# MYCORRHIZOSPHERE COMMUNICATION: MYCORRHIZAL FUNGI AND ENDOPHYTIC FUNGUS-PLANT INTERACTIONS

EDITED BY : Erika Kothe and Katarzyna Turnau PUBLISHED IN : Frontiers in Microbiology

#### Frontiers Copyright Statement

© Copyright 2007-2019 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-739-7 DOI 10.3389/978-2-88945-739-7

#### 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

## MYCORRHIZOSPHERE COMMUNICATION: MYCORRHIZAL FUNGI AND ENDOPHYTIC FUNGUS-PLANT INTERACTIONS

Topic Editors: Erika Kothe, Friedrich Schiller University Jena, Germany Katarzyna Turnau, Jagiellonian University in Krakow, Poland

Ectomycorrhizal root exemplifying the intimate association between plant and fungus in a mutually beneficial symbiosis. Image: Piotr Mleczko.

Cover image: Main background: Leszek Glasner/Shutterstock.com. Bottom right image: Piotr Mleczko.

The specific interactions of fungi with plants include the mutually beneficial mycorrhizal symbioses and an increasing number of case studies, where endophytic fungi communicate with their host plant to allow for beneficial interactions. The omics methods development has allowed for a substantial increase in knowledge that emphasized in many cases the intricate interplay between the symbiotic partners. In addition to the direct interactions, the mycorrhizosphere comes into view, as the fungal soil mycelium is interacting with the community outside the host plant, transferring signals also to the host. This Research Topic encompasses research on both major types of mycorrhizal interactions, endo- and ectomycorrhiza, and includes communication with the environment in which both partners interact with soil microbes. The mycorrhizosphere is in the center of molecular biology and modern ecological research, greatly fostered by the possibilities of genetic manipulation.

Citation: Kothe, E., Turnau, K., eds. (2019). Mycorrhizosphere Communication: Mycorrhizal Fungi and Endophytic Fungus-Plant Interactions. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-739-7

# Table of Contents

*06 Editorial: Mycorrhizosphere Communication: Mycorrhizal Fungi and Endophytic Fungus-Plant Interactions* Erika Kothe and Katarzyna Turnau

#### SECTION I

#### MYCORRHIZAL FUNGI

*10 Recent Insights on Biological and Ecological Aspects of Ectomycorrhizal Fungi and Their Interactions*

Antonietta Mello and Raffaella Balestrini

*23 Secretome Analysis From the Ectomycorrhizal Ascomycete* Cenococcum geophilum

Maíra de Freitas Pereira, Claire Veneault-Fourrey, Patrice Vion, Fréderic Guinet, Emmanuelle Morin, Kerrie W. Barry, Anna Lipzen, Vasanth Singan, Stephanie Pfister, Hyunsoo Na, Megan Kennedy, Simon Egli, Igor Grigoriev, Francis Martin, Annegret Kohler and Martina Peter


#### SECTION II

#### ENDOPHYTIC FUNGI


Madhunita Bakshi, Irena Sherameti, Doreen Meichsner, Johannes Thürich, Ajit Varma, Atul K. Johri, Kai-Wun Yeh and Ralf Oelmüller

#### SECTION III

#### PHYTOPATHOGENIC INTERACTIONS

*107 Genetic Diversity Studies Based on Morphological Variability, Pathogenicity and Molecular Phylogeny of the* Sclerotinia sclerotiorum *Population From Indian Mustard* (Brassica juncea)

Pankaj Sharma, Amos Samkumar, Mahesh Rao, Vijay V. Singh, Lakshman Prasad, Dwijesh C. Mishra, Ramcharan Bhattacharya and Navin C. Gupta

*125 Integrated Translatome and Proteome: Approach for Accurate Portraying of Widespread Multifunctional Aspects of* Trichoderma

Vivek Sharma, Richa Salwan, P. N. Sharma and Arvind Gulati

*138* Verticillium dahliae-Arabidopsis *Interaction Causes Changes in Gene Expression Profiles and Jasmonate Levels on Different Time Scales* Sandra S. Scholz, Wolfgang Schmidt-Heck, Reinhard Guthke, Alexandra C. U. Furch, Michael Reichelt, Jonathan Gershenzon and Ralf Oelmüller

#### SECTION IV

#### ENVIRONMENTAL INTERACTIONS

*157 Metabolomics Investigation of an Association of Induced Features and Corresponding Fungus During the Co-Culture of* Trametes versicolor *and*  Ganoderma applanatum

Xiao-Yan Xu, Xiao-Ting Shen, Xiao-Jie Yuan, Yuan-Ming Zhou, Huan Fan, Li-Ping Zhu, Feng-Yu Du, Martin Sadilek, Jie Yang, Bin Qiao and Song Yang

*171 How Does Salinity Shape Bacterial and Fungal Microbiomes of* Alnus glutinosa *Roots?*

Dominika Thiem, Marcin Gołębiewski, Piotr Hulisz, Agnieszka Piernik and Katarzyna Hrynkiewicz

*186 Two P1B-1-ATPases of* Amanita strobiliformis *With Distinct Properties in Cu/Ag Transport*

Vojtěch Beneš, Tereza Leonhardt, Jan Sácký and Pavel Kotrba

# Editorial: Mycorrhizosphere Communication: Mycorrhizal Fungi and Endophytic Fungus-Plant Interactions

Erika Kothe<sup>1</sup> \* and Katarzyna Turnau<sup>2</sup>

1 Institute of Microbiology, Friedrich Schiller University, Jena, Germany, <sup>2</sup> Institute of Environmental Sciences, Jagiellonian University in Krakow, Kraków, Poland

Keywords: mycorrhiza, endophyte, fungus-plant interactions, mycorrhizophere, communication

#### **Editorial on the Research Topic**

**Mycorrhizosphere Communication: Mycorrhizal Fungi and Endophytic Fungus-Plant Interactions**

#### PLANT MYCOBIOMES

#### Edited by: Anna Maria Pirttilä,

University of Oulu, Finland

Reviewed by: Tamás Papp, University of Szeged, Hungary

> \*Correspondence: Erika Kothe erika.kothe@uni-jena.de

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 12 September 2018 Accepted: 22 November 2018 Published: 05 December 2018

#### Citation:

Kothe E and Turnau K (2018) Editorial: Mycorrhizosphere Communication: Mycorrhizal Fungi and Endophytic Fungus-Plant Interactions. Front. Microbiol. 9:3015. doi: 10.3389/fmicb.2018.03015 Plants do not exist as single entities but should rather be considered to form a complex community with microbes and other organisms where plant tissues form diverse niches for microbes. One major relationship concerns plant-fungal interactions that range from pathogenicity to mutually beneficial symbioses. A balanced state (homeostasis) of these interactions is essential for maintaining the plant as well as an overall healthy state of the environment. Mycorrhizal associations are well-studied examples of root-fungal mutually beneficial symbiosis (Ferlian et al., 2018; Gehring and Johnson, 2018). The reasons for establishing a mutual symbiosis are only just beginning to be understood at the molecular level (Mello and Balestrini). Communication between endo- and ecto-mycorrhiza and their respective host plants (Raudaskoski and Kothe, 2015; Luginbuehl and Oldroyd, 2017; Garcia et al., 2018, and citations therein) and the effects on phytohormone levels and localized delivery (see Boivin et al., 2016; MacLean et al., 2017) have been the focus of several recent reports. But even so, a full understanding of these relationships will only be gained by investigating the effects of different strains of the same fungal species (Sharma et al.).

For roots, as well as above-ground plant tissues, endophytic fungi can be considered as examples of specific co-evolution, provided the term "endophytic" is used in its sensu strictu (for a detailed comment, see Kothe and Dudeja, 2011). To prove endophytic behavior, Koch's postulates need to be observed, and tissue specificity for re-infection may be used as a method to discriminate real endophytes from mere co-occurrence (We¸zowicz et al., 2017; Domka et al., 2018; Wa ˙ zny ˙ et al., 2018). The traits of endophytes that do not lead to symptoms in a healthy plant clearly delineate them from phytopathogenic fungi, however, caution is necessary because their effect on the symbiosis can vary with the species/variety of the partner and environmental conditions. For endophytic fungi, knowledge is much more limited as compared to mycorrhiza, although a role for strigolactone signaling is presented by Rozp ˛adek et al.

Temporal shifts in plant-associated fungal populations are known to occur. An example from mycorrhizal symbiosis is for young trees with endomycorrhizal symbionts that are later replaced by specific ectomycorrhizal associations (Knoblochová et al., 2017; Bachelot et al., 2018). Within 4 weeks, vesicles and hyphae are visible in the roots of Picea abies and Pinus sylvestis leading to increased main-root development and up to a 300% increase in secondary roots. These effects alone will increase the potential at a later stage for formation of ectomycorrhiza, which is the only form of mycorrhiza seen in mature pine and spruce. After the ectomycorrhiza is established, a succession of fungal partners appears. First, fast-growing, broad host-range, reproductionstrategy fungi are attracted; later, slower growing, but more specific, ectomycorrhizal fungi, acting by capacity strategy, are recruited. For spruce, that would lead to replacement of, e.g., Cenococcum geophilum (see de Freitas Periera et al.) or Pisolithus tinctorius by host-specific fungi like T. vaccinum.

#### MULTI-OMICS IN FUNGAL-PLANT MOLECULAR COMMUNICATION

The mycobionts and their hosts will constantly communicate to establish and maintain the symbiosis. Signals are perceived and result in changes in gene expression. With excreted proteins or metabolites, the partner is stimulated. A multi-level interaction thus will be visible with changed transcriptome, proteome, and metabolome patterns. These can be visualized with techniques such as transcriptomics (Fiorilli et al., 2016; Nagabhyru et al., 2018), proteomics (Sebastiana et al., 2017; Shrivastava et al., 2018), metabolomics (Hill et al., 2018; Maciá-Vicente et al., 2018), or combinations thereof (e.g., Larsen et al., 2016). Since secreted proteins may be important for the signal exchange, secretomics can be expected to identify effector proteins exchanged between symbiont and host (Doré et al., 2015; Wagner et al., 2015). Volatiles exchanged address a subgroup of metabolites for signal exchange (Ditengou et al., 2015; Pistelli et al., 2017). And a combination of translatome and proteome data (Sharma et al.), as well as multi-omics have proven to substantially improve the quality of prediction for the symbiotic molecular network (Vijayakumar et al., 2016).

Interactions between symbiotic fungus and plant in production of secondary metabolites (Ludwig-Müller, 2015; González-Menéndez et al., 2016) shows the intricate relationship between the symbiotic partners. In addition to the dual interaction, the communication is expanded with additional fungi or bacteria which co-occur in the environment.

#### MULTI-PARTNER INTERACTIONS

Usually, plant and fungus are not alone in the partnership, but additional interactions with bacteria or other fungi will influence the outcome of these associations in nature. This complex relationship is reflected in the concept of the mycorrhizosphere, where plant roots and hyphae of the mycorrhizal partner encounter other soil microorganisms. These additional interactions in the vicinity of the root need to be considered in studies of cross-talk (Vannini et al., 2016; Wagner et al., 2016).

Phytohormones produced in the mycorrhizosphere may alter the physiology of the symbiotic partners and aid formation of new mycorrhiza (Wagner et al., 2015). For example, the ectomycorrhizal Tricholoma vaccinum produces the auxin indole-3-acetic acid (IAA), which stimulates mycorrhization (Krause et al., 2015). In addition, the exogenous presence of the phytohormone promotes branching, which leads to an increased Hartig-net formation during symbiosis (Krause et al., 2015). Moreover, ectomycorrhization can be reversed by associated fungi that produce IAA-inhibiting compounds (Hause and Saarschmidt, 2009). With soil zygomycetes, a tripartite interaction occurs in which the zygomycete-derived metabolite, D-orenone, induces a transporter that allows for increased excretion of IAA by the mycorrhizal fungus, T. vaccinum (Wagner et al., 2016). Soil bacteria are also capable of auxin biosynthesis, mostly upon tryptophan induction via root exudates. Since IAA increases branching of ectomycorrhizal fungi, some have been termed mycorrhiza-helper bacteria (Frey-Klett et al., 2016). It is now clear that multi-partner communication systems have evolved and are present in any habitat.

Like symbionts, phytopathogens also act via phytohormes, e.g., by inducing systemic-acquired resistance in host plants (Scholz et al.). Similar mechanisms of cross-talk can be inferred from comparison of pathosystems to endophytic or mycorrhizal symbiosis.

### MOLECULAR RESPONSES TO ENVIRONMENTAL STRESSES

In addition to biotic interactions shaping the plant mycobiome, abiotic factors certainly influence the mutual, commensal or pathogenic interaction. Environmental conditions may influence plant mycobiomes, and these effects are more likely observed under detrimental conditions, such as nutrient limitation, drought, salinity, or other stresses, for example metal toxicity (Kumar and Verma, 2018; Shi et al., 2018). The molecular background of stress recognition and signal transduction within endomycorrhiza is reported by Sun et al. and an example of endophytes altering phosphate mobility in the mycorrhizosphere is given by Baum et al. The molecular background of stress recognition and signal transduction with an endomycorrhizal association is reported by Sun et al. as is an example of endophytes changing phosphate mobilities in the mycorrhizosphere (Baum et al.).

Environmental stresses, like metal contamination in the ground, to name just one example, can be buffered by mycorrhizal and endophytic associations. A molecular role has been shown for hydrophobins, small amphipatic proteins that decorate the cell wall of air-exposed mycelium. A study by Sammer et al. on the up-regulation of different hydrophobins during the life cycle of the mycorrhizal fungus exposed to metal contamination illustrates their protective effect. In that study, the biotic interaction is also characterized by volatiles and exudates from the host tree inducing mycorrhiza-associated hydrophobin genes (Sammer et al., 2016).

Transporter production, as well as intracellular storage of metals, illustrate specific adaptive responses in fungi. Many fungi are able to carry increased metal loads if grown on metalcontaminated substrate. A prominent example is the mushrooms that showed high cesium content when collected from the fallout areas after the Chernobyl accident. A molecular explanation is now available for this observation (see Benes et al.). For metals that are not only toxic, but essential for growth at lower concentrations, e.g., zinc, require the presence of a specific uptake and storage system. Indeed, for zinc, such a system has been described with the ectomycorrhizal fungus Suillus luteus (Coninx et al.). As an additional twist, there are fungi that influence mutual symbioses of bacteria with plants. An example of this is shown by Thiem et al. where the influence of microbiome, including the mycobiome, on the actinorrhizal interaction between Frankia and alder (Alnus glutinosa) under salinity is demonstrated.

#### APPLICATIONS FOR GROWING DEMANDS ON SUSTAINABLE AGRICULTURE AND FORESTRY

Anthropogenic impact, including industrial pollution and both conventional and organic agriculture, has already affected the soil microbiome, leading to decreases in soil quality and the nutritional value of crops. In doing so, it has created the necessity to use a range of chemicals, such as fertilizers and pesticides, to avoid the spread of unwanted pathogenic microbes. These substances not only affect plant-microbial foes but also the friendly microbes that help the plant to establish homeostasis and attain the nutritional quality of products that support the health of the consumers. The twenty first century brings us possibilities to develop new and innovative methods to rebuild soil tilth and to renovate the plant microbiota (see, e.g., Verzeaux et al., 2017; Campos et al., 2018). However, to be successful, we need an increased understanding by both food and wood producers

#### REFERENCES


on the molecular communication between fungi and the host plant, resulting in competitive advantages specifically under abiotic stress. The results may provide solutions for the problems aggravating sustainable agriculture and forestry, especially under the ever-changing environmental conditions (Shinde et al., 2018).

The re-introduction of protective microbes can be achieved through bioaugmentation strategies, which allow them to create optimal conditions for crop growth under harsh conditions (see Treu and Falandysz, 2017) and a reduction in the use of water, fertilizers, and pesticides. The most common plant inhabitants are endophytes that, when properly selected, can be a potent tool against pathogens and abiotic factors (see French, 2017). They also support mycorrhizae, which in turn contribute to plant growth and induce tolerance to salinity, pollution, drought, extreme temperatures, elevated CO2, etc. (see, e.g., Dhawi et al., 2017). The provision of well-selected microbes and the application of appropriate agricultural/forestry practices, both of which are feasible now, can decrease the current intensive use of fertilizers while maintaining an environment that is conducive to human health. Although nowadays the interaction of fungi and bacteria with plants is better understood, there is still a need for greater insight into the interplay between bacteria and fungi, fungi with other fungi, and their interactions with plants. Furthermore, building stronger bridges between bacteriologists and mycologists will help to benefit from their complementary skills, as exemplified by the work on the roles of strigolactones (see, De Cuyper and Goormachtig, 2017).

#### AUTHOR CONTRIBUTIONS

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


**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 © 2018 Kothe and Turnau. 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.

# Recent Insights on Biological and Ecological Aspects of Ectomycorrhizal Fungi and Their Interactions

#### Antonietta Mello\* and Raffaella Balestrini

Institute for Sustainable Plant Protection (IPSP), Torino Unit, National Research Council, Turin, Italy

The roots of most terrestrial plants are colonized by mycorrhizal fungi. They play a key role in terrestrial environments influencing soil structure and ecosystem functionality. Around them a peculiar region, the mycorrhizosphere, develops. This is a very dynamic environment where plants, soil and microorganisms interact. Interest in this fascinating environment has increased over the years. For a long period the knowledge of the microbial populations in the rhizosphere has been limited, because they have always been studied by traditional culture-based techniques. These methods, which only allow the study of cultured microorganisms, do not allow the characterization of most organisms existing in nature. The introduction in the last few years of methodologies that are independent of culture techniques has bypassed this limitation. This together with the development of high-throughput molecular tools has given new insights into the biology, evolution, and biodiversity of mycorrhizal associations, as well as, the molecular dialog between plants and fungi. The genomes of many mycorrhizal fungal species have been sequenced so far allowing to better understanding the lifestyle of these fungi, their sexual reproduction modalities and metabolic functions. The possibility to detect the mycelium and the mycorrhizae of heterothallic fungi has also allowed to follow the spatial and temporal distributional patterns of strains of different mating types. On the other hand, the availability of the genome sequencing from several mycorrhizal fungi with a different lifestyle, or belonging to different groups, allowed to verify the common feature of the mycorrhizal symbiosis as well as the differences on how different mycorrhizal species interact and dialog with the plant. Here, we will consider the aspects described before, mainly focusing on ectomycorrhizal fungi and their interactions with plants and other soil microorganisms.

#### Keywords: ectomycorrhizae, plant–microbe interactions, symbiosis, cell wall, mycorrhizal fungi

### INTRODUCTION

The roots of most terrestrial plants are colonized by mycorrhizal fungi. They play a key role in terrestrial environments providing to plants an improvement in mineral nutrient uptake and earning in return carbon compounds (Brundrett, 2009). Mycorrhizal interactions are usually classified on the basis of the features of the symbiotic interfaces and of the taxonomic identity of the plant and fungal symbionts (Smith and Read, 2008). Among mycorrhizal symbioses

#### Edited by:

Erika Kothe, Friedrich Schiller University Jena, Germany

#### Reviewed by:

Mika Tapio Tarkka, Helmholtz-Zentrum für Umweltforschung (UFZ), Germany Gwen-Aelle Grelet, Landcare Research, New Zealand

> \*Correspondence: Antonietta Mello antonietta.mello@ipsp.cnr.it

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 17 October 2017 Accepted: 30 January 2018 Published: 15 February 2018

#### Citation:

Mello A and Balestrini R (2018) Recent Insights on Biological and Ecological Aspects of Ectomycorrhizal Fungi and Their Interactions. Front. Microbiol. 9:216. doi: 10.3389/fmicb.2018.00216

(see van der Heijden et al., 2015 for a review), ectomycorrhizae are established by the mycelia of fungi almost exclusively belonging to the so called "higher fungi," i.e., Basidiomycetes and Ascomycetes, whose ecological strategies have been revisited by Tedersoo and Smith (2013). Ectomycorrhizal (ECM) fungi are present all over the world, and their host plants include most angiosperm and gymnosperm trees, as well as shrubs (Bonfante, 2010). Some ECM plants are economically important timber-producing tree species, while some ECM fungi are represented by the economically important truffles and porcini (Mello et al., 2015a). Between plant and soil there is a very specific environment, the ectomycorrhizosphere, in which diverse communities of microorganisms – fungi and bacteria – interact. It is known that ECM fungi have a key role in nitrogen cycling, particularly in boreal and temperate forests, and that they can help their host plants to tolerate abiotic stresses. ECM assemblages provide benefits for inorganic nitrogen uptake under environmental constraints, through stress activation of distinct ECM fungal taxa. This suggests that these taxa are functionally diverse and opens new opportunities to characterize the ECM fungal identities (Pena and Polle, 2014). Furthermore, Gehring et al. (2017) demonstrated that tree genetics determines fungal partner communities that confer drought tolerance, highlighting the interlinked importance of the genetics of a tree and its microbiome.

The development of an ECM symbiosis requires morphological changes in the two partners, to allow the formation of the symbiotic structures, through the regulation of several genes (Martin et al., 2007; Kohler et al., 2015). From the first work in which cDNA arrays were used to study gene expression in the ECM symbiosis between Eucalyptus globulus and Pisolithus tinctorius (Voiblet et al., 2001), important progress has been done in the comprehension of the mechanisms involved in the ectomycorrhiza development. Information on the functional diversity of the ECM interactions has been highlighted, leading to the discovery of many genes coding for plant/fungus symbiosis-regulated proteins. Among them, several mycorrhiza-induced small-secreted proteins (MiSSPs) that may act as effectors and are required for symbiosis establishment have been identified (Plett et al., 2011, 2014a; Kohler et al., 2015; Martin et al., 2016 for a review). Additionally, Pellegrin et al. (2015) showed, through a bioinformatics pipeline, that the secretome of ECM fungi is enriched in SSPs in comparison to other species with a different life style. Shared- and lifestylespecific SSPs have been identified in saprotrophic and ECM fungi, and the ECM-specific SSPs could be a signature of the ECM symbiosis lifestyle. This would suggest they have a role in a molecular dialog with host plants, leading to the formation of a functional ectomycorrhiza (Pellegrin et al., 2015; Garcia and Ané, 2016 for a commentary).

Despite similar anatomical patterns, the sequenced ECM genomes showed that differences are present in symbiosis regulated genes, revealing a diversity in the manner by which symbiotic fungi interact with their partners and suggesting the use of different molecular toolboxes to dialog with the host plant (Martin et al., 2008, 2010; Kohler et al., 2015; Peter et al., 2016). Remarkably, the role of MiSSPs (such as MiSSP7) to control host plant defense reactions has been elegantly demonstrated in Laccaria bicolor and Populus trichocarpa interaction (Plett et al., 2011, 2014b), while such fungal effectors have not been found among the upregulated transcripts in Tuber melanosporum ECMs (Martin et al., 2010), suggesting that different mechanisms may be involved in the development and maintaining of the ECM symbiosis. Transcript profiling of ECM roots from different plant/fungus interactions suggests that similar functional gene categories appear to be up-regulated, although these genes are not the same in the several ECM fungal species (Kohler et al., 2015). The availability of more genome sequences from ECM fungi also confirms that they have a reduced set of genes encoding plant cell wall degrading enzymes (PCWDEs) (Kohler et al., 2015), as already suggested from the genome sequence of the first two sequenced mycorrhizal fungi, i.e., L. bicolor and the black truffle T. melanosporum, respectively (Martin et al., 2008, 2010). In addition to genomic features and transcriptomic profiles, epigenetic variation is considered an important player in the evolution of biological diversity, and epigenetic regulatory systems have an important role in the response to environmental stimuli and stress factors (Zhong, 2016). The availability of the genome sequences from several fungal species will allow the understanding on how DNA methylation regulatory components are evolved in ECM fungi, and the role of the epigenetic mechanisms to cope with different environmental conditions through modifications of gene expression mediated by DNA methylation and transposon activity profiles. Considering that DNA methylation in fungi lead to transposable elements (TEs) silencing, comparative methylome and transcriptome analyses have been performed in a TE-rich organism such as the ECM fungus T. melanosporum (Chen et al., 2014), suggesting that a reversible methylation mechanism functions in truffles to cope with the multitude of TEs present in its genome. Information derived from these analyses, whether extended to individuals from different geographical areas, may also provide a new tool to explain intraspecific variability and adaptation to different environments and, in the case of truffles, commercially organoleptic properties (i.e., aroma).

An ECM root is a complex organ, formed not only by two individuals, plant and fungus, but also by two fungal pseudotissues: the mantle (i.e., the sheat), which develops outside the root, and the Hartig net, which colonizes the apoplastic space between root cells (Balestrini et al., 2012; Balestrini and Kottke, 2016). The two ECM compartments are thought to be functionally different. This has been first demonstrated by a study on Amanita muscaria ectomycorrhizae, where the mantle was manually dissected from the ectomycorrhizal root, revealing a differential expression for two fungal genes coding for a phenylalanine ammonium lyase (AmPAL) and a hexose transporter (AmMst1) (Nehls et al., 2001). While the first (AmPAL) was mainly expressed in the mantle, the expression of AmMst1 increased in the Hartig net. More recently, the combination of a laser microdissection (LMD) approach, which allows the collection of the two ECM fungal compartments, with microarray gene expression analysis, revealed a specificity in the transcript profiles, reflecting a functional specificity for these two ECM compartments, e.g., that the mantle is the

responsible for the mineral elements (i.e., nitrogen) and water uptake from soil, while the expression of several transporters is enhanced in the Hartig net (Hacquard et al., 2013). In the last years, different reviews have been focused on the molecular signals (mechanisms) underlying the ECM development and functioning (Garcia et al., 2015; Martin et al., 2016). A role of flavonoids and hormones in the signaling pathway during the early stages of the ECM establishment has been proposed since several years (Garcia et al., 2015 and references therein). More recently, two plant flavonoids have been suggested to trigger the expression of a fungal effector (MiSSP7, see below) in L. bicolor (Plett et al., 2014a). It has been also reported that accumulation at the root apex and redistribution of auxin, which is a hormone produced by both the symbiotic partners, may play a role to stimulate lateral short root development required for the ECM formation (Felten et al., 2009). Martin et al. (2016) have recently speculated that secreted fungal MiSSPs may interact with auxin, gibberellin, and salicylate receptors to alter root development. Moreover, increased concentrations of ethylene and jasmonic acid repressed fungal colonization, with an impact on the development of the ECM roots (Plett et al., 2014b). However, the effective involvement of plant hormones in ECM establishment and maintenance has to be still fully elucidated (Garcia et al., 2015).

Here, some specific aspects related to the biology and ecology of the ECM fungi will be considered, starting with their in situ dynamics to the symbiotic interface creation, before and after their genome sequencing and the advent of the environmental genomics.

Given that truffles are of high economic interest, crossing several research fields ranging from taxonomy to truffle cultivation, and are the first edible ECM fungi whose genome has been sequenced, extensive research has been focused on them in order to understand their life cycle and thus to increase their production. For this reason, particular attention will be given to some insights highlighted by the sequencing of the black truffle T. melanosporum genome.

### IDENTIFICATION OF ECTOMYCORRHIZAL FUNGI: FROM THE PAST TO THE PRESENT

The identification of ECM fungi has generally been focused on the macro- and microscopic examination of fruiting bodies and only since the early 1990s these fungi have also been characterized by DNA-based methods. At the beginning, most of the identification of fruiting bodies has involved restriction analyses of the internal transcribed spacer (ITS) region producing ITS-RFLP database from sporocarp samples (Horton and Bruns, 2001). The next step has been the direct sequence analysis of the ITS region and its deposit in GenBank or EMBL. Specific primers have been then developed for the identification of many fungal species upon increase of sequences number (Gardes et al., 1991). Amicucci et al. (1998) designed ITS primers for the identification of five species of white truffles, T. magnatum Pico, T. borchii Vittad., T. maculatum Vittad., T. dryophilum Berk. & Br. and T. puberulum Berk. & Br. that have similar morphological characteristics, but different organoleptic qualities and economic value. At this regard, Mello et al. (2006) designed ITS primers for the identification of the marketable boletes Boletus edulis Bull.: Fr. sensu stricto, B. aereus Bull.: Fr, B. pinophilus Pila ìt et Dermek and B. aestivalis Fr. (all classified as B. edulis s.l.), which are hardly distinguishable on the basis of their morphology and considered as the most frequently eaten fungi among those harvested in natural conditions in Europe. Once the molecular tools as sequencing and specific primers have been available, they have allowed typing the ECM tips, usually after sorting these in morphotypes. This method has, thereafter, been used in many studies on EM community structure and spatial distribution since the pioneering work of Gardes et al. (1991). According to Kaldorf et al. (2004), roughly 90% of all ectomycorrhizas of aspen clones in experimental fields was represented by Cenococcum geophilum, Laccaria sp., Phialocephala fortinii, two different Thelephoraceae, and one member of the Pezizales. Murat et al. (2005) sorted 335 mycorrhizal root tips collected in a truffleground into 39 morphotypes, on the basis of color, mantle shape and surface texture, presence of cystidia, and EM branching pattern, providing novel information on the ectomycorrhizal and endophytic species living in a T. magnatum truffle-ground. Above all, the finding of the few T. magnatum mycorrhizae in a non-productive period for this fungus, and in a nonproductive area, suggested that there is not a direct linkage between mycorrhizae and fruiting bodies. Since mycorrhizal networks permit interactions among trees, their architecture has been investigated by multi-locus microsatellite DNA, leading to the identification of the trees and fungal genets connected in a multi-aged old-growth forest (Beiler et al., 2010). In order to study functional diversity among ECM fungi in situ, the activities of enzymes involved in the degradation and nutrient release from soil organic matter have been used (Courty et al., 2010; Pritsch and Garbaye, 2011). Combining enzymatic activities and stable isotope assays of root tips Tedersoo et al. (2012) have tried to assess the functional aspects of tropical ECM fungi. This study demonstrates that the ECM fungus may affect both potential enzymatic activities and δ <sup>15</sup>N patterns of ECM tips in relation to phylogeny and exploration type (i.e., contact, short distance, medium-distance fringe and long-distance types; cf. Agerer, 2001).

As each ECM species is specialized in exploiting specific resources of the soil ecosystem, investigations have been thereafter focused on the spatial distribution of the extraradical mycelium. It interconnects plant rootlets in the forest ecosystems, forming the 'wood wide web' (Martin et al., 2016). Tracking the distribution of a given ECM fungus is considered difficult, since fruiting bodies do not reflect the distribution of ground networks (Dahlberg, 2001). The detection of the mycelium in soil has been possible thanks to the advent of new methods that have led to the direct extraction and amplification of DNA from this environment. Hebeloma cylindrosporum was the first ECM fungus to be detected in soil, within 50 cm from the fruiting bodies (Guidot et al., 2002). Using the β-tubulin gene as a marker, Zampieri et al. (2010) could show that the mycelium of T. magnatum is more widespread than was inferred

from the distribution of its fruiting bodies and ectomycorrhizae. Thanks to the progress of real-time PCR techniques, that has been optimized to quantify ECM mycelium of several ECM fungi (Iotti et al., 2014), De la Varga et al. (2012) quantified B. edulis extraradical mycelium in a Scots pine forest and found positive correlation between the concentration of mycelia and the presence of mycorrhizae of B. edulis, but not with the productivity of fruiting bodies, in the investigated samples. Given that knowledge of the annual dynamics of the mycelium of ectomycorrhizal fungi in forests soils is important in the carbon cycle, the mobilization of soil nutrients, and in the interactions of different components (plants, fungi, microfauna, and microorganisms) of the ecosystem, De la Varga et al. (2013) investigated with the same approach, the seasonal dynamics of B. edulis and Lactarius deliciosus extraradical mycelium in pine forests of central Spain. Soil mycelial dynamics of both species resulted to be strongly dependent on the weather, with an increase of biomass during the coldest months of the year.

