuosku et al.,2018). In contrast, the Arabidopsis acl5-1 spms-1 double mutant, which contains neither Spm nor T-Spm, is viable and shows no phenotypic change except for the reduced stem growth due to acl5-1 (Imai et al., 2004). In conifers PA contents and ratios followed developmental stage dependent profiles during zygotic embryogenesis (Minocha et al., 1999; Astarita et al., 2003; Vuosku et al., 2006; de Oliveira et al., 2017). Likewise, the triggering of somatic embryogenesis pathway modulated the PA metabolism at gene expression, enzyme activity, and metabolite levels (Minocha et al., 1999; Gemperlová et al., 2009; Jo et al., 2014; Salo et al., 2016). However, the reactions of embryogenic cell masses to exogenous PAs and PA biosynthesis inhibitors have proved to be complex and dependent on both the conifer species and the developmental stage of somatic embryos (Santanen and Simola, 1992; Minocha et al., 1993; Sarjala et al., 1997; Kong et al., 1998; Niemi et al., 2002; Steiner et al., 2007).

The extreme evolutionary conservation of PAs indicates their necessity in organism survival. On the other hand, different evolutionary processes as well as the high flexibility of PA metabolism in response to internal and environmental demands, especially in plants, implicate that PAs may have acquired a wide variety of different functions during evolution. Pinus species of the gymnosperms present an evolutionarily old group of vascular plants that last shared a common ancestor with angiosperms about 300 million years ago (Zhang et al., 2004). We hypothesized that the enzymes in the PA metabolic pathway have acquired different functions during their evolution after the divergence of the seed plant lineages. This hypothesis was supported by our previous findings showing that the Scots pine has a bifunctional SPDS also possessing SPMS activity, which is contrary to angiosperms, which rely on separate enzymes in Spd and Spm biosynthesis (Vuosku et al., 2018). In the present study we are looking for other specific functional characteristics possibly existing in the PA metabolism of Scots pine. Therefore, we compared the sequence conservation of the PA biosynthesis genes and DAO between angiosperms and gymnosperms for screening out dissimilarities that potentially indicate functional diversification of the enzymes (Fay and Wu, 2003). After that, the expression of the genes showing different selective constraints between the plant lineages was examined in Scots pine seeds and seedling tissues. Our findings reveal the preference of the ADC pathway in Put biosynthesis in conifers and emphasize the importance of ACL5 and DAO in development by suggesting novel roles in cellular functions for them. Furthermore, the results underline the strict developmental regulation of the PA metabolism in Scots pine.

### MATERIALS AND METHODS

### Plant Material

One-year-old immature seed cones were collected from two open-pollinated elite Scots pine (Pinus sylvestris L.) clones, K881 and K884, growing in the Scots pine clone collection in Punkaharju, Finland (61°48′ N; 29°17′ E). For both clones one representative graft was selected and used repeatedly for cone collection in different years of the PA research (e.g. Vuosku et al., 2006; Vuosku et al., 2012; Muilu-Mäkelä et al., 2015a; Salo et al., 2016). For the present study, cones were collected four times during the growing season of 2004, on July 5 (sampling date I), July 12 (sampling date II), July 19 (sampling date III), and July 26 (sampling date IV). The effective temperature sums (i.e. the heat sum unit based on the daily mean temperatures minus the adapted +5°C base temperature) were 436.3, 509.1, 587.4, and 678.9 d.d. on sampling dates I, II, III, and IV, respectively. In Scots pines the time of fertilization and, consequently, embryo development vary between years in the same locality according to the effective temperature sum (Sarvas, 1962; Vuosku et al., 2006). The sequence of embryo development is divided into three phases, which include proembryogeny, early embryogeny, and late embryogeny (Singh, 1978). Previously, Scots pine embryos were found to follow a developmental pattern in which a great majority of embryos were still at the early embryogeny stage when the effective temperature sum was between 600 and 700 d.d. (Vuosku et al., 2006), as was the case on sampling date IV in the present study. Immature seeds were removed from the developing cones and seed coats were removed. Each pooled sample contained about 20 seeds. Mature seed cones were collected from clone K884 in late autumn of the same growing season. Seeds were sterilized overnight in 3% Plant Preservative Mixture TM (Plant Cell Technology, USA) and germinated on moist filter papers in petri dishes (100 seeds/petri dish) in a growth chamber (Weiss Technik, type 266532/1/−SKS30058/+5…+45 3 µPa) at +20 °C, in 100% moisture and under continuous light. After 2 days of germination embryos and megagametophytes were sampled and, after 16 days of germination, the cotyledons, hypocotyls, and roots of the seedlings were excised. Samples (consisting of 20 embryos, 10 megagametophytes or needles, hypocotyls, or roots from tree seedlings) were stored in liquid nitrogen for RNA extraction. For the mRNA in situ hybridization assays of ACL5 and DAO transcripts immature seeds (without seed coat) and roots were fixed immediately after sampling as described previously in Vuosku et al. (2015).

Seeds collected from 24 open-pollinated (mostly halfsibs) families from three different populations (Kolari, Northern Finland: latitude 67°11′N, 24°03′E; Punkaharju, Southern Finland: latitude 61°48′N, 29°19′E and Radom, Poland: latitude 50°41′N, 20°05′E) were used for the sequencing of ADC and DAO genes.

#### Estimation of Nonsynonymous to Synonymous Substitution Rate Ratios in PA Genes

The nonsynonymous to synonymous substitution rate ratios (Ka/Ks) were used to infer the strength of purifying selection pressure (Hughes and Nei, 1988) on the genes in the PA biosynthesis pathway and the DAO genes. High selective constraints and strong purifying selection should result in low numbers of nonsynonymous substitutions (Ka) between species. As the mutation rate varies across the genome, the rate of neutral synonymous substitutions (Ks) will also vary. Thus, the ratio of Ka/Ks provides information on the level of constraint in proteincoding sequences across related species (Fay and Wu, 2003). For Ka/Ks calculations the PA gene sequences were retrieved from NCBI GenBank (Supplementary Table S1). Due to the different appearance of the ODC, SPMS, and ACL5 genes in plant species, the availability of sequence data, the reliability of sequence annotations and nucleotide substitution saturation between some sequences, the species used in pairwise Ka/Ks calculations varied slightly for different genes. Two pairs consisting of gymnosperm species and four pairs consisting of angiosperm species were used for all genes except SPMS. The Ka/ Ks ratios for the ADC, AIH, CPA, SAMDC, and SPDS genes were calculated between the following pairs: Scots pine (Pinus sylvestris L.)–white spruce (Picea glauca (Moench) Voss), Scots pine–loblolly pine (Pinus taeda L.), rice (Oryza sativa L.)– sorghum (Sorghum bicolor (L.) Moench), Arabidopsis–black cottonwood (Populus trichocarpa L.), Arabidopsis–vine (Vitis vinifera L.), and black cottonwood–vine. The pairs were the same for the DAO genes except that white spruce was replaced with Sitka spruce (Picea sitchensis (Bong.) Carr.). Ka/Ks ratios for the ACL5 genes were calculated between Scots pine–white spruce, Scots pine–loblolly pine, rice–sorghum, Arabidopsis–clementine (Citrus clementina Hort. ex Tan), Arabidopsis–apple (Malus domestica Borkh.), and clementine–apple. The set of the ODC sequence pairs included Scots pine–sitka spruce (Picea sitchensis (Bong.) Carr.), Scots pine–loblolly pine, rice–sorghum, tobacco (Nicotiana tabacum L.)–black cottonwood, black cottonwood– vine, and apple–vine. The five compared sequence pairs for SPMS were rice–sorghum, Arabidopsis–vine, Arabidopsis– clementine, Arabidopsis–apple, and clementine–apple. The Scots pine ACL5 (HM236828), ADC (HM236823), AIH (HM236824), CPA (HM236825), DAO (HM236829), ODC (HM236831), SAMDC (HM236826), and SPDS (KX761190) sequences were used in the analyses. In the case of other conifers, where no unigene sequences were available, EST information was acquired by BLAST searches against the Scots pine PA genes and used to reconstruct a contig containing the complete coding sequence. The sequences were aligned based on codon boundaries i.e. they were translated to amino acid sequences, aligned, and then transformed back to DNA. Synonymous and non-synonymous changes were calculated with MEGA 5.05 using the Pamilo-Bianchi method (Pamilo and Bianchi, 1993).

#### Sequencing of ADC and DAO Genes and Estimation of Gene Copy Numbers

Genomic DNA was extracted from the haploid megagametophyte tissue of the seeds using the Nucleospin Plant II kit (Macherey-Nagel, Germany) with the lysis buffer PL1. DNA quality was checked by agarose gel electrophoresis and DNA quantity was measured using a NanoDrop ND1000 spectrophotometer (NanoDrop Technologies, USA). All the PCR fragments of a gene were amplified from the same DNA template which was extracted from a single megagametophyte. The PCR products were purified with the MinElute PCR purification kit (Qiagen, USA) or the FastAP Thermosensitive Alkaline Phosphatase and Exonuclease I (Fermentas, Lithuania). The Genome Walker Universal kit (Clontech, USA) was used according to the manufacturer's instructions to increase the length of the sequenced region to the promoter. Sequencing reactions were carried out with an ABI Prism 3730 DNA Analyzer (Applied Biosystems) with a Big Dye Terminator kit v3.1 (Applied Biosystems). The obtained sequences were verified and edited manually using Sequencher 4.7 (Gene Codes Corporation, USA). The sequencing primers are presented in the Supplementary Table S2.

For estimating the ADC, ODC, and DAO gene copy numbers in pine genomes the complete cDNA sequence for ODC and the obtained full gene nucleotide sequences for ADC and DAO were used for BLAST searches against the complete genome of loblolly pine via the website http://congenie.org/blast. BLAST searches were also performed using the predicted protein sequences of the ADC, ODC, and DAO genes.

### Phylogenetic Analysis

The multiple CuAO protein alignment (Supplementary Figure S2) was created using MUSCLE (Edgar, 2004) in Geneious version 11.1.5 (created by Biomatters, available from http:// www.geneious.com). The phylogenetic tree was constructed with maximum likelihood using PhyML 3.3.20180621 (Guindon et al., 2010) as a plug-in in Geneious. The analysis was run with the default parameters, using the Jones, Taylor, and Thornton substitution matrix for protein sequences (Jones et al., 1992) and optimization on tree topology, branch length, and substitution rate. Branch support was obtained by 500 bootstrap replicates.