Once it has been possible detecting mycelium of a given species and tracing its distribution, research has moved toward genet localization not only of sporocarps but also of the subterranean parts, i.e., the ectomycorrhizae and the extraradical mycelia. Studies based on the analysis of the genet structure of sporocarps, have proved that early stage fungi such as Hebeloma and Laccaria formed many small genets, and late stage fungi such as Cortinarius formed a few large genets (Hirose et al., 2004). Using a polymorphic microsatellite marker specific for Suillus grevillei, Zhou et al. (2001) demonstrated that the development of S. grevillei sporocarps is correlated with that of extra-radical mycelia and ectomycorrhizae of the same genet, which are distributed in a narrow area, however, no S. grevillei mycelia and mycorrhizae were detected close to the area where S. grevillei sporocarps emerged in the previous year, thus suggesting that subterranean genets change location year after year. Also Guidot et al. (2001) found a spatial congruence of above- and belowground distribution for H. cylindrosporum, and the tendency of the same genets to be dominant above and below ground. Interestingly, it has been possible proving the interconnection between a single genet and different plants. At this regard, Lian et al. (2006) showed that each genet detected in the mycelial mats of Tricholoma matsutake colonized from three to seven trees in a natural Pinus densiflora forest.

All these studies have been focused on the individual recognition of ectomycorrhizal fungi clarifying many aspects of their population biology (for a review see Dowhan et al., 2011). The introduction of high-throughput sequencing techniques and the suitability of studying (micro)organisms directly in situ (metagenomics or environmental genomics) has provided new information on ECM fungal communities by 'barcodes' of ITS regions in several biomes/ecosystems, e.g., tropical African forests (Tedersoo et al., 2010); Swedish spruce plantations (Wallander et al., 2010); truffle grounds (Mello et al., 2011); transgenic poplar plantations (Danielsen et al., 2012); ECM roots in the Svalbard (Blaalid et al., 2012, 2014); an urban landscape (Lothamer et al., 2014); boreal and tropical forests (Clemmensen et al., 2013; McGuire et al., 2013a); a forest dominated by oaks in Japan (Toju et al., 2013); Pinus sylvestris-dominated plots across three study areas in Estonia (Hiiesalu et al., 2017).

In parallel with the development of the next-generation sequencing systems such as 454 Genome sequencer (introduced in 2005, it uses real-time sequencing-by-synthesis pyrosequencing technology), the Illumina platform (utilizes a sequencing-by-synthesis approach coupled with bridge amplification on the surface of a flow cell) and Ion Torrent PGM (relies on the real-time detection of hydrogen ion concentration, released as a by-product when a nucleotide is incorporated into a strand of DNA by the polymerase action), new tools such as FUNGuild, have been developed to taxonomically parse fungal OTUs by ecological guild independent of sequencing platform or analysis pipeline. Using a database and an accompanying bioinformatics script, Nguyen et al. (2015) demonstrated the application of FUNGuild to three high-throughput sequencing datasets from different habitats: forest soils, grassland soils, and decomposing wood. Several pipelines provided as web services have been produced for processing fungal ITS metabarcoding using 454-sequenced amplicons, such as CLOTU (Kumar et al., 2011), SCATA<sup>1</sup> , PLUTOF (Abarenkov et al., 2010). Once the Illumina MiSeq platform for fungal metabarcoding has become very popular (starting with research by Bokulich et al., 2013; McGuire et al., 2013b; Schmidt et al., 2013), Bálint et al. (2014) developed a pipeline for cleaning up fungal ITS metabarcoding data generated on this platform.

An investigation of community–environment relationships in truffle grounds of the ECM fungus T. melanosporum, sampled in two areas, one devoid of vegetation (known as brulé in French and where fruitingbodies of this fungus are usually collected), and outside the brulé, has shown that Ascomycota were the dominant phylum in the brulé, and that their number decreased moving from inside the brulé to outside, while the number of Basidiomycota increased (Mello et al., 2011). Furthermore, this work provides comparison of the two ITS regions, ITS1 and ITS2, for fungal communities assessment. Changes in ECM fungal communities have been registered in many investigations (Mello et al., 2015b). Hui et al. (2011) have observed, in a Finnish forest, that a long-term exposition to Pb contamination can result in a shift in the composition of the ECM community associated with P. sylvestris L., as well as an increase in the abundance of the OTUs corresponding to the Thelephora genus and a decrease in the frequency of OTUs assigned to Pseudotomentella, Suillus, and Tylospora. In the Siberian tundra Gittel et al. (2013) have verified a decrease in ECM fungi abundance and an increase of bacteria in buried soils because of the low temperature and anoxia of these sites. ECM fungal communities of a temperate oak forest soil resulted to be affected by seasonality and soil depth (Voˇrìšková et al., 2014). Rincón et al. (2015) revealed a highly compartmentalized and contrasted response of fungal communities of Pinus sylvestris in France and Spain with different response of fungal sub-assemblages in soil vs. roots and lifestyle. High-throughput sequencing analysis of fungal communities in temperate beech forests in Germany showed that distance decays of soil-inhabiting and root-associated fungal assemblages differ,

<sup>1</sup>http://scata.mykopat.slu.se/

and identified explanatory environmental variables (Goldmann et al., 2016).

Although many sophisticated bioinformatics tools are available, high-throughput assessment of species richness and evenness in a fungal community (including ectomycorrhizal fungi) remains still a technical challenge, because of the methodological biases and the limitations of markers. According to Lindahl et al. (2013) the major benefit of high-throughput methods relies in their capacity to unearth the main fungal colonizers in large numbers of samples, since not always singletons represent authentic rare taxa. A global soil sampling in 365 sites across the world, followed by DNA metabarcoding, revealed representatives of all major phyla and classes of fungi in all ecosystems but with relative proportions variable several fold across biomes, in addition to several deeply divergent class-level fungal lineages that had not yet been described or previously sequenced (Tedersoo et al., 2014). The overall richness of soil fungi increased toward the equator, however, functional differences were observed between fungal communities in forested and tree-less ecosystems. In fact, richness of ECM fungi head a peak at mid latitudes, especially in temperate forests and Mediterranean biomes of the Northern Hemisphere, in accordance with the dominance at mid latitudes of Pinaceae, which is the oldest family of ECM plants (**Figure 1**). Data from this paper clearly determine climatic factors as the main drivers of fungal diversity and community composition, and greatly advance our understanding of global fungal diversity patterns. Beside this, they alert us on the impact of climate change on the consequences of altered soil microorganism communities and highlight the lack of data from understudied tropical and subtropical ecosystems (Tedersoo et al., 2014). Moving from a global scale analysis to the tripartite associations between roots, fungi and bacteria, that are known to influence plant health and growth (Bonfante and Anca, 2009), Marupakula et al. (2017) identify the different bacterial communities associated with different types of ECM associations, providing clue for more detailed functional studies of specific combinations of ECM fungi and bacteria. Furthermore the study shows as N additions impact fungal–bacterial interactions at the ectomycorrhizal root tip level in different soil horizons, likely influencing patterns of carbon allocation to roots. The study of microbe–microbe interactions is recently also taking advantage of a combination of -omics with direct process measurements (e.g., stable isotope probing 'SIP') to map functions and relationships in complex communities. Musat et al. (2016) review how is might be possible now to track microbial interactions with NanoSIMS (Nano secondary ion mass spectrometry), that has the potential to provide quantitative measures of organic matter-mineral-microbial interactions and biogeochemical processing at the macro- and microaggregate or single-cell scale. Regarding this approach, Worrich et al. (2017) demonstrate that mycelium-forming fungi and oomycetes provide nitrogen, carbon and water to bacteria in dry and oligotrophic environments, thus helping them and contributing to ecosystem functioning in stressed conditions.

Anyway, the parallel increase of data output from highthroughput sequencing and of databases of entire genomes will move the research toward direct analysis of meta-genomes and meta-transcriptomes of complex fungal communities (Kuske and Lindahl, 2013). However, Marmeisse et al. (2013) pointed out that ribosomal genes, whose copy number is potentially variable, do not reflect the real pool of organisms in a community and auspicate more rigorous methodologies to assess the ecological questions related to fungal communities. Another limit of DNA metabarcoding is that the number of sequenced species is still a limitation to the precise taxonomic identification of soil fungal sequences.

So far, the meta-genomes studied, included the soil metagenome, have isolated only bacterial and archaeal genes but not eukaryotic ones (Marmeisse et al., 2017). These authors clearly explain in their review that the reason of this lack is due to the dilution of eukaryotic genes of interest in the total metagenome, the large size of eukaryotic genomes and the strategy of cloning and expressing environmental DNA in bacteria. Besides listing the reasons to care about eukaryotic environmental nucleic acids (most of commercial enzymes and metabolites come from eukaryotes; eukaryotes are highly diverse at both local and global scales and have diverse gene repertoires), Marmeisse et al. (2017) review how metatranscriptomics is an alternative approach to access to environmental eukaryotes, starting with the isolation of their mRNAs that are distinguishable from those of bacteria having the poly-(A) tails lacking in bacteria messages. The systematic sequencing of eukaryotic metatranscriptomes from forest soil has been first applied by Damon et al. (2012), so resulting to be still in its infancy. Although most sequences coming from this approach cannot be affiliated to any taxon, it provides functional data despite of the barcoding of soil communities, in addition to unique products of biotechnological interest. A novel fungal family of oligopeptide transporters has been identified by functional metatranscriptomics of soil eukaryotes by Damon et al. (2011). Within the framework of the DOE Joint Genome Institute Community Sequencing Program, a challenging largescale metatranscriptomics project - 'Metatranscriptomics of Forest Soil Ecosystems'<sup>2</sup> – to explore the interaction of forest trees with communities of soil fungi, including ectomycorrhizal symbionts, has started. This project aims to sequence the metatranscriptome of soil fungi of ecosystems from the boreal, temperate and mediterranean forests. But fungal ecology is also taking benefits from environmental proteomics (Bastida et al., 2009; Schneider et al., 2012; Zampieri et al., 2016). This approach provides insights into the metabolically active species and the composition and functionality of microbial communities. Keiblinger et al. (2016) eloquently review the opportunities and limits of this approach and discuss how linking phylogenetics and functionality can help learn more on microbial ecology and on potential soil metabolic pathways. Within the –omics, also metabolomics has been applied to soil. Jones et al. (2014) obtained the metabolic profiles of communities living in soils from a range of former mine sites in the United Kingdom to assess the effects of pollution. Although each –omics approach provides valuable information separately, only network-based approaches and combination of data can

<sup>2</sup>http://mycor.nancy.inra.fr/blogGenomes/?page\_id=3262

lead to the understanding of microbiomes. Below an example of a combination of data from genomics, metagenomics and metaproteomics.

#### TOOLS FROM T. melanosporum GENOME SEQUENCE FOR DECIPHERING ITS IN SITU DYNAMICS

In the last years, the biology and the ecology of truffles have greatly increased, thanks to many new scientific insights and technologies as the sequencing of T. melanosporum genome.

Regarding the life cycle, truffles have been considered for long-time self-fertile (Bertault et al., 1998). This opinion could not be tested in absence of an experimental system, based on spore germination, and therefore of the classical breeding of the resulting mycelia (Mello et al., 2005). Only thanks to the T. melanosporum genome sequencing, it has been possible to discover that T. melanosporum has a heterothallic organization (Martin et al., 2010; Rubini et al., 2011a). Heterothallic organization with a MAT locus structured in two idiomorphs harbored by different strains was also found in other truffles: T. borchii and T. indicum (Belfiori et al., 2013, 2016). That means that for truffle reproduction it is necessary that strains of opposite mating type meet. Since these discoveries, the spatio-temporal distribution of these strains in soil and in mycorrhizae has been investigated by numerous authors. The distribution of mating type genes of T. melanosporum has been first investigated in T. melanosporum natural plantation by Rubini et al. (2011b) who reported that, contrary to what is expected, strains with opposite mating types were never present on the same root apparatus, while both mating types were detected in the soil of the plantation. The same authors showed in experiments of inoculation of host plants in controlled conditions that the coexistence of both types on the roots of the same host plant can happen, but lasts until their competition excludes one of the two mating types. According to Iotti et al. (2012) the competition between strains of different mating types seems related to a self-/non-self-recognition system acting before hyphal contact rather than to the presence of a heterokaryon incompatibility (HI) system which leads to the death of the heterokaryotic cells in incompatible reactions. In support of this fact, Rubini et al. (2014) reports that orthologs of the genes controlling HI in other filamentous ascomycetes are present also in the T. melanosporum genome, but they lack the key functional domains involved in the HI process. Zampieri et al. (2012) could detect mating type genes for T. melanosporum under productive and formally productive trees but, generally, not under unproductive trees, so suggesting that the presence of the two mating types in soil can be a promising predictor of the fertility of truffle orchards and, hopefully of the T. melanosporum production when other abiotic and biotic factors are favorable. Mating type distribution of T. melanosporum has also been investigated in artificially planted truffiéres in Australia to increase ascoma

Mello and Balestrini Aspects of Ectomycorrhizal Fungi and Their Interactions

production (Linde and Selmes, 2012). Since the discovery of the T. melanosporum heterothallism, inoculation techniques for production of seedlings with mycelia of opposite mating type are envisaged to improve truffle productivity (Rubini et al., 2014). This modern approach has been recently applied by Iotti et al. (2016) who produced T. borchii fruiting bodies starting from the mycorrhization of plants with mycelial pure culture.

The spatial genetic structure of T. melanosporum populations at a small scale has been investigated in two productive T. melanosporum orchards, one located in the northern France and the other in central Italy thanks to polymorphic SSR markers searched in the T. melanosporum genome and mating type genes (Murat et al., 2013). The analysis of the genetic profiles of ectomycorrhizae using both SSR and mating type markers, and the monitoring of the distribution of T. melanosporum mycelia of the two mating types in the soil allowed the authors to demonstrate a pronounced spatial genetic structure of T. melanosporum, characterized by non-random distribution of small genets. Several small T. melanosporum genets that shared the same mating types could be found on the same host plant, suggesting that the genet distributional pattern is related to the allelic configuration of the MAT locus. However, the factors involved in truffle sexual reproduction are difficult to search due to the impossibility of manipulating truffle in vitro (Le Tacon et al., 2015). In T. melanosporum, the female gametes are ascogonia (MAT1-1 or MAT1-2) produced by a haploid mycelium forming the ectomycorrhizal root tips from which the peridium and the sterile tissue of the gleba constituting the truffle develop. After the fertilization from germinating ascospores acting as male genotypes, a diploid transitory phase occurs (that cannot be detected in mature ascocarps), followed by a meiosis phase that ends in the formation of a mature truffle (De la Varga et al., 2017). **Figure 2** shows a laser microdissection (LMD) experiment coupled with RT-PCR analysis using matingtype genes on different cell-type populations collected from a T. melanosporum fruiting body, i.e., vegetative hyphae and reproductive structures (asci containing the ascospores). As confirmation of the heterotallism in T. melanosporum (Rubini et al., 2011a), transcripts corresponding to the first mating type gene (MAT1-1-1) can be observed in both LMD samples, while those corresponding to the second mating type gene (MAT 1-2-1) can only be detected in the reproductive compartment.

From the ecological point of view, the development of the ECM symbiosis and of the fruiting bodies of T. melanosporum is associated to the formation of a burnt area (known by the French word brulé), characterized by little vegetation around their host plants because of the phytotoxic effects generated by the truffle metabolites and volatile organic compounds (Splivallo et al., 2011). Metagenomics data applied to French trufflegrounds have showed a reduced fungal biodiversity, a dominance of T. melanosporum and a reduced presence both of the ECM Basidiomycota and of bacteria belonging to Pseudomonas and Flavobacteriaceae inside the brulé, together with a reduction of richness of arbuscular mycorrhizal fungi (Napoli et al., 2010; Mello et al., 2011, 2013, 2015b; Mello and Zampieri, 2017). In order to relate microbial community composition to ecological processes happening in the brulé, Zampieri et al. (2016) applied a

metaproteomics analysis to the brulé previously characterized by metagenomics, and cross-referenced the resulting proteins with a database they constructed, incorporating the metagenomics data for the organisms previously identified in this soil, including the black truffle T. melanosporum. The resulting proteins were categorized and assigned to the organisms living in the brulé, leading to discover that the soil inside the brulé contained a larger number of proteins compared with the soil outside the brulé, of which more proteins from herbaceous plants (despite the scarce vegetation typical of such a niche), and more biological processed, mostly of them related to responses to multiple types of stress from most of the brulé components. Thus, although the brulé has a reduced diversity of plant and microbial species, it seems to be a very active environment, characterized by broad stress responses and in particular by herbaceous plants. From these results Zampieri et al. (2016) hypothesize that volatile organic compounds, may elicit stress and defense responses in fungi, bacteria, and above all in the herbaceous plants inside the

brulé. At this regard, already Splivallo et al. (2007) had showed that Arabidopsis, exposed to volatile organic compounds under laboratory conditions, produced an oxidative burst.

Taking in the all, the combination of metagenomics and metaproteomics has provided a powerful tool to reveal functioning of a complex phenomenon associated to an ECM fungus, as the brulé. Since metaproteomics is the study of all the proteins expressed by the organisms within an ecosystem at a specific time will surely help, together with different - OMIC approaches, to understand the ecological regulation of environmental processes.

#### FROM THE MORPHOLOGICAL OBSERVATIONS TO THE IDENTIFICATION OF GENES POTENTIALLY INVOLVED IN THE SYMBIOTIC INTERFACE CREATION

For many years, the interest of researchers has been dedicated to reveal, through morphological observations, the changes in hyphal growth and remodeling of the root and hyphal cell walls during ECM development (Balestrini and Kottke, 2016). At morphological level, the symbiotic interface in an ectomycorrhiza is formed by the plant and fungal cell walls in direct contact, because the ECM fungus remains apoplastic (Peterson and Massicotte, 2004; Balestrini and Bonfante, 2014; Balestrini and Kottke, 2016; **Figure 3**). The use of in situ affinity techniques that utilize specific probes for fungal and plant cell wall components has allowed information to be obtained on the cell wall composition at the plant/fungus interface (Balestrini et al., 1996; Peterson and Massicotte, 2004; Balestrini and Bonfante, 2014; Balestrini and Kottke, 2016). Several fungal proteins localized on the fungal cell wall in the ectomycorrhizal basidiomycete P. tinctorius have been observed to be highly increased during eucalypt root colonization, such as symbiosis-regulated acidic polypeptides (SRAPs) and hydrophobins (Laurent et al., 1999; Martin et al., 1999; Tagu et al., 2001). In the ectomycorrhizal ascomycete T. borchii, a secreted phospholipase A2 (TbSP1) has been also localized on the fungal cell wall and a role during ECM development has been proposed (Soragni et al., 2001; Miozzi et al., 2005). A homolog gene (TmelPLA2) has been also identified in T. melanosporum genome and this gene was one of the most upregulated transcripts during the colonization of Corylus avellana roots (Balestrini et al., 2012), in agreement with the previous data. More recently, the information derived from the several mycorrhizal genomes sequencing and the transcriptomics data on different ECM symbioses allowed the identification of novel fungal cell wall components with a putative role in the interaction with the host plant. A genome-wide inventory of hydrophobins, i.e., fungal small secreted proteins associated with the outer surface of the cell wall and able to mediate the interaction between the fungus and the environment (Whiteford and Spanu, 2002), have been obtained from L. bicolor where it has been demonstrated that the expression of these genes changed depending on the life-cycle stage and on the host root environment (Plett et al., 2012). A weak up-regulation of one of the four putative hydrophobin genes identified in T. melanosporum genome, has been also reported in its ECMs (Balestrini et al., 2012). A role for these proteins in the formation of the symbiotic interface and/or in the hyphal aggregation required for the formation of the symbiotic structures has been hypothesized (Balestrini and Bonfante, 2014). Interestingly, genes coding for putative chitin deacetylases (CDAs), which are enzymes belonging to the carbohydrate esterase 4 (CE4) family<sup>3</sup> that are involved in the chitin conversion to chitosan, have been also reported as upregulated in T. melanosporum (Balestrini et al., 2012) and L. bicolor (Veneault-Fourrey et al., 2014) ectomycorrhizae, suggesting a role in the symbiosis establishment. The role(s) of CE4 enzymes in ectomycorrhizae is still unknown, but on the basis of their expression profiles, it has been suggested that some of them are involved in cell wall synthesis, whereas others are perhaps involved in fungal colonization to avoid plant defense responses (Veneault-Fourrey et al., 2014). CE4 can be in fact involved in cell wall formation, but a role in the reduction of chitin oligomers elicitor activity through their de-acetylation has been also proposed during plant–pathogen interactions. Additionally, chitin de-acetylation to chitosan in pathogen fungal structures could also have a function to protect fungal cell wall plant from chitinases (Tsigos et al., 2000). The presence of a chitin-binding domain in TmelPDA3, one of the T. melanosporum upregulated genes, should also be highlighted, considering that a role for chitin-binding proteins has been also proposed in pathogenic fungi to protect the fungal cell wall from chitinases produced by host plants. This has been reported for the biotrophic fungal pathogen Cladosporium fulvum that secrets the apoplastic effector Avr4, which is a chitin-binding lectin that functions to protect the integrity of the fungal cell wall against chitinases (van den Burg et al., 2006; Malinovsky et al., 2014). Genome-wide transcriptome profiling allowed to demonstrate that several genes related to cell wall modification, significantly regulated during ectomycorrhiza formation, are involved in fungal cell wall processing, suggesting an extensive remodeling when the fungus is in contact with the host plant cells in agreement with the view of fungal cell wall as a highly dynamic structure (Veneault-Fourrey et al., 2014). Additionally, several expansin-like genes have been identified in the L. bicolor genome and several of them were found to be regulated during ECM development, suggesting a role as complement of the enzyme set involved in fungal and/or plant cell wall modification. One expansin-gene (LbEXP1) was first showed as the most highly induced CAZyme in ECM tissues (Martin et al., 2008) and then localized on the fungal cell wall, both in the Hartig net and mantle hyphae, suggesting a role in the fungal cell wall remodeling during symbiosis structures development (Veneault-Fourrey et al., 2014).

A subtle remodeling of the root cell wall in response to the contact with the ECM fungus has been reported years ago (Balestrini et al., 1996; Balestrini and Bonfante, 2014 for a review). A localized degradation of pectin has been suggested

<sup>3</sup>http://www.cazy.org/CE4.html

FIGURE 3 | Hazelnut-T. melanosporum ectomycorrhizal root. Ectomycorrhizal roots have been prepared for the transmission electron microscopy using the high-pressure freezing and freeze substitution. (A) A fungal hypha (F) penetrates between two host cells (H). The asterisk marks the plant cell wall. m, mitochondria; w, fungal cell wall. Bar, 0.60 µm. (B) Hartig net (Hn) in a fully developed mycorrhiza. Hyphae develop among plant cells (H), and their cell walls are in direct contact with the plant cell walls, showing a simple interface. Bar, 1.2 µm. (Inset) Magnification of the contact zone between plant (asterisk) and fungal cell wall (arrows). F, fungus. Bar, 0.50 µm.

during fungal colonization, according with the growth of the ECM fungus through the middle lamella and with the expression of fungal genes acting on these plant cell-wall components (Balestrini and Kottke, 2016; Sillo et al., 2016). In addition to a remodeling of the middle lamella, a soft remodeling of the plant cell wall through the loosening of cellulose has been also suggested during L. bicolor ECM development. Considering ectomycorrhizae at different stages of the interactions it was possible to verify that L. bicolor CAZymes acting on several plant cell wall components are expressed at a different developmental stage (Veneault-Fourrey et al., 2014). Recently, Doré et al. (2017), working on different developmental stages of the ECM interaction between H. cylindrosporum and Pinus pinaster, showed that genes coding for extracellular proteins, such as MiSSPs and CAZymes, were over-represented among the genes up-regulated upon pre-infectious interaction, suggesting that these specific proteins are host-induced and might play essential function(s) in the early fungal response to host root. Remarkably, the expression of some of these genes has been reported in the Hartig net compartment in T. melanosporum and C. avellana ECMs (Hacquard et al., 2013). Looking at the plant genes, Sebastiana et al. (2014) also reported that several genes coding enzymes involved in cell wall biosynthesis and modification were found to be differentially expressed in ectomycorrhizal cork oak roots with respect to non-colonized roots. In detail, cell wall-related glycosylhydrolases (GH), which are required for the modification of cell wall polysaccharides and are involved in wall loosening and elongation, were mostly up-regulated in oak ECM roots. By contrast, cell wall-related glycosyltransferases (GT), involved in the synthesis of noncellulosic polysaccharides as part of the biosynthetic machinery to synthesize the complex plant cell-wall polysaccharides, resulted to be mostly down-regulated. Overall, these plant regulated genes might be involved in the remodeling of the plant cell wall required to facilitate hyphae penetration between cells and fungal accommodation (Sillo et al., 2016), but they might be also involved in maintaining the cell wall thickness during the changes in the architecture of host colonized cells, i.e., radial elongation (Sebastiana et al., 2014), requiring the addition of newly synthesized polysaccharides. Interestingly, an activation of cellulose synthesis in oak colonized roots has been also suggested, and transcripts corresponding to plant expansins, which are known to be involved in cell wall loosening and cell enlargement, were found to be up-regulated in ECM roots (Sebastiana et al., 2014).

However, although many studies suggest a role for the proteins regulated in symbiosis, functional analyses with the aim to highlight the function of these proteins are still lacking, as well as the nature of the cell wall remodeling during the ectomycorrhiza establishment and development.

#### CONCLUSION AND PERSPECTIVES

In conclusion, genomic and transcriptomic sequencing projects starting with the first mycorrhizal genome sequencing (i.e., that of L. bicolor) have allowed the identification of the common core of ECM symbiosis-related genes, as determinants of the symbiotic lifestyle, as well as the identification of species-specific traits. However, genome sequencing is only the first step to obtain information on how an organism interacts with the environment and with other organisms. The combination of functional, structural, cellular, and bioinformatics approaches is providing knowledge on the function of genes/proteins and permits to reconstruct the pathways of an organism in specific growth conditions, and in its natural environment. In fact, metagenomics, metatranscriptomics and metaproteomics studies are currently and fast providing a powerful mean for the analysis of environmental microorganisms without the need of culturing them. At the question: "can -omics provide insight into microbial ecology that cannot be achieved using traditional methods?", Jansson and Prosser (2013) and Prosser (2015) reply that although -omics generate a large amount of 'big data' experiments that test hypotheses on microbe-environment associations may allow more direct identification and analysis of the ecological processes. The large volume of sequence data involved in the ECM symbiosis, provide a reference database for an estimation of the ECM fungal taxa number and their ecology.

The next crucial research will be linking molecular and metabolic data to key processes such as the exchange of nutrients, the plant protection against stresses and diseases and the genes responsible of the symbiosis.

#### AUTHOR CONTRIBUTIONS

RB has contributed on the molecular and cellular interactions in the ectomycorrhizal symbiosis, with particular attention on the cell walls of the two symbionts. AM has contributed on the biodiversity of ectomycorrhizal fungi, their population dynamics and their interactions with other soil microorganisms.

#### ACKNOWLEDGMENTS

fmicb-09-00216 February 13, 2018 Time: 18:50 # 10

The authors thank Fabiano Sillo for providing the results illustrated in the **Figure 2** and Antonella Faccio for the preparation of the ectomycorrhizal roots with high-pressure

#### REFERENCES


freezing and freeze substitution in the laboratory of Robert Roberson (Arizona State University) during a Short Term Mobility program funded by the CNR. They are grateful to Marco Chiapello for modifications of original Figure 1 from Tedersoo et al. (2014).



nitrogen addition. Environ. Microbiol. 19, 4736–4753. doi: 10.1111/1462-2920. 13939



**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 © 2018 Mello and Balestrini. 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 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.

# Secretome Analysis from the Ectomycorrhizal Ascomycete *Cenococcum geophilum*

Maíra de Freitas Pereira1,2†, Claire Veneault-Fourrey 1,3†, Patrice Vion<sup>1</sup> , Fréderic Guinet 1,3 , Emmanuelle Morin<sup>1</sup> , Kerrie W. Barry <sup>4</sup> , Anna Lipzen<sup>4</sup> , Vasanth Singan<sup>4</sup> , Stephanie Pfister <sup>2</sup> , Hyunsoo Na<sup>4</sup> , Megan Kennedy <sup>4</sup> , Simon Egli <sup>2</sup> , Igor Grigoriev <sup>4</sup> , Francis Martin<sup>1</sup> , Annegret Kohler <sup>1</sup> \* and Martina Peter <sup>2</sup> \*

1 Institut National de la Recherche Agronomique, Unité Mixte de Recherche 1136 Interactions Arbres, Microorganismes, Laboratoire D'excellence Recherches Avancés sur la Biologie de l'Arbre et les Ecosystémes Forestiers, Centre Institut National de la Recherche Agronomique-Lorraine, Champenoux, France, <sup>2</sup> Swiss Federal Research Institute WSL, Forest Dynamics, Birmensdorf, Switzerland, <sup>3</sup> Université de Lorraine, Unité Mixte de Recherche 1136 Interactions Arbres-Microorganismes, Vandoeuvre les Nancy, France, <sup>4</sup> United States Department of Energy Joint Genome Institute, Walnut Creek, CA, United States

#### *Edited by:*

Erika Kothe, Friedrich Schiller Universität Jena, Germany

#### *Reviewed by:*

Stefano Ghignone, Istituto per la Protezione Sostenibile delle Piante (CNR), Italy Nuria Ferrol, Consejo Superior de Investigaciones Científicas (CSIC), Spain

#### *\*Correspondence:*

Annegret Kohler annegret.kohler@inra.fr Martina Peter martina.peter@wsl.ch

† These authors have contributed equally to this work.

#### *Specialty section:*

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

*Received:* 22 November 2017 *Accepted:* 22 January 2018 *Published:* 13 February 2018

#### *Citation:*

de Freitas Pereira M, Veneault-Fourrey C, Vion P, Guinet F, Morin E, Barry KW, Lipzen A, Singan V, Pfister S, Na H, Kennedy M, Egli S, Grigoriev I, Martin F, Kohler A and Peter M (2018) Secretome Analysis from the Ectomycorrhizal Ascomycete Cenococcum geophilum. Front. Microbiol. 9:141. doi: 10.3389/fmicb.2018.00141 Cenococcum geophilum is an ectomycorrhizal fungus with global distribution in numerous habitats and associates with a large range of host species including gymnosperm and angiosperm trees. Moreover, C. geophilum is the unique ectomycorrhizal species within the clade Dothideomycetes, the largest class of Ascomycetes containing predominantly saprotrophic and many devastating phytopathogenic fungi. Recent studies highlight that mycorrhizal fungi, as pathogenic ones, use effectors in form of Small Secreted Proteins (SSPs) as molecular keys to promote symbiosis. In order to better understand the biotic interaction of C. geophilum with its host plants, the goal of this work was to characterize mycorrhiza-induced small-secreted proteins (MiSSPs) that potentially play a role in the ectomycorrhiza formation and functioning of this ecologically very important species. We combined different approaches such as gene expression profiling, genome localization and conservation of MiSSP genes in different C. geophilum strains and closely related species as well as protein subcellular localization studies of potential targets of MiSSPs in interacting plants using in tobacco leaf cells. Gene expression analyses of C. geophilum interacting with Pinus sylvestris (pine) and Populus tremula × Populus alba (poplar) showed that similar sets of genes coding for secreted proteins were up-regulated and only few were specific to each host. Whereas pine induced more carbohydrate active enzymes (CAZymes), the interaction with poplar induced the expression of specific SSPs. We identified a set of 22 MiSSPs, which are located in both, gene-rich, repeat-poor or gene-sparse, repeat-rich regions of the C. geophilum genome, a genome showing a bipartite architecture as seen for some pathogens but not yet for an ectomycorrhizal fungus. Genome re-sequencing data of 15 C. geophilum strains and two close relatives Glonium stellatum and Lepidopterella palustris were used to study sequence conservation of MiSSP-encoding genes. The 22 MiSSPs showed a high presence-absence polymorphism among the studied C. geophilum strains suggesting an evolution through gene gain/gene loss.

**23**

Finally, we showed that six CgMiSSPs target four distinct sub-cellular compartments such as endoplasmic reticulum, plasma membrane, cytosol and tonoplast. Overall, this work presents a comprehensive analysis of secreted proteins and MiSSPs in different genetic level of C. geophilum opening a valuable resource to future functional analysis.

Keywords: *Cenococcum geophilum*, small secreted proteins, ectomycorrhiza, symbiosis, interaction

#### INTRODUCTION

Symbiotic plant–fungal interactions are predominant in worldwide soils and have important roles in the global colonization by land plants. In forest soils, the ectomycorrhizal (ECM) symbiosis is the dominant form of a mutualistic interaction between the fine roots of trees and fungal hyphae. This interaction allows the exchange of nutrients and water between partners and increases the disease resistance of host plants (Smith and Read, 2010). Approximately 20,000 ECM fungi from diverse fungal clades and about 6,000 tree species worldwide are able to form this association (van der Heijden et al., 2015; Martin et al., 2016). Although the ECM lifestyle evolved independently several times from ancestral saprotrophs (Hibbett et al., 2000; Kohler et al., 2015) the arisen symbiotic organ and mutualistic interaction is surprisingly similar each time. Recent genomic and transcriptomic studies indicate the convergent evolution of a symbiosis toolkit with two major features of the ECM lifestyle: a reduced number of plant cell wall degrading enzymes as compared to saprotrophic ancestors in the genome and in the transcriptome the accumulation of lineage-specific transcripts possibly involved in the biotic interaction (Kohler et al., 2015; Martin et al., 2016).

Ectomycorrhiza formation is a process controlled by different genetic and environment factors (Tagu et al., 2002; Smith and Read, 2010; Kohler et al., 2015; Martin et al., 2016). A molecular communication between fungi and plant is a prerequisite for establishment of a symbiotic interaction (Plett and Martin, 2011; Martin et al., 2016). In order to manipulate host defenses and enable colonization, the secretion of small proteins is a known mechanism of pathogenic fungal–host interactions. It has been observed in mycorrhizal interactions as well but their role is still poorly understood (Martin et al., 2008, 2016; Garcia et al., 2015; Plett and Martin, 2015). Fungal genome availability allowed comparative analyses across different lifestyles including saprotrophic, mycorrhizal, pathogenic and endophytic ones revealing that all fungal genomes encode for small-secreted proteins (SSPs), which are defined as proteins of <300 aminoacids containing a signal-peptide (Martin et al., 2008, 2010, 2016; Kohler et al., 2015; Pellegrin et al., 2015; Kamel et al., 2017). Despite some overlap among shared SSPs in ECM fungi and saprotrophic fungi based on sequence similarities, many genes encoding SSPs are orphan genes and are unique to each ECM species (Kohler et al., 2015; Pellegrin et al., 2015). To understand the role of SSPs in the mycorrhiza development, both gene expression studies as well as functional analyses are necessary. So far, only two SSPs, a Mycorrhizae induced Small Secreted Protein of 7Kda (MiSSP7) in Laccaria bicolor, an ECM fungus and the secreted protein 7 (SP7) in Rhizophagus irregularis (previously known as Glomus intraradices), an arbuscular mycorrhizal fungus have been functionally characterized, and in both cases, the secreted effector targeted to the host plants nucleus and reshuffled plant defense pathways (Kloppholz et al., 2011; Plett et al., 2011, 2014).

One of the most abundant ECM fungi is the ascomycete Cenococcum geophilum Fr. showing a worldwide distribution through numerous habitats, environments and geographic regions and associating with a large variety of host species including gymnosperms and angiosperms (Trappe, 1962; LoBuglio, 1999; Obase et al., 2017). C. geophilum forms characteristic black monopodial or dichotomous ectomycorrhizas with darkly pigmented, emanating hyphae, as well as resistance propagules known as sclerotia, but sexual structures have never been found (Trappe, 1962; LoBuglio, 1999; Obase et al., 2017). Although being a broadly distributed fungus, the biology of C. geophilum is poorly understood. Studies on the fine-scale diversity of C. geophilum populations revealed a high level of genetic polymorphism and this can help to explain the large amount of physiological and phenotypic differences reported among C. geophilum isolates from similar as well as diverse geographic regions (LoBuglio, 1999; Douhan et al., 2007; Obase et al., 2017). Likewise, the variability in genome size, ploidy level and gene polymorphism among C. geophilum isolates support the evidence of possible cryptic sexual recombination and speciation (Spatafora et al., 2012; Bourne et al., 2014). C. geophilum is the only ECM fungus belonging to the clade of Dothideomycetes, the largest class of Ascomycota with a high level of ecological diversity, including many devastating plant pathogens and saprotrophs (LoBuglio, 1999; Ohm et al., 2010).

The recent genome sequencing of a C. geophilum strain revealed a large size of 178 Mbp and is predicted to encode for 14,748 gene models (Peter et al., 2016). Transcript profiling of C. geophilum genes expressed in pine ectomycorrhizal root tips revealed the upregulation of genes encoding membrane transporters, including aquaporin water channels and sugar transporters in symbiosis. Also, MiSSPs were highly induced or even specifically expressed in symbiotic tissues as seen for

**Abbreviations:** CAZymes, Carbohydrate-Active Enzymes; CBM, Carbohydrate binding modules; CDS, Coding DNA sequence; CEG, Core eukaryotic genes; ECM, Ectomycorrhiza; ER-Endoplasmic reticulum; FDR, False discovery rate; FIR, Intergenic flanking region; FPKM, Fragment Per Kilobase of exon model per Million mapped reads; GAPDH, Glyceraldehyde-3-phosphate dehydrogenase; GEO, Gene Expression Omnibus; GFP, Green fluorescent protein; GH, Glycoside hydrolase; GDR, gene-dense repeat-poor region; GSR, gene-sparse repeatrich region; GT, glycosyl transferase; ITS, Internal transcribed spacer; MiSSP, Mycorrhizae induced Small Secreted Protein; SP, Secreted Protein; SiP-Signal Peptide; SSP, Small Secreted Proteins; TE, Transposable Elements; TPM, Transcripts per million.

other ECM fungi (Kohler et al., 2015; Peter et al., 2016). Furthermore, comparative genome analysis of C. geophilum with sequenced Dothideomycetes and a set of other fungi revealed that eight of the symbiosis-induced (>5 fold) SSPs are unique to Cenococcum and might play an important role in the fungal-plant interaction as seen for effector genes (Peter et al., 2016). One of the most striking features of the C. geophilum genome is its massively increased size compared to other sequenced Dothideomycetes (Peter et al., 2016). The 3–4 times larger genome of this ECM species is explained by the proliferation of transposable elements (TE), which make up 75% of the genome (Peter et al., 2016). Increased genome sizes due to TE bursts have been observed for other mycorrhizal fungi such as Tuber melanosporum and Rhizophagus irregularis (Kohler and Martin, 2016), but also for many biotrophic plant pathogens (Raffaele and Kamoun, 2012; Stukenbrock and Croll, 2014). In plant pathogens, these TEs are often not randomly spread over the genome but cluster in repeat-rich chromosomal segments that evolve at accelerated rates than the rest of the genome due to diverse mechanisms such as TEactivity and TE silencing machineries (Raffaele and Kamoun, 2012). Also, genes implicated in virulence and host adaptation such as effector genes tend to localize in repeat-rich, faster evolving regions (Raffaele and Kamoun, 2012). Even within species, substantial presence/absence polymorphisms have been observed for such genes in proximity of TEs for a plant pathogen, being a source of variation and driving local adaptation (Hartmann and Croll, 2017). Such a two-speed genome has convergently evolved in plant pathogenic fungi in independent lineages such as the oomycetes and the Dothideomycetes (Dong et al., 2015) but has not been observed for mycorrhizal fungi so far.

In order to better understand the biotic interaction of C. geophilum with its host plants, the goal of this work was to analyze whether C. geophilum is secreting MiSSPs as mean of communication with its host plants and narrow down the repertoire of candidate effectors for further demonstration. The specific objectives were (i) to assess the regulation of the C. geophilum secretome in ectomycorrhizal root tips formed with two different host plants, the gymnosperm Pinus sylvestris (pine) and the angiosperm Populus tremula × Populus alba-INRA clone 717-1-B4 (poplar) through transcriptomic analyses (ii) to identify candidate symbiosis effector genes and study their genomic localization, (iii) to study presence-absence polymorphism in candidate effectors by analyzing 15 re-sequenced C. geophilum strains and two closely related Dothideomycetes genomes to elucidate their evolution and conservation and (iv) to obtain a first glimpse of the possible role as effectors for a selection of MiSSPs by studying their potential target within the host plant cell through sub-cellular localization experiments in Nicotiana benthamiana leaf cells.

### MATERIALS AND METHODS

#### Microorganisms Growth Condition

Cenococcum geophilum isolates originating from different sites (**Supplementary Table S1**) were kept in Petri dishes (100 × 20 mm) containing Cenococcum medium, a modified MMN medium containing casein (Trappe, 1962), at 25◦C and transferred to new culture medium every 20 days. Escherichia coli (subcloning efficiency DH5a competent cells; Invitrogen, Carlsbad, CA, U.S.A.) and Agrobacterium tumefaciens (electrocompetent strain GV3101) were conserved at −80◦C and they were grown in LB and YEPD medium at 37 and 28◦C, respectively.

### Plant Growth Condition and Ectomycorrhiza Formation

In vitro interaction systems were established between C. geophilum isolate 1.058, of which the genome is available (http://genome.jgi.doe.gov/Cenge3/Cenge3.info. html) and Scots pine (Pinus sylvestris) or hybrid poplar (Populus tremula × Populus alba; INRA clone 717-1-B4) respectively. Pine seeds [P. sylvestris provenance VS/Leuk (31/10) WSL] were superficially disinfected in a laminar flux hood by immersion in H2O<sup>2</sup> for 30 min, followed by three rinses with sterile distilled water. The seeds were germinated in modified MMN medium described by Brun et al. (1995), with low nitrogen and phosphorus during seven days for observation of contamination. After seed germination, the plants were transferred to Petri dishes containing modified MMN and covered with a cellophane membrane (135 mm). Approximately ten agar disks containing fungal mycelium of C. geophilum 1.58 were placed in the vicinity of the roots. The dishes were incubated in a growth chamber at 25◦C with 16 h light/day for 90 days (**Supplementary Figures S1A,B,E**).

The hybrid poplar (Populus tremula × Populus alba; INRA clone 717-1-B4) was micropropagated in vitro in Murashige and Skoog (MS) medium (Murashige and Skoog, 1962), with hormone supplements to synchronize rhizogenesis as described by Felten et al. (2009). In parallel, the MNM medium with low phosphorus and nitrogen (Brun et al., 1995) was covered with cellophane membranes and inoculated with 10–12 agar disks containing fungal mycelium at 25◦C for 20 days. Following this, two hybrid poplar plants per dish were transferred onto the fungal mycelium. The Petri dishes were incubated in a growth chamber at 25◦C with 16 h light/day for 60 days (**Supplementary Figures S1C–E**).

For both experiments, pure cultures of C. geophilum, pine and hybrid poplar grown in identical conditions were used as experimental controls, and the assays were conducted in minimum of three replicates. After the indicated period of time, Petri dishes were opened and the different tissues were collected for RNA analyses as follows: Single mycorrhizal root tips were collected in a 1.5 ml tube and immediately frozen in liquid nitrogen. Extramatrical mycelium surrounding roots and emanating from pine ECMs was scratched from the cellophane using a scalpel and if present, sclerotia formed in these dishes were separately collected and also immediately frozen in liquid nitrogen. For pure culture controls, free-living mycelium or fine root tips, respectively, were collected at the same time and manner as indicated for synthesis Petri dishes.

### RNA Extraction and Illumina Sequencing and Data Analysis

Total RNA from mycorrhizal roots, sclerotia, extramatrical mycelium, fungal, and plant controls were extracted with the RNeasy Plant Mini kit (Qiagen, Courtaboeuf, France), including a DNase I (Qiagen) treatment, according to the manufacturer's instructions to eliminate traces of genomic DNA. Assays for the quantification and integrity check were conducted using an Experion Automated Electrophoresis Station (Bio-Rad, Hercules, CA, USA) or Agilent 2100 Bioanalyzer system (Agilent, Santa Clara, CA, USA).

Preparation of libraries and 2 × 150 bp Illumina HiSeq2000/2500 mRNA sequencing (RNA-Seq) was performed by the Joint Genome Institute (JGI) facilities. Raw reads were filtered and trimmed using the JGI QC pipeline (see **Supplementary Table S2**). Using BBDuk, raw reads were evaluated for artifact sequence by kmer matching (kmer = 25), allowing 1 mismatch and detected artifact was trimmed from the 3′ end of the reads. RNA spike-in reads, PhiX reads and reads containing any Ns were removed. Quality trimming was performed using the phred trimming method set at Q6. Finally, following trimming, reads under the length threshold were removed (minimum length 25 bases or 1/3 of the original read length—whichever is longer). Filtered reads from each library were aligned to C. geophilum v 2.0 reference transcripts available at the JGI database (http://genome.jgi.doe.gov/Cenge3/Cenge3. info.html). FeatureCounts was used to generate the raw gene counts and only primary hits assigned to the reverse strand were included in the raw gene counts (Liao et al., 2014). Raw gene counts were used to evaluate the level of correlation between biological replicates using Pearson's correlation and determine which replicates would be used in the DGE analysis. FPKM (Fragment Per Kilobase of exon model per Million mapped reads) and TPM (transcripts per million) normalized gene counts were also provided. DESeq2 (version 1.10.0), including an independent filtering procedure by default, was used to determine which genes were differentially expressed between pairs of conditions (Love et al., 2014). The parameters used to call a gene differentially expressed between conditions were fold change > log1 and FDR p < 0.05. A gene with a FPKM > 1 was considered as expressed. The complete RNA-Seq data was submitted to GEO (GSE108831 and GSE108866). For selecting MiSSPs as well as for comparisons among different synthesis systems, we added RNA-Seq data of a semi-sterile greenhouse trial growing P. sylvestris with C. geophilum 1.58 in pots (Peter et al., 2016; GEO Accession GSE83909). Here, pine trees were pre-grown for 2 months in pots containing a 1:2 double-autoclaved mixture of quartz sand and sieved forest topsoil before being inoculated by C. geophilum 1.58 mycelia and grown for another 3 months before harvesting ECMs. As pure culture fungal control, 2-months-old mycelium grown as indicated above on agar Petri dishes was used (Peter et al., 2016).

#### Genome Architecture and Gene Density Analysis

Genomic distances between two genes and genome architecture heatmaps were generated according to Saunders et al. (2014). These results were binned according to log (length) and plotted as a 2-dimensional heatmap using Excel. Plotting the abundance of genes according to their 5′ and 3′ flanking intergenic lengths indicate local gene density (**Figure 3**). In C. geophilum genome, we defined two contrasting regions: one gene-dense repeatpoor (GDR) containing a high number of genes (gene-dense) combined with short 3′ and 5′ flanking regions indicating a low level of repeats (repeat-poor), whereas the gene-sparse repeat-rich (GSR) region is characterized by a low number of genes displaying long 5′ and/or 3′ flanking regions. We also represent, according to local gene density, the distribution of gene expression induction in ECM root tips compared to freeliving mycelium (log2 fold change) or their level of expression (fpkm values).

### DNA Extraction, Genome Re-sequencing of *Cenococcum geophilum* Isolates and Presence–Absence Analyses of Selected SSPs

To study the presence/absence polymorphism of selected SSPs, data of 15 recently re-sequenced strains of C. geophilum was used. The 15 strains originated from diverse locations in Switzerland, France, Poland and Finland (**Supplementary Table S1**). For genomic DNA sequencing, mycelia were grown in liquid culture containing Cenococcum medium for 3–4 weeks after which they were harvested, pulverized in liquid nitrogen and stored at −80◦C until processing. DNA was extracted using the PowerMax Soil DNA isolation kit (MOBIO/QIAGEN CA, USA) according to the manufacturer's instructions and using around 2 g of mycelia. Library construction and sequencing was performed at the Joint Genome Institute (JGI) using Illumina HiSeq 2500 and 2 × 100 bp read length sequencing in two different lanes. Between 26 and 48 million raw reads were generated corresponding to a 15–27x coverage. CLC genomic workbench 10 was used to de novo assembly the 15 genomes with the following parameters: Mapping mode: map reads to contigs; minimum contig lenght: 500; Mismatch cost = 2; Insertion cost = 3; Deletion cost = 2; length fraction = 1.0; Similarity fraction = 0.9. A summary is given in **Supplementary Table S3** and sequence contigs for the different strains and for all MiSSPs studied are compiled in **Supplementary File 1** (http://mycor.nancy.inra.fr/IMGC/ CenococcumGenome/download/Supplementary\_data\_1.fa.gz).

Screening for presence–absence polymorphism of the 22 selected C. geophilum MiSSPs and 22 Core eukaryotic genes– CEG (**Supplementary Table S4**) in the 15 re-sequenced strains was done by conducting a BLASTN search against the de novo assemblies and the reference genome 1.58 (https://genome.jgi. doe.gov/Cenge3/Cenge3.home.html). Further, the genome data of the closest related species, Glonium stellatum (https://genome. jgi.doe.gov/Glost2/Glost2.info.html) and Lepidopterella palustris (https://genome.jgi.doe.gov/Leppa1/Leppa1.home.html) was used to compare C. geophilum SSP sequences for polymorphism (Peter et al., 2016). A gene was considered as affected if the deletion event was overlapping >90% of the gene. To check presence–absence polymorphism in gene duplications, manual alignments was done using the INRA Multalin interface (Corpet, 1988). The presence of a C. geophilum MiSSP in the respective de novo assembly contigs was defined as the lowest e-value accession (E-value) combined with the greatest HSP length (number of nucleotides in the reference genome–Cg1.58).

The variability in presence–absence patterns of the 22 selected MiSSP genes among C. geophilum isolates was examined with principal coordinate analyses (PCO) using the Jaccard similarity index. Variation explained in these patterns by phylogenetic clade (3 levels), country (4 levels) and forest type (4 levels) of isolate origin were assessed using the PERMANOVA routine (Anderson, 2001) implemented in the software Primer7 using 9,999 unrestricted permutations of raw data as well as by Monte Carlo tests (Clarke and Gorley, 2015). Phylogenetic analysis was performed using the online software phylogeny.fr from concatenated sequences of C. geophilum GAPDH and ITS using default parameters (Dereeper et al., 2008). In short, MUSCLE was used to align sequences and Gblocks for curation. Phylogeny was performed using a maximum likelihood algorithm using PhyML and branch confidence indices were calculated based on an approximate likelihood ratio test. ITS and GAPDH sequences are given in **Supplementary Table S5**.

#### Validation of SSP Gene Presence–Absence in Different *Cenococcum Geophilum* Isolates by PCR

We validated gene presence–absence polymorphism for some selected C. geophilum MiSSPs using direct amplification of target genes including upstream and downstream regions. The primers were designed using Primer 3.0 (Untergasser et al., 2012) from a conserved flanking sequences of each gene (**Supplementary Table S6**). PCR reactions were performed with OneTaq <sup>R</sup> DNA Polymerases according to the manufacturer's instructions (New England Biolabs, Mass, USA) and amplicons run on 1% agarose gels. Each PCR reaction was purified with QIAquick PCR Purification Kit (Qiagen, Courtaboeuf, France) and the PCR product verified by sequencing (Eurofins, Ebersberg, Germany).

#### Cloning Procedures and Plasmids Used for Localization Experiments

The open reading frame (ORF) coding the mature form (i.e., without the signal peptide) of 22 C. geophilum selected MiSSPs were synthetized by GeneCust Europe (Ellange, Luxembourg). The vectors were designed with att sites accomplish to gene sequence to be compatible with PCR Cloning System with Gateway <sup>R</sup> Technology. The entry clone (C. geophilum MiSSP vectors) was utilize in LR recombination reaction with pB7WGF2 (C-terminal fusion with GFP) destination vector to create an expression clone (Karimi et al., 2002). The vectors were amplified in E. coli (DH5a competent cells; Invitrogen, Carlsbad, CA, USA). Sequences of DNA fragments inserted in vectors obtained by PCR were verified by sequencing (Eurofins genomics, Ebersberg, Germany) before to clone in A. tumefaciens (electrocompetent strain GV3101). For colocalization studies, we used a set of markers fused to mCherry protein developed by Nelson et al. (2007).

## Transient Protein Expression in *Nicotiana benthamiana* Leaf Cells

N. benthamiana plants were grown in phytotron at 22◦C under 16-h day and 8-h night conditions. A. tumefaciens GV3101 was used to deliver T-DNA constructs into leaf cells of 4–6 weeks-old N. benthamiana plants, following the agroinfiltration method previously described (Win et al., 2011). Overnight-grown bacterial cultures were resuspend into 10 ml of infiltration buffer (10 mM MgCl2, 10 mM MES, pH 5.6, 200µM acetosyringone), optical density at 600 nm (OD600) adjusted at 0.1. Bacteria were incubated at 28◦C during 2 h under 50 rpm. For all cotransformations, A. tumefaciens strains were mixed in a 1:1 ratio in infiltration buffer to a final OD 600 of 0.2. The leaves were collected 2 days after infiltration for further protein isolation or microscopy analysis.

## Live-Cell Imaging by Laser-Scanning Confocal Microscopy

Small pieces of leaves were mounted in Perfluorodecalin 95% (Sigma-Aldrich, Saint Louis, MO, USA) and water between a slide and a coverslip and were immediately observed. Livecell imaging was performed with a Zeiss LSM780, confocal microscope system, using 10× (air) and 40× (water immersion) objectives. The GFP was excited at 488 nm, whereas the mCherry was excited at 561 nm. Specific emission signals corresponding to the GFP and the mCherry were collected between 505– 525 and 580–620 nm, respectively. Each construct gave a similar localization pattern across at least three independent observations. After observation, leaves were frozen in liquid nitrogen and were conserved at −80◦ C for further use.

## Total Protein Isolation and Immunoblotting

N. benthamiana leaves were harvested 2 days after infiltration, were frozen in liquid nitrogen, and were ground into powder with mortar and pestle. Total protein extraction was performed by reducing and denaturing proteins from the leaf powder 10 min at 95◦C in Laemmli buffer (0.5 M Tris-HCl, pH 6.8, 10 mM dithiothreitol [DTT], 2% SDS, 20% glycerol) in order to avoid in vitro nonspecific degradation of the fusion proteins. Proteins were separated by 15–20% SDS-PAGE (Mini-PROTEAN <sup>R</sup> TGXTM Gels) and transferred onto a nitrocellulose membrane using Trans-Blot Turbo Transfer System (Bio-Rad, CA, USA). Transfert efficiency was assessed by Red Ponceau staining. GFP detection was performed in a single step using a GFP (B2): sc-9996 horseradish peroxidase (HRP)-conjugated antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Protein bands on immunoblots were detected using Clarity ECL Western Blot Substrate (Bio-Rad, CA, USA) using the manufacturer's protocol.

### RESULTS

#### Host-Dependent Gene Expression Changes of *Cenococcum geophilum* Secreted Protein-Encoding Genes

The C. geophilum genome contains a total of 595 predicted secreted proteins (SP) including 227 Small Secreted Proteins (SSPs, <300 aa), 120 Carbohydrate-Active Enzymes (CAZymes), 13 lipases, 27 proteases, and 208 other SPs (Peter et al., 2016).

To study host dependent changes in the gene expression of secreted proteins we performed RNA-Seq analyses on C. geophilum ECM roots from P. sylvestris and P. tremula x alba and C. geophilum free living mycelium grown in in vitro systems. We complemented the analysis with samples from extramatrical mycelium of pine ECM and sclerotia.

The majority of SPs were expressed in all tissues (88–93%). For 30 (5%) of them, no transcripts were detected in any of the conditions studied. The expression of 221 transcripts was significantly regulated in ECM root tips as compared to freeliving mycelium in the in vitro systems (FC > log1, FDR p < 0.05; **Figure 1**; **Supplementary Table S7**). In interaction with pine roots, 114 SSPs were up- and 70 down-regulated, while in contact with poplar roots 107 SSPs were up- and only 20 down-regulated compared to control free-living mycelium (**Figure 1B**). The majority of genes were similarly regulated in the interaction with both host trees (**Figure 1A**). Among the most highly up-regulated transcripts in both interactions were SSPs (e.g., Cenge3:660401, Cenge3:693798, Cenge3:698167), but also secreted CAZymes (GH131, CBM1-GH45, CBM18- CE4-CBM18) (**Supplementary Table S7**). Interestingly, the upregulated transcripts of pine ECM were significantly enriched in CAZymes, whereas for poplar ECM, they were enriched in SSPs (**Figure 1B**). We further analyzed the host-dependent expression levels of these 221 SPs in ECM of the different hosts. If the expression values varied less then five times between the two hosts, we considered a transcript as used in both interactions; with a more then five-fold difference, the transcript was considered as more important for either of the two host interactions (**Supplementary Table S8**). The majority

of the genes (134/147 up-regulated, 76/83 down-regulated) were similarly expressed in ECM root tips of both host plants (**Figure 2**). None of the SPs was specific for one interaction; that is, showing no expression (value 0) when interacting with the other host tree. However, significantly higher expression was observed for two SSPs in interaction with poplar (Cenge3:573854; Cenge3:294776) and for one SSP and several CAZymes in interaction with pine roots (Cenge3:28058; CBM1-GH45, GH12, CBM1-GH5-4, GH28, GT90) (**Supplementary Table S7**). Interestingly, when comparing the different synthesis systems used, i.e., in vitro agar Petri dishes for pine and poplar and a semi-sterile pot system for pine (Peter et al., 2016), it seems that the system had a more pronounced impact on gene expression changes in interactions than had the host identity. Clearly more

difference was <5-fold. The expression was considered as more specific for one host tree if the expression differences was >5-fold. The expression of genes showing a more then 5-fold difference is shown in heatmaps for pine and poplar ECM. Note that the expression in some cases is <1 fpkm but never zero. Note that eight transcripts that were up-regulated in one and down-regulated in the other ECM were counted two times.

C. geophilum genes were commonly up-regulated in ECM of pine and poplar from the in vitro system (41 genes) as compared to commonly up-regulated genes in pine ECM using the different systems (6 genes; **Supplementary Figure S2**).

### Selecting Mycorrhiza Induced Small Secreted Proteins (MiSSPs) for Further Characterization

We further focused on C. geophilum MiSSPs in order to identify candidate effector proteins for the interaction between C. geophilum and its host trees. A set of 22 MiSSPs, induced (>2.5 fold) in the interaction between C. geophilum and P. sylvestris under semi-sterile greenhouse conditions was selected from Peter et al. (2016) (**Table 1**). The predicted protein size of these MiSSPs ranged from 58 to 275 aa, containing no (e.g., Cenge3:664950 and Cenge3:679266) to 10.61% of cysteine residues (Cenge3:666290). Only six MiSSP sequences contain known domains or sequence homology to known proteins such as the SnoaL domain (PF12680, PF13577; Cenge3: 677330), the cupin domain (PF00190, PF07883; Cenge3:552209), the Ubiquitin 3 binding protein But2 C-terminal domain (PF09792; Cenge3:677232 and Cenge3:658610) or the "secreted in xylem 1" (Six1) protein of Fusarium oxysporum (Cenge3:698167; Rep et al., 2004). Eight MiSSPs were specific to C. geophilum while the others share sequence similarity with genes from other Dothideomycetes fungi (**Table 1**). Three pair of duplications were present within the selected MiSSPs showing sequence similarities from 77–95 to 72–93% for nucleotide and protein sequences, respectively (**Supplementary Figure S3**). Expression studies showed that some of these MiSSPs were also upregulated in sclerotia formed in in vitro synthesis dishes and in extramatrical mycelium emanating from the ECM root tips (**Table 1**).

#### Genome of *Cenococcum geophilum* Displays a Bipartite Architecture with MiSSPs Present in Both Regions

Due to richness in transposable elements found in the C. geophilum genome, we measured for each gene the distance to the neighboring genes at both 5′ and 3′ end. This method is used as a proxy to detect repeat-rich regions, assuming that the larger the intergenic region is the more repetitive sequences are present (Raffaele et al., 2010b). C. geophilum genome displayed two types of regions: repeat-rich, gene-sparse regions (GSR) and repeat-poor, gene-dense (GDR) regions, with a cut-off for 5′ and/or 3′ intergenic region length at >6,495 bp (**Figure 3**). This indicates a "two-speed" genome for C. geophilum as seen for some pathogenic fungi. In order to test whether gene position and environment could impact the in planta gene regulation, we measured the distribution of all C. geophilum genes for their expression induction and repression in ectomycorrhizal root tips compared to free-living mycelium according to local gene density. We observed that in planta regulated (either induced or repressed) genes are scattered all over the genome independently of the type of region. Neither did the host (pine vs poplar) nor the environmental condition (greenhouse vs. in vitro) influence this observation (**Supplementary Figures S4A–C**). Furthermore, gene location had an impact on the median level of gene expression (rpkm). Genes located in GSR (repeat-rich) tended to be expressed at a lower level than genes located in GDR (repeatsparse) (**Supplementary Figure S4D**). This suggests an impact of repeats on the gene expression level.

One third (34%) of the genes encoding for the predicted C. geophilum secretome were located in GSR, which parallels the proportion found for the full-predicted proteome (33%; **Supplementary Figure S5**). Within the secretome, the categories "SSPs" and "other SPs" tended to show more members in GSR (37%) compared to the secreted CAZymes, lipases or proteases (28, 23, and 33%, respectively; **Supplementary Figure S5**) but no significant enrichment was observed. Again, we did not notice a difference in in planta gene regulation whether secretomeencoding genes were located in GDR or GSR (data not shown).

The 22 C. geophilum MiSSPs were present in both types of regions (**Figure 3**). Interestingly, when MiSSPs are duplicated, one copy is located in the GDR and one in GSR. Two duplications likely occurred at the same event since the genes were neighbors in both compartments but with invaded repeats in the repeat-rich region (**Supplementary Figure S6**).

### MiSSP Encoding Genes Show Presence–Absence Polymorphism across *Cenococcum geophilum* Isolates

Re-sequencing data from C. geophilum isolates originating from different European countries (Switzerland, France, Poland and Finland) and genome data of closely related species, Glonium stellatum and Lepidopterella palustris allowed to compare SSP sequences for polymorphism (presence–absence) among fungal strains. For seven MiSSPs, PCR amplifications were performed to verify the presence or absence and PCR products were sequenced. Six C. geophilum MiSSPs (Cenge3:666290, Cenge3:723230, Cenge3:664950, Cenge3:28058, Cenge3:661585 and Cenge3:667330) were present in all C. geophilum isolates but were missing in either G. stellatum or L. palustris or in both (**Figure 4**, **Supplementary Table S9**). Two MiSSPs (Cenge3:668273 and Cenge3:552209) and all analyzed CEGs were present in all C. geophilum isolates as well as in G. stellatum and L. palustris. All other MiSSPs were dispersed among C. geophilum isolates. Most conserved MiSSPs (6 of 8; 75%), i.e., those that are present in all C. geophilum strains, localized in GDR, whereas only 36% (5 of 14) of the non-conserved MiSSPs did (**Figure 4C**). The entire set of 22 MiSSPs were present in only three of the four isolates originating from the same site as the sequenced strain. Based on phylogeny analysis using the internal transcribed spacer (ITS) and the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as marker genes (Obase et al., 2016), these three strains were closely related to the sequenced one and clustered within the clade 5 according to the nomenclature of Obase et al. (2016) (**Figure 4A**; **Supplementary Figure S7**). The clades 5 and 6 likely correspond to cryptic species and within clade 5, even more subdivisions are indicated based on species delimitation analyses (Obase et al., 2016; here divided in clade 5a and 5b). When looking at similarities in presence–absence of the 22 MiSSPs



FC, Fold change; SiP, Signal Peptide; ECM,

ectomycorrhizal

 roots. GDR, Gene dense region; GSR, Gene sparse region.

among the 16 C. geophilum isolates, clade affiliation explained best the polymorphism, whereas country origin marginally and the forest type (and therefore potential plant host) did not significantly explain these patterns (**Supplementary Figure S8**).