#### RNA Extraction and Reverse Transcription

Total RNA was extracted from the Scots pine tissues using the KingFisher™ mL method (Thermo Electron Corporation, Finland) with the MagExtractor® total RNA purification kit (Toyobo, Japan) according to the manufacturer's instructions. The samples consisting of immature seeds, mature embryos, megagametophytes, needles, stems, or roots were homogenized in liquid nitrogen using a mortar and pestle and 30 mg of the powder was subsequently used for RNA extraction. The RNA samples were treated with RNase-free DNase (Invitrogen, USA) at RT for 15 min for the elimination of contaminating genomic DNA. Thereafter, the RNA samples were purified with the NucleoSpin® RNA Clean-Up kit (Macherey-Nagel, Germany). The RNA yields were measured three times with OD260 analysis using NanoDrop ND1000 spectrophotometer (NanoDrop Technologies, USA) and 1 µg of each RNA sample was subsequently used for the cDNA synthesis. cDNA was reversetranscribed from an anchored oligo-dT primer by SuperScript II reverse transcriptase (Invitrogen, USA) using standard methods in a reaction volume of 20 µl. PCR with actin (ACT) primers, which have been designed so that the amplicon contains an intron, was used for revealing possible genomic contamination in the cDNA samples (Jaakola et al., 2004).

### Quantitative PCR Analysis of mRNA Transcripts

The absolute quantification of ADC and ODC mRNA transcripts in developing zygotic embryos at early (n = 5) and late (n = 4) developmental stages, in mature embryos (n = 5) and megagametophytes (n = 4) as well as in cotyledons (n = 7), hypocotyls (n = 8), and roots (n = 5) was performed with quantitative real-time PCR (qPCR) analysis using synthesized RNA molecules as standards. The DNA templates from which the RNA molecules could be transcribed were amplified by basic PCR procedure. The PCR primers for the ADC standard were 5′- AGAAATTGGGGATGCTGGAT-3′ and 5′-GCCATCACCGA CTGGTATTCACC-3′ and for the ODC standard 5′-TTGCGT TGCAGACGTATTTC-3′ and 5′-CAGCGCAAAAGGACGT AGAT-3′. The upstream primers contained T7 promoter sequence (TAATACGACTCACTATAGGG) and the downstream primers contained a poly (T) tail at their 5′ end, which enabled both the synthesis of RNA molecules and the reverse transcription (RT) of synthesized RNA molecules to cDNA with anchored oligo-dT primers. The DNA molecules were used as templates for in vitro transcription by T7 RNA polymerase. The standard curves were generated using serial 10 fold dilutions of synthesized RNA molecules to control variability during both the RT and PCR steps of the analysis. The number of RNA molecules added to the RT reaction in the first standard was 1011. The weight of standard RNA (in ng) equivalent to 1011 was calculated using the molecular weight of oligonucleotide and Avogadro's constant (6.022 · 1023 mol−<sup>1</sup> ). The quantification of the target mRNA (reverse transcribed to cDNA) was based on a standard curve constructed in the same quantification assay. For the relative quantification of ACL5, ADC, AIH, CPA, DAO, SAMDC, and SPDS expression during seed development in Scots pine clones K818 and K884 five biological replicates per clone and per sampling date were performed except for sampling dates III and IV when there were only two and four biological replicates for K884, respectively. The geometric mean (Vandesompele et al., 2002) of three independently regulated reference genes actin (ACT), ubiquitin (UBI), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) from different functional classes was used for the normalization of the gene expression levels according to the Pfaffl method (2001).

The PCR amplification conditions of the gene fragments were optimized for the LightCycler® 2.0 instrument (Roche Diagnostics, Germany), and the subsequent PCR runs showed a single PCR product in melting curve analysis (The Tm Calling Analysis of Lightcycler® 480 Software release 1.5.0 SP3) and agarose gel electrophoresis. The real-time PCR amplifications were performed using the LightCycler® 480 SYBR Green I Master (Roche Diagnostics, Germany), 50 nM gene specific primers (Supplementary Table S3), and 2 ml cDNA in the reaction volume of 20 ml. The real-time PCR amplification was initiated by incubation at 95 °C for 10 min followed by 45 cycles: 10 s at 95 °C, 10 s at 58 °C, and 5 s at 72 °C. Two technical replicates of each PCR reaction were performed to control for the variability of PCR amplification. The Abs Quant/2nd Derivative analysis of Lightcycler® 480 software release 1.5.0 SP3 was utilized to generate the crossing point and concentration values for each sample.

#### Localization of ACL5 and DAO Transcripts by mRNA In Situ Hybridization

The localization of ACL5 and DAO transcripts was performed by mRNA in situ hybridization assay as previously described in Vuosku et al. (2015). The 332-bp long ACL5 probe was amplified with the primers 5′-GCCGAGCTCGAGAGTAGAGA-3′ (upstream) and 5′-TCGATTTCTTCGGCGTCTAT-3′ (downstream). The primers used for the synthesis of a 342 bp probe for DAO transcripts were 5′-ATTTCAGGCATGGA GATTCG-3′ (upstream) and 5′-ATTCTTCACCGTTT GCTTGG-3′ (downstream). The DIG-labeled probes were detected by Anti-DIG-AP Fab fragments and NBT/BCIP substrate (Dig Nucleic acid detection Kit, Roche Molecular Biochemicals, Germany). The sections were examined with a light microscope (Nikon Optiphot 2, Japan) and imaged with an Infinity1-3C camera (Lumenera Corporation, Canada) using the IMT iSolution Lite image-processing program (IMT i-Solution Inc., Canada). Adobe Photoshop CS5 was used to adjust contrast, brightness, and color uniformly to entire images.

#### Statistical Analysis

The effects of the four sampling dates, the two clones and their interactions on the relative gene expression of the ADC, AIH, CPA, DAO, SAMDC, SPDS, and ACL5 genes, each in turn, were analyzed as follows. Because the distribution of the relative gene expression values was highly skewed to the right, we first transformed them to the logarithmic scale. We then fitted a linear regression model on the log-transformed relative gene expression for each gene separately. The sampling date and clone were included as categorical factors with fixed effects, such that their reference levels were sampling date I and clone K818, respectively. Thus, our model is equivalent to a two-way analysis of variance model with interaction. The models were fitted using the function lm (Chambers, 1992) of the R environment version 3.6.1 (R Core Team 2019). The results from estimation of the model parameters on the log-scale were back-transformed onto the original scale, and are presented in terms of the relative contrasts of relative gene expression associated with the levels of the two factors vs. the reference level of that factor, and supplemented with the 95% confidence intervals for the respective contrasts.

### RESULTS

#### Fast- and Slow-Evolving Genes in Scots Pine PA Biosynthesis Pathway

The sequence conservation and the type of selection operating on the protein coding regions of the ACL5, ADC, AIH, CPA, DAO, SAMDC, SPDS, and SPMS were studied using the ratio of Ka/Ks, which provides information on the level of evolutionary constraint across related species (Fay and Wu, 2003). Under neutral evolution Ka equals Ks, whereas a low Ka/Ks ratio indicates purifying (negative) selection and a high Ka/Ks ratio directional (positive) selection. Over the PA genes and the pairs of plant species the arithmetic mean values (ranges in parentheses) of Ks, Ka, and Ka/Ks were 0.89 (0.004 to 2.83), 0.10 (0, 0.33), and 0.14 (0, 0.35), respectively (Supplementary Table S4). The wide range of Ks estimates reflects the range of very close and very distant phylogenetic comparisons included. There were no sequence pairs with Ka/Ks value greater than one and thus no strong evidence for the contribution of positive selection to the interspecific sequence divergence over the whole gene on any of the PA genes. The highest Ka/Ks values in the PA biosynthesis pathway, 0.32 and 0.30, were found when the Scots pine ODC sequence was compared with the Sitka spruce (Picea sitchensis (Bong.) Carr.) and loblolly pine (Pinus taeda L.) ODC sequences, respectively. No nonsynonymous substitutions were found between the Scots pine and loblolly pine ACL5 sequences leading to the lowest Ka/Ks value. Interestingly, both the highest and lowest Ka/Ks values between the DAO sequences, 0.35 and 0.05, were detected in conifers: the Scots pine DAO sequence compared to the Sitka spruce and loblolly pine DAO sequences respectively (Figure 1). Note that the comparisons between the two pine species were based on rather low numbers of substitutions, as average Ks estimates were less than 0.05 (Supplementary Table S4).

Despite their position in the pathway or evolutionary origin, all the PA biosynthesis genes showed evidence of strong purifying selection across evolutionary time, as their Ka values were much lower than Ks values (Figure 1 and Supplementary Table S4). However, the heterogeneity of the regression coefficients of Ka over Ks (b) suggested different substitution rates at the amino acid level, and thus different strength of purifying selection. The SAMDC gene had the highest rate of amino acid change, (b = 0.15), whereas the substitution rate appeared lowest (b = 0.07) for the CPA and SPDS genes (Figure 2). Strong positive correlations were observed between Ka and Ks in all the PA biosynthesis genes with correlation coefficients ranging from 0.91 to 1. The high correlations for individual genes could bea resultof the samemutation ratesinfluencing bothKa

FIGURE 1 | Graphical presentation of the ratios (Ka/Ks) of the nonsynonymous substitution rate (Ka) to the synonymous substitution rate (Ks) in the polyamine genes. Ka/Ks ratios were estimated for the thermospermine synthase (ACL5), arginine decarboxylase (ADC), agmatine iminohydrolase (AIH), N-carbamoylputrescine amidohydrolase (CPA), diamine oxidase (DAO), ornithine decarboxylase (ODC), S-adenosyl methionine decarboxylase (SAMDC), spermidine synthase (SPDS), and spermine synthase (SPMS) genes. Each dot refers to a single pair of angiosperms or gymnosperm species in the comparison. The pairs consisting of angiosperm species are marked with blue symbols and the pairs consisting of gymnosperm species with red symbols. The vertical blue and red line segments indicate the arithmetic mean values of Ka/Ks in angiosperms and gymnosperms, respectively. The angiosperms used for the comparisons were apple, Arabidopsis, black cottonwood, clementine, rice, sorghum, tobacco, and vine. The gymnosperms were Scots pine, loblolly pine, white spruce, and Sitka spruce (see Supplementary Table S4 for details).