### *C. geophilum* MiSSPs Accumulate in Distinct Plant Subcellular Compartments

To determine possible in planta subcellular location of the 22 C. geophilum MiSSP, we cloned their coding DNA sequence (CDS) without the signal peptide (mature form of protein) in an expression vector to obtain MiSSPs fused to a green fluorescent protein (GFP) and expressed them into tobacco leaf cells. 21 C. geophilum MiSSP::GFP fusions emitted a detectable fluorescent signal using confocal microscope (**Figure 5**). The fluorescent signals of Cenge3:552209-GFP, Cenge3:667330-GFP, and Cenge3:659858-GFP accumulated in the plasma membrane, endoplasmic reticulum, and tonoplast, respectively. The signals of Cenge3:679266-GFP and Cenge3:634429-GFP accumulated in small cytosolic bodies (**Figure 5**). The displayed fluorescent signal in specific subcellular compartments was markedly different from GFP controls and the localization was confirmed by co-expression of specific organelle plant markers (Nelson et al., 2007; **Figure 5**). All other C. geophilum SSP-GFP showed an uninformative subcellular distribution in the nucleus and cytosol as did the GFP control (**Supplementary Figure S9**). It is important to consider that these localizations were obtained using a 35S promoter and GFP (a protein with triple size of our protein of interest) as a tag. Both actions could result to different localization from those observed when MiSSPs are delivered by the symbiont. Immunoblotting experiments demonstrated both protein production and the integrity for 20 fusion proteins displayed a band at the expected protein size, confirming their integrity. In contrast, one fusion protein (Cenge3:698167) showed no detectable fluorescent signal and no bands on the immunoblots and another (Cenge3:679266) was localized at cytosolic bodies but the integrity could not be confirmed (**Supplementary Figure S10**). In conclusion, we showed that four distinct subcellular compartments are targeted

eukaryotic genes (CEGs) across the C. geophilum strains and the two closely related species. (C) Gene density localization is indicated for 22 MiSSPs present in gene sparse (blue) or gene dense region (red). Asterisks next to the Protein IDs indicate that the presence–absence was confirmed by PCR. Presence–absence patterns are

by the selected C. geophilum MiSSPs, including the plasma membrane, tonoplast, cytosol and endoplasmic reticulum.

hierarchically clustered (top). Color code for gene duplication was indicated (top).

### DISCUSSION

Cenococcum geophilum is a cosmopolitan ECM fungus well known for its extremely wide range of host plants and habitats (LoBuglio, 1999). Although some intraspecific variation in host specificity may exist among different isolates of C. geophilum as indicated by a re-synthesis experiment of this species (Antibus et al., 1981) and as known for other generalist ECM species (Le Quéré et al., 2004), the studied C. geophilum 1.58 isolate forms ECMs with diverse hosts both gymnosperms (Pinus sylvestris, Picea abies) and angiosperms (Populus spp., Quercus spp.; M. Peter unpublished). To understand its communication strategy with different symbiotic partners, we compared the gene regulation of secreted proteins and found that the same gene sets were used and similarly expressed in ECMs of C. geophilum formed with either host. Only very few genes were differentially expressed, which is astonishing for so different trees as are the gymnosperm pine and the angiosperm poplar. Only few studies about host specific interactions are available from ECM systems. The generalist Laccaria bicolor also expressed a core gene regulon when interacting with the two different hosts Pseudotsuga menziesii and Populus trichocarpa but almost 80% of the about 4,000 up-regulated genes were specific to one of the host trees, among which many secreted proteins (Plett et al., 2015). On the contrary, a very similar set of genes was expressed in compatible interactions between species of the Suillus genus, considered as specialists on Pinaceae, interacting with different Pinus species and significant differences were mainly observed in incompatible interactions (Liao et al., 2016). Generalists such as the root endophyte Pirifomospora indica (Lahrmann et al., 2013) and the plant pathogen Sclerotinia sclerotiorum (Guyon et al., 2014) clearly induced host-specific gene sets whereas the specialist barley powdery mildew pathogen Blumeria graminis f. sp. hordei expressed very similar gene sets when interacting with divergent hosts such as monocots and dicots (Hacquard et al., 2013). Being a very broad generalist, we therefore expected C. geophilum to rather show a host-specific regulon, which was not the case with only 8% of all up-regulated genes being hostspecific. More work is needed to see whether such a uniform response of C. geophilum holds true for other host trees and in more natural systems such as in greenhouse trials. The small set of differentially regulated genes likely corresponds to the fine-tuning necessary for the interaction with each host tree. A more than five-fold higher difference in transcript abundance was detected mainly for CAZymes during the interaction between C. geophilum and pine and for MiSSPs in C. geophilum poplar ECM. Since pine and poplar roots have different cell wall compositions (Sarkar et al., 2009), a different set of cell-wall loosening enzymes could be necessary for the penetration of hyphae and development of the Hartig net. For instance, the two GH28 acting on pectins, a GH5-4 acting on hemi-/cellulose and a GH45 a cellulase with similarity to plant expansins that have an important role in plant cell wall loosening (Cosgrove, 2000)

are among the more highly expressed genes in pine ECM. The colonization of the gymnosperm P. menziesii by L. bicolor was also accompanied by a high number of differentially expressed CAZymes with increased abundances; among these were GH5 and GH28. On the opposite, L. bicolor interacting with poplar roots under semi-sterile greenhouse conditions expressed a dozen of SSPs specifically in interaction with poplar (Plett et al., 2015), which fits well with our observations.

cytosolic bodies in Nicotiana benthamiana leaf cells.

Expression profiling data from in vitro ECM produced in this work revealed that several MiSSPs up-regulated in greenhouse (Peter et al., 2016) were not regulated in the in vitro syntheses even when interacting with the same host tree. This suggests that environmental factors are equally important for the regulation of MiSSPs. Proteomic analysis of the secretome of Hebeloma cylindrosporum free-living mycelium revealed that 17% of the secreted proteins were SSPs (Doré et al., 2015). These SSPencoding genes were differentially regulated in ECM root tips and depending on the environmental conditions. Likewise, gene profiling of extramatrical mycelium and sclerotia of C. geophilum performed in the present study showed that some MiSSPs were not only up-regulated in the ECM root tip itself, but also in other fungal tissues in the presence of a host plant, which indicates that they are induced by the host plant but play a role in biological processes not directly linked to the fungal-plant communication at the symbiotic interface. All these data strengthen the concept that fungal SSPs not only play an important role as candidate effector genes in fungal-plant interactions, but also in the adaptation to their environment and saprotrophic growth (Doré et al., 2015).

C. geophilum is the first ectomycorrhizal fungus showing a bipartite genome architecture with repeat-rich, gene-poor regions and vice versa. This genome compartmentalization refers to regions with uneven mutation rates, GC-content and gene density with the gene-sparse, repeat-rich compartments evolving at higher rates (Raffaele and Kamoun, 2012; Plissonneau et al., 2017). This phenomenon has first been described for the oomycete Phytophtora infestans genome (Haas et al., 2009) and recently, a convergence toward similar genome architecture has been demonstrated in phylogenetically unrelated fungal plant pathogens such as Leptosphaeria maculans (Rouxel et al., 2011; Grandaubert et al., 2014) and Zymoseptoria triticii (Stukenbrock et al., 2010). In all these plant-pathogens, the two-speed-genome explained genomic plasticity in order to increase and adapt the repertoire of effector/avirulence genes (Möller and Stukenbrock, 2017). The rapidly evolving compartment in pathogen genomes is largely controlled by transposable elements (Plissonneau et al., 2017) being hot spots for duplication, deletion, and recombination as well as local mutagenesis through TE-silencing mechanisms such as repeat induced point mutation (RIP; Dong et al., 2015). The high content of transposable elements in C. geophilum (75% of genome; Peter et al., 2016) might therefore also play an important role in genome adaptation and since TEs are arranged in different compartments, these might also evolve at different rates. Moreover, in plant pathogens, genes implicated in virulence and host adaptation such as effector genes tend to localize in repeat-rich, faster evolving regions (Raffaele and Kamoun, 2012). The genes coding for the selected 22 MiSSPs of C. geophilum are localized not only in repeat rich but also in gene dense regions. This is true also for other in plantainduced genes and differs from what has been found for the pathogenic oomycete P. infestans (Raffaele et al., 2010a). We noticed that conserved MiSSPs are rather located in gene-dense regions, whereas those that are dispersed among C. geophilum isolates are more often found close to repeats. This indicates that these MiSSPs might evolve at different rates. Likewise, duplicated MiSSPs that each have a member in both regions show different presence–absence patterns among C. geophilum isolates, and are therefore evolving differentially. Subcellular localization analyses for one of the three duplications indicate that the duplicated genes might have different functions since one member targeted the tonoplast within the plant cell (Cenge3:659858) whereas the other showed no specific localization (Cenge3:660401). Interestingly, gene expression and regulation for duplicated genes located in either repeat-rich regions or gene-dense regions were different, indicating that TEs might affect promotor regions and thereby gene regulation and/or that duplicated genes play different roles in the fungal-host interaction. Clearly more population genomic and functional studies are needed to elucidate the evolutionary rate of change of these duplications and its functional significance for adaptation.

Intra- and interspecies comparisons of MiSSP presence revealed that some MiSSPs are conserved not only in C. geophilum isolates but also in saprotrophic relatives. Distinct factors can contribute to secretome variation and evolution such as host specificity, phylogenetic history and lifestyle (Kohler et al., 2015; Pellegrin et al., 2015; Liao et al., 2016; Kamel et al., 2017). The genomes of ECM fungi analyzed so far, share a large set of SSPs with brown and white rot fungi and litter decayers (Kohler et al., 2015; Pellegrin et al., 2015; Martin et al., 2016) and these are likely involved in conserved processes such as developmental changes (e.g., hyphal aggregation in mycelia and fruiting body formation), saprotrophic growth or soil environmental interactions, as indicated by the present and other gene expression studies (Doré et al., 2015). Further, intraspecific analyses show that several species-specific MiSSPs found in the reference genome 1.58 are absent in other C. geophilum strains. This dispersion of MiSSPs within C. geophilum isolates, which to some extend reflect phylogenetic sub-clades, suggests that gene gain and loss may be an important driver of evolution as shown for pathogenic fungi (Syme et al., 2013; van Dam et al., 2016; Hartmann and Croll, 2017). Whether sub-clades identified with commonly used phylogenetic marker genes do reflect cryptic species, needs to be evaluated using genome-wide polymorphism analyses of additional isolates of C. geophilum populations. Likewise, the role mating, but also TE activity, which has been suggested as possible driver of cryptic speciation in plant systems (Bonchev and Parisod, 2013), play in microevolutionary processes within this taxon remains to be determined.

Over the past 5 years, information about secretome repertoires with a particular emphasis on SSPs became available for fungi with different ways of life (Guyon et al., 2014; Lo Presti et al., 2015; Pellegrin et al., 2015; Kamel et al., 2017). However, only a few studies provide data regarding the role of these fungal SSPs in mutualistic plant-microbe interactions. One limitation is that the majority of SSPs are orphan genes that have no domains of known function. Tools to assess their functional role are scarce for species that cannot be transformed and easily handled in the laboratory, as are mycorrhizal fungi. One available approach to elucidate their role is to first identify their subcellular localization in an heterologous system and then try to identify their potential targets (Alfano, 2009). Confocal microscopy assays revealed that five C. geophilum MiSSPs over the 22 tested target distinct sub-cellular compartments, such as plasma membrane, cytosol, endoplasmic reticulum and tonoplast.

Only six MiSSPs showed known domains or homology to other proteins. For example, the cupin domain containing MiSSP, which was conserved in all fungal strains studied here, localized to the plasma membrane and was only induced in ECM of pine. This domain is found in a wide variety of functionally diverse proteins in eukaryotes and prokaryotes (Dunwell et al., 2000) which does not allow speculating about possible functions of this MiSSP. A second MiSSP contains a NTF2 super family domain, probably similar to the one in SnoaL polyketide cyclase. These domains are found in several organisms, including filamentous fungal phytopathogens and present different functions within the proteins, including both enzymatic and non-enzymatic versions (Eberhardt et al., 2013; Deng et al., 2017). This MiSSP was conserved among C. geophilum isolates and L. palustris and localized in plant endoplasmic reticulum. The third MiSSP was the most highly expressed and up-regulated (Cenge3:698167) in ECM roots and to a lower degree in root-associated sclerotia in all ECM experiments. Unfortunately, sub-cellular localization experiments were unsuccessful for this particular MiSSP. It shows similarity (66%) to the "secreted in xylem 1" (Six1) effector of the asexual, soil inhabiting ascomycete Fusarium oxysporum that can switch from a saprotrophic to a pathogenic lifestyle infecting plant roots (Rep et al., 2004). Although several studies have been performed on this protein, the exact function of it is still unclear. This protein (and homologs of it) has been shown to be secreted in root xylem vessels, the gene expression being induced in early root infection only by living cells and that it plays a role in virulence (Rep et al., 2004; Van Der Does et al., 2008; Li et al., 2016). Only strains of the polyphyletic formae speciales lycopersici causing tomato wilt have the genomic region containing Six1 (Van Der Does et al., 2008). In C. geophilum, this MiSSP was only present in a few strains and it remains to be determined, whether it is highly expressed in ECM roots of all these strains and what role it could play in the symbiotic fungal-plant interaction.

None of the MiSSPs analyzed in this work were predicted to localize in the nucleus. We expected such a localization as the two symbiotic effectors characterized in mycorrhizal fungi so far, MiSSP7 and SP7, and many effectors from pathogens target the nucleus (Kloppholz et al., 2011; Plett et al., 2011; Petre et al., 2015). Two MiSSPs (Cenge3:679266 and Cenge3:634429) formed cytosolic bodies and the irregularity of the bodies suggesting that they might be artefactual aggregates, as proposed for localization of rust effectors (Petre et al., 2015; Qi et al., 2018). It is important to consider that these localizations were obtained using GFP as a tag, which can interfere with MiSSPs localization as it is a large fluorescent tag of ∼27 kDa (Varden et al., 2017). However, the proportion of informative localizations as well as identified compartments are consistent with similar studies on SSPs of filamentous plant-pathogenic fungi (Caillaud et al., 2012; Chaudhari et al., 2014; Petre et al., 2015, 2016; Germain et al., 2017; Varden et al., 2017; Qi et al., 2018). Next steps now are to localize the MiSSPs in root cells of host plants and to search for direct plant targets using co-immunoprecipitation/mass spectrometry. The confirmation that a MiSSP is an authentic symbiotic effector requires the demonstration that it is essential for symbiosis development and that it has the ability to interfere with a host component to conclusively support symbiosis. A promising approach in this respect is to use double stranded interfering (dsi) RNA to knock down transcription of MiSSPs, a method successfully applied in fungal-plant systems for which efficient transformation protocols are lacking (Wang et al., 2016).

### AUTHOR CONTRIBUTIONS

Conceived and designed the experiments: MP, AK, CV-F, FM, MdFP, and KB; Performed the experiments: MdFP, PV, FG, SP, HN, and MK; Analyzed sequence data: MdFP, EM, AK, CV-F, VS, AL, and MP; Drafted the manuscript: MdFP, AK, CV-F, and MP; Revised the manuscript: MdFP, MP; AK, CV-F, SE, IG, and FM. All authors read and approved the final manuscript.

### ACKNOWLEDGMENTS

We would like to thank Barbara Meier and Ursula Oggenfuss for the great help in the set-up of the mycorrhization systems at WSL. This project was supported by grants from the French National Agency of Research (ANR) as part of the "Investissement d'Avenir program (ANR-11\_LABX-0002-01) of Labex ARBRE (CFP15) and the WSL within the frame of the ARBRE/WSL project 'Blacksecret' as well as by the Region Lorraine Research Council who provided a 6-month researcher grant to MFP and the EC-supported Network of Excellence Evoltree (GOCE-016322 to MP).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00141/full#supplementary-material

Supplementary Figure S1 | Ectomycorrhiza formed by Cenococcum geophilum and their host plants. Morphological characteristics of typical ectomycorrhiza formed by C. geophilum in interaction with Pinus sylvestris (A) and Populus tremula L. × Populus alba L.-INRA 717) (C). Cross-sections of ectomycorrhiza roots of both system shows a presence of the Hartig net between epidermal and cortex cells in both interactions (B,D). Percentage of ectomycorrhiza formation between C. geophilum and P. sylvestris and C. geophilum and P. tremula × P. alba (E).

Supplementary Figure S2 | Comparison of C. geophilum gene expression changes in greenhouse pine ECM, in vitro pine ECM and in vitro poplar ECM. Venn diagram based on the comparison of secreted proteins regulated in at least one experiment: ECM formed by C. geophilum with Pinus sylvestris [in vitro synthesis or greenhouse (Peter et al., 2016)] and C. geophilum with Populus tremula × Populus alba L.-INRA 717 (in vitro synthesis). Note that the age of control mycelium was different: C. geophilum with P. sylvestris [in vitro synthesis = 90 days, C. geophilum with P. sylvestris greenhouse = 15 days; C. geophilum with Populus tremula × P. alba in vitro synthesis = 60 days. Data are provided in Supplementary Table S6.

Supplementary Figure S3 | Nucleotide (A–C) and protein (D–F) alignments of duplications of candidate MiSSPs in the C. geophilum genome. (A,D) Cenge3:636312 and Cenge3:660403, (B,E) Cenge3:679266 and Cenge3:693798, (C,F) Cenge3:660401 and Cenge3:659858. Protein ID from Joint Genome Institute (JGI).

Supplementary Figure S4 | Distribution of gene expression induction in ectomycorrhizal root tips compared to free-living mycelium according to local gene density for all genes. The median (A), minimum (B) or maximum (C) induction (log2 ratio ECM vs. FLM) values associated to genes in each bin are shown as a color-coded heat map. (D) Distribution of the average gene expression level in ectomycorrhizal root tips according to local gene density. The median values for gene expression in each bin are shown as a color-coded heat map. Data are presented for ECM root tips of C. geophilum and P. sylvestrissemi-sterile under greenhouse conditions (left column) or in vitro system (middle column) and for C. geophilum- Populus tremula x alba in vitro (right column).

Supplementary Figure S5 | Percentage and number of genes found in gene-dense repeat sparse or gene sparse repeat rich regions for the proteome and the secretome of C. geophilum. The secretome was categorized into functional categories (proteases, lipases, CAZymes, SSPs and other secreted proteins). Enrichment tests were not significant.

Supplementary Figure S6 | Genomic landscape of compartments on scaffold 21 and 23 of Cenococcum geophilum harboring duplications of MiSSPs in gene-dense and gene-poor, repeat-rich regions. Displays are extracted from the genome viewer of the Joint Genome Institute (JGI) website (https://genome.jgi. doe.gov/Cenge3/Cenge3.home.html) showing tracks of base position, GC content, predicted genes (GeneCatalog; dark blue), and predicted repetitive regions (black, 3 tracks) discovered by RepeatScout and masked by RepeatMasker.

Supplementary Figure S7 | Phylogenetic tree of C. geophilum strains and the closest relative Glonium stellatum reconstructed based on concatenated nucleotide sequences of the internal transcribed spacer (ITS) and the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) using PhyML-maximum likelihood. In addition to the 15 C. geophilum strains from the present study, six representative strains of the six clades from the study of Obase et al. (2016) were included in the analysis. Branch confidence indices were calculated using an approximate likelihood ratio test. The scale bar indicates the number of nucleotide substitutions per site. Three distinct clades are indicated and numbered according to Obase et al. (2016) including a possible subdivision of clade 5 (left). Glonium stellatum was designated as the outgroup.

Supplementary Figure S8 | Variability in presence/absence of 22 MiSSP genes among 16 C. geophilum isolates. The first two axes of a principal coordinate analysis based on the Jaccard similarity index are provided. Each symbol represents an isolate originating from the given country with isolates closer to each other showing more similar presence/absence patterns. In (A), different symbols indicate the phylogenetic clade the isolate are grouped into based on a concatenated dataset of the ITS and GADPH regions (Obase et al., 2016). In (B), different symbols indicate the forest type with the dominating tree species: Mx, mixed forest; Pa, Picea abies; Fs, Fagus sylvatica; Ps, Pinus sylvestris. (C) PERMANOVA table showing the effects of phylogenetic clade, country of origin

#### REFERENCES


and forest type of isolation on the MiSSP presence/absence patterns in the 16 C. geophilum isolates. Analyses were performed with the Primer-E software (Clarke and Gorley, 2015).

Supplementary Figure S9 | Candidate effectors with no informative localization in planta. Representative images corresponding to the 13 fusion proteins accumulating in the nucleoplasm and the cytosol. The fusion proteins were transiently expressed in Nicotiana benthamiana leaf cells by agroinfiltration. Live-cell imaging was performed with a laser-scanning confocal microscope 2 days after infiltration. The green fluorescent protein (GFP) was excited at 488 nm. GFP (green) fluorescence was collected at 505–525 nm.

Supplementary Figure S10 | Immunoblots of CgMiSSPs:GFP fusion proteins in N. benthamiana leaves. GFP detection was performed in a single step by a GFP-HRP conjugated antibody. The theoretical size of each fusion protein (SSP+GFP) is indicated between parentheses in kiloDalton (kDa). Page rulers and corresponding sizes in kiloDalton (kDa) are indicated on the blots. White asterisks indicate specific protein bands.

Supplementary Table S1 | Cenococcum geophilum and other fungal strains used in this work.

Supplementary Table S2 | Main features of C. geophilum RNAseq data.

Supplementary Table S3 | Main features of C. geophilum re-sequencing data.

Supplementary Table S4 | Core eukaryotic genes selected for presence/absence polymorphism analysis.

Supplementary Table S5 | Concatenated and separated sequences of C. geophilum glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and internal transcribed spacer (ITS).

Supplementary Table S6 | List of primers used to amplify seven MiSSPs in C. geophilum mycelium.

Supplementary Table S7 | Genes encoding small secreted proteins regulated in the interaction between Cenoccoccum geophilum and Pinus sylvestris or Populus tremula × alba as compared to free-living mycelium (FLM).

Supplementary Table S8 | Expression comparison of genes encoding secreted proteins significantly regulated in the interaction between Cenococcum geophilum and Pinus sylvestris or Populus tremula × alba. A fold change between Pine and Poplar ECM was calculated and a cut-off 5 fold was established to show specificity expression in each tissue.

Supplementary Table S9 | Blast results of C. geophilum 1.58 MiSSP candidates against de novo assemblies of C. geophilum strains. The genes that are absent in respective strains based on e-value and HSP length are marked in red. A fasta file with respective contigs is available (Supplementary File 1).

Supplementary File 1 | C. geophilum nucleotide sequences contig containing MiSSPs of respective C. geophilum strains.


Clarke, K., and Gorley, R. (2015). PRIMER v7, User Manual/Tutorial.


an emphasis on small-secreted proteins. Front. Microbiol. 6:1278. doi: 10.3389/fmicb.2015.01278


**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 © 2018 de Freitas Pereira, Veneault-Fourrey, Vion, Guinet, Morin, Barry, Lipzen, Singan, Pfister, Na, Kennedy, Egli, Grigoriev, Martin, Kohler and Peter. 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 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 SlZRT1 Gene Encodes a Plasma Membrane-Located ZIP (Zrt-, Irt-Like Protein) Transporter in the Ectomycorrhizal Fungus Suillus luteus

Laura Coninx<sup>1</sup> , Anneleen Thoonen<sup>1</sup> , Eli Slenders<sup>2</sup> , Emmanuelle Morin<sup>3</sup> , Natascha Arnauts<sup>1</sup> , Michiel Op De Beeck<sup>1</sup>† , Annegret Kohler<sup>3</sup> , Joske Ruytinx<sup>1</sup> \* and Jan V. Colpaert<sup>1</sup>

<sup>1</sup> Environmental Biology, Centre for Environmental Sciences, Hasselt University, Hasselt, Belgium, <sup>2</sup> Biomedical Research Institute, Hasselt University, Hasselt, Belgium, <sup>3</sup> Institut National de la Recherche Agronomique, Laboratoire d'Excellence ARBRE, UMR 1136, Université de Lorraine Interactions Arbres/Microorganismes, Champenoux, France

#### Edited by:

Erika Kothe, Universität Jena, Germany

#### Reviewed by:

Marcela Claudia Pagano, Universidade Federal de Minas Gerais, Brazil Oswaldo Valdes-Lopez, Universidad Nacional Autónoma de México, Mexico

#### \*Correspondence: Joske Ruytinx

joske.ruytinx@uhasselt.be

#### †Present address:

Michiel Op De Beeck, Microbial Ecology, Department of Biology, Lund University, Lund, Sweden

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 24 August 2017 Accepted: 10 November 2017 Published: 28 November 2017

#### Citation:

Coninx L, Thoonen A, Slenders E, Morin E, Arnauts N, Op De Beeck M, Kohler A, Ruytinx J and Colpaert JV (2017) The SlZRT1 Gene Encodes a Plasma Membrane-Located ZIP (Zrt-, Irt-Like Protein) Transporter in the Ectomycorrhizal Fungus Suillus luteus. Front. Microbiol. 8:2320. doi: 10.3389/fmicb.2017.02320 Zinc (Zn) is an essential micronutrient but may become toxic when present in excess. In Zn-contaminated environments, trees can be protected from Zn toxicity by their rootassociated micro-organisms, in particular ectomycorrhizal fungi. The mechanisms of cellular Zn homeostasis in ectomycorrhizal fungi and their contribution to the host tree's Zn status are however not yet fully understood. The aim of this study was to identify and characterize transporters involved in Zn uptake in the ectomycorrhizal fungus Suillus luteus, a cosmopolitan pine mycobiont. Zn uptake in fungi is known to be predominantly governed by members of the ZIP (Zrt/IrtT-like protein) family of Zn transporters. Four ZIP transporter encoding genes were identified in the S. luteus genome. By in silico and phylogenetic analysis, one of these proteins, SlZRT1, was predicted to be a plasma membrane located Zn importer. Heterologous expression in yeast confirmed the predicted function and localization of the protein. A gene expression analysis via RT-qPCR was performed in S. luteus to establish whether SlZRT1 expression is affected by external Zn concentrations. SlZRT1 transcripts accumulated almost immediately, though transiently upon growth in the absence of Zn. Exposure to elevated concentrations of Zn resulted in a significant reduction of SlZRT1 transcripts within the first hour after initiation of the exposure. Altogether, the data support a role as cellular Zn importer for SlZRT1 and indicate a key role in cellular Zn uptake of S. luteus. Further research is needed to understand the eventual contribution of SlZRT1 to the Zn status of the host plant.

Keywords: Suillus luteus, mycorrhiza, zinc transporter, zinc homeostasis, zinc deficiency, metal uptake

### INTRODUCTION

Zinc (Zn) is an essential micronutrient required by all living organisms (Eide, 2006). Due to its unique set of chemical properties, Zn plays both a functional and a structural role in many proteins. The metal is characterized by a small radius and has thus a highly concentrated charge (Clemens, 2006a). Furthermore, Zn is a Lewis acid with a high affinity for ligands with sulfur- (S),

**40**

nitrogen- (N), and oxygen- (O) containing functional groups (Clemens, 2006a). Due to its full d-subshell, Zn is also able to interact with these ligands more flexibly than other transition metals (Berg and Shi, 1996). By supporting a tetrahedral coordination geometry, Zn allows proteins to quickly shift conformations in biological reactions. These properties combined with its lack of redox activity, explain why Zn is a valuable structural element (e.g., Zn finger proteins) and a catalyst for many enzymes (e.g., hydrolytic enzymes) (Frausto da Silva and Williams, 2001). However, the same properties that make Zn an indispensable nutrient can also induce toxicity (Clemens, 2006b). When present in excess, Zn can cause protein damage and inactivation by uncontrolled high-affinity binding to functional groups within proteins (Clemens, 2006b). For this reason, it is necessary for living cells to tightly regulate Zn concentrations in the cytoplasm. The physiological concentration range of Zn between deficiency and toxicity is extremely narrow and organisms are consequently equipped with a number of homeostatic mechanisms to tightly regulate cytoplasmic Zn concentrations (Eide, 2006). Especially transporter proteins play a crucial role in maintaining Zn homeostasis (Eide, 2006).

In eukaryotes most of the Zn transport is achieved by two protein families: the ZIP (Zrt/Irt-like protein) and CDF (Cation Diffusion Facilitator) transporter families (Gaither and Eide, 2001). Proteins belonging to the ZIP transporter family increase cytoplasmic Zn levels by transporting Zn across the plasma membrane or by mobilizing stored Zn from intracellular compartments. Whereas members of the CDF family transport Zn in the direction opposite to that of the ZIP proteins. Efflux or compartmentalization of Zn is promoted by transporting Zn from the cytoplasm into the lumen of an organelle or out of the cell (Gaither and Eide, 2001).

Transporters belonging to the ZIP family typically possess 5 to 8 transmembrane domains (TMDs). The protein sequence is most conserved in TMD IV and the region adjacent to TMD IV (Eng et al., 1998). The ZIP family can be divided into four subfamilies based on a higher degree of sequence similarity: the ZIP I, ZIP II, GufA and LIV-I subfamily (Guerinot, 2000; Gaither and Eide, 2001). Members of the ZIP family are well-studied in Saccharomyces cerevisiae, which is an excellent fungal model system to investigate Zn uptake and efflux (Zhao and Eide, 1996). Currently, four ZIP Zn transporter genes have been identified in baker's yeast: ZRT1, ZRT2, ZRT3, and YKE4. The yeast ZRT1 gene was the first influx Zn transporter gene from any organism to be characterized at the molecular level (Zhao and Eide, 1996). The ZRT1 gene encodes a high-affinity Zn uptake system induced by Zn limitation, whereas the ZRT2 transporter corresponds to a low-affinity uptake system that is active in Zn repleted cells (Eide, 1996; Zhao and Eide, 1996). Zn uptake in yeast is predominantly governed by these two plasma membrane-located transporters (Eide, 1996; Zhao and Eide, 1996). Both transporters are included in the ZIP I subfamily (Gaither and Eide, 2001). The third characterized yeast ZIP protein, ZRT3, belongs to the GufA ZIP subfamily (Gaither and Eide, 2001). This transporter localizes to the vacuolar membrane and mobilizes Zn under Zn deficiency (MacDiarmid et al., 2000). Lastly, Kumánovics et al. (2006) characterized YKE4, a bidirectional Zn transporter in the endoplasmic reticulum (ER) of S. cerevisiae, which regulates Zn concentrations in the ER and cytoplasm. YKE4 is a LIV-I subfamily transporter (Gaither and Eide, 2001).

Additional to the research in yeast, other ZIP transporters and mechanisms of Zn homeostasis in fungi are primarily characterized and studied in human fungal pathogens. Membrane Zn importers of the ZIP I subfamily have been shown to be crucial for the acquisition of Zn and the virulence of several human pathogenic fungi (Crawford and Wilson, 2015). This was observed in Candida albicans for the Zn transporter CaZRT1 (Citiulo et al., 2012), in Cryptococcus neoformans for CnZIP1 (Do et al., 2016) and in Aspergillus fumigatus for AfZrfC (Amich et al., 2014). These transporters enable pathogenic fungi to overcome Zn deficiency within the Zn-limited host environment (Jung, 2015). Zn and fungal ZIP transporters are therefore considered to be key players in this kind of pathogenic host-microbe interactions.

In the current study we aim to identify plasma membrane localized Zn importers and their role in cellular Zn homeostasis in the ectomycorrhizal fungus Suillus luteus. Ectomycorrhizae are mutualistic host-microbe interactions between tree roots and ectomycorrhizal fungi. The mycobiont offers the tree a balanced nutrient supply in exchange for photosynthetic sugar (Martin et al., 2016). Although Zn is not expected to be a key regulator of ectomycorrhizal development nor to be extremely scarce at the symbiotic interface, availability of this element may have an impact on the fitness of both individual symbiotic partners and the mutualism in particular environments. Micronutrient deficiencies are rarely observed in natural forests but severe Zn deficiency in tree plantations has been reported previously (Thorn and Robertson, 1987; Boardman and McGuire, 1990). Moreover, trees are sensitive to high soil Zn concentrations. We previously demonstrated that well-adapted ectomycorrhizal fungi can protect host trees from Zn toxicity when Zn is present in excess (Adriaensen et al., 2004, 2006). An improved knowledge on the mechanisms of cellular Zn homeostasis in ectomycorrhizal fungi, going beyond the general focus on detoxification by vacuolar sequestration and including Zn uptake and deficiency, will be the first step toward a better understanding of the contribution of ectomycorrhizal fungi to host tree Zn homeostasis.