FIGURE 2 | The relationship between nucleotide substitutions at synonymous (Ks) and nonsynonymous (Ka) sites in polyamine biosynthesis genes. Ks and Ka values were estimated for the thermospermine synthase (ACL5), arginine decarboxylase (ADC), agmatine iminohydrolase (AIH), N-carbamoylputrescine amidohydrolase (CPA), ornithine decarboxylase (ODC), S-adenosyl methionine decarboxylase (SAMDC), spermidine synthase (SPDS), and spermine synthase (SPMS) genes. The compared plant pairs consisting of angiosperm species (Arabidopsis, black cottonwood, clementine, rice, sorghum, tobacco, and vine) are marked with blue symbols and the pairs consisting of gymnosperm species (loblolly pine, Scots pine, Sitka spruce, and white spruce) with red symbols (see Supplementary Table S4 for details). r, correlation coefficient; b, slope of the fitted regression line.

and Ks. Moreover, due to the very long evolutionary time scales (especially within the angiosperm comparisons), the average strength of purifying selection may have been similar in different lineages (Mouchiroud et al., 1995; Ohta and Ina, 1995). Note that the comparisons between conifers represent very short evolutionary time scale and the positive correlations are mainly driven by the longer-term comparisons of angiosperms.

#### Preference of ADC Pathway in Putrescine Biosynthesis in Conifers

The sequencing of Scots pine ADC in 24 samples showed that the gene structure is very simple. PsADC (HM236823) is an intronlacking gene composed of a single exon of 2139 bp. BLAST searches against the loblolly pine genome retrieved one full gene sequence for both ADC (protein sequence: lcl|PITA\_000014798 and nucleotide sequence: lcl|tscaffold9160) and ODC (protein sequence: lcl|PITA\_000064865 and nucleotide sequence: lcl| tscaffold2337) suggesting that ADC and ODC are single-copy genes in pine genomes.

We hypothesized that the observed differences in the strength of purifying selection acting on the ADC and ODC sequences in conifers reflect the different use of the ADC and ODC for Put production. To test the hypothesis in the Scots pine, we determined the mRNA copy numbers of the ADC and ODC genes in developing seeds, mature embryos, and megagametophytes as well as in the cotyledons, hypocotyls, and roots of young seedlings (Supplementary Figure S3). The ratios of the ADC and ODC mRNA transcripts in the samples were used to evaluate the roles of ADC and ODC in Put production in the Scots pine tissues (Figure 3). The lowest ADC/ODC-ratio, seven, was found in megagametophytes as well as in the cotyledons of young seedlings. The highest ratio of 681 was observed in mature embryos. Over the whole data set the geometric mean of the ADC/ODC-ratio was 66. Thus, Put was almost solely produced via the ADC pathway in both developing and mature embryos, but the ADC pathway was also strongly preferred in young seedlings. The Scots pine ODC sequence could only be amplified using the second round PCR with hypocotyl cDNA as an original template, which also suggested low expression of the ODC gene in the Scots pine tissues. In the BLAST searches against NCBI databases there were more than 100 ESTs from coniferous species, which may represent either homologues or paralogues to the Scots pine ADC, whereas there were only 9 ESTs representing ODC. The results indicated that the ADC pathway is the principal route for Put biosynthesis in Scots pine but also generally in conifers.

#### PA Gene Expression During Scots Pine Zygotic Embryogenesis

The expression of the ADC, AIH, CPA, DAO, SAMDC, SPDS, and ACL5 genes was determined during the Scots pine zygotic embryogenesis (Figure 4), which provides a favorable target for studies on PA gene expression in developmental processes due to the simultaneously ongoing cell division, cell specification, and programmed cell death processes (reviewed by Vuosku et al., 2009b). The expression of the genes in the ADC pathway, ADC, AIH, and CPA, remained relatively stable during the embryo development in Scots pine clones K818 and K884. DAO expression increased in both clones when the embryo development proceeded. In clone K884 it increased above the value of the baseline (K818, sampling date I) by sampling date III and in clone K818 the value of the baseline was passed by sampling date IV. In the overall decreasing trend of SAMDC expression there emerged a difference between the baseline and the sampling date III. SPDS expression showed a relatively stable trend in K818 but in K884 a clear drop from the baseline occurred on sampling date II after which the expression stabilized. ACL5 expression increased with embryo development in both clones and was higher on sampling dates III and IV compared to the baseline (Figure 4 and Supplementary Table S5). Altogether, the results indicated that embryo growth and development was characterized by a consistent increase in both ACL5 and DAO expression, whereas the expression of the genes responsible for the Put biosynthesis remained quite stable during the seed development.

FIGURE 4 | Polyamine gene expression during Scots pine seed development. The relative expression of the arginine decarboxylase (ADC), agmatine iminohydrolase (AIH), N-carbamoylputrescine amidohydrolase (CPA), diamine oxidase (DAO), S-adenosyl methionine decarboxylase (SAMDC), spermidine synthase (SPDS), and thermospermine synthase (ACL5) genes is presented relative to the sampling date in clones K818 (red symbols) and K884 (blue symbols). The effective temperature sums were 436.3, 509.1, 587.4, and 678.9 d.d. on sampling dates I, II, III, and IV, respectively. The geometric mean values of the five biological replicates per clone and per sampling date (except for sampling dates III and IV when there are only two and four biological replicates for K884, respectively) are connected with lines. The results of the statistical analyses of these data are presented in Supplementary Table S5. Based on them, we found evidence for an increasing trend over time in the expression of DAO and ACL5 in both clones, and some evidence for a decreasing trend in that of SAMDC, more pronounced in clone K884. For CPA, the gene expression level was consistently higher in K818 than in K884. No other statistically discernible trends or contrasts could be found.

### Association of ACL5 Expression With Morphologically Necrotic Cell Death

The role of ACL5 in Scots pine zygotic embryogenesis was studied further by localizing ACL5 mRNA transcripts. In a developing Scots pine seed the embryo lies within the corrosion cavity of the megagametophyte which houses most of the storage reserves of the seed. The developmental stage of early embryogeny (Figures 5A, B) initiates with the elongation of the suspensor system and terminates with appearance of the root meristem, whereas late embryogeny (Figure 5C) includes the establishment of root and shoot meristems and the maturation of the embryo (Singh, 1978). At the early embryogeny stage, ACL5 expressed only in few specific cells in the embryo, whereas strong ACL5 expression was detected in the megagametophyte cells in the embryo surrounding region (ESR) and in the region in the front of the expanding corrosion cavity (Figures 5A, B). In our previous study we showed that megagametophyte cells in the ESR are destroyed by morphologically necrotic cell death to nourish the developing embryo throughout embryogenesis (Vuosku et al., 2009a). At the

late embryogeny stage ACL5 expressed specifically in the procambial cells of the embryo and in the ESR of the megagametophyte (Figure 5C). The specificity of the antisense ACL5 probe was confirmed by the absence of signals in the seed section hybridized with the sense ACL5 probe (Figure 5D). The non-specific signal observed in the ESR was generated by fragmented nucleic acids as previously described in Vuosku et al. (2010). To conclude, ACL5 expression was associated with the earliest events of vascular specification in the embryo and with morphologically necrotic cell death in megagametophyte cells.

#### Accumulation of DAO Transcripts Between the Cell Walls of Adjacent Cells

Sequencing of Scots pine putative PsCuAO (HM236829), here called DAO, in 24 samples showed that the gene is composed of five exons (exon1: 1320 bp, exon2: 114 bp, exon3: 450 bp, exon4: 98 bp, and exon5: 208 bp). The introns are relatively short, with the exception of the first one (intron1: 1240 to 1265 bp, intron2: 106 bp, intron3: 307 bp, and intron4: 394 bp), leading to a total gene size between 4245 and 4262 bp. The coding sequence of the PsCuAO gene is 2190 bp long and the predicted protein of 729 amino acids contains the domain structure of the coppercontaining amine oxidases. BLAST searches against the loblolly pine genome retrieved only a single full gene copy of DAO (protein sequence: lcl|PITA\_000016951 and nucleotide sequence: lcl|tscaffold4097). In the phylogenetic analysis of the Scots pine, Arabidopsis and apple CuAO amino acid sequences Scots pine DAO belonged to the same main group with the Arabidopsis CuAO1 and apple MdAO2 proteins with 100% bootstrap support (Figure 6).

For studying the role of DAO in the Scots pine zygotic embryogenesis DAO mRNA transcripts were localized in developing Scots pine seeds. DAO expression was detected only in the outermost cells of the early embryo (Figure 7A) in which DAO transcripts were localized in the vicinity of the cell walls (Figure 7B). At high magnification DAO transcripts were also found to accumulate between the cell walls of neighboring cells where the transcript groups were localized with equal distances from each other forming a string-of-pearls-like appearance (Figure 7C). DAO expression was minor in zygotic embryos. Therefore, we compared the intensity of DAO expression in developing and mature embryos, megagametophytes, cotyledons, hypocotyls, and roots to be able to study the interesting observation further in a Scots pine tissue showing stronger DAO expression. The highest DAO expression was detected in roots, which were also selected as a target for the localization of DAO expression. Similar accumulation of DAO transcripts between two adjacent cells was detected in roots (Figure 7D). The specificity of the antisense DAO probe was

confirmed by the absence of signals in the seed and root sections hybridized with the sense DAO probe (Supplementary Figure S4). In conclusion, in Scots pine tissues DAO expression was minor and localized very specifically close to cell walls and between the cell walls of adjacent cells.

#### DISCUSSION

In the present study the Ka/Ks ratios (0 to 0.35) over all the PA genes and plant species were in line with the Ka/Ks ratios previously found between A. thaliana and Arabidopsis lyrata (0.21) (Barrier et al., 2003), between A. thaliana and Chinese cabbage (Brassica rapa ssp. pekinensis) (0.14) (Tiffin and Hahn, 2002), between Pinus and Picea species (0.10–0.15) (Palmé et al., 2008), between Sitka spruce (Picea sitchensis (Bong.) Carr.) and loblolly pine (0.31) (Buschiazzo et al., 2012) in angiosperms (0.09–0.13), in gymnosperms (0.17–0.67) (De La Torre et al., 2017), and among higher plants (Embryophytes) (0.21) (Roth and Liberles, 2006). Generally, PA genes have evolved under strong selective constraint in seed plants (about 80% of Ka/Ks estimates were lower than 0.2), which underlines the fundamental role of the PA metabolism. Furthermore, there was no strong evidence for the contribution of positive selection. However, the criterion of Ka/Ks > 1 is very stringent and several studies with large number of genes have suggested that genes with the lowest Ka/Ks values evolve under strong purifying selection pressure, whereas genes with the highest Ka/ Ks values may evolve due to positive selection in addition to relaxed constraints (Charlesworth et al., 2001; Tiffin and Hahn, 2002; Palmé et al., 2008). In conifers ACL5 genes were found to evolve under stronger purifying selection than in flowering plants, whereas in ODC genes selection pressure is relaxed in conifers compared to flowering plants.

In a biosynthetic pathway the existence of alternative routes may lead to weaker purifying selection than in the case of a pathway controlled by a single locus (Ohno, 1970). Our results suggest that this applies in the PA biosynthesis pathway of conifers where ADC and ODC genes have evolved differently. ADC genes have undergone stronger purifying selection likely because of the clear preference of the ADC pathway in Put production, whereas ODC genes have evolved more freely due to the reduced use of the ODC enzyme. As in conifers, ADC is the primary gene involved in Put biosynthesis in some angiosperms, such as Arabidopsis and apple (Hanfrey et al., 2001; Hao et al., 2005), emphasizing the general importance of the ADC pathway in plants. Thus, we propose that Put biosynthesis from ornithine may not be essential for normal plant growth.

Our results suggest that the PA biosynthesis pathway has less flexibility in Scots pine than generally in flowering plants. According the BLAST searches against loblolly pine genome there is only one copy of both the ADC and ODC genes in pines, although the presence of other copies cannot be definitely excluded due to the imperfect annotation of conifer genomes.

Arabidopsis, for example, possesses two paralogues of ADC which has allowed the specialization of the AtADC1 and AtADC2 genes (Galloway et al., 1998; Hummel et al., 2004). In addition to the production of Put almost solely via the ADC pathway Scots pine possesses a single bifunctional SPDS enzyme for both Spd and Spm synthesis (Vuosku et al., 2018). In flowering plants the duplications of SPDS genes has led to presence of more than one SPDS genes and, further, to the evolution of separate SPDS and SPMS genes (Minguet et al., 2008). Thus, the different evolution of the enzyme genes in the PA biosynthesis pathway in the two plant lineages may have resulted in increased adaptability of PA homeostasis in flowering plants compared to conifers.

Both ACL5 and DAO expression increased with embryo development. ACL5 expression was connected to the earliest events of vascular specification in the developing embryo. Furthermore, intense ACL5 expression was found in the megagametophyte cells destined to die to provide nourishment and space for the growing embryo. Previously, ACL5 expression was associated with the later development of the vascular structures during the seed germination and early seedlings growth in Scots pine (Vuosku et al., 2018). T-Spm has been identified as a plant growth regulator that represses xylem differentiation and promotes stem elongation by preventing premature death of developing xylem vessel elements in Arabidopsis (Kakehi et al., 2008; Muñiz et al., 2008; Vera-Sirera et al., 2010). However, ACL5 expression and the presence of T-Spm were recently found in non-vascular land plants and in the unicellular green alga Chlamydomonas reinhardtii (Solé-Gil et al., 2019), which suggested that T-Spm plays also other roles in addition to xylem development in vascular plants. Our present results show a novel function for ACL5 in the morphologically necrotic cell death and support the view of the genetic regulation of necrosis in Scots pine embryogenesis and further in plant development. Furthermore, the results suggest that higher sequence conservation of the ACL5 genes in conifers compared to flowering plants results from the necessity of ACL5 for both embryogenesis and vascular development and, therefore, may reflect differences in the roles of ACL5 in those crucial developmental events between the plant lineages.

Interestingly, only one DAO gene was found from the loblolly pine genome, whereas DAO genes form large gene families in flowering plants (Tavladoraki et al., 2016). In Arabidopsis, 10 genes have been annotated asCuAOsfromwhichATAO1,AtCuAO1, AtCuAO2, AtCuAO3, and AtCuAO8 have been also characterized at the protein level (Møller and McPherson, 1998; Wimalasekera et al., 2011; Planas-Portell et al., 2013; Groß et al., 2017).

Furthermore, five CuAO genes were identified from apple fruit cDNA and the two most abundant (MdAO1 and MdAO2) were biochemically characterized (Zarei et al., 2015). In the phylogenetic tree the Scots pine DAO protein belonged to the same main branch as AtCuAO1 and MdAO2 which have been reported to be extracellular proteins (Planas-Portell et al., 2013; Zarei et al., 2015). In Scots pine zygotic embryos and roots DAO expression was localized in the vicinity of cell walls and between the cell walls of adjacent cells, where they formed string-ofpearls-like appearances suggesting that DAO transcripts accumulated in the plasmodesmata. We therefore suggest that DAO, in addition to the roles in the regulation of cellular Put content and production of hydrogen peroxide (H2O2) for cell wall loosening/stiffening events (Tavladoraki et al., 2016), may also participate in cell-to-cell communication at the mRNA level. The role of H2O2 in signaling has received much attention but it has been also studied for its toxic effects (Smirnoff and Arnaud, 2018). Also, mRNAs may not be constrained to their cell of origin but potentially move to and act in other cells and, thus, serve as messengers between cells (Ham and Lucas, 2017; Morris, 2018). In the light of our results, we suggest that DAO transcripts may be transported between cells instead of H2O2 to avoid harmful effects especially in sensitive developing tissues.

In conifers the detailed roles of the enzymes in the PA pathway have been poorly known, which has hindered the development of PA-based applications in forest biotechnology. Our results suggest that the adaptability of PA homeostasis may be restricted in Scots pine and, thus, the manipulation of PA levels may not provide as practical tools for the enhancement of stress tolerance as in many flowering plants (Hussain et al., 2011; Marco et al., 2016). In both Scots pine seedlings and proembryogenic cells the expression of the PA biosynthetic and catabolic genes was down regulated rather than up regulated during drought/ osmotic stress, whereas PA contents remained quite stable (Muilu-Mäkelä et al., 2015a; Muilu-Mäkelä et al., 2015b). Therefore, we suggest that PAs may have more potential for biotechnological applications in development related processes such as somatic embryogenesis and the control of wood formation in Scots pine. During the induction of Scots pine somatic embryogenesis, the triggering of the embryo-producing pathway was connected with consistent changes in PA gene expression (Salo et al., 2016). Thus, the specific manipulation of PA gene expression might provide a way to enhance somatic embryo production in recalcitrant Scots pine lines. In the present study ACL5 expression was associated with the earliest events of vascular specification during the Scots pine zygotic embryogenesis, and ACL5 expression has been localized in procambial cells also during seed germination and early seedlings growth (Vuosku et al., 2018). The role of auxin as main regulator of vascular differentiation is well documented (Miyashima et al., 2013) and findings on xylem differentiation have proposed a model of complex functional interaction between auxin, T-Spm, and HD-ZIP III genes (Milhinhos et al., 2013; Baima et al., 2014). Furthermore, Yoshimoto et al. (2012) found that the auxin signaling that promotes xylem differentiation is normally limited by SAC51-mediated T-Spm signaling but can be continually stimulated by exogenous auxin analogs in the absence of T-Spm. Thus, the opposite action between T-Spm and auxin seems to be essential for xylem differentiation and the combined use of these growth regulators might provide opportunities to control wood formation in Scots pine.

#### DATA AVAILABILITY STATEMENT

All datasets generated/analyzed for this study are included in the manuscript and the supplementary files.

### AUTHOR CONTRIBUTIONS

TS, HH, and JV conceived the project. JV mainly designed the study, participated in the molecular evolutionary and gene expression analyses, and acted as a principal author of the manuscript. RM-M performed the statistical analyses of the gene expression results and participated in the writing of themanuscript. KA performed thefull gene sequencing, the BLAST search analyses, and participated in the writing of the manuscript. MS performed the computational analysis of the molecular evolution data. JK carried out the in situ mRNA hybridization analyses. EL created the statistical graphs and performed the statistical analysis of the molecular evolution data. HH participated in the coordination of the study and in the writing of the manuscript. OS provided expertise in the molecular evolutionary analyses and interpretation of the results. TS participated in the coordination of the study and in the writing of the manuscript. All the authors read and approved the final manuscript.

#### FUNDING

The Research was funded by the Academy of Finland (Project 121994 to TS) and by grants from the Finnish Cultural Foundation and the Niemi Foundation (to JV). OS and KA acknowledge EU project Noveltree (FP7211868) and Biocenter Oulu.