#### MATERIALS AND METHODS

#### Fungal Strains and Culture Conditions

The monokaryotic S. luteus isolate UH-Slu-Lm8-n1 (Kohler et al., 2015) and the dikaryotic isolate UH-Slu-P4 (Colpaert et al., 2004) were used in this study. Cultures were maintained on solid Fries medium according to Colpaert et al. (2004). Preceding Zn exposure assays, 1-week-old exponentially growing mycelia were harvested and liquid cultures were initiated and maintained according to Nguyen et al. (2017). Three gram of spherical mycelia grown for 1 week in liquid culture were transferred to petri dishes containing 25 ml modified liquid Fries medium supplemented with 0, 20, 500, or 1000 µM ZnSO4·7H2O. These Zn concentrations were chosen to induce Zn deficiency, Zn

sufficiency and mild Zn toxicity (Ruytinx et al., 2017). The petri dishes were incubated on a shaking incubator at 23◦C. Metal exposure was performed in triplicate. Mycelia (400 mg) were sampled at 0, 1, 2, 4, 8, and 24 h after initiation of exposure, flash frozen in liquid nitrogen and stored at −70◦C.

### ZIP Identification and Phylogenetic Tree Construction

The S. luteus reference genome was searched for ZIP transporter encoding genes. A BLASTp search using characterized fungal ZIP transporters (Supplementary Table 1) and a Pfam domain search were performed at the S. luteus genome portal at MycoCosm of the Joint Genome Institute (JGI)<sup>1</sup> (Grigoriev et al., 2012; Kohler et al., 2015). Full-length amino acid sequences of previously characterized ZIP transporters were obtained from the transporter classification database<sup>2</sup> , the Swissprot database<sup>3</sup> and the National Center for Biotechnology Information (NCBI) server<sup>4</sup> . All sequences, including the newly identified S. luteus ZIP sequences, were aligned with the Multiple Alignment using Fast Fourier Transform (MAFFT) alignment logarithm version 7 (Katoh and Standley, 2013) and imported into the Molecular Evolutionary Genetics Analysis (MEGA) package version 6.06 (Tamura et al., 2013). A phylogenetic tree was constructed using the neighbor-joining (NJ) method (Poisson correction model for distance computation) to infer evolution of the identified S. luteus ZIP transporters and to predict their function more precisely.

#### Cloning of SlZRT1

Total RNA was extracted from S. luteus mycelium ground in liquid nitrogen using the RNeasy Plant Mini kit (Qiagen, Germany) and a cDNA library was constructed using the SMARTer cDNA synthesis kit (Clontech, United States) according to the manufacturer's instructions. Specific primers were designed for amplification of the full-length coding sequence of SlZRT1 (left: 5<sup>0</sup> CCT CAAACTATGTCAGATTTAAATT 3<sup>0</sup> ; right: 5<sup>0</sup> TGCCCA ACGCCCCAGGAGC 3<sup>0</sup> ). The PCR reaction contained: 10x High Fidelity PCR buffer, 0.2 mM dNTP-mixture, 2 mM MgSO4, 0.2 µM SlZRT1 forward and reverse primer, 5 ng cDNA and 0.5 U Platinum Taq High Fidelity DNA polymerase (Invitrogen, United States). RNase-free water was added to obtain a final reaction volume of 30 µl. The following PCR cycling conditions were used: 2 min at 95◦C; 35 cycles of 30 s at 95◦C + 30 s at 56◦C + 1 min at 68◦C, and 1 cycle of 3 min at 68◦C. 5 µl of the PCR product was visualized on an agarose gel to verify the reaction specificity and the length of the amplicon. The remaining 25 µl PCR product was purified using the GeneJET PCR Purification Kit (ThermoScientific, United States). The purified PCR-product was cloned into the gateway entry vector pCR8/GW/TOPO (Invitrogen) and subsequently transferred by the Gateway LR-clonase II Enzyme Mix (Invitrogen) to destination vectors pYES-DEST52 (Invitrogen, United States) and pAG426GAL-ccdB-EGFP (Alberti et al., 2007) for functional analysis in yeast. The insert was sequenced in both directions to verify correct orientation and fusion.

#### Yeast Mutant Complementation and Subcellular Localization

SlZRT1 was heterologous expressed in S. cerevisiae. Yeast strains used are CM30 (MATα, ade6, can1-100, his3-11, 15 leu2-3, trp1-1, ura3-52) and CM34 or 1zrt11zrt2 (CM30, zrt1::LEU2, zrt2::HIS3) (MacDiarmid et al., 2000). Yeast cells were transformed according to the LiAc/PEG method as described by Gietz and Woods (2002). Transformed yeast cells were selected on synthetic defined medium without uracil [SD-URA; 0.7% w/v yeast nitrogen base (Difco), 2% w/v D-glucose, and 0.2% w/v Yeast Synthetic Drop-out Mix without uracil (Sigma)]. Plates were incubated at 30◦C.

For metal tolerance assays, transformed yeasts were grown to mid log phase (OD<sup>600</sup> ± 1.5) in liquid SD-URA medium with 2% w/v D-galactose instead of D-glucose (induction medium). Yeast cells were pelleted, washed with sterile distilled water, and adjusted to OD<sup>600</sup> = 1. A 1/10 dilution series was prepared (10<sup>0</sup> , 10−<sup>1</sup> , 10−<sup>2</sup> , and 10−<sup>3</sup> ). Drop assays were performed for three independent yeast clones on SD-URA control induction medium (1 mM Zn) and induction medium supplemented with 50, 100, or 200 µM ethylenediaminetetraacetic acid (EDTA) (MacDiarmid et al., 2000). For subcellular localization of SlZRT1::EGFP fusion proteins, yeast transformants were grown to mid-log phase OD<sup>600</sup> = 1 on induction medium. Plasma membrane of the cells was stained at 0◦C by FM4-64 (Molecular Probes, Invitrogen) according to Vida and Emr (1995). Afterward, a 3 µl droplet of yeast cells was analyzed at 0◦C with a Zeiss LSM 510 META laser scanning confocal microscope (Carl Zeiss, Jena, Germany), using a Zeiss 40x NA1.1 water immersion objective (C-Apochromat 40x/1.1 W Corr., Carl Zeiss). Enhanced green fluorescent protein (EGFP) fluorescence analysis was performed with the 488 nm excitation line of an argon-ion laser and a band-pass 500–550 nm emission filter. FM4-64 (ThermoFisher) fluorescence analyses were performed with a 543 nm HeNe laser and a long-pass 560 nm emission filter. Image processing was carried out with ImageJ (NIH, Bethesda, MD, United States) software.

### Zn Content Analysis of Transformed Yeast

Transformed yeast cells were cultured at 30◦C in liquid induction medium without Zn until culture saturation. Three rounds of Zn deprivation were completed by re-inoculating 0.5 ml of saturated yeast suspension to new Zn-less induction medium. Zn-starved cells were grown to mid log phase (OD<sup>600</sup> ± 1.5) and diluted to OD<sup>600</sup> = 1. One ml of yeast suspension was added to Erlenmeyer flasks containing 20 ml liquid induction medium without Zn and medium supplemented with 500 µM Zn (repletion). Zn treatments were performed for five independent yeast clones. Cultures were allowed to grow for 24 h at 30◦C. Yeast cells were collected by centrifugation, washed three times with 20 mM PbNO<sup>3</sup> and milli-Q water. Afterward cells were resuspended

<sup>1</sup>http://genome.jgi.doe.gov/Suilu2/Suilu2.home.html

<sup>2</sup>http://www.tcdb.org/

<sup>3</sup>http://www.uniprot.org/

<sup>4</sup>http://www.ncbi.nlm.nih.gov/

in 0.5 ml of milli-Q water, frozen (−20◦C) and lyophilized. Lyophilized cells were acid digested (HNO3/HCl) and Zn content was determined by inductively coupled plasma optical emission spectrometry (ICP-OES).

#### RNA Extraction, cDNA Synthesis, and qPCR

Total RNA was extracted from mycelium pulverized in liquid nitrogen using the RNeasy Plant Mini Kit (Qiagen). The TURBO DNA-free kit [Invirtogen (Life Technologies), United States] was used to perform a desoxyribonuclease (DNase) treatment to remove residual genomic DNA. RNA integrity and absence of DNA contamination was verified via agarose gel analysis. RNA concentration and purity were evaluated on a NanoDrop ND-1000 Spectrophotometer (Isogen Life Science, the Netherlands). One µg of each RNA sample was converted to cDNA with the Primescript RT Reagent Kit (Perfect Real Time) (TaKaRa Clontech, United States). A 10-fold dilution of the cDNA was prepared in 1/10 diluted Tris-EDTA (TE) buffer and stored at −20◦C.

Real-time reverse transcription polymerase chain reaction (qRT-PCR) was performed in a 96-well optical plate with an ABI PRISM 7500 Fast Real-Time PCR System (Life Technologies) according to Ruytinx et al. (2016). A SlZRT1 specific primer pair was designed using Primer3 (Rozen and Skaletsky, 2000) (5<sup>0</sup> GCCAAACGGACAAACTGG 3<sup>0</sup> ; 5 <sup>0</sup> GACAGGCACGGAGATGAAAG 3<sup>0</sup> ; efficiency = 92.1%). Data were expressed relative to the sample with the highest expression level via the formula 2−(Ct−Ctmin) and normalized using five reference genes (TUB1, ACT1, GR975621, AM085296, and AM085296). Reference genes were selected previously by Ruytinx et al. (2016) and their stability was confirmed within the current experimental conditions using GeNorm (Vandesompele et al., 2002). A normalization factor was calculated as the geometric mean of the relative expression levels of the reference genes. Mean values of four biological replicates were calculated, rescaled to the control condition (20 µM Zn) within each time point and log2 transformed. A two-way analysis of variance (ANOVA) followed by a Tukey's HSD test was run in "R" version 3.2.2 (R Core Team, 2015) to assess differences in SlZRT1 expression level.

#### RESULTS

#### Identification of a ZIP Transporter in S. luteus

The BLASTp search of the S. luteus genome (UH-Slu-Lm8 n1 v2.0) with characterized fungal ZIP transporters identified four S. luteus genes (protein IDs 720881, 22926, 229544, and 811220) predicted to encode ZIP proteins. A NJ phylogenetic tree including previously characterized ZIP transporters was constructed to predict the function of the newly identified S. luteus genes. The different ZIP subfamilies are well-supported in the tree as indicated by high bootstrap values (>90; **Figure 1**). Three of the identified S. luteus proteins (IDs 720881, 22926, and 229544) cluster within the ZIP I subfamily. Within this subfamily, proteins with ID 720881 and 22926 cluster together with the S. cerevisiae Zn importers ScZRT1 and ScZRT2; the protein with ID 229544 clusters with S. cerevisiae ATX2, a Golgi transporter involved in manganese (Mn) homeostasis. The fourth identified S. luteus gene encodes a protein (ID 811220) clustering close to the S. cerevisiae YKE4 (ER localized Zn transporter) within the LIV-1 subfamily of ZIP transporters. We were not able to detect a member of the Guf A and ZIP II subfamily of ZIP transporters within the S. luteus reference genome.

One identified S. luteus gene, encoding the protein with ID 22926 was selected for further analysis because of its high sequence similarity with the yeast Zn importers ScZRT1 and SpZRT1. Reciprocal BLASTp suggest the S. luteus protein to be orthologous to the high-affinity Zn importers ScZRT1 and SpZRT1 and was therefore named SlZRT1. SlZRT1 is predicted to have a 1398 bp open reading frame with eight exons encoding a 338 amino acid polypeptide. The encoded peptide shows several characteristics that are typical for proteins belonging to the ZIP family (**Figure 2**). Eight TMDs were predicted by the topology program TMHMM and a long variable cytoplasmic loop is present between TMD3 and TMD4. A histidine rich motif HXX(HX)3, suggested to function as Zn binding site, is present in the variable cytoplasmic loop of SlZRT1 and two other histidines that are typically conserved in ZIP transporters were identified (**Figure 2**). One of these conserved histidines is located in the conserved TMD4, which contains the ZIP signature sequence described by Eng et al. (1998). SlZRT1 matches 13 of the 15 amino acids of this ZIP signature sequence.

### Functional Analysis of SlZRT1 in Yeast

SlZRT1 was heterologous expressed in yeast to confirm that it encodes a plasma membrane-located ZIP Zn importer, which was predicted by the phylogenetic analysis. **Figure 3** and **Supplementary Figure 1** illustrate that transformation with SlZRT1 partly restored the growth of the zinc-uptake-deficient yeast strain 1zrt11zrt2 on medium supplemented with different concentrations of EDTA. Transformation with the empty vector did not result in complementation of the Zn deficient phenotype (**Figure 3** and **Supplementary Figure 1**). Expression of the SlZRT1::EGFP fusion protein indicates a localization of SlZRT1 on the plasma membrane. Yeast cells transformed with SlZRT1::EGFP showed a bright green fluorescent ring surrounding the cells, which co-localized with FM4-64 plasma membrane staining (**Figure 4**).

#### Zn and Fe Content Analysis of Transformed Yeast

Zn and Fe content were measured in Zn starved (**Figure 5A**) and Zn replete (**Figure 5B**) yeast cells in order to obtain more insight into the function of SlZRT1. **Figure 5A** illustrates that 1zrt11zrt2 yeast mutants transformed with SlZRT1 contained the same amount of Zn as the wild type (WT) yeast after starvation (0 µM Zn) while 1zrt11zrt2 mutants transformed with the empty vector had a significantly lower

Zn content. Similarly, 24 h after Zn repletion (500 µM Zn) SlZRT1 transformed yeast mutants and WT yeast accumulated significantly more Zn than empty vector transformed yeast mutants (**Figure 5B**). A small difference in Zn accumulation was observed between WT yeast cells and SlZRT1 transformed 1zrt11zrt2 yeast cells. Additionally the Fe content in the yeast transformants was analyzed, since some ZIP transporters can also use Fe as a substrate. Yet, no significant differences in Fe content were observed among the yeast transformants exposed to 0 or 500 µM Zn (**Supplementary Figures S2A,B**).

### SlZRT1 Gene Expression Analysis in S. luteus

In S. luteus, SlZRT1 gene expression was determined at early time points (0, 1, 2, 4, 8, and 24 h) after exposure to different concentrations of Zn [0, 20 (control), 500, and 1000 µM]

FIGURE 3 | Functional complementation of the zinc-uptake-deficient yeast strain 1zrt11zrt2 by SlZRT1. Wild type (WT) and mutant yeast cultures with an OD<sup>600</sup> = 1 were 10-fold serial diluted (10<sup>0</sup> , 10−<sup>1</sup> , 10−<sup>2</sup> , and 10−<sup>3</sup> ) and spotted on control (1 mM Zn) or ethylenediaminetetraacetic acid (EDTA) supplemented synthetic drop-out (SD) medium. The WT strain was transformed with the empty vector (EV, pYES-DEST52; Invitrogen), the mutant strain 1zrt11zrt2 with either the EV or the vector containing SlZRT1. The experiment was carried out for three independent clones and pictures were taken after 4 days of growth.

FIGURE 4 | Localization of the SlZRT1:EGFP fusion protein to the plasma membrane in yeast (a–d). (a) Bright field image, (b) EGFP fusion protein, (c) FM4-64 plasma membrane staining, and (d) merged images. SlZRT1:EGFP and FM4-64 plasma membrane staining co-localize.

to assess the role of the SlZRT1 in Zn homeostasis. Results clearly illustrate that mRNA levels of SlZRT1 are dependent of external Zn concentration (**Figure 6**). Exposure to mildly toxic Zn concentrations (500 and 1000 µM) results in an almost immediate significant downregulation of SlZRT1 gene expression. The expression patterns upon exposure to 500 and 1000 µM Zn are similar regardless differences in external Zn concentrations. In contrast, in the absence of external Zn, SlZRT1 expression is quickly induced to reach a maximum level after 2 h, declines to control levels after 4 h and tends to be higher again in the long term (24 h).

### DISCUSSION

Transition metals, such as Zn, Fe, Mg, are essential to all living organisms. However, when present in excess these metals may become toxic. To overcome metal toxicity, it is crucial for cells to tightly control cytoplasmic metal concentrations (Eide et al., 2005). Metal transporter proteins play a crucial role in the regulation of cytoplasmic metal concentrations and cellular metal homeostasis (Migeon et al., 2010). Among fungi, mechanisms involved in Zn homeostasis are mostly studied in S. cerevisiae. Transporters of the ZIP family were shown to be vital to prevent Zn deficiency in this species (Eide, 2006) and several other fungi (Kiranmayi et al., 2009; Jung, 2015). S. cerevisiae possess two plasma membrane localized Zn importers of the ZIP family (ScZRT1 and ScZRT2) and one tonoplast localized ZIP transporter (ScZRT3) for re-mobilization of vacuolar stored Zn. In the current study, we identified four ZIP transporter encoding genes in the genome of the ectomycorrhizal fungus S. luteus. Three of the newly identified proteins are members of the ZIP I subfamily of ZIP transporters, one belongs to the LIV-1 subfamily (**Figure 1**). With the exception of ScZRT3, a tonoplast transporter

involved in Zn mobilization from the vacuole, homologs for all characterized S. cerevisiae ZIP transporters were identified within the S. luteus genome. So far, no homologs for the ScZRT3 protein have been identified in members of the Basidiomycota. Nevertheless, several basidiomycetes including S. luteus store excess Zn into their vacuoles (Sacky et al., 2016; Ruytinx et al., 2017). Transporters belonging to other protein families likely evolved in these species to re-mobilize stored Zn in absence of external environmental Zn. In accordance with what has been found in other fungi, there was no member of the ZIP II subfamily of ZIP transporters detected in S. luteus. This subfamily consists mainly of metazoan representatives (Guerinot, 2000).

Reciprocal BLASTp suggested the S. luteus protein with ID 22926, named SlZRT1 to be orthologous to the S. cerevisiae ScZRT1 transceptor. ScZRT1 functions as a high-affinity Zn uptake transporter and receptor (Schothorst et al., 2017). Together with its homolog, the plasma membrane transporter ScZRT2, ScZRT1 is responsible for Zn uptake in Zn deficient yeast cells (Gaither and Eide, 2001). SlZRT1 and ScZRT1 show 39% sequence identity. An important difference in the sequence of SlZRT1 and ScZRT1 is found within the putative Zn binding domain (histidine rich domain, HRD) localized within the cytoplasmic loop between TMD3 and TMD4 (**Figure 2**). SlZRT1's binding domain (HDVHGHGHG) shows a HXX additional to the classical (HX)<sup>3</sup> domain of ScZRT1 (HDHTHD). This difference might correspond to an altered affinity toward Zn and/or a modified function of the protein. Mutation of the histidines in the HRD of ScZRT1 results in a 70% reduction in the maximum uptake rate of ScZRT1 (Vmax), whereas the substrate concentration at which the reaction rate is half of Vmax (Km) remains unaffected (Gitan et al., 2003). Also for other ZIP1 subfamily transporters a reduction in Zn uptake due to mutation of histidines in the HRD was observed (Mao et al., 2007) and some of these histidines are even necessary for the protein to be functional, i.e., able to transport Zn across the plasma membrane (Milon et al., 2006).

Heterologous expression and subcellular localization in yeast are common experimental procedures to study eukaryotic gene function and protein localization (Zhao and Eide, 1996; Mokdad-Gargouri et al., 2012). Heterologous expression of SlZRT1 in the 1zrt11zrt2 yeast double mutant, which is defective in Zn uptake, resulted in an almost complete restoration of the phenotype (**Figure 3**) and SlZRT1::EGFP fusion proteins localize at the plasma membrane of yeast cells (**Figure 4**). These results support a role as plasma membrane localized Zn transporter for the SlZRT1 protein. However, kinetics of the transporter might be different from the ScZRT1 protein. SlZRT1 did not fully complement ScZRT1 as observed in the drop assays (**Figure 3**) and Zn starved SlZRT1 transformed 1zrt11zrt2 yeast cells accumulate less Zn within 24 h after Zn replenishment than WT yeast cells do (**Figure 5**). No significant differences in Fe content were observed (**Supplementary Figure 2**), indicating a high Zn specificity of the transporter.

In yeast, ScZRT1 expression is regulated both at the transcriptional and the post-transcriptional level by Zn (Gitan and Eide, 2000). Post-translationally, Zn induces the removal of ScZRT1 from the plasma membrane via ubiquitination (Gitan et al., 2003). After endocytosis the protein is degraded in the vacuole. This regulatory system ensures a rapid shutdown of Zn uptake in yeast cells exposed to high concentrations of Zn (Gitan and Eide, 2000). In S. luteus SlZRT1 expression is regulated by excess Zn. SlZRT1 expression level is significantly lower after exposure to potentially toxic concentrations of Zn (500 and 1000 µM) as compared to the control (20 µM) and this already 1 h after initiation of the exposure (**Figure 6**). In contrast, absence of external Zn results in a rapid accumulation of SlZRT1 mRNA. Two hours after initiation of Zn starvation in S. luteus mycelium, SlZRT1 gene expression peaks and declines again to reach control levels after 4 h of growth in absence of Zn. After 24 h of growth in the absence of Zn, the SlZRT1 expression level in S. luteus mycelium is slightly higher again compared to the level in mycelium grown in control conditions. These fluctuations in expression level could possibly reflect the cell's Zn status. A similar expression pattern, though delayed in time was detected by Schothorst et al. (2017) in S. cerevisiae for ScZRT1

in conditions of Zn deprivation. ScZRT1 transcripts peak at 2 days under Zn deprivation and decline again afterward. A fast transcriptional response on limited environmental Zn concentrations is common for plasma membrane localized Zn transporters of the ZIP I subfamily. Induction of transcription in the absence of external Zn was reported previously for fungal ZIP I subfamily Zn importers which were identified in Ascomycota (ScZRT1 of Schizosaccharomyces pombe, ZrfA, ZrfB and ZrfC of Aspergillus fumigatus, Tzn1 and Tzn2 of Neurospora crassa, CaZRT1 and CaZRT2 of Candida albicans) and in Basidiomycota (CgZIP1 and GgZIP2 of Cryptococcus gattii) (Dainty et al., 2008; Kiranmayi et al., 2009; Jung, 2015).

Altogether, our data support a function as plasma membrane localized Zn importer with an important role in Zn homeostasis of S. luteus for SlZRT1. Likely, SlZRT1 is responsible for an adequate supply of Zn to the cell when environmental Zn is limited. With our current data, we cannot conclude on a role for SlZRT1 as Zn receptor for signaling in order to adjust primary metabolism to external Zn availability. Such a role was reported recently for ScZRT1 (Schothorst et al., 2017) and is certainly worth investigation in S. luteus and mycorrhizal fungi in general. Ectomycorrhizal fungi are wellknown to offer their host plant a balanced nutrient supply by efficiently collecting limited nutrients and reducing the transfer of excess, potentially toxic elements. In relation to Zn, ectomycorrhizal fungi in general, and S. luteus in particular, are reported to protect their host plant from Zn toxicity (Colpaert et al., 2011). As trees in general do not tolerate high Zn soil concentrations, this protective feature of S. luteus is interesting for phytoremediation purposes. Further research is needed to better understand the regulation and function of SlZRT1 within the S. luteus – host ectomycorrhizal association and to assess the contribution of SlZRT1 to the Zn status of the host plant.

### AUTHOR CONTRIBUTIONS

LC, JR, and JC designed the study. LC, AT, ES, and NA performed the experiments. LC, EM, AK, and JR analyzed the data. LC and JR wrote the manuscript. LC, ES, NA, MODB, JR, and JC contributed in manuscript editing. All authors read and approved the final version of the manuscript.

### REFERENCES


### FUNDING

This work was financially supported by the Research Foundation Flanders (FWO Project G079213N). LC holds a Flanders Innovation & Entrepreneurships Ph.D. fellowship (IWT Project 141461) and her research visit at INRA Grand Est Nancy was funded by the Laboratory of Excellence Advanced Research on the Biology of Tree and Forest Ecosystems (ARBRE; grant No. ANR-11-LABX-0002-01). Part of the computations were performed at the INRA Grand Est-Nancy Ecogenomics facilities. The Mycorrhizal Genomics Initiative is supported by the French National Institute for Agricultural Research (INRA), the US Department of Energy (DOE) Joint Genome Institute (JGI; Office of Science of the US Department of Energy), the Region Lorraine Research Council and the European Commission [European Regional Development Fund (ERDF)].

### ACKNOWLEDGMENTS

We thank Carine Put, Ann Wijgaerts, and Brigitte Vanacken for their technical assistance. Stefan Gobert for his assistance with the confocal microscopy. We are also grateful to Prof. Dr. David Eide for kindly providing the yeast mutant.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2017.02320/full#supplementary-material

FIGURE S1 | Functional complementation of the zinc-uptake-deficient yeast strain 1zrt11zrt2 by SlZRT1::EGFP. Wild type (WT) and mutant yeast cultures with an OD<sup>600</sup> = 1 were 10-fold serial diluted (10<sup>0</sup> , 10−<sup>1</sup> , 10−<sup>2</sup> , and 10−<sup>3</sup> ) and spotted on control (1 mM Zn) or ethylenediaminetetraacetic acid (EDTA) supplemented synthetic drop-out (SD) medium. The WT strain was transformed with the empty vector (EV, pAG306GAL-ccdB-EGFP; Alberti et al., 2007), the mutant strain 1zrt11zrt2 with either the EV or the vector containing SlZRT1:GFP. The experiment was carried out for three independent clones and pictures were taken after 4 days of growth.

FIGURE S2 | Fe concentration in transformed yeast cells (A,B). The WT strain was transformed with the EV (pYES-DEST52, Invitrogen), the mutant strain with either the EV or the vector containing SlZRT1. Data are the average ± SE of five biological replicates, significant differences (p < 0.05) are indicated by different letters. (A) In control conditions (0 µm Zn), (B) after exposure to Zn (500 µM).

fumigatus virulence and its growth in the presence of the Zn/Mn-chelating protein calprotectin. Cell. Microbiol. 16, 548–564. doi: 10.1111/cmi.12238



Zhao, H., and Eide, D. (1996). The yeast ZRT1 gene encodes the zinc transporter protein of a high-affinity uptake system induced by zinc limitation. Proc. Natl. Acad. Sci. U.S.A. 93, 2454–2458. doi: 10.1073/pnas.93.6.2454

**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 © 2017 Coninx, Thoonen, Slenders, Morin, Arnauts, Op De Beeck, Kohler, Ruytinx and Colpaert. 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) or licensor 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.

# Arbuscular Mycorrhizal Fungal 14-3-3 Proteins Are Involved in Arbuscule Formation and Responses to Abiotic Stresses During AM Symbiosis

Zhongfeng Sun<sup>1</sup> , Jiabin Song<sup>1</sup> , Xi'an Xin<sup>1</sup> , Xianan Xie<sup>2</sup> \* and Bin Zhao<sup>1</sup> \*

<sup>1</sup> State Key Laboratory of Agricultural Microbiology, College of Life Science and Technology, Huazhong Agricultural University, Wuhan, China, <sup>2</sup> State Key Laboratory for Conservation and Utilization of Subtropical Agro-Bioresources, College of Forestry and Landscape Architecture, South China Agricultural University, Guangzhou, China

#### Edited by:

Erika Kothe, Friedrich-Schiller-Universität Jena, Germany

#### Reviewed by:

Raffaella Balestrini, Consiglio Nazionale delle Ricerche (CNR), Italy Maria Rapala-Kozik, Jagiellonian University, Poland

#### \*Correspondence:

Bin Zhao binzhao@mail.hzau.edu.cn Xianan Xie 30004537@scau.edu.cn

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 18 September 2017 Accepted: 16 January 2018 Published: 05 March 2018

#### Citation:

Sun Z, Song J, Xin X, Xie X and Zhao B (2018) Arbuscular Mycorrhizal Fungal 14-3-3 Proteins Are Involved in Arbuscule Formation and Responses to Abiotic Stresses During AM Symbiosis. Front. Microbiol. 9:91. doi: 10.3389/fmicb.2018.00091 Arbuscular mycorrhizal (AM) fungi are soil-borne fungi belonging to the ancient phylum Glomeromycota and are important symbionts of the arbuscular mycorrhiza, enhancing plant nutrient acquisition and resistance to various abiotic stresses. In contrast to their significant physiological implications, the molecular basis involved is poorly understood, largely due to their obligate biotrophism and complicated genetics. Here, we identify and characterize three genes termed Fm201, Ri14-3-3 and RiBMH2 that encode 14- 3-3-like proteins in the AM fungi Funneliformis mosseae and Rhizophagus irregularis, respectively. The transcriptional levels of Fm201, Ri14-3-3 and RiBMH2 are strongly induced in the pre-symbiotic and symbiotic phases, including germinating spores, intraradical hyphae- and arbuscules-enriched roots. To functionally characterize the Fm201, Ri14-3-3 and RiBMH2 genes, we took advantage of a yeast heterologous system owing to the lack of AM fungal transformation systems. Our data suggest that all three genes can restore the lethal Saccharomyces cerevisiae bmh1 bmh2 double mutant on galactose-containing media. Importantly, yeast one-hybrid analysis suggests that the transcription factor RiMsn2 is able to recognize the STRE (CCCCT/AGGGG) element present in the promoter region of Fm201 gene. More importantly, Host-Induced Gene Silencing of both Ri14-3-3 and RiBMH2 in Rhizophagus irregularis impairs the arbuscule formation in AM symbiosis and inhibits the expression of symbiotic PT4 and MST2 genes from plant and fungal partners, respectively. We further subjected the AM fungus-Medicago truncatula association system to drought or salinity stress. Accordingly, the expression profiles in both mycorrhizal roots and extraradical hyphae reveal that these three 14-3-3-like genes are involved in response to drought or salinity stress. Collectively, our results provide new insights into molecular functions of the AM fungal 14-3-3 proteins in abiotic stress responses and arbuscule formation during AM symbiosis.

Keywords: arbuscular mycorrhiza, abiotic stresses, Funneliformis mosseae, Rhizophagus irregularis, Fm201, host-induced gene silencing, 14-3-3 proteins

### INTRODUCTION

fmicb-09-00091 March 1, 2018 Time: 15:53 # 2

Arbuscular mycorrhizal (AM) fungi, belonging to the ancient phylum Glomeromycota, are soil-borne microbes and capable of establishing the most widespread mutualistic association, namely AM symbiosis, with more than 80% terrestrial flowering plant species (Simon et al., 1993; Remy et al., 1994). Due to the obligate biotrophic nature, AM fungi need to consume plant photosynthates (Bago et al., 2000) and lipids to complete their life cycle (Bravo et al., 2017; Jiang et al., 2017), and reciprocally AM fungi significantly contribute to plant growth not only by enhancing mineral nutrient uptake and water acquisition from surrounding soil, but also protecting plants against fungal pathogens (Smith and Read, 2008; Jung et al., 2012; Chitarra et al., 2016) and a variety of abiotic stresses (Augé, 2001; Schützendübel and Polle, 2002; Lenoir et al., 2016). Therefore, AM fungi are key endosymbionts of the plant symbiosis and have significant impacts on plant productivity and ecosystem function (Van der Heijden et al., 1998), and are of great interest for the sustainable agricultural development (Gianinazzi et al., 2010).

The formation of a functional AM symbiosis requires successive stages between AM fungal and host symbionts at both physiological and molecular levels (Genre et al., 2005; Bonfante and Genre, 2010). Specifically, the development of arbuscular mycorrhiza consists of three major distinct stages through the progression of AM fungal hyphae during root colonization (Genre et al., 2005; Harrison, 2012; Gutjahr and Parniske, 2013). Arbuscules are generally thought to be the primary sites for nutrients exchange between the two symbionts (Parniske, 2008; Bonfante and Genre, 2010). In this symbiotic interface, the host membrane surrounding an arbuscule, known as the periarbuscular membrane (PAM), harbors AM-specific Pi transporters that acquire Pi released from the arbuscule (Harrison et al., 2002; Javot et al., 2007a). Outside the roots, the extraradical mycelia of AM fungi can extend the soil substratum beyond the depletion zone of the rhizosphere to uptake nutrients (particularly Pi and N) and water from the surrounding soils (Govindarajulu et al., 2005; Javot et al., 2007b; Li et al., 2013).