### ACKNOWLEDGMENTS

We thank the personnel of the Natural Resources Institute Finland at the Punkaharju and Parkano Research Units for conducting the collections of the research material and Dr. Juhani Hopkins for English revision.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.01600/ full#supplementary-material

### REFERENCES


Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Vuosku, Muilu-Mäkelä, Avia, Suokas, Kestilä, Läärä, Häggman, Savolainen and Sarjala. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Compatible and Incompatible Pollen-Styles Interaction in Pyrus communis L. Show Different Transglutaminase Features, Polyamine Pattern and Metabolomics Profiles

Manuela Mandrone<sup>1</sup> , Fabiana Antognoni<sup>2</sup> , Iris Aloisi<sup>3</sup> , Giulia Potente<sup>2</sup> , Ferruccio Poli<sup>1</sup> , Giampiero Cai<sup>4</sup> , Claudia Faleri<sup>4</sup> , Luigi Parrotta<sup>3</sup> and Stefano Del Duca<sup>3</sup> \*

#### Edited by:

Antonio F. Tiburcio, University of Barcelona, Spain

#### Reviewed by:

Penglin Sun, University of California, Riverside, United States Milagros Bueno, Universidad de Jaén, Spain

> \*Correspondence: Stefano Del Duca stefano.delduca@unibo.it

#### Specialty section:

This article was submitted to Plant Metabolism and Chemodiversity, a section of the journal Frontiers in Plant Science

Received: 15 March 2019 Accepted: 17 May 2019 Published: 07 June 2019

#### Citation:

Mandrone M, Antognoni F, Aloisi I, Potente G, Poli F, Cai G, Faleri C, Parrotta L and Del Duca S (2019) Compatible and Incompatible Pollen-Styles Interaction in Pyrus communis L. Show Different Transglutaminase Features, Polyamine Pattern and Metabolomics Profiles. Front. Plant Sci. 10:741. doi: 10.3389/fpls.2019.00741 <sup>1</sup> Department of Pharmacy and Biotechnology, University of Bologna, Bologna, Italy, <sup>2</sup> Department for Life Quality Studies, University of Bologna, Rimini, Italy, <sup>3</sup> Department of Biological, Geological and Environmental Sciences, University of Bologna, Bologna, Italy, <sup>4</sup> Department of Life Sciences, University of Siena, Siena, Italy

Pollen-stigma interaction is a highly selective process, which leads to compatible or incompatible pollination, in the latter case, affecting quantitative and qualitative aspects of productivity in species of agronomic interest. While the genes and the corresponding protein partners involved in this highly specific pollen-stigma recognition have been studied, providing important insights into pollen-stigma recognition in selfincompatible (SI), many other factors involved in the SI response are not understood yet. This work concerns the study of transglutaminase (TGase), polyamines (PAs) pattern and metabolomic profiles following the pollination of Pyrus communis L. pistils with compatible and SI pollen in order to deepen their possible involvement in the reproduction of plants. Immunolocalization, abundance and activity of TGase as well as the content of free, soluble-conjugated and insoluble-bound PAs have been investigated. <sup>1</sup>H NMR-profiling coupled with multivariate data treatment (PCA and PLS-DA) allowed to compare, for the first time, the metabolic patterns of not-pollinated and pollinated styles. Results clearly indicate that during the SI response TGase activity increases, resulting in the accumulation of PAs conjugated to hydroxycinnamic acids and other small molecules. Metabolomic analysis showed a remarkable differences between pollinated and not-pollinated styles, where, except for glucose, all the other metabolites where less concentrated. Moreover, styles pollinated with compatible pollen showed the highest amount of sucrose than SI pollinated ones, which, in turn, contained highest amount of all the other metabolites, including aromatic compounds, such as flavonoids and a cynnamoil derivative.

Keywords: self-incompatibility, Pyrus communis, transglutaminase, <sup>1</sup>H NMR-metabolomics, polyamines

### INTRODUCTION

fpls-10-00741 June 12, 2019 Time: 19:21 # 2

In order to prevent self-fertilization, plants have evolved different strategies, from the temporally asynchronous development of male and female reproductive organs, to their specific localization within the flower, up to genetic-based strategies called "selfincompatibility" (SI) (Barrett, 2003; Ashman et al., 2004). The latter process prevents self-fertilization by the rejection of pollen of the same species. Based on whether the gametophytic or the sporophytic genome will determine pollen rejection, SI is classified into gametophytic SI (GSI) or sporophytic SI (SSI) (Fujii et al., 2016). Most of SI systems are controlled by a single gene locus, the so-called "S locus" which presents multiple S-alleles, i.e., the pistil-S and pollen-S genes; however, several other genes, proteins and external factors are involved in the process of pollen acceptance/rejection (Serrano et al., 2015; Qu et al., 2017).

In the last decades, both genetic and molecular factors involved in SI have been studied in Malinae, as GSI affects both quantitative and qualitative aspects of productivity in agricultural species of great economic interest (Del Duca et al., 2019). In the GSI of Malinae, the stylar S locus plays an important role, since it encodes for glycoproteins with ribonuclease (S-RNase) activities that enter inside the pollen tube. Once inside, they are degraded in case of compatible pollen, allowing pollen tubes to grow, while in SI pollen the S-RNases are kept active, causing the degradation of pollen RNA and the cell death of pollen tube, which generally reaches only the upper third part of the style (De Franceschi et al., 2012).

Several evidences highlight the involvement of transglutaminase (TGase) in the process of pollen rejection, not only in Malinae (Gentile et al., 2012; Aloisi et al., 2016a; Del Duca et al., 2019). TGases are ubiquitous enzymes that catalyze the posttranslational modification of proteins, through the incorporation of primary amines or by protein cross-linking, resulting often in high molecular mass products (Griffin et al., 2002). In plants, TGases are distributed in different cell compartments where they exert a structural or conformational role (Del Duca and Serafini-Fracassini, 2005; Serafini-Fracassini and Del Duca, 2008; Serafini-Fracassini et al., 2009). The post-translational modification of proteins by polyamines (PAs) and the catalysis of isopeptide bonds are the main TGase reaction studied in plant (Del Duca et al., 2000; Aloisi et al., 2016b). Although TGase could also catalyze the deamidation of endoglutamine residues of protein substrates, up to now this reaction has been scarcely considered in plants. TGase is a crucial factor for the growth of pollen tubes as is involved in the cross-link of several substrates (mainly cytoskeleton proteins) and in the post-translational modification of proteins (Di Sandro et al., 2010; Aloisi et al., 2016a). TGase activity increases during the SI response (Del Duca et al., 2010; Gentile et al., 2012) and programmed cell death (PCD) related processes, when the stimulation of TGase activity is mainly due to the increase of Ca2<sup>+</sup> (Della Mea et al., 2007). Recently, also in Malineae, PCD phenomena have been shown to occur as the consequence of SI response (Wang et al., 2010; Li et al., 2018).

Polyamines are essential for cell growth, and their contents in cells are maintained by biosynthesis, degradation and transport; PAs transport systems have been identified in different organisms, from bacteria to plants (Antognoni et al., 1999; Fujita and Shinozaki, 2014). Despite the wide array of investigations on PA roles in plant cells, little information is available on their role during the apical growth of pollen within the styles. PAs biosynthesis increased during pollen tube emergence and elongation and they are released in the external space (Bagni et al., 1981) where they affected RNAse activity (Speranza et al., 1984). Together with the decrease in free PAs, changes in the levels of PAs bound to low- and high-mass molecules take place inside the pollen tube (Antognoni and Bagni, 2008). Moreover, PAs are directly involved in ROS regulation, whose concentration is essential for pollen tube growth (Wu et al., 2010; Aloisi et al., 2015).

Metabolomics is a novel inductive approach, relying on untargeted analysis protocols, whose results are handled by multivariate data treatment. This workflow proved to be successful in several research fields, from human diagnostics and epidemiology to plant science (Wolfender et al., 2015; Anąelkovic et al., 2017 ´ ), where it resulted particularly helpful to face the challenges posed by the complexity of natural product chemistry (Pauli et al., 2012). For instance, metabolomics has been employed to facilitate the identification of medicinal plants active principles (Goąevac et al., 2018; Mandrone et al., 2018).

In this work, styles of Pyrus communis L. cv Abbé Fétel were pollinated with compatible Williams pollen (AxW) and with incompatible Abbé Fétel pollen (AxA) in order to induce SI response.

As many changes occur (i.e., in the cell wall) during the growth of pollen tube inside the style, it is interesting to check if this is complemented by variation in metabolomics profile that has been analyzed in not-pollinated (NP) styles and in styles pollinated with compatible and SI pollen, at 48 h after pollination. Besides, the metabolomic analyses, performed for the first time on SI systems, will add knowledge enlarging the picture of factors involved in the SI process and will help to clarify what are the main actors playing into the SI response in the complex molecular networks of pear SI system. For example, it is reasonable to hypothesize changes in cell wall as well as in phenolic compounds or hydroxycinnamoyl-derivatives that could take place because of the pollen-style interaction.

As the cytoskeleton is a target of SI response (Thomas et al., 2006), and TGase affects the organization of microfilaments and microtubules (Del Duca et al., 1997, 2009), TGase amount, activity, and localization have been checked. In particular, analysis were carried out at different time intervals, from the first hours until 48 h after pollination. Moreover, given the role of PAs in fertilization process (Aloisi et al., 2016a), and the role of TGase as a mediator of PAs action (Del Duca et al., 2014a), their profiles and their changes in styles pollinated with compatible and SI pollen have been investigated.

#### MATERIALS AND METHODS

#### Materials

All reagents were purchased from Sigma-Aldrich (Milan, Italy), except the deuterated solvents, which were purchased by Eurisotop (Cambridge, United Kingdom).

### Plant Material, in vitro and in vivo Pollen Germination and Sampling

Mature pollen of pear (Pyrus communis L. cv. "Abbé Fétel" and "Williams" was collected from plants grown in experimental plots at the University of Bologna (Dipartimento di Scienze e Tecnologie Agro-Alimentari, University of Bologna). Handling, storage, pollen hydration and 2 h germination were performed as previously reported (Aloisi et al., 2015, 2017). Styles were collected from flowers of Pyrus communis "Abbé Fétel" at the balloon stage. For pollination, about 15 sprigs, each with at least 50 flowers at the balloon stage, were collected from a maximum of 20 different Abate trees. These were grown in different positions in the orchard and flowers were collected from various positions on the same tree. Sprigs with flowers were placed in a beaker with water, and anthers were removed from each flower in the laboratory. Once emasculated, they were pollinated with compatible and incompatible pollen. In the compatible crosses, styles of Abbé Fétel (S-genotype: S104/S105) were pollinated with William (S101/S102) pollen (AxW); in the incompatible ones, styles of Abbé Fétel were self-pollinated (AxA). After 48 h, pollinated pistils were collected and immediately frozen in liquid nitrogen, then transferred at −80◦C. Equal numbers of pistils were randomly chosen to create pooled samples of 20 pistils. The procedure was performed four times to generate four biological replicates for the analysis of metabolomics, total flavonoids, PAs and TGase activity. Three technical replicates were performed.

### Staining of Callose With Aniline Blue

Styles of pear were immediately fixed in 3% glutaraldehyde in 44 mM cacodylate buffer, pH 7.2, for 2 h at room temperature. Samples were washed rapidly with the cacodylate buffer and placed in a solution of 8 N NaOH for 2 h at room temperature. Samples were then washed in distilled water for 10 min and stained with 0.1% aniline blue in 0.1 M KH2PO<sup>4</sup> for 2 h; then, samples were squashed by placing them between a glass slide and a coverslip and by applying a gentle pressure. Observations were made with a 340 nm UV filter-equipped fluorescence microscope (Zeiss Axiovision).

### Transglutaminase Quantification and Enzyme Activity Assay

Pollen and pistils proteins were extracted according to literature (Gentile et al., 2012). Briefly, proteins were solubilised at 4 ◦C in extraction buffer (10 mg ml−<sup>1</sup> ) containing 100 mM Tris–HCl pH 8.5, 2 mM dithiothreitol (DTT), 0.5 mM ethylenediaminetetraacetic acid (EDTA) and 0.2 mM phenylmethylsulphonylfluoride (PMSF) in a Potter–Elvehjem homogenizer. Large cell debris were removed from the total homogenate by centrifugation at 10,000 g for 10 min at 4◦C. Protein concentration was estimated on the supernatant by the Bradford (1976) method with bovine serum albumin (BSA) as the standard protein.

ELISA assay was carried out in triplicate as described previously (Paris et al., 2017). Briefly, a 96-wells plate was incubated overnight at 4◦C with extracted proteins (50 µl/well). Wells were washed twice with PBS (Phosphate buffered saline) buffer then incubated 1 h at RT with 200 µl/well of 5% defatted milk dissolved in PBS. Wells were washed twice and the mouse monoclonal antibody ID10 (Nottingham Trent University) was added after dilution (1:500) in PBS for 2 h at RT. Wells were washed with 0.05% Tween 80 in PBS three times and incubated with the secondary peroxidase conjugated antibody (1:500) for 1 h at RT. After washing, the substrate solution of 0.3 mM 3,3<sup>0</sup> ,5,5-Tetramethylbenzidine (TMB) (10 mg/mL of dimethyl sulfoxide) and 0.03% (v/v) hydrogen peroxide (H2O2) in 100 mM sodium acetate (CH3COONa) pH 6.0 was added. The staining development was stopped after maximum 30 min with 50 µl per well of 5 N sulfuric acid (H2SO4). The absorbance was read at 450 nm using a Wallac Victor Multiscan ELISA (Perkin Elmer).

The in vitro TGase activity was measured on extracts of ungerminated pollen (UGP) and germinated (GP) in vitro for 2 h as well as on NP (not pollinated) and pollinated styles after 6, 24 and 48 h from pollination, by the conjugation of biotinylated cadaverine to exogenous substrates N, N 0 -dimethylcasein or fibronectin as previously described (Di Sandro et al., 2010; Del Duca et al., 2018). Specific activity was determined as a change in A<sup>450</sup> of 0.1 per hour per mg of pollen after subtraction of the value of the controls treated with 20mM EGTA [ethylene glycol-bis(β-aminoethyl ether)-N,N,N<sup>0</sup> ,N0 -tetraacetic acid].

### Immuno-Localisation of TGase

Compatible (AxW) and incompatible (AxA) styles of pear were pollinated and after 24 and 48 h frozen at −80◦C. Samples were directly thawed in a buffer solution (100 mM Pipes pH 6.8, 10 mM EGTA, 10 mM MgCl2, 0.1% NaN3) plus detergent and fixative (0.05% Triton X-100, 1.5% paraformaldehyde, 0.05% glutaraldehyde) for 30 min on ice and then at 4◦C for additional 30 min. For localization of TGase, fixed pear styles were cut along their length and placed in the buffer solution containing 0.75% cellulysin and 0.75% pectinase for 7 min. For immunofluorescence microscopy, samples were washed in the above buffer and incubated with the anti-TGase antibody ID10 (Nottingham Trent University) diluted 1:20 in the buffer; incubation was 1 h at 37◦C. This antibody has been shown to immunoreact with partially purified TGase extracted from apple pollen (Del Duca et al., 2009; Di Sandro et al., 2010). After washing with the buffer, styles were incubated with the secondary antibody goat anti-mouse FITC-conjugated, diluted 1:50 in the buffer solution, for 45 min at 37◦C in the dark. Samples were observed with a Zeiss Axiophot fluorescence microscope equipped with a MRm video camera and a 63× oil-immersion objective. For electron microscopy, samples were embedded in resin and sectioned as described (Parrotta et al., 2018). In this case, the ID10 antibody was used at 1:5 dilution while the goat anti-mouse 10 nm gold-conjugated secondary antibody was used at 1:100. Samples were observed with a Philips Morgagni 268D electron microscope equipped with a MegaView II video camera.

### HPLC Analysis of Polyamines

HPLC PAs analysis was performed to investigate the content of the diamines Putrescine (Put) and Cadaverine (Cad), the triamine Spermidine (Spd) and the tetramine Spermine (Spm). Dried styles (0.025 g) were extracted in 100 vol. of 4% (w/v)

cold perchloric acid (PCA), left on ice for 1 h and centrifuged at 15.000 × g for 30 min at 4◦C. Supernatant was measured and used for analysis of Free (F) and soluble-conjugated (SC) PAs. After washing the pellets twice with cold PCA, it was resuspended in the original volume, and used for analysis of insoluble-bound (IB) PAs.

For SC and IB fractions, replicates (0.3 mL) of supernatant and pellet suspension, respectively, were subjected to hydrolysis with 6N HCl for 24 h at RT, in order to free polyamines from their conjugates, as described previously (Torrigiani et al., 1989). After hydrolysis, samples were brought to dryness at 140◦C and 0.3 mL of 4% PCA were added.

Aliquots (0.2 mL) of supernatant, hydrolysed supernatant and hydrolysed resuspended pellet were dansylated according to Smith and Davies (1985) with minor modifications. 0.2 mL of dansylchloride (5 mg/mL in acetone) and 0.2 mL of a saturated solution of Na2CO<sup>3</sup> were added to samples and incubated at 60◦C for 1 h in the dark. Then, after addition of 0.1 mL proline (15 mg/mL), samples were incubated for 30 min at RT in the dark. Dansylpolyamines were then extracted in toluene, the solvent was evaporated and samples were resuspended in 0.2 mL acetonitrile. Standard polyamines and heptamethylendiamine, as an internal standard (all by SIGMA-Aldrich) were subjected to the same procedure.

HPLC analysis was carried out on a Jasco system (Jasco– Tokyo, Japan) consisting of PU-4180 pump, FP-821 detector, and an AS-4050 autosampler. Stationary phase was an Agilent (Santa Clara, CA, United States) Zorbax Eclipse Plus C18 reversed-phase column (100 mm × 3 mm I.D., 3.5 µm), and mobile phase was a mixture of acetonitrile and water. Elution was carried out with a step gradient as follows: 60 to 70% acetonitrile in 5.5 min, 70 to 80% acetonitrile in 1.5 min, 80 to 100% acetonitrile in 2 min, 100 to 100% acetonitrile in 2 min, 100 to 70% in 2 min, and 70 to 60% in 2 min at a flow rate of 1.5 mL/min. Eluted peaks were detected by the FP fluorimeter at 365 nm excitation, and 510 nm emission, and data signals were acquired and processed through the software Chromnav 2.0 (Jasco). Results were expressed as nmol/g DW.

### Extraction for Metabolomic Analysis and NMR Measurements

Thirty milligrams of freeze-dried styles were extracted using 1 mL of a blend (1:1) of phosphate buffer (90 mM, pH 6.0) in D2O containing 0.01% trimethylsilylpropionic-2,2,3,3-d<sup>4</sup> acid sodium salt (TMSP) and CD3OD by ultrasonication for 25 min (Verpoorte et al., 2007; Kim et al., 2010). After this procedure, samples were centrifuged for 10 min (1700 × g) and 700 µL of the supernatant were transferred into NMR tubes for the analysis. The <sup>1</sup>H NMR spectra were recorded at 25 ◦C on a Varian Inova 600 MHz NMR instrument (600 MHz operating at the <sup>1</sup>H frequency) equipped with an indirect triple resonance probe, CD3OD was used for internal lock. Relaxation delay was 2.0 s, observed pulse 5.80 µs, number of scans 256, acquisition time 16 min, and spectral width of 9595.78 Hz (corresponding to δ 16.0). A presaturation sequence (PRESAT) was used to suppress the residual H2O signal at δ 4.83 (power = −6dB, presaturation delay 2 s).

### NMR Processing and Multivariate Data Treatment

Free Induction Decays (FIDs) were Fourier transformed, and the resulting spectra were phased, baseline-corrected and calibrated to TMSP at δ 0.0, spectral intensities were reduced to integrated regions of equal width (δ 0.04) corresponding to the region from δ 0.0 to 10.0, with scaling on standard at δ 0.0 using the <sup>1</sup>H NMR Mestrenova software (Mestrelab Research, Spain). The metabolites were identified on the bases of an in-house library and comparison with literature (Verpoorte et al., 2007; Scognamiglio et al., 2015; Mandrone et al., 2017).

The regions of δ 5–4.5 and 3.34–3.30 were excluded from the analysis because of the residual solvents signals. For multivariate analysis (PCA, PLS-DA), data were subjected to Pareto scaling. The models were developed using SIMCA-P software (v. 15.0, Umetrics, Sweden).

#### Total Flavonoids Assay

The assays was performed in Spectrophotometer Jasco V-530 as described by Chiocchio et al. (2018) with slight modifications. Thirty milligrams of freeze-dried sample were extracted using

1 mL of a MeOH/H2O (50:50) by ultrasonication for 25 min. After this procedure, samples were centrifuged for 10 min (1700 × g) and 50 µL of the supernatant were transferred were mixed with 450 µL of MeOH and 500 µL of AlCl<sup>3</sup> (2% w/v in methanol). Rutin stock solutions (from 1 to 100 µg/mL) were prepared in MeOH (from 1 mg/mL rutin solution in DMSO) and 50 µL of each stock were treated as above described for the samples, in order to obtain a calibration curve. The absorption at 430 nm was recorded after incubation (15 min) at room temperature. Total flavonoid content of the extracts was calculated by interpolation in the calibration curve, and expressed in terms of mg RE (rutin equivalent)/g of samples (dried weight of plant material).