Despite their great importance, the underlying signaling events during initiation and formation of AM symbiosis are not well understood (Paszkowski, 2006; Bonfante and Requena, 2011; Gutjahr and Parniske, 2013; Oldroyd, 2013; Schmitz and Harrison, 2014; Bonfante and Genre, 2015). In contrast to a plethora of discoveries on morphological and chemical features in AM fungi, the molecular basis involved is still largely unknown, partially due to the limited available genomic resources. Many genome-wide gene expression analysis have been employed recently in order to understand the underlying molecular mechanisms of the AM formation. These studies mainly focused on the host plants (recently reviewed in Salvioli and Bonfante, 2013), whereas only a few investigations addressed the fungi partners (Requena et al., 2002; Breuninger and Requena, 2004; Cappellazzo et al., 2007; Kikuchi et al., 2014). Major progress has been recently achieved using transcriptomics and genomics data of Rhizophagus irregularis (Tisserant et al., 2012; Tisserant et al., 2013; Lin et al., 2014) and Gigaspora genus (Salvioli et al., 2016; Tang et al., 2016).

Using the suppression subtractive hybridization library (SSH) strategy, Breuninger and Requena (2004) firstly found some ESTs of fungal genes which were induced in the appressorium stage may display potential roles in this stage of Funneliformis mosseae. In this case, an EST tag termed 201, which encodes a 14-3-3 like protein in fungi, shows a significant up-regulation in the appressorium stage of AM symbiosis (Breuninger and Requena, 2004). Recently, Tisserant et al. (2012) released the first genome-wide overview of the transcriptional profiles of the various fungal tissues of R. irregularis. Particularly, a large number of fungal non-redundantly expressed transcripts was investigated in spores, intraradical mycelia (IRM), extraradical mycelia (ERM), and arbuscules. Interestingly, the transcripts encoding R. irregularis 14-3-3 proteins were inducible in both IRM and ERM.

14-3-3 proteins are highly conserved and dimeric proteins with a subunit mass of approximate 30 KDa (van Heusden and Steensma, 2006). These proteins are named based on the fraction number after EDTA-cellulose chromatography and the position after subsequent starch gel-electrophoresis (Moore, 1967). The first description of the function of 14-3-3 protein is substantially comparable to the 'activator' protein, that is important in the regulation of serotonin and noradrenaline biosynthesis in the brain (Ichimura et al., 1987). Moreover, 14- 3-3 proteins form homo- or hetero-dimers by two subunits harboring the independent ligand-binding channels. Until now, it is extensively studied that these proteins generally serve as adapters, chaperones, activators, or repressors in the regulation of signal transduction pathways by reorganization of specific phosphoserine/phosphothreonine-inclusive binding motifs phosphorylated by protein kinase A (Smith et al., 1998; van Heusden, 2009; Smith et al., 2011; Parua and Young, 2014). Additionally, 14-3-3 proteins also play important roles in the pseudohyphal growth of Saccharomyces cerevisiae and the pathogenic fungal infection, such as Ustilago maydis (Gancedo, 2001; Rispail et al., 2009; Ballou et al., 2013; Liu et al., 2015). These known 14-3-3 proteins have also been implicated in several signaling cascades responding to biotic and abiotic stresses in plants (Roberts et al., 2002; Lozano-Duran and Robatzek, 2015; Li et al., 2016), suggesting that these proteins may display distinct roles during eukaryotes life cycle (Liu et al., 2015). So far, at least two distinct 14-3-3 subunits have been characterized in fungi (Darling et al., 2005; Hermeking and Benzinger, 2006). Porcel et al. (2006) identified a gene Gi14-3-3 (currently Ri14-3-3) from the AM fungus R. irregularis, encoding a 14-3-3 protein subunit that is enhanced under drought stress during AM symbiosis, being the first 14-3-3 protein from AM fungus reported so far. Additionally, recent work has provided new evidence for the potential involvement of Ri14-3-3 gene in the interaction between maize and R. irregularis under drought stress (Li et al., 2016). However, the molecular mechanisms of Ri14-3-3 gene in enhancing plant resistance to drought stress are still unclear.

To further advance our understanding of the roles of 14-3- 3 proteins in fungal symbionts during AM symbiosis, we here report three novel fungal genes, so called Fm201, Ri14-3-3 and RiBMH2, which encode 14-3-3-like proteins from F. mosseae (BEG12) and R. irregularis (DAOM197198), respectively. 14-3-3

genes are strongly induced in the early stage of AM symbiosis. Moreover, the expression of 14-3-3 genes are regulated in response to drought and osmotic stresses. To further characterize these AM fungal 14-3-3 genes, we validated the capability of these genes to complement the metabolic deficient 1bmhs mutant (bmh1 and bmh2 double mutant) in a yeast heterologous expression system. We also provided insights into the regulatory mechanism between 14-3-3 protein and Msn2 transcription factor from AM fungi and further proved the existence of two distinct 14-3-3 subunits in AM fungi. More importantly, in the absence of stable transformation protocols for AM fungi (Helber and Requena, 2008; Helber et al., 2011), host-induced gene silencing (HIGS) of the two 14-3-3 genes in R. irregularis, whereby these genes are silenced in the AM fungal symbiont by expressing an RNA interference construct in the host, provides a potential tool to address the function of 14-3-3 proteins in obligate biotrophic AM fungi. Collectively, our results provide new insights into molecular functions of the AM fungal 14-3-3 proteins in stress responses and arbuscule formation during AM symbiosis.

## RESULTS

### Identification of Fm201 Gene From Funneliformis mosseae

In the previous study, transcript abundance of 201-tag was significantly enhanced at the early appressorium stage of AM symbiosis (Breuninger and Requena, 2004). The amino acid sequence of 201-tag exhibits a high similarity (∼97%) with the Ri14-3-3 protein from R. irregularis (Breuninger and Requena, 2004; Tisserant et al., 2012). With the aim to confirm if this fungal 14-3-3 protein is involved in AM symbiosis, a DNA clone of 1.5 kb in length was obtained by inverse PCR from the cloning procedures on F. mosseae genomic DNA based on the 201-tag. The isolated fragment with 5<sup>0</sup> end and upstream region was highly similar to the sequence of Ri14-3-3 gene from R. irregularis (Porcel et al., 2006). Since the 5<sup>0</sup> and 3<sup>0</sup> end sequences of this gene are not available, 5<sup>0</sup> and 3<sup>0</sup> RACE experiments on RNA pools of F. mosseae germinating spores were subsequently performed to obtain the full-length CDS sequence. A 1,401 bp full-length cDNA sequence of Fm201, covering the 50UTR (188 bp) and 30UTR (411 bp), was thus identified (Accession number: KM258580). The corresponding genomic sequence of Fm201 gene is 1,685 bp in length, containing seven exons and seven introns (Supplementary Figure S1). Interestingly, Fm201 gene contains a 100 nt intron in the 5<sup>0</sup> UTR and two transcriptional variants of 30UTRs (61 and 411 nt in length, respectively). These unusual features of AM fungal 14-3-3 gene firstly reported in the present study may suggest important roles in the regulation of Fm201 expression during AM symbiosis.

### Fm201 Protein Is Conserved Among Eukaryotes

To further investigate the phylogenetic and structural features of the Fm201 protein from AM fungi, we exploited the phylogenetic placement and 3D structure of Fm201 protein using bioinformatics strategy. The in silico analysis revealed that the open reading frame (ORF) of Fm201 gene consists of 804 bp corresponding to 267 amino acids with a predicted molecular weight of approximate 30 kDa. A phylogeny of basal fungi and 14-3-3 proteins from Homo sapiens clearly supports Fm201 as a sister clade to Ri14-3-3 (**Figure 1A**), indicative of the conserved evolutionary origin of the 14-3-3 genes in AM fungi, whereas the RiBMH2 protein from R. irregularis belongs to the closer relative of the yeast BMH2 (**Figure 1A**). Compared to the 14-3-3 proteins from H. sapiens, Fm201 protein still shares a very high homology. This also demonstrates that 14-3-3 proteins are highly conserved in eukaryotes. As a conserved protein, Fm201 protein shares 97% similarities with Ri14-3-3 protein from AM fungi. The amino acid sequence of Fm201 was compared with BMHs from S. cerevisiae, 14-3-3s from R. irregularis and H. sapiens 14-3-3 epsilon and a high homology with over 72% identity at the amino acid level was observed (Supplementary Figure S2).

We further carried out the Homology modeling using Homo sapiens 14-3-3 epsilon (80.52% identity in amino acid sequences) as a model. The predicted three-dimensional conformation of Fm201 indicates that Fm201 is a typical 14-3-3 protein with 9 alpha helices and 8 loops, with the highly homologous α3, α5, α7, and α9 putatively forming its amphipathic ligand-binding grooves (**Figure 1C**). Fm201 homologous dimers could form a typical C-shape cup, which provides a basic structure of 14-3-3 dimers for implementing its function (**Figure 1D**).

### 14-3-3 Proteins Restore Metabolic Activity of S. cerevisiae 1bmhs Mutant

To gain further insights into the function of 14-3-3, a S. cerevisiae heterologous expression system was exploited. Since Fm201 shares 82.3% identity at the amino acid sequence with both BMH1 and BMH2 in S. cerevisiae (see Supplementary Figure S2), the ORF of Fm201 was cloned into pMR-12 under the control of the Gal7 promoter and replaced S. cerevisiae BMH1. To test if Fm201 can restore the metabolic activity of S. cerevisiae 1bmhs mutant, as referred in Materials and Methods (also see Supplementary Table S1). S. cerevisiae 1bmhs mutant with pMR-12-Fm201 cannot grow on YPD with 2% glucose as the sole carbon source. However, cell growth was recovered when using 2% galactose as the sole carbon source (**Figure 2**). Similar results were observed when replacing Fm201 with Ri14-3-3 or RiBMH2 (**Figure 2**). These data confirmed that Fm201 has similar function as BMH1 in S. cerevisiae.

### Functional Dissection of CREs on Promoter of Fm201 (pFm201) in S. cerevisiae

14-3-3 proteins have been reported to participate in pseudohyphal growth and resistance in yeast (Roberts et al., 1997; Hurtado and Rachubinski, 2002). However, the roles of these 14-3-3 proteins in mycorrhizal fungi are largely unknown. It has been observed that expression specificity of plant 14-3-3 genes in response to various stresses is largely promoter

software. Bootstrap values were calculated using 1,000 replicates. 14-3-3 proteins from mycorrhizal fungi are highlighted in bold and their homologs from H. sapiens are colored red. (B) Predicted structure of Homo sapiens 14-3-3 epsilon (2br9A) dimers. (C,D) Homology modeling of Fm201 monomer and homodimer using Homo sapiens 14-3-3 epsilon (2br9A) as reference.

FIGURE 2 | 14-3-3 proteins from mycorrhizal fungi restore the growth of S. cerevisiae 1bmhs mutant on galactose-contained media. Ten-fold serial dilutions of yeast cells (the wild-type BY4741, 1bmh1 mutant or 1bmhs mutant) carrying different constructs (the empty vector pSH47, the full-length cDNA of Fm201, Ri14-3-3, or RiBMH2) were grown on SD/-Ura plates with 2% glucose or galactose as carbon source. Expression of Fm201, Ri14-3-3, or RiBMH2 gene is controlled by galactose-inducible Gal7 promoter.

dependent (Aksamit et al., 2005). To analyze the putative CREs (Cis-Regulatory Elements) located in the promoter region of Fm201, a 1.5 kb length promoter sequence upstream of Fm201 coding region (pFm201) was analyzed via Yeastract database<sup>1</sup> . The CREs present in pFm201 were compared with pBMH1, pBMH2 from S. cerevisiae and pRiBMH2 in R. irregularis (van Heusden, 2009) (**Table 1**). Many common CREs in corresponding sites shared by pFm201 and pBMHs have been shown to be recognized by many transcriptional factors (Bruckmann et al., 2004; van Heusden, 2009). In the present study, two CREs possiblely recognized by Msn2 and STE12 were chosen for further investigations. Msn2 is an STRE element (AGGGG/CCCCT) binding transcription factor, which is supposed to be related to fungal infection and resistance to abiotic stress in other filamentous fungi ( Schmitt and Mcentee, 1996; Seidl et al., 2004; Elfving et al., 2014; Zhang et al., 2014). STE12, so-called GintSTE, is the transcriptional factor that has been reported in mycorrhizal fungi and is believed to be an indispensable component in the early process of mycorrhizal fungi infection (Tollot et al., 2009; Tang et al., 2016). The common and shared CREs present upstream of Fm201, RiBMH2 and S. cerevisiae BMHs imply that 14-3-3 proteins in AM fungi may be involved in the regulation of resistance to abiotic stress and hyphal growth in AM fungi as BMHs in S. cerevisiae.

<sup>1</sup>http://www.yeastract.com/

TABLE 1 | Predicted motifs on the Fm201 gene promoter compared with that of RiBMH2, ScBMH1 and ScBMH2<sup>∗</sup>


<sup>∗</sup>1.5 kb length promoter sequences of RiBMH2 and Fm201 were chosen to analysis the predicted motifs. The promoter sequence of Ri14-3-3 has not yet been released. ∗∗ Words in bold are transcription factor verified in the report.

Due to the lack of stable genetic transformation approaches in the AM fungi, it is technically challenging to knockout Fm201 gene to confirm the biological function in the early stages during symbiosis (Sanders, 1999; Maldonado-Mendoza et al., 2001). To probe the possible function of Fm201 protein, we employed the site-specific mutagenesis and yeast one-hybrid system to initially explore the essential region of Fm201 promoter. Compared to the site-specific mutagenesis of STRE (CCCCT/AGGGG) located in pFm201(pFm201-1STRE), pFm201 is more sensitive to abiotic stresses, osmotic pressure and drought stress (**Figure 3A**). This result indicates that the STRE element of pFm201 could be recognized by Msn2 in S. cerevisiae. Although there is no any report, to our knowledge, about the functional properties of RiMsn2 factor mentioned as RiMsn4 in R. irregularis, it may play a major role in eukaryotic abiotic stress response and hyphae differentiation (Tisserant et al., 2013; Zhang et al., 2014).

A yeast one-hybrid system was also performed to confirm the interaction between GintSTE and RiMsn2 with CREs on pFm201 (**Figure 3B**). As shown in **Figure 3B**, yeast cells harboring pFm201 with STRE element and RiMsn2 or GintSTE protein grew well, whereas growth of the cells carrying RiMsn2 or GintSTE and Fm201 promoter without STRE element as negative controls were severly inhibited under the same conditions. These data suggest that GintSTE and RiMsn2 proteins interact with the STRE element of Fm201 gene, the similar results were also acquired from the promoter of RiBMH2 (data not shown). It is therefore reasonable to speculate that GintSTE and RiMsn2 proteins positively regulate the expression of 14-3-3 genes in presymbiotic stage through the binding between GintSTE/RiMsn2 and pFm201.

### 14-3-3 Genes Are Highly Induced in Germinating Spores and Early Stages of Symbiosis

Due to the obligate biotrophic and asexual multinucleate nature of the AM fungi (Sanders, 1999; Maldonado-Mendoza et al., 2001), it is difficult to generate mutants and overexpression

strains to analyze the biological functions of Fm201 protein during AM fungal infection. To obtain further insights into the expression profile of Fm201 gene during the colonization process, we performed a time-course analysis of Medicago truncatula roots inoculated with F. mosseae in the pot system, then sampled at 12, 18, 25, and 50 days post-inoculation (dpi) and also collected the quiescent spores and germinated spores. Morphological analyses of mycorrhizal roots showed almost majority of the appressoria and intraradical hyphae at 12–18 dpi. More arbuscules were obviously detected starting from 25 dpi, while the abundance of arbuscules decreased at 50 dpi (**Figure 4A**), the mycorrhizal colonization of the root samples was also calculated as described by Trouvelot et al. (2015) (data not shown). The transcript abundance of Fm201 gene in F. mosseae at different stages were also analyzed by qRT-PCR. As shown in **Figure 4B**, the transcript abundance of Fm201 was obviously higher in germinating spores than in quiescent spores, the transcript abundance of Fm201 is ∼40% lower in 50 dpi than in 25 dpi when arbuscules decreased. This expression pattern was similar with fungal colonization in early stages of symbiosis, especially during the hyphopodium formation and root penetration (see **Figures 4A,D**). The similar results of RiBMH2 and Ri14-3-3 were also obtained from quiescent spores, germinated spores and sampled at 8, 12, 18, 40 days post-inoculation (dpi) by R. irregularis (**Figure 4C**). The expression levels of both RiBMH2 and Ri14-3-3 obviously increased accompanying the infection process and arbuscules initiation, as demonstrated by the parallel increased transcriptional levels of MtStbM1 (**Figure 4E**), the host plant subtilase-encoding gene which is considered as a molecular marker of arbuscular mycorrhiza development (Wegel et al., 2007; Takeda et al., 2009).

#### Knock-Down of 14-3-3 Impairs the Arbuscule Formation in AM Symbiosis

Because of the obligate property of mycorrhizal fungi, the effect of 1Fm201 mutant on the establishment and maintenance of mycorrhizal symbiosis cannot be confirmed in vivo. RNAi technique has been successfully utilized to inhibit Ri14-3-3 encoding a homologous protein of Fm201 as described above in R. irregularis induced by hairy root lines of M. truncatula during symbiosis (Nowara et al., 2010; Helber et al., 2011). A 262 bp cDNA sequence from R. irregularis Ri14-3-3 gene was cloned into pK7GWIWG2 (II) RR according to the approach mentioned in Materials and Methods. The Host-Induced Gene Silencing (HIGS) results of Ri14-3-3 showed no significant influence on the intraradical structures of R. irregularis within the roots, when compared with the control roots (**Figure 5A**).

Since the draft of R. irregularis genome was recently released (Tisserant et al., 2013; Lin et al., 2014), we cloned and identified the coding sequence of another 14-3-3 protein subunit termed RiBMH2 (EXX69786.1). The existence of this novel 14-3-3 protein subunit may explain the nice arbuscule observed in HIGS of Ri14-3-3. Thus, the HIGS experiment targeting both Ri14-3-3 and RiBMH2 was designed to address this issue. Mycorrhizal phenotype analysis uncovered that the arbuscules are defective. The almost collapsed arbuscules were present in the hairy root of Ri14-3-3/RiBMH2 RNAi plants, when these two genes were both strongly repressed. Furthermore, the mycorrhizal colonization of each group was also calculated. The data suggests that the abundance of arbuscules in RNAi roots was also significantly lower than that in the control roots (**Figures 5A–C**). Moreover, the expression levels of symbiotic MtPT4 and RiMST2, which are considered as molecular markers of the functioning of arbuscules (Harrison et al., 2002; Helber et al., 2011), are significantly reduced in the Ri14-3-3/RiBMH2 RNAi roots relative to the control roots (**Figures 5D,E**), indicating that knock-down of both Ri14-3-3 and RiBMH2 has a significant effect on the symbiotic phenotype of AM symbiosis. These results also suggest that RiBMH2 may be required for arbuscule formation in AM symbiosis. It also provides a direct evidence that AM fungal 14-3-3 proteins play important roles during AM symbiosis.

### 14-3-3 Genes Are Up-Regulated in Response to Salinity and Drought Stresses During AM Symbiosis

To further investigate the potential roles of 14-3-3 proteins in response to salinity and drought stresses, the transcript profiles of 14-3-3s in mycorrhizal roots and external hyphae were analyzed by qRT-PCR after 150 mM NaCl treatment for various time (**Figures 6A,C**). The transcript abundance of 14-3-3 shows slight but significant increase after 1.5 h and relatively stable within 24 h in mycorrhizal roots. In addition, the transcription profile of Fm201 in extraradical hyphae treated with NaCl shows more rapid induction than in intraradical mycelia. To determine whether 14-3-3 genes are responsive to drought stress, the transcript abundance of 14-3-3 genes under 1/2 water holding capacity of drought treatment was also compared (**Figures 6B,D**). Unlike salinity stress treatment, the transcript abundance of 14-3-3 genes show a ∼4 fold and ∼7 fold up regulation in mycorrhizal roots and extraradical hyphae, respectively. These findings suggest that Fm201 may be responsible for the crosstalk between plant and R. intraradices under salinity and/or drought stresses.

## DISCUSSION

In terrestrial ecosystems, AM symbiosis is considered to be the most widespread ecologically and agriculturally mutualistic beneficial association among plant symbioses. Despite their great importance in both ecology and agriculture, advance in understanding the molecular basis of AM symbiosis from the fungal aspect is slow until the release of the transcriptomic data of several AM fungal species (Tisserant et al., 2012; Salvioli et al., 2016; Tang et al., 2016) and genomic data of R. irregularis (Tisserant et al., 2013; Lin et al., 2014), biological functions of only a few genes have been characterized during interaction with plants. In such a context, we focus on the characterization of the 14-3-3 genes from AM fungi based on its higher expression during the colonization process.

### AM Fungal 14-3-3 Proteins Are a Conserved Feature of Glomeromycota

According to bioinformatics analyses, 14-3-3s from AM fungi are typical 14-3-3 proteins with higher similarity to the known 14-3-3 sequences of yeast and human. Additionally, Fm201 protein is conserved across eukaryotes based on the phylogenetic relationships among AM fungi and other basal eukaryotic species as well as the conserved 3D homology structures between F. mosseae and human (Yang et al., 2006). Therefore, it is of interest to find that two similar sequences were found in the recently released genome and transcriptome of another AM fungus, R. irregularis (Tisserant et al., 2012, 2013). Since the sequences of the two additional genes, the so called Ri14-3-3 and RiBMH2, are complete with the full-length of CDSs, and the percentage of identity is relatively high (∼97%), the three AM

FIGURE 5 | Mycorrhizal symbiotic phenotypes of Host-Induced Gene Silencing of Ri14-3-3 and RiBMH2. (A) Hairy root transformation of M. truncatula with empty vector (EV), Ri14-3-3 RNAi vector, or Ri14-3-3/BMH2 RNAi vector. Transgenic hairy roots were infected by AM fungi and the mycorrhizal phenotypes were observed with fluorescence microscope. a, mature arbuscules; ad, arbuscule degradation; ih, internal hyphae. (B) Transcript abundance change of Ri14-3-3 and RiBMH2 in transgenic hairy roots as measured by qRT-PCR using RiActin gene as the reference gene. (C) Mycorrhization level was analyzed by WGA 488 staining of hairy roots at 30 dpi with R. irregularis. F%, frequency of colonization; M%, intensity of mycorrhiza; A%, arbuscule abundance. (D) Expression levels of MtPT4 in control (EV) and RNAi lines were determined by real-time RT-PCR. The M. truncatula MtTEF gene was used as the reference gene. (E) Transcript accumulation of RiMST2 in control (EV) and RNAi mycorrhizal roots measured by real-time RT-PCR. The R. irregularis RiActin gene was used as endogenous control. Three technical replicates were analyzed. Asterisks indicate statistically significant differences from respective control lines. Error bars indicate the means of three biological replicates with SD values. Data shown are averages ± SD; n = 3. (#, <sup>∗</sup>p < 0.05, ##, ∗∗p < 0.01).

mycorrhizal roots, external hyphae under drought stress. (C) Expression fold change of Ri14-3-3, RiBMH2 and MtCBF4 in mycorrhizal roots, external hyphae after exposure to osmotic stress treated by 150 mM NaCl for different hours. Lines with a significant ratio to the express rate of Ri14-3-3, RiBMH2 or MtCBF4 in 0 h. (D) Expression fold change of Ri14-3-3, RiBMH2, MtCBF4 in mycorrhizal roots, external hyphae. The FmActin, RiActin or MtTEF was used as the reference gene. Three biological replicates were analyzed. Asterisks indicate statistically significant differences from respective control lines. Lines with a significant ratio to the express rate of Ri14-3-3, RiBMH2 or MtCBF4 in CK. Error bars indicate the means of three biological replicates with SD values. Data shown are averages ± SD; n = 3. (#, <sup>∗</sup>p < 0.05, ##, ∗∗p < 0.01).

fungal proteins share the same nine α-helix domain topologies. Among them, Ri14-3-3 gene from R. irregularis has been firstly reported by Porcel et al. (2006). Moreover, RNA-seq data presented a significant induction in planta phase compared to spores (Tisserant et al., 2012). Only the investigation within genomic and transcriptomic data in AM fungi will clarify whether Fm201-related sequences are a general feature among fungi. Consistent with the previous in silico analyses, these three Fm201,

Ri14-3-3 and RiBMH2 are able to complement the yeast BMH1 and BMH2 double mutants. This finding is in agreement with those data reported in the earlier studies (van Heusden et al., 1995, 1996), indicating that these genes identified above encode the functional 14-3-3-like proteins in AM fungi. Further studies need to be carried out to confirm whether these 14-3-3-like proteins identified are a conserved feature of Glomeromycota and whether they may have an essential role in the intraradical phase during interaction with the host plants.

The transcription of 14-3-3 genes show a clear increase in the germinating spores as well as the intraradical phase in both R. irregularis and F. mosseae. The data stemming from the time-course experiment presented that the relatively higher transcription levels were achieved in the phases of root penetration and arbuscules formation, while the expression levels of Fm201 and Ri14-3-3 are obviously reduced compared with RiBMH2 in the degenerating mycorrhizal roots. In addition, we also correlated the Fm201 mRNA abundance with the morphological structures of F. mosseae inside the roots (at 12–50 dpi). The results of Fm201 transcription patterns also suggest that it may play an important role in the germination and hyphopodium formation of F. mosseae, which was also proposed by Breuninger and Requena (2004) through SSH of AM symbiosis at the early stage. In addition, transcript levels of Fm201 remain higher during the symbiotic stage (see **Figure 4B**), suggesting that this 14-3-3 protein may also play important roles during AM symbiosis, especially the formation of arbuscule besides the root penetration stage. It is thus speculated that the expression of 14-3-3s are, to some extent, related to root penetration and arbuscules formation. This hypothesis is supported by the evidence that 14-3-3 transcripts were present in both the laser micro-dissected arbuscule-containing cells and the IRM including intercellular hyphae (Tisserant et al., 2012). Overall these data implicate a relationship between AM fungal 14-3-3 related genes and intraradical hyphal growth and arbuscule differentiation.

### Two AM Fungal 14-3-3 Protein Subunits Have the Impacts on the Success of Arbuscular Mycorrhizal Colonization and Arbuscule Formation

The potential involvement of AM fungal 14-3-3 genes Ri14-3-3 and RiBMH2 in the in planta phase of the colonization process was also supported by the HIGS of Ri14-3-3 and/or RiBMH2 during the M. truncatula–R. irregularis mycorrhizal symbiosis. Lacking the stable genetic transformation protocols for AM fungi, HIGS was confined to AM fungi (Helber et al., 2011; Xie et al., 2016).

The data of the knock-down of both Ri14-3-3 and RiBMH2 genes by HIGS resulting in the impaired arbuscule formation of R. irregularis suggest the significance of these AM fungal 14-3-3 proteins for AM symbiosis. Connecting with the transcripts of RiBMH2 during M. truncatula–R. irregularis mycorrhizal symbiosis, RiBMH2 may be required for the development of AM symbioses and the arbuscule differentiation within roots. However, the Ri14-3-3 RNAi roots colonized by R. irregularis exhibited a considerable arbuscule abundance as compared with control mycorrhizal roots. These findings suggest that the AM functionality or arbuscule formation is redundantly regulated by the two 14-3-3-like genes in R. irregularis. Nevertheless, we here propose that RiBMH2 is essential for arbuscule formation, whereas Ri14-3-3 could be involved in the colonization process but not AM functionality. This hypothesis is supported by the evidence that the transcripts of MtPT4 and RiMST2, two symbiotic genes responsible for arbuscule functionality, were strongly reduced in Ri14-3-3/RiBMH2 RNAi roots, while they were not repressed in Ri14-3-3 RNAi roots. Although Ri14-3-3 homologous gene RiBMH2 is identified in the R. irregularis draft genome (Tisserant et al., 2013; Lin et al., 2014) and RiBMH2 was not down-regulated in Ri14-3-3 RNAi roots (see **Figure 5B**), the normal AM fungal structures observed in this HIGS system indicate a novel but unknown role for Ri14-3-3 in the establishment of AM symbiosis. Based on the above findings and the previous study (Liu et al., 2015), we hypothesize that RiBMH2-mediated signal could be an important signal in the control of arbuscules formation and R. irregularis hyphal growth within roots. This unknown signal relayed by RiBMH2 serves as the essential signal to ensure the metabolic activity of R. irregularis in the hyphal growth and/or arbuscule differentiation during symbiosis. In the absence of this RiBMH2-mediated signal, the arbuscules are impaired, and growth of the fungus is prevented. The R. irregularis itself needs to activate 14-3-3 protein RiBMH2 in response to the environmental clues to meet demands during fungal growth and division. In addition, our functional analysis in yeast cells suggested that Ri14-3-3 and RiBMH2 encode functional signal proteins involved in growth induction (see **Figures 2B,C**), indicating that these two proteins may play potential roles in signal transduction during the colonization process and arbuscule formation, respectively. Thus, we can speculate the involvement of Ri14-3-3 in fine-tuning fungal growth in the intraradical phase responding to the external stimuli, moreover, RiBMH2 may be indispensable for arbuscules differentiation. This complex mechanism by which arbuscular mycorrhizas are formed in roots requires the elaborate control of the two AM fungal 14-3-3 proteins in the intraradical phase during cross-talk with host plant.

Remarkably, these results from the HIGS experiments revealed that one 14-3-3 protein subunit can adjust its own expression quantity to offset the adverse influence caused by the lack of another 14-3-3 protein subunit. This conclusion is consistent with the previous results derived from yeast system (van Heusden et al., 1995). Based on this point, it is reasonable to hypothesize that the AM fungal 14-3-3 proteins are indispensable for the symbiosis functioning.

#### Involvement of AM Fungal 14-3-3 Proteins in Msn2/STRE Element-Mediated Signaling Pathway

The knockdown of both Ri14-3-3 and RiBMH2 by HIGS exhibits somewhat distinct phenotypes, i.e., fewer arbuscule abundance and impaired arbuscules (see **Figures 5A,C**), repression of the

endosymbiosis functioning with regard to transcription of the symbiotic MtPT4 and MST2 genes (see **Figures 5D,E**). We hypothesize that there exists a positive feedback mechanism in the potential signaling pathway in R. irregularis. It is also proposed that the CREs upstream of a gene always show close relationship with its function, especially for the regulatory proteins (Carey et al., 2012; Petrov et al., 2012), although most CREs are composed of short sequences which may be very abundant in eukaryotic genomes (van Heusden, 2009). As expected, we observed some conserved motifs including STRE elements in the promoters of two AM fungal 14-3-3 genes Fm201 and RiBMH2 (see **Table 1**), as predicted by YEASTRACT database, in comparison with the promoters of yeast BMH1 and BMH2. Interestingly, the deletion of STRE (CCCCT/AGGGG) element in the promoter of Fm201 showed significantly reduced levels of the reporter gene mRNAs when expressed in yeast cells (see **Figure 3A**). These hypotheses mentioned above are also supported by the fact that the orthologous Fm201 gene promoter with STRE element (pFm201) directly interacts with transcription factor Msn2 in yeast cells (see **Figure 3B**). Furthermore, this recognition between RiMsn2 and pFm201 may contribute to the induction of Fm201 in extraradical hyphae in response to the salinity (150 mM NaCl treatment) stress (see **Figure 6A**). Therefore, based on the site-specific mutagenesis and the yeast one-hybrid analyses, the transcription of AM fungal 14-3-3-like genes during AM symbiosis is Msn2/STRE-element dependent. The zinc finger DNA-binding proteins Msn2 and Msn4 serve as the key factors that controlling fungal growth and stress responses in different fungal species (Martinez-Pastor et al., 1996; Schmitt and Mcentee, 1996; Liu et al., 2013; Zhang et al., 2014). In addition, the Msn2-controlled and STRE-driven gene Fm201 and RiBMH2 from F. mosseae and R. irregularis, respectively, are positively regulated in response to drought stress during AM symbiosis (see **Figures 6B,D**), reinforcing that AM fungal 14-3-3 genes participate in the Msn2/STRE elementmediated signaling pathway in AM fungal symbiont during AM symbiosis.