#### Statistical Analysis

Data from polyamine, total flavonoids analysis and TGase activity were statistically analyzed by the one-way Analysis of Variance (ANOVA), followed by Tukey's Multiple Comparison Test, using Graph Pad Prism package (v. 5.01 for Windows; GraphPad Software, San Diego, CA, United States). Metabolomics data were subjected to Pareto scaling and the models (PCA, PLS-DA) were developed using SIMCA-P software (v. 15.0, Umetrics, Sweden).

### RESULTS

#### Callose Staining With Aniline Blue in Styles and Pollinated Styles

A preliminary analysis was carried out to check the timing of SI response, thus styles of pear (NP and pollinated with compatible and incompatible pollen) were stained with aniline blue for callose reaction 48 h after pollination. Compatible pollinated styles (AxW) showed several pollen grains (PG) while adhering on the stigmatic papillae (P). Most of pollen grains showed a pollen tube (arrows) that penetrate along the style of flowers grown on entire plants (**Supplementary**

**Figures 1A–C**). The self-fertilized styles (AxA) showed several pollen grains (PG) deposited on the surface of stigmatic papillae (P). Unlike the compatible cross, only a few pollen grains

on AxA pollinated styles; the images show details of pollen tubes with cell wall-associated labeling in the form of distinct clusters (curly brackets in E,

square bracket in F). Bars in E,F: 500 nm.

emitted a pollen tube (arrows) that were hardly observable along the style; in addition, the few pollen tubes appeared as bent and distorted (**Supplementary Figures 2A–C**). In NP styles, pollen grains as well as pollen tubes were not visible (**Supplementary Figures 3A,B**). Definitely, no differences in timing of SI response have been observed when pollinated styles belong to sprigs (**Supplementary Figures 4A–C**) or to entire planta (**Supplementary Figures 1–3**).

#### Transglutaminase Activity

Transglutaminase activity was measured in NP and pollinated styles using both N, N 0 -dimethylcasein and fibronectin, two wellknown TGase substrates from mammal sources (**Figures 1A,B**). Since it is not possible to distinguish if the measured activity derives from the enzyme present in the germinated pollen or in the styles, also hydrated and in vitro germinated pollens were analyzed for TGase activity (**Figures 1A,B**, inserts). No differences were observed 6 h after pollination neither in AxA, nor in AxW pollinated styles. After 24 h, both types of pollinated styles showed a significant decrease in TGase activity compared with NP styles, with a more pronounced effect in AxW than AxA. After 48 h, TGase activity in SI model was much higher than in the compatible one, reaching a ten-fold difference when dimethylcasein was used as a substrate (**Figure 1A**). In case of fibronectin (**Figure 1B**), the activity was higher when compared to N, N 0 -dimethylcasein for all samples, allowing to hypothesize that the former is a better substrate for pollen TGase than the latter. In both cases, AxA showed much higher (5- to 10-fold) TGase activity compared to AxW.

Transglutaminase activity was partially mirrored by the enzyme amount. In fact, the amount of TGase present during the pollination phases from 0 to 48 h was evaluated with an ELISA test (**Figure 2**) by ID10. Results showed that, at all-time points (6, 24, and 48 h), AxA showed a significantly higher TGase amount compared to NP styles. Conversely, in AxW, only after 6 h a significantly higher amount compared to control was observed, whereas at 24 and 48 h the TGase amount definitely decreased compared to both control and AxA (**Figure 2**). The greater presence of TGase observed at 6 h both in AxA and AxW in respect to control did not translate into a higher value of enzymatic activity. On the contrary, at 48 h the amount of TGase both in AxA and AxW mirrored the level of its activity, showing the highest level of TGase activity in AxA and the lowest in AxW (**Figures 1A,B**). These data showed that SI stimulated TGase activity. Finally, TGase present in pollen was stimulated by germination, being this enzyme activity higher in GP than UGP. This activity was higher in William in respect

to Abate pollen, probably due to different germination rates (**Figures 1A,B**, inserts).

#### Immuno-Localization of TGase

The presence of TGase was investigated in pear styles pollinated in both a compatible and incompatible way (**Figure 3**). In the case of compatible styles (**Figures 3A,B**), immunofluorescence investigations with the ID10 antibody did not produce specific signals, apart from some backgrounds. Conversely, incompatible styles (**Figures 3C,D**), when analyzed in indirect immunofluorescence with ID10 antibody, showed pollen tubes with an intense and specific labeling along their entire surface (arrows). The labeling had a predominantly punctiform appearance and was associated with the peripheral region of pollen tubes, i.e., the plasma membrane and/or cell wall. To get more details on the distribution of TGase, analyses were carried out using the immunogold labeling technique (**Figures 3E,F**). In compatible pollinated styles, no evidence of signals or specific signals occurred (data not shown). On the contrary, in incompatible pollinated styles, labeling with ID10 antibody showed evidence of signal associated with the cell wall of pollen tubes. The signal was distributed in the form of distinct clusters present in the thickness of the cell wall (**Figure 3E**) or on the external side (**Figure 3F**).

#### Polyamine Levels

The pattern of PAs in the F, SC, and IB forms was investigated in NP and pollinated styles AxA and AxW, respectively (**Figures 4A–C**). Put, Spd, and Spm were the only polyamines detected, while Cad was not found, neither in the former, nor in the latter types of styles (data not shown).

As concerns the F fraction, Put was the most abundant polyamine in all samples; AxW showed the highest levels, reaching values about 4.9 and 3.1 higher than AxA and NP styles, respectively (**Figure 4A**). As regards Spd and Spm, they were present at similar levels in NP styles, whereas in pollinated ones Spm was higher than Spd, with a ratio of 9.1 and 4.8 in AxW and AxA, respectively (**Figure 4A**).

Spermine was the major polyamine present in the SC fraction, i.e., conjugated through covalent bonding to low molecular weight compounds, followed by Spd and Put (**Figure 4B**). Spm levels turned much higher in pollinated compared to NP styles; in particular, SI styles (AxA) had a 2.7 times higher SC Spm than AxW. These styles also had a higher SC Spd level compared to AxW styles, while rather similar compared to NP ones. Thus, the predominant form of Spd and Spm in pollinated styles was the soluble-conjugated one, which turned much higher, compared to the free form (**Figure 4B**). In particular, in AxA styles, solubleconjugated Spm was more than 40 times higher than the free form. Conversely, Put was present at rather similar levels in the F and SC form in all types of styles.

As concerns the IB fraction, this represent, generally speaking, the less abundant form of PAs in all styles (**Figure 4C**). Put in the bound form was much lower than the free and SC ones, with very similar levels in NP, AxA and AxW styles. Conversely, much pronounced differences were observed in bound Spd and Spm, both being more concentrated in pollinated than in NP styles. Spd bound to high molecular mass compounds was more than 2-fold higher in AxW compared to AxA, while for bound Spm an opposite trend was observed, with higher levels in the latter than in the former (**Figure 4C**).

### <sup>1</sup>H NMR-Based Metabolomic Analysis

The <sup>1</sup>H NMR-based metabolome of four samples for each class (NP, AxA and AxW) were analyzed and compared. Each sample was constituted of pooled material from twenty different flowers.

The <sup>1</sup>H NMR spectra were registered and processed in order to be suitable for multivariate data analysis. A representative spectrum for each class of samples is showed in **Figure 5** and diagnostic signals of all metabolites are reported in **Table 1**. Firstly, PCA (an unsupervised model) was built with observation (N) of 12 and 212 variables (x) constituted by bucketed NMR signals, obtaining a R2X(cum) of 0.942 and a Q2X(cum) of 0.782. As showed by the PCA Score Scatter Plot (**Figure 6A**), a clear grouping of the three classes was identified (**Figure 6A**), indicating the occurrence of metabolomic differences among them.

To deeply and easily analyze these differences, a supervised model was built (PLS-DA) with N = 12 and 215 variables (x = 212 and y = 3, namely the three above mentioned classes). The first five components explained the maximum of variance

TABLE 1 | <sup>1</sup>H NMR spectral references of the metabolites mentioned in this work, and resulting, according to multivariate data analysis, highly variated among the three classes NP, AxW and AxA of Pyrus communis samples.


<sup>a</sup>Changeable by pH and concentration.

classes indicated by the Score Scatter Plot. Black squares represent the three classes, dots are the 12 analyzed samples, inverted triangles are binned NMR signals and, among them, the red ones are the diagnostic signals of the most important metabolites. A general increase of all metabolites is observed in AxA, followed by AxW, while NP is less enriched in all metabolome, except glucose, which signals increase along component t[1]. Pollinated samples variate along the component t[2], indicating that AxW is more enriched in sucrose than AxA (this metabolite increases on positive component t[2]). 1, fatty acids; 2, α-linolenic acids; 3, valine; 4, isoleucine; 5, alanine; 6, quinic acid; 7, GABA; 8, malic acid; 9, succininc acid;10, glutamine; 11, β-glucose; 12, sucrose; 13, α-glucose; 14, quercetin like flavonoids; 15, cynnamoyl derivative; 16, shikimic acid; 17, fumaric acid; 18, p-hydroxybenzoyc acid; 19, kaempherol like flavonoid; 20, trigonelline.

and the model yield a R2X(cum) of 0.952, R2Y(cum) of 0.979 and a Q2(cum) of 0.873. The model was further validated by permutation test (using 100 permutations), which gives R2Y(cum) of 0.98 and Q2(cum) of 0.84, while the intercepts were 0.64 and −0.42 for R2 and Q2, respectively.

Samples belonging to the NP class were strongly distanced from the other two classes, being clustered along the negative component t[1] (**Figures 6B,C**). This indicated the globally less enriched spectrum (lowest amount of both primary and secondary metabolites) of this class in comparison to pollinated classes (AxA and AxW). The only exception was constituted by signals related to glucose, which increase in NP samples. More specifically, NP showed a lowest content of aliphatic amino acids (alanine, valine, isoleucine), glutamine, fatty acids, α-linoleic-acid, organic acid (quinic acid, GABA, succinic acid, shikimic acid, malic acid, p-hydroxybenzoyc acid), sucrose, trigonelline alkaloid and aromatic spectral signals probably due to cynnamoil derivatives and flavonoids, such as quercetin and kaempferol derivatives.

Moreover, samples belonging to AxA and AxW classes were distanced along the component t[2], indicating that class AxW was more enriched in sucrose than AxA, while the latter showed highest concentration of all the other metabolites, including aromatic signals ascribable to cynnamoil derivatives

and flavonoids. To further confirm this latter data, total flavonoid content of the three classes was also analyzed (**Figure 7**) showing, in fact, a significant increasing of these metabolites in AxA samples, which differ by (P < 0.05) from NP.

### DISCUSSION

Cytosolic and cell wall-associated TGases activity during pollen tube growth have been reported (Del Duca et al., 2014b), suggesting its potential role in the modification of both the cytoskeleton (Del Duca et al., 2010) and components of the cell wall matrix, such as proteins, polysaccharides or glycoproteins (Waffenschmidt et al., 1999). Moreover, previous evidences have also demonstrated an alteration in TGase activity occurring during the SI reaction (Del Duca et al., 2010).

In this work, the staining of NP and pollinated styles with aniline blue confirmed the timing of the SI response according to literature data (Del Duca et al., 2010). In order to extend knowledge on TGases activity during SI, the distribution of this enzyme and its activity were monitored during 48 h. The obtained results highlighted changes in TGase localization after SI induction, generating aggregates in the cell wall, visible with confocal analysis. These aggregates were not present in compatible pollinated styles. These data might suggest the involvement of TGase in cell wall stiffening, leading to the arrest of pollen tube growth.

A key event following stresses or insults, including SI induction is the increase of Ca2<sup>+</sup> levels up to mM concentrations (Griffin et al., 2002; Ranty et al., 2016). Being TGase a Ca2+-dependent enzyme, this ion is likely involved in up-regulation of TGase activity, whose catalysis takes place from 20 nM Ca2<sup>+</sup> onward, leading, when the concentration increase up to mM, to a massive protein crosslinking, and consequent formation of protein aggregates (Griffin et al., 2002). The increased activity of TGase strongly affects the structure of cytosolic proteins such as actin and tubulin (Del Duca et al., 2009), resulting in cytoskeleton rearrangements, already reported in Pyrus and other SI models (Staiger and Franklin-Tong, 2003; Del Duca et al., 2009). These TGase-mediated rearrangements of cytosolic proteins are necessary for the proper growth of the pollen tube, and an upregulation of TGase could play a crucial role in SI response.

In addition to protein cross-linking, TGase activity modifies proteins conformation and surface charge also by PAsprotein conjugation. PAs are found both as free amines in cell compartments and conjugated to low molecular weight metabolites (e.g., hydroxycinnamoyl-derivatives), which are involved in pollen cell wall organization (Yang et al., 2007; Grienenberger et al., 2009) and fertilization (Elejalde-Palmett et al., 2015). Among these PAs-conjugated metabolites, hydroxycinnamoyl-amides (HCAs) are of great importance, since they modulate the strength of pollen cell wall, creating a bridge among cell wall molecules, especially lignins and hemicelluloses (Lam et al., 1992). Moreover, HCAs reduce the cytotoxicity of free hydroxycinnamic acids, due to their lipophilic character, known to inhibit growth (Kefeli et al., 2003).

Together with other metabolites, such as flavonol glycosides, HCAs are highly conserved in Angiosperm, conferring pollen coat structural characteristics (Fellenberg and Vogt, 2015). The amine moiety of HCAs derives from aromatic monoamines (such as tyramine, tryptamine) or aliphatic PAs (such as Put, Spd, Spm). In M. domestica HCAs are present in pollen coat (Elejalde-Palmett et al., 2015), and exhert also a protective role toward biotic and abiotic stressors.

Since the highest activity of TGase was detected at 48 h, at this time point PAs content was analyzed, showing a different pattern of PAs between AxA and AxW, with a higher level of free Put and Spm in compatible pollination compared to SI.

In Pyrus communis, previous results showed that PAs content changes during in vitro pollen germination, depending on both type of PA and their form (F, SC, IB) (Del Duca et al., 2010). According to these data, Put, Spd, and Spm were the only PAs detected in styles, as confirmed by recent bibliography about PAs in the early stages of fruit development (Quinet et al., 2019). This suggests a possible role for these compounds in the pollen-style interaction during the fertilization process (Aloisi et al., 2016a). These data are in agreement with the protective effect of PAs toward RNA, exerting an inhibitory role on the of S-RNase activity (Speranza et al., 1984; Li et al., 2018), which has been reported in different plant models (Altman, 1982; MccCre and Franklin-Tong, 2006; Wang et al., 2010; Li et al., 2018). PAs are also regulators of cytosolic Ca2<sup>+</sup> (Alcázar et al., 2010), leading to the above- mentioned rearrangements of the actin cytoskeleton mediated by TGase (Aloisi et al., 2017).

Contrary to free PAs pattern, AxA showed a three-fold increase in SC-Spm compared to AxW, allowing hypothesizing its involvement in pollen tube growth inhibition. In fact, SC-PAs represent the product of PAs linkage to HCAs or to amino acids

and/or peptides with MW < 5 kDa. The highest level of SC-Spm in AxA pollination model could affect the stiffening of pollen tube wall or of stylar cell wall, participating in the inhibition of tube growth. Moreover, the conjugation of PAs to HCAs, may represent a mechanism to counteract the cytotoxic effect due to HCAs (Kefeli et al., 2003).

The <sup>1</sup>H NMR metabolomic analysis, which allowed comparing the metabolic patterns of the three classes, showed that pollination increased the overall metabolism, inducing a consumption of stored glucose, and increasing production of fatty acids, amino acids and organic acids, including also GABA and p-hydroxybenzoyc acid. The shikimate pathway was also enhanced, resulting in an increased production of phenolics, in particular flavonoids, confirmed by the analysis of total flavonoid content showing a significant increasing of these metabolites in AxA samples.

Moreover, both pollinated styles showed an increased concentration of alkaloid trigonelline. This compound is a nicotinamide metabolite involved in plant cell cycle regulation and oxidative stress (Perchat et al., 2018). Differences were also found between compatible (AxW) and incompatible (AxA). In particular, AxA showed generally higher metabolites content than AxW, except for sucrose, which was more concentrated in AxW. Signals due to a cynnamoil-derivative appeared also increased in AxA. This can reflect the huge accumulation of SC-Spm observed in SI pollinated styles.

Recent data from literature report that flavonols enhanced pollen development by acting as ROS scavengers and reducing the abundance of ROS that occurred when plant are exposed to stress (Muhlemann et al., 2018). The observation that maize and petunia mutants lacking chalcone synthase (CHS) activity were not only deficient in flavonoids, but were also male sterile suggested that flavonoids might be involved in pollen fertility (Mo et al., 1992; Muhlemann et al., 2018). These results reveal that flavonol metabolites regulate plant sexual reproduction under both normal and stress conditions (i.e., high temperatures) by maintaining ROS homeostasis. When pollen of CHS-deficient plants land on stigmas of wild-type plants, the mutant pollen is partially functional; this is because stigma release a diffusible factor that allows pollen to be functional at pollination. Moreover, this evidence led to the isolation and identification of kaempferol as a pollen germination-inducing constituent in wild-type petunia stigma extracts (Mo et al., 1992; Muhlemann et al., 2018).

Being flavonoids antioxidant molecules (Muhlemann et al., 2018), their increased level in AxA in respect to AxW, might represent a response to oxidative stress, which takes place during SI (Serrano et al., 2015). Although ROS are necessary for pollen tube growth, a misregulated ROS homeostasis, leading to exceeding accumulation of these radical species, causes damage to cell structures (Daher and Geitmann, 2011; Speranza et al., 2012; Sewelam et al., 2016). Increasing PAs, which were also found in AxA, might be related to this ROS imbalance. In fact, exogenous PAs alter ROS levels in Pyrus pyrifolia (Wu et al., 2010; Aloisi et al., 2015) and during the SI response (Wang et al., 2010; Jiang et al., 2014).

### CONCLUSION

The results of this study show that pollination process produces an intense metabolic activity, as emerged from the high accumulation of primary and secondary metabolites in pollinated styles compared to not pollinated ones. Moreover, a clear difference in the metabolomics profile exists between compatiblepollinated and SI-pollinated styles. In particular, during the self-pollinated response TGase activity increases, as well as the level of polyamines conjugated to hydroxycinnamic acids and other small molecules, and the shikimate pathways is induced, resulting in an increased production of phenolics, in particular flavonoids. These could be overproduced in the attempt to overcome the oxidative stress occurring during the SI response. On the other hands, the conjugation of hydroxycinnamic acids with aliphatic polyamines can moderate the reported cytotoxicity of free hydroxycinnamic acids. Taken together, the activation of TGase and the increase in conjugated and bound forms of polyamines can contribute to cause the rigidification of the cell wall and the decrease in its extensibility, thus impairing the pollen tube growth, which is the final event in the SI response.

### DATA AVAILABILITY

All datasets for this study are included in the manuscript and the **Supplementary Files**.

### AUTHOR CONTRIBUTIONS

SD and FA designed the experiments. IA and SD produced the plant material. MM, FA, IA, GP, GC, LP, and CF performed the experiments, analyzed the data, and prepared the figures. MM, SD, FA, FP, and IA planned the research and interpreted the data. SD and MM wrote the manuscript with contributions from all other authors.

#### FUNDING

This work was supported by PRIN 2015 ISIDE (Investigating Self Incompatibility Determinants in fruit trees) (http://prin.miur.it/) to SD and GC.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2019.00741/ full#supplementary-material

#### REFERENCES

fpls-10-00741 June 12, 2019 Time: 19:21 # 12


different polyamine levels and transglutaminase activity. Amino Acids 38, 659–667. doi: 10.1007/s00726-009-0426-5



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Mandrone, Antognoni, Aloisi, Potente, Poli, Cai, Faleri, Parrotta and Del Duca. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.