Overall these data presented in this study provided new insights into the signaling function of the 14-3-3 proteins in AM fungal cells during crosstalk with host plants. Based on the aforementioned data, we also propose the hypothesis that abiotic stresses such as salinity and drought affect a Msn2/STRE-mediated signaling pathway governing the expression of AM fungal 14-3-3 proteins that promoted fungal colonization and arbuscule formation within roots (see **Figure 7**). In the first version of the scheme for abiotic stresses induced signaling, it has been proposed that 14-3-3 proteins preferentially expressed in the intraradical phase are involved in AM fungal colonization process and arbuscule functionality by the regulation of Msn2/STRE-mediated signaling pathway that may control the fungal growth and arbuscule lifespan during AM symbiosis.

Further studies, such as characterizing the precise roles of the novel RiMsn2 gene identified in this work, validating the protein-protein interactions in the Msn2-mediated signaling pathway and the biochemical functions of core components of this pathway, and determining the direct evidence of

Schematic representation of the abiotic stresses-induced signaling cascades in AM fungi and the involvement of Ri14-3-3 and RiBMH2 genes in the RiMsn2/STRE-mediated signaling pathway during AM symbiosis. In AM fungi, the unknown signaling cascades are triggered by the external stress stimuli such as salinity or drought; then the core component RiMsn2 factor is activated by the upstream of potential cascades. The functional RiMsn2 is translocated to the nuclear to recognize the STRE (Stress response element) (CCCCT/AGGGG) on the AM fungal genome. Meanwhile, in the in planta phase, transcription of the STRE genes, including Ri14-3-3 and RiBMH2, are clearly induced after this interaction between RiMsn2 and STRE. Thus, the expression of Ri14-3-3 and RiBMH2 proteins are involved in or essential for AM colonization and arbuscule differentiation within roots. The black arrows indicate the positive interactions, while the red arrow suggests the induction of Ri14-3-3 and RiBMH2 genes in the in planta phase. The images 1 and 2 represent the intraradical hyphae and arbuscule from WGA488 staining. ih, intraradical hyphae; t, trunk; a, arbuscule. Scale bars represent 25 µm.

Msn2-dependent mechanisms in R. irregularis, are needed to define the underlying stress response mechanisms in AM symbionts. Furthermore, the RNA-seq data and gene expression analyses show that both R. irregularis and Gigaspora margarita contain multiple distinct MAPK (Mitogen-activated protein kinase)-related proteins (Tisserant et al., 2012; Salvioli et al., 2016; Xie et al., 2016), indicative of the presence of MAPK signaling cascade in AM fungi to respond to external stresses stimuli and adapt to environmental fluctuation. Thus, a major goal in this field will be to uncover a master MAPK protein regulating the AM fungal growth and differentiation during symbiosis under various abiotic stresses.

In summary, we showed that Fm201, Ri14-3-3 and RiBMH2, three genes from two different AM fungi, are preferentially expressed in the intraradical phase and may have impacts on the success of AM colonization and arbuscule formation. Our data also presented that Msn2 protein governs the Fm201

gene transcription, indicating that AM fungal 14-3-3 gene identified is involved in Msn2 factor/STRE element-mediated signaling pathway. Importantly, host-induced gene silencing of both Ri14-3-3 and RiBMH2 impairs the arbuscule differentiation within roots, indicating that the two AM fungal 14-3-3 protein subunits are required for arbuscule formation. Additionally, these AM fungal 14-3-3 genes are up-regulated in response to salinity and drought stresses during AM symbiosis. Based on these new findings, we propose that the AM fungal 14-3-3 genes are essential for the interaction between AM fungi and host plants, and are potentially involved in enhancing plant salinity and drought tolerance by Msn2/STRE element-controlled signaling.

### MATERIALS AND METHODS

### Biological Materials and Growth Conditions

A-grade spores of R. irregularis DAOM197198 were purchased from Agronutrition (Carbonne, France). Spores of F. mosseae BEG12 were kindly provided by the International Bank of Glomeromycota (IBG, Dijon, France) and collected from Medicago truncatula pot cultures by wet sieving to isolate genomic DNA and total RNA. Spores surface sterilized by 2% chloramine T, and then immersed in a solution containing 0.02% streptomycin and 0.02% gentamycin for 10 min (Besserer et al., 2008). Germinated spores of F. mosseae were selected from acetone solution containing 10−8mol/L GR24 (10−9mol/L GR24 for R. irregularis in two days) in 25◦C dark incubator. Quiescent spores, germination spores and mycorrhizal roots (at 10, 18 25, 50 dpi for F. mosseae; 8, 12, 18, 40 dpi for R. irregularis) were harvested. After washing in sterile water, all materials described above were immediately frozen in liquid nitrogen and stored at −80◦C before nucleic acid extraction.

The water-holding capacity of the soil was computated before planting. The soil was weighed before plant, watered uniformly until flowed from the bottom. Keep the pot suspended in midair for 1 day before weighing. The increase weight of soil after waterd is water-holding capacity. M. truncatula mycorrhizal roots inoculated with AM fungi were treated with NaCl (0.5 M) to 150 mM in final concentration (Calculate according to 70% of water holding capacity in soil) (Giovannetti et al., 2001; Li et al., 2011). Drought treatment was measured by 1/2 water-holding capacity treatment which was proved drought stress treatment to M. truncatula in previous experiments. Mycorrhizal roots and extraradical hyphae from sandwich system were harvested at 1.5, 5, 24 h after treatments (Abdel Latef and Chaoxing, 2011; Estrada et al., 2012) to monitor the transcript profiles of 14-3-3 genes under different abiotic stresses by qRT-PCR analysis.

Fm201 promoter-YGFP chimeric gene in S. cerevisiae BY4741 was treated with NaCl (500 mM), CuSO<sup>4</sup> (50 mM), CdCl<sup>2</sup> (0.2 mM), PEG4000 (25%), and 37◦C abiotic stresses in YPD medium and then harvested after 1h treatments (Vido et al., 2001).

## DNA and RNA Extraction, RT-PCR and Real Time RT-PCR

The total DNA was isolated from AM fungal sporescarps as described by Zézé et al. (1994). Total RNA of different AM fungal tissues was extracted with TRIzol reagent (Invitrogen) according to the protocol. Surface-sterilized spores were placed into 1.5 ml RNase free microtube and then frozen in liquid nitrogen, 0.3ml TRIzol solution was immediately added to the microtube. Electric mill (TIANGEN OSE-Y20, Beijing, China) and Phase Lock Gel (TIANGEN, Beijing, China) were used to make sure the quality of RNA. Total RNA yields and concentrations were measured by the Thermo NanoDrop 2000 spectrophotometer (Thermo). To remove residual genomic DNA, each total RNA sample was treated with RNase-free DNaseI (Thermo) according to the manufacturer's instructions. The first cDNA strand was synthesized as described in RevertAid First Strand cDNA Synthesis Kit (Thermo).

Transcript profiles of AM fungal genes Fm201, Ri14-3-3, and RiBMH2 as well as host plant genes MtSbtM1 and MtCBF4 in different symbiotic stages and under abiotic stresses were studied by qRT-PCR using ViiA 7 system (Life Technologies, United States), three biological replications were performed. The expression levels were normalized to transcripts of the β-actin gene of F. mosseae or R. irregularis and to transcripts of the MtTEF gene of M. truncatula (Hohnjec et al., 2005). Before real time RT-PCR, gene-specific primers for all target genes were validated on genomic DNA and cDNA. Total RNA was isolated from AM roots comprised plant and fungal materials. The specificity of the primer pairs were also confirmed via PCR method on M. truncatula total DNA. No amplification signals were present on plant DNA. The primers sequences for all genes studied in this work are provided in Supplementary Table S2. qRT-PCR was performed using SYBR Green Real-time PCR Master Mix (TOYOBO, Japan) according to the manufacturer's instructions. Each 10 µl reaction contained 1 µl of the synthesized cDNA (cDNA pool was diluted to 200 µl), 5 µl SYBR Green Real-time PCR Master Mix, 0.5 µl each primer(10 µM), 3 µl ddH2O. PCR program consisted of a 30 s incubation at 95◦C to active the hot-start recombinant Taq DNA polymerase, followed by 40 cycles of 10 s at 95◦C, 15 s at 57◦C, and 20 s at 72◦C. The relative levels of transcripts were calculated by using the 2−11ct method (Livak and Schmittgen, 2001).

### Cloning of Fm201 Gene From F. mosseae

The Fm201 EST sequence was obtained from NCBI (Accession number: CF803281), (Breuninger and Requena, 2004). Reverse PCR was utilized to get the 5<sup>0</sup> flanking sequence of Fm201 gene. The gene-specific primers 201F and 201R were designed to amplify the partial DNA fragment of Fm201 according to the available sequence of Fm201 EST. Genomic DNA of F. mosseae was digested by FastDigest restriction enzyme XhoI (Thermo). DNA fragments were self-ligated by T<sup>4</sup> DNA ligase, and the reaction was carried out in a final volume of 20 µl containing 0.5 µl digested DNA fragments, 2 µl 10× buffer, 0.5 µl T<sup>4</sup> DNA ligase (Thermo), 17 µl ddH2O, incubated for 12 h at 10◦C. Nest-PCR was performed in this experiment, 0.5 µl ligated

production was used as PCR template, the specific primers used in the first PCR reaction were 201RF1 and 201RR1, products from first PCR reaction were diluted to 1/1000 as the template for the second PCR reaction, and specific primer 201RF2 and 201RR2 were used.

RACE as a classic method to rapidly obtain the 5<sup>0</sup> and 3<sup>0</sup> ends of the Fm201 gene (Scotto-Lavino et al., 2006a,b). 30RACE was carried out on the total RNA from F. mosseae sporecarps by using primer QT. Two pairs of primers RACE201F/Q<sup>O</sup> and RACE201F2/Q<sup>I</sup> were used for the subsequent nest PCR reactions, respectively. Due to the high A/T containing feature of AM fungi genomic DNA, the dGTP and Q<sup>C</sup> replaced the dATP and Q<sup>T</sup> used in classic 50RACE, responsively. The first cDNA strand was obtained from F. mosseae sporocarps by using specific primer RACE201R1. Primers QC, Q<sup>o</sup> and RACE201R2 were used as the first PCR cycle primers, while primers Q<sup>I</sup> and RACE201R3 were used as the second PCR cycle. Transfast pfu DNA polymerase (Transgen, Beijing, China) was used in the PCR reactions mentioned above, PCR products were cloned into pEASY-Blunt vector (Transgen, Beijing, China) and sequenced.

#### Plasmids Construction

Plasmid pMR12 was generated from pMRI-11(Xie et al., 2014), the promoter of Gal7 was amplified from the genome DNA of S. cerevisiae BY4741 by PCR using the specific primers PGal7F/PGal7R. PGal7 and pMR-11 were digested with both SacI and SpeI, respectively, then the digests were cloned into target vector pMR-11. To address the regulation of CREs located in the promoter of Fm201, the expression profiles of YGFP reporter were conducted in S. cerevisiae BY4741. Monoclonal vector pUG35 carrying a YGFP reporter is used in yeast heterologous systems (Cormack et al., 1997). Two restriction sites were SacI and XbaI, which were added to the start codon upstream sequence of Fm201 (pFm201) by using primers 201PFn (n = 1,2,3,4) and 201PR1, then cloned into pUG35 to replace PMET−<sup>25</sup> to produce a series of 5<sup>0</sup> truncated promoters-reporter vectors. pFm201 targeted deletion of cis-element was conducted by using SOE-PCR method (Ho et al., 1989). The specific primers used are provided in Supplementary Table S2). The site-directed mutation promoter sequences were also cloned into pUG35 as the promoter truncated verification vectors. qRT-PCR was used for monitoring transcriptional efficiency of Fm201, the Ura gene of pUG35 was chosen as the internal standard, primers UraF, UraR, YGFPF, YGFPR were used in this experiment (Peter et al., 2006).

#### Yeast One-Hybrid Screening

Yeast one hybrid experiment was carried out using the Matchmaker One-hybrid System (Clontech), the ORF of GintSTE was cloned from cDNA of Rhizophagus irregularis by using primers RiSTE12F and RiSTE12R, and cloned into pGADT7 rec2. Two same 272 bp length promoter fragments contained STE12 targeted cis-element STRE were cloned into pHIS2 in tandem, the same fragment only lacking the cis-element STRE was also inserted into the pHIS2 as a negative control. Yeast onehybrid experiment of Msn2 was carried out in the same way and the primers RiMsn2F and RiMsn2R were used as mentioned in Supplementary Table S2.

Plasmids for yeast one-hybrid were co-transformed into yeast Y187 strain. Y187 cells carrying the target plasmids were cultivated in SD medium lacking leucine and tryptophan, and were also gradiently inoculated at 1.0 OD<sup>600</sup> on SD medium lacking leucine, histidine and tryptophan and supplemented with 30 mM 3-AT, which is a competitive inhibitor of the His3 protein.

### HIGS of R. irregularis 14-3-3 Genes in Hairy Root Lines of M. truncatula

The RNAi-target sequences of Ri14-3-3 and RiBMH2 were amplified by the specific primers Ri14-3-3F/Ri14-3-3R and RiBMH2ATG/RiBMH2F. The PCR products were cloned to the linearized pDONR221 used CloneExpressII (Vazyme, Nanjing, China), then the LR reaction was done to recombine the target sequences into the pK7GWIWG2(II)RR according to the instructions in Gateway protocol.

Agrobacterium rhizogenes Msu440-mediated root transformation was performed following the method as described in Medicago Truncatula Handbook (International Committee, 2006). In vitro hairy roots were cultured on EM plates containing Benzyl penicillin (200 mg/L) for three times. The root tip (2∼3 cm in length) was used for each subculture. The root was re-cultured in the M medium without antibiotics for half a month. The hairy root lines without bacteria were re-cultured in new M medium for mycorrhization. Mycorrhizal hairy root of P. crispum without DsRed tag was cut into small pieces (∼3 mm) and placed around the hairy root of M. truncatula harboring DsRed marker as described in Supplementary Figure S3. The mycorrhizal roots with red fluorescence were harvested in one month until the external hyphae of R. irregularis beyond the hairy root surface of M. truncatula.

#### Quantification of Arbuscular Mycorrhizal Colonization

Mycorrhizal roots collected from pot cultures were stained with 0.1% Typan blue, while the mycorrhizal hairy roots expressing red fluorescence grown on plates were stained with WGA488, and the estimation of AM colonization was performed as described by Trouvelot et al. (2015) using MYCOCALC program<sup>2</sup> .

## In Silico Analysis of Fm201 Protein

The deduced amino acid sequence of Fm201 was analyzed with the computer program DNAstar. Multiple sequence alignments were performed by DNAMAN8. The unrooted phylogenetic tree constructed by neighbor-joining algorithm was carried out using MEGA6. The computation of physical and chemical parameters was conducted by using ProtParam tool<sup>3</sup> . Homology modeling of the three-dimensional structure of Fm201 protein was done with the program Swiss Model<sup>4</sup> using Homo sapiens 14-3-3ε protein (2br9A) as the template (**Figure 1B**; Yang et al., 2006). The ciselements of the promoters Fm201 and RiBMH2 were analyzed on YEASTRACT<sup>5</sup> using S. cerevisiae S288c as the reference.

<sup>2</sup>http://www2.dijon.inra.fr/mychintec/Mycocalc-prg/download.html

<sup>3</sup>http://www.expasy.ch/tools/protparam.html

<sup>4</sup>http://swissmodel.expasy.org/

<sup>5</sup>http://www.yeastract.com/

Yeast mutant strains used in this article are constructed with the methods mentioned by van Heusden (van Heusden et al., 1995, 1996). The detail information for each strain is available at Supplementary Table S1. In the construction of bhms-Fm201(MATa; his311; leu210; met1510; ura310; BMH1::KanMX (Gal7[Fm201]); BMH2::ura3), fragments of pMRI-12 which contain a KanMX and a Gal7 promoter and pSH47 which contain a Ura3 marker were used to replace BY4741, BMH1 and BMH2, respectively, by primers PMRI-12F1, PMRI-12R1 and PSH47F, PSH47R. The bmhs-Ri14-3-3 and bmhs-RiBMH2 were also built in the same way.

#### Statistical Analyses

fmicb-09-00091 March 1, 2018 Time: 15:53 # 14

Statistical analyses were performed through one-way ANOVA. Following ANOVA, Tukey's test was performed to make comparisons between treatments, using a probability level of p < 0.05(<sup>∗</sup> , #), 0.05 ≤ p < 0.01 (∗∗, ##). All statistical analyses were performed using SPSS statistical package (version 23.0, SPSS Inc., United States).

#### ACCESSION NUMBERS

The sequence data can be found in the GenBank data libraries under accession numbers. Nucleic acid sequence: MtCBF4 (HQ110079.1), MtStbM1 (XM\_003611148.1), MtPT4 (AY116211.1), Fmactin (KM360085.1), Fm201 (KM258580.1), Riactin (EXX64987.1), RiBMH2 (JEMT01016782.1), Ri14-3-3 (AM049264.1), RiMST2 (HM143864.1), Ri14-3-3 (CAJ16742.1). Amino acid sequence: Fm201(KM258580), RiBMH2 (EXX69786.1), R. oryzae 14-3-3 (EIE87660.1), M. medusa 14-3-3 (ABS86241.1), M. circinelloides 14-3-3 (EPB82885.1), A. oryzae 14-3-3 (XP\_001819291.2), A. niger 14-3-3 (XP\_001399080.1), S. borealis 14-3-3 (ESZ95350.1), M. oryzae 14-3-3 (XP\_003710925.1), A. nidulans (CBF81292.1), A. terreus 14-3-3 (XP\_001212078.1), P. strigosozonata 14-3-3

#### REFERENCES


(XP\_007382290.1), S. musiva 14-3-3 (EMF09853.1), B. bassiana 14-3-3 (XP\_008601347.1), R. solani 14-3-3 (CCO32840.1), S. cerevisiae BMH1 (CAA46959.1), S. tuberosum 14-3-3 (XP\_004250139.1), O. sativa 14-3-3 (NP\_001047234.1), S. cerevisiae BMH2 (CAA59275.1), R. norvegicus 14-3-3ε (NP\_113791.1), H. sapiens 14-3-3ε (NP\_006752.1).

#### AUTHOR CONTRIBUTIONS

BZ and XXie conceived this research. ZS and JS prepared the biological material for gene expression analysis. ZS and XXin performed the data analysis. ZS and XXie wrote the manuscript. BZ and XXie revised the manuscript.

#### FUNDING

This study was financially supported by grant from the Natural Science Foundation of China (Grant No. 31270159).

#### ACKNOWLEDGMENTS

We are grateful to Professor Ton Bisseling, Ph.D. U. Gueldener and Ph.D. Wenping Xie for kindly providing the pK7GWIWG2 (II) RR, pUG35 and pMR-11 plasmids, respectively. We also thank Professor Deqiang Duanmu for the constructive discussions and language corrections during the manuscript preparation.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00091/full#supplementary-material


structures of the endomycorrhizal fungus Glomus intraradices. Curr. Genet. 51:59. doi: 10.1007/s00294-006-0101-2


fmicb-09-00091 March 1, 2018 Time: 15:53 # 15



**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 © 2018 Sun, Song, Xin, Xie and Zhao. 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 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 Role of Strigolactone in the Cross-Talk Between Arabidopsis thaliana and the Endophytic Fungus Mucor sp.

Piotr Rozp ˛adek<sup>1</sup> \*, Agnieszka M. Domka<sup>2</sup> , Michał Nosek<sup>3</sup> , Rafał Wazny ˙ 1 , Roman J. J ˛edrzejczyk<sup>1</sup> , Monika Wiciarz<sup>4</sup> and Katarzyna Turnau<sup>2</sup>

<sup>1</sup> Małopolska Centre of Biotechnology, Jagiellonian University, Kraków, Poland, <sup>2</sup> Institute of Environmental Sciences, Jagiellonian University, Kraków, Poland, <sup>3</sup> Institute of Biology, Pedagogical University of Kraków, Kraków, Poland, <sup>4</sup> Faculty of Biochemistry, Biophysics and Biotechnology, Jagiellonian University, Kraków, Poland

#### Edited by:

Mohamed Hijri, Université de Montréal, Canada

#### Reviewed by:

George Newcombe, University of Idaho, United States Mika Tapio Tarkka, Helmholtz-Zentrum für Umweltforschung (UFZ), Germany

> \*Correspondence: Piotr Rozp ˛adek piotr.rozpadek@uj.edu.pl

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 31 August 2017 Accepted: 26 February 2018 Published: 19 March 2018

#### Citation:

Rozp ˛adek P, Domka AM, Nosek M, Wazny R, J ˛edrzejczyk RJ, Wiciarz M ˙ and Turnau K (2018) The Role of Strigolactone in the Cross-Talk Between Arabidopsis thaliana and the Endophytic Fungus Mucor sp. Front. Microbiol. 9:441. doi: 10.3389/fmicb.2018.00441 Over the last years the role of fungal endophytes in plant biology has been extensively studied. A number of species were shown to positively affect plant growth and fitness, thus attempts have been made to utilize these microorganisms in agriculture and phytoremediation. Plant-fungi symbiosis requires multiple metabolic adjustments of both of the interacting organisms. The mechanisms of these adaptations are mostly unknown, however, plant hormones seem to play a central role in this process. The plant hormone strigolactone (SL) was previously shown to activate hyphae branching of mycorrhizal fungi and to negatively affect pathogenic fungi growth. Its role in the plant–endophytic fungi interaction is unknown. The effect of the synthetic SL analog GR24 on the endophytic fungi Mucor sp. growth, respiration, H2O<sup>2</sup> production and the activity of antioxidant enzymes was evaluated. We found fungi colony growth rate was decreased in a GR24 concentration dependent manner. Additionally, the fungi accumulated more H2O<sup>2</sup> what was accompanied by an altered activity of antioxidant enzymes. Symbiosis with Mucor sp. positively affected Arabidopsis thaliana growth, but SL was necessary for the establishment of the beneficial interaction. A. thaliana biosynthesis mutants max1 and max4, but not the SL signaling mutant max2 did not develop the beneficial phenotype. The negative growth response was correlated with alterations in SA homeostasis and a significant upregulation of genes encoding selected plant defensins. The fungi were also shown to be able to decompose SL in planta and to downregulate the expression of SL biosynthesis genes. Additionally, we have shown that GR24 treatment with a dose of 1 µM activates the production of SA in A. thaliana. The results presented here provide evidence for a role of SL in the plant– endophyte cross-talk during the mutualistic interaction between Arabidopsis thaliana and Mucor sp.

Keywords: strigolactone, fungal endophytes, Arabidopsis thaliana, salicylic acid, symbiosis, jasmonic acid

#### Rozp ˛adek et al. Dual Role of Strigolactone

### INTRODUCTION

fmicb-09-00441 March 15, 2018 Time: 18:1 # 2

A growing number of evidence indicates that endophytic fungi play a significant role in plant biology (Schulz, 2006; Yuan et al., 2009; Johnson et al., 2013). Endophytic fungi facilitate water and nutrient acquisition, resistance to abiotic stress such as drought, salinity and metal stress and provide protection against pathogenic microorganisms and herbivores. Their popularity is growing, for their potential in agriculture and bioremediation (Oelmüller et al., 2009; Li et al., 2012; Johnson et al., 2013). The potential for application of beneficial fungi seems very optimistic, however, understanding the mechanisms of the interactions between plants and endophytic fungi requires extensive research.

Plant adaptation to biotic and abiotic constraints requires several adjustments in plant metabolism, morphology, life cycle etc. These adaptations are often mediated by phytohormones. Recently, the plant hormone strigolactone (SL) has been recognized due to its role in root and shoot architecture determination and plant interactions in the rhizosphere (Rochange, 2010; Foo and Reid, 2013; Koltai and Kapulnik, 2013; Pandya-Kumar et al., 2014; Chen et al., 2015; van Zeijl et al., 2015). SLs are carotenoid derivates synthesized from β-carotene by consecutive action of a β-carotene isomerase, two carotenoid cleavage dioxygenases: CCD7 and CCD8 (MORE AXILARY GROWTH-MAX3 and 4) respectively and an enzyme from the P450 cytochrome family: MAX1 (MORE AXILARY GROWTH1). Downstream, a LATERAL BRANCHING OXYREDUCTASE (LBO) was recently shown to convert a carlactone intermediate in the process of SL biosynthesis (Brewer et al., 2016).

Strigolactones are important in plant responses to nutrient and water deficiency (López-Ráez et al., 2008; Yoneyama et al., 2012; Foo et al., 2013; López-Ráez, 2016; Visentin et al., 2016). SL biosynthesis mutants max3 and max4 and the SL signaling mutant max2 are more sensitive to drought, due to a relationship between SL and ABA (Bu et al., 2014; Ha et al., 2014). Similarly, osmotic stress had a more severe effect on the Lotus japonicus SL biosynthesis mutant: Ljccd7 (Liu et al., 2015). Nitrogen and phosphorus starvation activated SL biosynthesis and exudation and available reports indicate that SL plays a role in plant adaptation to P deficiency (Umehara et al., 2008; Kohlen et al., 2011; Ruyter-Spira et al., 2011; Andreo-Jimenez et al., 2015). Symbiotic microorganisms including endophytic fungi facilitate adaptation to environmental challenges including nutrient deficiencies and drought (reviewed in Bacon and White, 2016).

Strigolactones are signaling molecules involved in plantsoil microorganism interactions (reviewed in López-Ráez, 2016; Waters et al., 2017). The best described is its action in the plant–AMF interaction. In this mutual relationship the fungus provides the plant with necessary nutrients in exchange for reduced carbon. The AMF symbiosis is widespread throughout the plant kingdom; according to available reports, the roots of over 80% of terrestrial plants are colonized by AMF (Smith and Read, 2008; Brundrett, 2009). Only a few plant families, including the Brassicaceae have lost the ability to engage in mutual symbiosis with AMF (reviewed in Venkateshwaran et al., 2013). Nevertheless, recent reports indicate that numerous members of this family harbor a wide variety of fungal symbionts, including beneficial endophytic fungi that may play a similar role in plant physiology as AMF (Barzanti et al., 2007; García et al., 2013; Card et al., 2015; Hong et al., 2015). This also allows to study the mechanisms of symbiosis with well-established plant models such as Arabidopsis thaliana, Thlaspi caerulescens etc.

Strigolactones are secreted from plant roots into the rhizosphere and act as a signal for directional growth of the hyphae, thus SL seems to facilitate the plant-AMF interaction in the pre-symbiotic stage of symbiosis (Akiyama et al., 2005; Besserer et al., 2006; Kretzschmar et al., 2012; Mori et al., 2016). Fungal mycelium treated with synthetic SL analogs exhibits a number of changes such as: hyphal branching and growth, increased respiratory activity and ATP and NADPH production, mitosis, expression of effector genes and spore germination. Additionally, SL treatment activates synthesis and release of short chain chitin oligomers which can activate the symbiotic (SYM) signaling pathway, which in turn trigger symbiotic responses in the plant (Lopez-Raez et al., 2017). According to studies with SL-biosynthesis and SL-exudation mutants of pea, petunia, rice and tomato, SL was not necessary for the establishment of the plant-AMF symbiosis, however, the colonization rate of these mutants was much lower compared to wild type plants (Gomez-Roldan et al., 2008; López-Ráez et al., 2008; Vogel et al., 2009; Koltai et al., 2010; Gutjahr et al., 2012; Kretzschmar et al., 2012). The response of different AMF species differs in respect to various SL molecules, however, doses as low as 10 nM of GR24 were shown to affect the growth pattern of the fungi (Besserer et al., 2008). Even though significant progress has been made in elucidating SL signaling and perception in A. thaliana and rice, the mechanism of SL perception nor signaling are not known in fungi. In A. thaliana the α/β hydrolase D14 was recognized as a SL receptor. SL signaling is mediated by the MAX2 (MORE AXILARY GROWTH2)/SMAXL (SUPRESOR OF MAX2 6, 7, and 8 in particular) signal transduction pathway, but no clear MAX2 or D14 homologs were found in sequenced fungal genomes including Rhizophagus irregularis (Waters et al., 2017).

Strigolactones are also involved in other interactions in the rhizosphere: act as a signal for rhizobacteria, as stimulants of parasitic plant seed (Striga sp. and Orobanche sp.) and pathogenic fungi (Rochange, 2010; Foo and Reid, 2013; Koltai and Kapulnik, 2013). The role of SL in plant–pathogenic fungi interactions is not clear. There are several, contradictory reports. A range of responses to the synthetic SL analog-GR24 on pathogenic fungi growth and branching were reported for the same species (Dor et al., 2011; Torres-Vera et al., 2014; Foo et al., 2016). Recently, Belmondo et al. (2017) has shown that a thioredoxin reductase is necessary for limiting Botrytis cinerea growth by GR24, indicating a relationship between SL and ROS (reactive oxygen species) metabolism.

The role of SL in biotic stress responses may be associated with its interaction with other phytohormones or hormone dependent signaling. In the SL deficient tomato, slccd8, reduced concentration of ABA, SA and JA were shown (Torres-Vera et al., 2014). In response to the parasitic plant Phelipanche

ramosa, the expression of SL biosynthetic D27 and CCD8 and SA, JA and ABA marker genes was upregulated (Torres-Vera et al., 2016). The SL signaling mutant max2 was more susceptible to Pectobacterium carotovorum and Pseudomonas syringe, probably due to alterations in ABA metabolism (Piisilä et al., 2015). However, studies performed on SL biosynthesis and signaling garden pea mutants contradict these reports, showing no increased sensitivity to infection by the necrotrophic soilborne oomycete Pythium irregulare (Foo et al., 2016).

Previously, the endophytic fungus Mucor sp. was found to accelerate Arabidopsis arenosa and A. thaliana growth (Rozp ˛adek et al., 2017). The fungus was also shown to improve A. arenosa toxic metal tolerance (Rozp ˛adek et al., 2017). Other members of this genus improved oilseed rape growth in heavily polluted environments (Zhu et al., 2015; Zahoor et al., 2017) As recently suggested by Martin and Plett (2015), the closely related with AMF Mucoromycetes associated with extant, basal land plants, such as liverworts, hornworts and lycopods, in a symbiosis whose mutualistic nature is suspected, making this group of fungi a good model for studying the mechanisms of symbiosis.

In this study, we evaluated the role of SL in the interaction between A. thaliana and its fungal symbiont Mucor sp. It was hypothesized that SL is necessary in the development of mutualism between the two interacting organisms both as a secretory signal adjusting the Mucor sp. metabolism, to the mutualistic mode and a plant, intrinsic regulatory molecule. Its role was assumed to be associated with its connection to SA synthesis or signaling. Additionally, the possibility that the fungi may have an effect on SL metabolism after colonization was tested.

#### MATERIALS AND METHODS

#### Plant Cultivation

Arabidopsis thaliana WT (N6000), max1 (N9564), max4-1 (N9568) and max2-2 (N9566) (more axillary branches 1, 4 and 2) mutants (all in Col-0 background) were obtained from NASC (The Nottingham Arabidopsis Stock Centre, United Kingdom). Seeds were surface sterilized with 8% NaOCl, 96% and 75% EtOH and sown to sterile <sup>1</sup>/<sup>4</sup> MS medium in a petri dish and placed in darkness (4◦C). After 48 h seeds were transferred to a growth chamber (Panasonic MLR-352H-PE, JP) with a 16 h photoperiod, 21/17◦C day/night temperature and 50% humidity. After 10 days seedlings were moved to MSR medium with no sugar (10 plants per petri dish) and inoculated with the fungus. To evaluate the effect of SL on plant biomass yield, MSR was supplemented with 1 µM of the synthetic SL analog: GR24 (StrigoLab, I) in acetone. Inoculation of in vitro cultures was performed by placing 2.1 × 10<sup>6</sup> Mucor sp. spores 5 mm from the tip of the main root. After 10–12 days of growth plants were harvested, frozen in liquid nitrogen and stored at −80◦C. For biomass yield evaluation 3 separate experiments with 25–30 plants were performed. Due to differences in plant growth in between experiments, fresh weight of treated plants (E+, GR24 and E+GR24) was presented in relation to appropriate control. In all GR24 feeding experiments acetone mock control was performed.

### Strigolactone Feeding Experiments, Fungi Growth, Respiration

Mucor sp. (KU234656, strain UNIJAG.PL.E50) spores (2.1 × 10<sup>6</sup> ) were placed in PDA (potato dextrose agar) medium supplemented with GR24. For colony growth evaluation spores were inoculated onto PDA containing 1, 10, 50, 100, 500, and 1000 nM of GR24 in 9 cm petri dishes. Colony surface area was measured after 48 h of growth in 24◦C in darkness. Fungi respiration was measured with a 30 channel Micro-oxymax respirometer (Columbus Instruments, United States) between the 24 and 48 h of growth. A single O<sup>2</sup> measurement was performed every 2 h. Spores were placed in PDA supplemented with 50, 500, and 1000 nM of GR24 in 250 ml Duran bottles in the darkness at 24◦C (growth chamber of Memmert, IPP400, United States). Fungi respiration was measured as O<sup>2</sup> consumption per 2 h for 24 h. The respiration rate was presented in relation to colony diameter. The experiment was run in 5 replicates. For all experiments mock (acetone) treated control was performed.

#### Enzyme Activity Assays

#### Protein Extraction and Quantification

Mucor sp. colonies grown in PDA supplemented with 1, 10, 50, 100, and 1000 nM of GR24 were harvested from media after 48 h of growth and grounded with a mortar and pestle in liquid nitrogen. For crude protein extraction, powdered mycelia were homogenized with molybdenum beads in a TissueLyzer LT (Qiagen, DE) in ice cold 100 mM HEPES-NaOH buffer (pH 7.5, 4 mM DTT, 1 mM EDTA) at 35 Hz for 15 min. The homogenizer adapter was precooled in liquid nitrogen to keep samples frozen. After extraction, samples were centrifuged for 10 min at 10,000 g at 4◦C. Protein content was quantified according to Bradford (1976) using BSA as a standard. A separate set of fungi mycelium was prepared for each enzyme activity assay. The experiment was run in 5 replicates. For all experiments mock (acetone) treated control was performed.

#### Catalase Activity

The spectrophotometric measurement was performed according to the modified method described by Aebi (1984). Crude tissue extracts (10 µl) were added to 990 µl of phosphate buffer pH 7.0 containing 3 mM H2O2. CAT activity was determined from the decrease in absorbance at 240 nm due to CAT dependent reduction of H2O2. Enzyme activity was defined as 1 µmol of H2O<sup>2</sup> decomposed by 1 mg of total soluble proteins per minute.

#### Glutathione Reductase Activity

The supernatants were analyzed for GR activity according to the modified method described by Foyer et al. (1995). Enzyme activity was determined from the decrease in absorbance at 340 nm in the reaction mixture containing TRIS-HCl (50 mM, pH 7.5) buffer, EDTA (1 mM) and GSSG (0.5 mM) in a total volume of 1 ml. Reaction was initiated with the addition of 0.15 mM NADPH.

#### Superoxide Dismutase Activity

fmicb-09-00441 March 15, 2018 Time: 18:1 # 4

Separations of soluble protein fractions were performed using native non-continuous PAGE in the buffer system described by Laemmli (1970) at 4◦C and 180 V. SOD bands on 12% polyacrylamide gels were visualized according to the staining procedure described by Beauchamp and Fridovich (1971). The gels were incubated in the staining buffer for 30 min, in darkness, at room temperature and subsequently exposed to white light until SOD activity bands became visible. The gels were scanned using the office scanner Epson V700 Photo, and densitometric analysis was performed with ImageJ (NIH, United States).

## Determination of H2O<sup>2</sup> Concentration

For H2O<sup>2</sup> assay powdered mycelia were homogenized with molybdenum beads in a TissueLyzer LT (Qiagen, DE) in ice cold 50 mM phosphate buffer pH 7.0 and at 40 Hz for 15 min. The homogenizer's adapter was precooled in liquid nitrogen to keep frozen. After extraction, samples were centrifuged for 15 min at 13,000 g at 4◦C. H2O<sup>2</sup> was assayed with Amplex <sup>R</sup> Red Hydrogen Peroxide Kit (Invitrogen) according to the manufacturer's instructions. The experiment was run in 5 replicates.

#### Fungi Staining, Confocal Microscopy and Plant Colonization Assessment

Plant colonization by the fungus was assessed according to Domka et al. (under review), by comparing the expression of the fungal TEF1α (Translation elongation factor 1-alpha) with plant ACT7 (Actin-7) with qPCR. To visualize mycelium in planta GFP-expressing strain of Mucor sp. (KU234656, strain UNIJAG.PL.E50) was generated (Domka et al., under review). Visualization was performed with a confocal microscopy (Nikon Eclipse, JP) equipped with GFP filter blocks.

#### SL Decomposition Assay

The ability of the fungi to decompose SL in planta was verified by transferring 10 day old seedlings from <sup>1</sup> ⁄<sup>4</sup> MS medium to MSR supplemented with 1 µM GR24 fluorescent analog: GR24- BODIPY (StrigoLab, I) for 24 h to allow the plant to uptake it. Subsequently, seedlings were transferred to fresh MSR and inoculated with Mucor sp. Plants were harvested after 48 h of growth. Visualization of the fluorescence signal was performed with confocal microscopy (Nikon Eclipse, JP). Fluorescence was excited by 490 nm The fluorescence signal intensity was measured with ImageJ (NIH, United States). Five petri dishes with 10 seedlings for treated and not treated plants were prepared.

#### Salicylic Acid and Jasmonic Acid Biosynthesis Induction

To test the relationship between SL and SA and JA production, A. thaliana 10 day old seedlings grown in MS medium were transferred to MSR supplemented with 0, 1, 10, 50, 100, 500, 1000 nM of GR24 and harvested after 10 days of vegetation. To evaluate the impact of the fungi on SA production in SL treated plants, A. thaliana seedlings were transferred from MS to MSR supplemented with 1 µM GR24 and simultaneously inoculated with Mucor sp. (as described in "Plant cultivation"). The temporal pattern of SA production dynamics was evaluated with E+ seedlings grown in medium supplemented with GR24 for 1, 2, 5, and 10 days.

#### Salicylic Acid and Jasmonic Acid Concentration Measurement

Salicylic acid concentration was measured 1, 2, 5, and 10 days after transferring seedlings to MSR supplemented with 1 µM GR24. Inoculation was performed simultaneously with transfer. Unlabeled SA was purchased from Sigma-Aldrich (D). Sample preparation and HPLC analysis were carried out according to Müller et al. (2011) with modifications. Frozen plant roots (about 200 mg Fw – 30 plants per sample) was powdered in liquid nitrogen with a metal pestle in polypropylene tubes and then extracted with methanol:isopropanol:glacial acetic acid (20:79:1; v/v/v) in a 10:1 v/w ratio for 20 min in 4◦C. During extraction sonification was applied. Subsequently, samples were centrifuged for 20 min in 15000 g. This procedure was performed 5 times to assure maximum, close to 100% extraction (from the second extraction 1 ml of extraction solution was used).

The HPLC analysis was performed using Shimadzu LCMS-2020 (JP) system equipped with an autosampler. Separation of plant extracts was performed with a Kinetex 2.6u C18 100x2.1 mm column. The total eluent flow was 0.400 ml min−<sup>1</sup> . Gradient profile described in Müller et al. (2011). The MS analysis was performed using quadrupole mass spectrometer (Shimadzu) negative mode. The following MS parameters were used for analysis: DL temp. 250◦C, HB temp. 200◦C, detector voltage 0.95 kV, oven temp. 35◦C, nebulizing gas flow 15 l min−<sup>1</sup> . The external standard calibration curve method was used for determination of hormone concentration in plant tissues. Five standard solutions were prepared ranging from 0.05 to 10 ng µl −1 . All samples were run in 3 replicates.

#### Gene Expression Analysis RNA Preparation

Total RNA was extracted from frozen in liquid nitrogen, ground leaves (from 5 plants per sample) with the Total RNA Mini Kit (Bio-Rad, United States). RNA purity and quantity was determined by Biospec-Nano (SHIMADZU, JP) The integrity of RNA was assessed with the Agilent 2100 Bioanalyzer (United States) and RNA 6000 Nano Kit (Agilent, DE).

#### qPCR

Reverse transcription was carried out on 1000 ng of total RNA, after digestion with DNase (DNA free kit, Ambion Bioscience, United States), with iScript cDNA synthesis kit (Bio-Rad, United States). For qPCR, probes were labeled with the EvaGreen (SsoFast EvaGreen Supermix, Bio-Rad, United States) fluorescent dye. For a single reaction 10 ng of cDNA and 150 nM of gene specific primers were used. To test amplification specificity a dissociation curve was acquired by heating samples from 60◦C to 95◦C. As house-keeping reference α-tubulin 5 (At5g19780) and ubiquitin 10 (At4g05320) was used. Reaction efficiency was tested by serial dilutions of cDNAs with gene specific primers (Supplementary Table 1). All samples were run in triplicates.

Expression was calculated according to Pfaffl (2001) with WT plants serving as calibrator. For gene expression analysis plants were harvested 10–12 days after inoculation. The experiment was repeated twice. In each experiment 3 samples (10 plants per sample) per variant were collected. Analysis were performed in triplicates.

#### Statistical Analysis

Data normality and variance homogeneity were evaluated by the Shapiro–Wilk and Levene's tests, respectively. If necessary, data were normalized with log or Box-Cox transformation. Statistical significance was determined by analysis of variance (ANOVA), followed by Tuckey or Fischer post hoc test (p ≤ 0.05) as indicated in figure captions. Differences between two groups were tested by t-test. Statistical analyses were conducted using Statistica ver. 12.5 (Statsoft).

#### RESULTS

#### GR24 Inhibits Mycelium Growth and Increases the Respiration Rate

Supplementation of growth medium with GR24 significantly inhibited the growth of Mucor sp. Colony diameter decreased with growing concentrations of GR24. The lowest concentration which had an inhibitory effect was 50 nM. Treatment with 100 nM inhibited mycelium growth in a similar fashion, whereas treatment with concentrations of 500 and 1000 nM resulted in a gradual decline in colony diameter (**Figures 1A,B**). Additionally, as shown in **Figure 1B** treatment with 1000 nM delayed spore formation by the fungi. The respiration rate was significantly decreased only in result of treatment with 1000 nM of GR24. Lower concentrations did not significantly affect it (**Figure 1C**). Mock treatments did not differ from control.

#### GR24 Activates H2O<sup>2</sup> Production and Alters the Activity of Antioxidant Enzymes

GR24 treatment with doses higher than 50 nM induced H2O<sup>2</sup> production in the fungi mycelium. There were no difference in H2O<sup>2</sup> concentration in mycelium treated with 50, 100, and 1000 nM (**Figure 2A**) The activity of GR was significantly decreased in all treatments (**Figure 2B**). The response of CAT and Cu/ZnSOD was dose dependent. Doses of 1 and 10 nM decreased enzyme activity (**Figures 2C,D**) whereas 10– 1000 nM significantly increased Cu/ZnSOD, but not CAT activity (**Figures 2C,D**). Mock treatments did not differ from control.

### Fungi Colonization of SL Biosynthesis Mutants

One day after inoculation fungal hyphae was detected on the surface of the root (**Figures 3A,B**). Later, the mycelium was present inside root hairs (**Figure 3C**) and inside other root

value ± SE.

cells (**Figure 3D**). Colonization of A. thaliana tissues by Mucor sp. was described in detail by Domka et al. (2017, submitted). No significant changes in colonization rate were observed (**Figure 3E**).

#### SL Biosynthesis Mutants Do Not Develop the Beneficial Growth Phenotype Upon Inoculation

Due to differences in growth tempo between experiments, bars in **Figure 4** represent fresh weight treated plants relative to appropriate control: WT, max1, max4 and max2. Inoculation with Mucor sp. improved biomass yield of WT and max2 A. thaliana, whereas SL biosynthesis mutants produced significantly less biomass (**Figure 4**).

#### Mucor sp. Decomposes GR24 in Planta

The fluorescence signal from GR24-BODIPY was significantly lower in E+ plants 48 h after inoculation (**Figures 5A,B**).

### High Doses of GR24 Induce SA Synthesis in A. thaliana, But Has No Effect on JA Production

To test the relationship between SL and SA and JA production, A. thaliana 10 day old seedlings grown in medium MS medium were transferred to MSR supplemented with 0, 1, 10, 50, 100, 500, 1000 nM of GR24. Synthetic SL treatment had no effect on JA accumulation in plants (data not shown). SA concentration was significantly increased (close to 30-fold) in plants treated with the highest dose of GR24 (**Figure 6A**). Lower concentrations did not affect SA synthesis in A. thaliana. No differences in SA concentration were shown 24 hpi (hours past inoculation) (**Figure 6B**). 48 hpi GR24 treated seedlings accumulated significantly more SA. Twelve days after inoculation GR24 treated plants accumulated 25-fold more SA compared to control. Inoculation with the fungi prevented SA production by the plant (**Figure 6B**). 48 hpi SA accumulation was slightly lower than in untreated (GR24+) plants, but not significantly

FIGURE 3 | Roots of A. thaliana colonized by Mucor sp. tagged with GFP. Fungal hyphae on the root surface, bars 100 and 50 µm respectively (A,B). Mycelium (m) inside root hair (h), bar 100 µm (C). Fungal mycelium (m) inside root cells, bar 50 µm (D). Root colonization rate by mycelium shown as fungal TEF1α gene expression in relation to plant ACT7 gene expression. Letters above bars indicate statistically significant differences according to one-way ANOVA and Tuckey post hoc test (N = 3, P ≤ 0.05, mean value ± SE) (E).

FIGURE 4 | Fresh weight of Arabidopsis thaliana WT and max1, max4 and max2 mutants 10 days after inoculation with Mucor sp., GR24 treated and inoculated with Mucor sp. and treated with GR24 (A). Three separate experiments with 25–30 plants per genotype were performed. Due to differences in plant growth in between experiments, fresh weight of treated plants (E+, GR24 and E+GR24) was presented in relation to appropriate control (WT, max1, max4, max2). Stars above bars represent statistically significant differences according to the student's t-test N = 5, P ≤ 0.05, mean value ± SE.

higher than in control and E+ seedlings. No differences in SA concentration were shown 12 dpi between control and E+ seedlings. Mock treatments did not differ from control.

### Salicylic Acid and Jasmonic Acid Accumulation in E+ WT and max1 and 2 Mutants

SA accumulation in max1 mutants was significantly higher, whereas max2 accumulated significantly less SA than in WT plants. Upon inoculation SA concentration increased in WT and max2. max1 mutants responded to inoculation with decreased SA accumulation, which was significantly lower than in WT and max2 (**Figure 6C**). No differences in JA accumulation were shown between WT, max1 and max2 mutants, nor did inoculation affect it (**Figure 6D**).

### The Expression SL Biosynthesis Was Downregulated in E+ Arabidopsis thaliana

The expression of genes encoding proteins involved in SL biosynthesis: CYP711A (cytochrome P450, family 711, subfamily A), D27 (beta-carotene isomerase D27-like protein) and CCD8 (carotenoid cleavage dioxygenase 8) were significantly downregulated in inoculated WT plants, whereas genes encoding proteins involved in SL signaling were either unaffected by inoculation: D14 (strigolactone esterase D14), BRC1 (branched 1) (**Figure 7C**).

#### The Expression Profiles of Plant Defense Related Genes Were Altered in max1 and max2 Mutants During the Interaction With Mucor sp.

In order to evaluate the role of SL in adjusting A. thaliana's defense to inoculation with Mucor sp. the expression of selected plant defensins was quantified (**Figure 7A**). Upon inoculation, the abundance of PR2 mRNA was not changed in WT and max2. max1 exhibited a significant upregulation of the expression of this gene. The expression of PR3 was upregulated in E+WT and E+max1 (in max1 more significantly then in WT), whereas in max2 no differences in PR3 expression after inoculation were shown. No differences in the expression of PR5 between E+ and appropriate controls were shown, however, in the SL biosynthesis mutant PR5 transcript abundance was significantly higher compared to WT and max2 (in both control and E+). Interestingly, PR5 expression in E+max4 was significantly higher (Supplementary Figure 1).

Inoculation with Mucor sp. resulted in a significant upregulation in the expression of PDF1.2 in relation to relevant controls in all plants examined. Expression levels in WT and max2 were similar, whereas PDF1.2 transcript abundance in max1 was significantly higher (in both control and E+ plants). The expression of two defense related transcription factors WRKY25 and WRKY33 was upregulated in max1 compared to WT plants. After inoculation their abundance rose in WT, whereas in max1 and max2 no differences were found. The response of SA biosynthesis genes was also altered in SL biosynthesis mutants. Genes encoding ICS1 and PAL1 were downregulated upon inoculation in max1 E+ in relation to respective control, whereas no differences in their expression were found in WT (**Figure 7B**). The expression of the examined genes in max4 closely resembled that in max1 (Supplementary Figure 1).

#### DISSCUSION

GR24 strongly inhibited the growth of a number of phytopathogenic fungi, suggesting a role in plant defense (Dor et al., 2011). Additionally, hyphal branching was shown to be activated upon treatment with the SL analog. In this study GR24 treatment had a similar effect on the endophytes growth, but it did not affect branching (data not shown). GR24 doses used in this study were, however significantly lower compared to the concentrations used by other authors. Previously, micromolar concentrations of GR24 were shown to limit growth of mycelium (Dor et al., 2011; Belmondo et al., 2017), whereas here, 50 nM of GR24 was sufficient enough to limit colony expansion. This suggests that SL in the mutualistic interaction, imposes its effects in concentrations relatively lower compared to plant-pathogen interactions. Higher pathogen resistance to SL may have evolved due to selective pressure of this group of fungi. This, however, requires more detailed research.

Mycelium growth inhibition was accompanied by increased production/accumulation of H2O<sup>2</sup> and inhibited the activity of GR which plays an important role in H2O<sup>2</sup> scavenging.

Lower concentrations of GR24, below 50 nM, reduced the activity of antioxidant enzymes examined, indicating that SL activates ROS production in the mycelium either directly by inducing an oxidative burst or indirectly, by reducing the activity of ROS scavenging enzymes. ROS were suggested to be necessary in limiting B. cinerea radial growth by GR24 (Belmondo et al., 2017). Previously, Tanaka et al. (2006) showed that by knocking out the NOXA gene encoding the plasmalemma bound, O•− 2 generating NADPH oxidase, the perennial ryegrass endophyte Epichloë festucae spreads throughout the plant causing disease symptoms. This indicated that by inducing ROS production the host plant may control its symbiotic partner. GR24 concentrations of 50 nM and above induced the production of H2O<sup>2</sup> in Mucor sp. and activated the H2O<sup>2</sup> producing/O•− 2 scavenging Cu/Zn SOD, suggesting that SL plays a role in activating ROS production during plant–fungi symbiosis.

Strigolactone are perceived by D14, a non-canonical α/β hydrolase receptor. Upon binding D14 was proposed to hydrolyse SL (Koltai and Kapulnik, 2013). The mechanism of SL perception by fungi is unknown, however, SL decomposition is not a unique plant feature. Recently it was shown that soil borne fungi from the Trichoderma and Fusarium genus were able to degrade 4 different natural and synthetic SLs including GR24. This ability is considered to be used in prevention of parasitic plant seed germination, but the biological relevance of this phenomenon is not known (Boari et al., 2016). In this study, we showed that Mucor sp. can also decompose GR24, both ex planta and more importantly in planta, thus it cannot be excluded that the fungi can modulate SL metabolism during the colonization process. Additionally, the results presented here show that upon inoculation, the expression of SL biosynthesis genes was downregulated in A. thaliana. This provides further evidence for the modulatory role of the endophytic Mucor sp. on SL metabolism during symbiosis establishment.

To evaluate the relevance of SL in the interaction between A. thaliana and Mucor sp. SL mutants were inoculated with the fungi. Biosynthesis mutants max1 and max4 responded negatively, in terms of growth to inoculation. The phenotype of max2 (which synthesizes, but does not respond to SL) after inoculation with Mucor sp. resembled WT, suggesting that SL may had induced changes in Mucor sp. metabolism by switching it from "beneficial to parasitic mode" and MAX2 dependent

test (N = 6, P ≤ 0.05, mean value ± SE).

signaling was not necessary for the development of the beneficial phenotype. What's interesting is that the colonization rate of max1 and 4 mutants did not differ from WT, indicating that in the absence of SL the plants did not lose ability to control colonization.

Plants control mycelium spread inside their tissues by activating a specific immune response. Selected defense mechanisms controlled by plant hormones JA, SA and ET are upregulated and others are downregulated (Jacobs et al., 2011; Lahrmann et al., 2015). A relationship between SL and these phytohormones has also been suggested (Marzec, 2016). According to the only report connecting SL to plant defense (Torres-Vera et al., 2014), SL deficiency alters defense related hormone profiles: the Slccd8 tomato RNAi line accumulated less JA and SA. According to results presented here JA accumulation was not affected by neither SL deficiency, lack of MAX2 dependent SL signaling nor inoculation. To verify weather SL acts in controlling plant defense (possibly in the suppression of plant defense) we quantified the accumulation of SA in WT and max mutants inoculated with the fungus. The results indicate that, SL may act in suppressing SA production: A. thaliana max1 and max4 (see Supplement) accumulated significantly more SA than WT plants. At the same time, SA concentration in WT E+ and max2 E+ increased what was coincident with deactivation of SL biosynthesis genes, suggesting that there may be a relationship between these two processes. SA biosynthesis gene expression was not changed in WT plants what indicates that other routes of SA accumulation regulation (SA catabolism for instance) had to control this process (Lahrmann et al., 2015). Exogenously applied synthetic SL activated SA synthesis in A. thaliana but only in concentrations of 1 µM, lower concentrations did not affect SA production in seedlings. This undermines the hypothesis of the downregulating role of SL in SA production. The physiological relevance of SL activated SA production is controversial. There are no reports indicating that plants can be exposed to such high concentrations in nature, nM and pM concentrations are usually found in planta (Seto et al., 2014), but 1 µM of GR24 is preferentially used in most studies using this SL analog (Marquez-Garcia et al., 2014; Passaia et al., 2014; Ito et al., 2015). In literature, there is only one report indicating that micromolar doses of GR24 can be toxic to the plant (Ito et al., 2015). No reports consider the fact that A. thaliana seedlings treated with GR24 can be exposed to stress (SA biosynthesis is stress responsive), nor consider the fact of SA synthesis activation upon GR24 treatment. Nevertheless, the results of SA accumulation seem

to confirm that SL serves in inhibition of SA accumulation. SA accumulated in max1 mutants, co-cultivation of A. thaliana with Mucor sp. resulted in downregulation of SL biosynthesis genes and an increase in accumulation of SA (probably via a max2 independent pathway). Additionally, we have shown that the fungus has the ability to degrade SL in planta what can be another routes of regulating SL content by the fungus. Several reports indicate that mycorrhizal plants prevent further root colonization by AMF possibly by altering SL production (reviewed in Steinkellner et al., 2007). The results presented here provide indirect evidence confirming that, indeed the endophytic fungus Mucor sp. possesses the ability to modulate SL metabolism/abundance in planta. Verifying the ecological significance of this phenomenon requires, however, further research. The relationship and the role of SL and SA in the establishment of the plant–fungi symbiosis is much more complicated than the picture presented above, the accumulation of SA in max1 E+ and max4 E+ mutants, as well as SA biosynthesis gene downregulation in E+ mutants indicates that other factors are involved in establishing the post infection equilibrium between SA and SL in plant–endophytic fungi interactions.

To further explore the relationship between SL and plant defense during the interaction between A. thaliana and Mucor sp. the expression of selected defense related genes in A. thaliana SL was measured. The expression of plant defensins is SA and JA inducible (Thomma et al., 1998, 1999). The expression of the tested defensins (PR2, PR3, PR5, PDF1.2) was significantly upregulated in inoculated plants under SL deficiency. Additionally, PDF1.2 and PR5 expression was significantly higher in the absence of the fungal stimulus providing further evidence for a link between SL and plant defense. The expression of the two defense related TFs examined also differed in WT and max1. WRKY25 and WRKY33 transcript abundance was higher in max1 and did not respond to inoculation, whereas WT exhibited an activation of their expression. Surprisingly, the response of the SL signaling mutant max2 differed from both WT and max1. Constitutive expression of the genes tested resembled WT, but, barely any of the genes responded to the fungi. Out of the tested genes only PDF1.2 exhibited a response similar to that found in WT. Nevertheless, inhibition of max2 dependent SL signaling resulted in almost complete non-responsiveness of the tested genes. The max2 mutant produces SL and developed the beneficial phenotype upon inoculation, what indicates that SL serves both as a signal influencing the fungi and a signaling molecule controlling plant defense in response to the endophyte. It seems as though, SL MAX2 dependent signaling and activation of defense that takes place in the WT is not a necessary condition in the development of a mutual symbiosis between A. thaliana and Mucor sp. However, defense related gene expression in max1 and max4 mutants indicates that SL plays a role in suppressing the expression of the tested genes. The MAX2 dependent SL signaling pathway is intact in these mutants, suggesting that other factors compensate for SL under SL deficiency. Additionally, MAX2 signaling is not a unique feature of SL signal transduction. Another group of plant signaling molecules: karrikins, derived from cellulose combustion and involved in fire follower plant seed germination (among other functions) utilize MAX2 dependent signaling (Waters et al., 2014). MAX2 also targets several other proteins including the brassinosteroid target protein BRI1-EMS SUPPRESSOR1 (BES1) and the proteins from the DELLA family of GRAS transcriptional regulators involved in gibberellin signaling (Wang et al., 2013). Thus, caution needs to be taken when interpreting the response of the MAX2 mutant.

During the A. thaliana–Mucor sp. interaction an accumulation of defense compounds took place in WT (PDF1.2 and PR3), but in significantly less quantities when compared to max1 and max4. Previously, Jacobs et al. (2011) have shown that some basal defense is necessary for the establishment of the beneficial interaction between an endophytic fungus and its host plant. The results presented here seem to confirm this statement, however, there are variations in the response of A. thaliana to Mucor sp. and other fungi species in terms of activation specific defense related genes.

Upregulated defense mechanisms may not necessarily indicate a direct link between SL and defense. We can also imagine that since SL affects the fungi limiting its growth, it can also affect some unknown recognition patterns rendering the fungi recognizable as an endophyte and the upregulation of plant defense results from not recognizing the fungi as potentially beneficial. Additionally, the negative growth response of SL biosynthesis mutants suggests that in the absence of SL, a shift in resource allocation from growth to defense necessary to restrain fungi spread may take place. This, however, requires further investigations.

### CONCLUSION

Recently, intense efforts are made to define the elements that allow plants to distinguish between potential symbionts and pathogens and to elucidate the plant-fungi cross-talk. Studying signals exchanged in the rhizosphere by plants and fungi is thus significantly important (Bonfante and Genre, 2010; Antolín-Llovera et al., 2014; Hayashi and Parniske, 2014). The role of SL in this process has been proposed previously. The data presented in this study indicates that SL plays a dual role in the interaction between plants and endophytic fungal symbionts. The synthetic SL analog GR24 was shown to affect the metabolism of Mucor sp. by limiting its growth and inducing ROS production. On the other hand, the fungus was shown to be able to decompose SL in planta and reduce SL biosynthesis gene expression. At the same time SL deficient A. thaliana mutants, but not the max2 signaling mutant were not able to benefit from its fungal symbiont. The expression of a number of defense related genes was upregulated in SL deficient plants, suggesting a role of SL in regulating the plants immune response. However, it seems that the postulated link between SL and JA, SA is not as unequivocal as previously thought. Nevertheless, we present several lines of evidence of a direct and/or indirect relation of SL and plant defense.

Additionally, we have shown that high doses of GR24 activate SA production in A. thaliana and that by inoculating the plant with Mucor sp. we were able to prevent SA accumulation most probably by limiting root exposition to GR24 (the fungi decomposed GR24). This model represents a good illustration of the role that fungi play in natural environments. SL decomposition and possible downregulation of SL biosynthesis in the host plant limits SL availability in the soil, what in turn may have a limiting effect on parasitic plant seed germination.

#### AUTHOR CONTRIBUTIONS

PR designed the study, participated in experiments, analyzed the data and prepared the text of the manuscript. AD performed the qPCR, microscopic analyses. RW performed the qPCR and statistical analysis of data. RJ performed the HPLC determination of hormones. MN performed enzyme activity assays. KT was involved in designing the study and interpretation of the results. All authors were involved in the manuscript preparation and approved the manuscript prior to the submission.

#### REFERENCES


#### FUNDING

This work was supported by NSC, Maestro Project, DEC – 2011/02/A/NZ9/00137, COST action FA1206, and Jagiellonian University funds (DS 758).

#### ACKNOWLEDGMENTS

The authors would like to acknowledge Teresa Anielska, Martyna Janicka, Weronika Janas (Institute of Environmental Sciences, Jagiellonian University, Kraków, Poland) and Agnieszka Bednarska (Institute of Nature Conservation, Polish Academy of Sciences, Kraków, Poland) for technical support.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00441/full#supplementary-material



to defense signaling pathways and strigolactone biosynthesis during the early interaction tomato- Phelipanche ramosa. Physiol. Mol. Plant Pathol. 94, 100–107. doi: 10.1016/j.pmpp.2016.05.007


**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 © 2018 Rozp ˛adek, Domka, Nosek, Wazny, J˛edrzejczyk, Wiciarz and ˙ Turnau. 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 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.

# Mixture of Salix Genotypes Promotes Root Colonization With Dark Septate Endophytes and Changes P Cycling in the Mycorrhizosphere

Christel Baum<sup>1</sup> \*, Katarzyna Hrynkiewicz<sup>2</sup> , Sonia Szymanska ´ 2 , Nora Vitow<sup>1</sup> , Stefanie Hoeber<sup>3</sup> , Petra M. A. Fransson<sup>4</sup> and Martin Weih<sup>3</sup>

<sup>1</sup> Soil Science, Faculty of Agricultural and Environmental Sciences, University of Rostock, Rostock, Germany, <sup>2</sup> Department of Microbiology, Faculty of Biology and Environmental Protection, Nicolaus Copernicus University in Torun, Toru ´ n, Poland, ´ <sup>3</sup> Department of Crop Production Ecology, Swedish University of Agricultural Sciences, Uppsala, Sweden, <sup>4</sup> Uppsala BioCenter, Department of Forest Mycology and Plant Pathology, Swedish University of Agricultural Sciences, Uppsala, Sweden

#### Edited by:

Erika Kothe, Friedrich-Schiller-Universität Jena, Germany

#### Reviewed by:

Fred Asiegbu, University of Helsinki, Finland Katarzyna Turnau, Jagiellonian University, Poland Johanna Witzell, Southern Swedish Forest Research Centre, Swedish University of Agricultural Sciences, Sweden

> \*Correspondence: Christel Baum christel.baum@uni-rostock.de

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 13 October 2017 Accepted: 30 April 2018 Published: 18 May 2018

#### Citation:

Baum C, Hrynkiewicz K, Szymanska S, Vitow N, Hoeber S, ´ Fransson PMA and Weih M (2018) Mixture of Salix Genotypes Promotes Root Colonization With Dark Septate Endophytes and Changes P Cycling in the Mycorrhizosphere. Front. Microbiol. 9:1012. doi: 10.3389/fmicb.2018.01012 The roots of Salix spp. can be colonized by two types of mycorrhizal fungi (ectomycorrhizal and arbuscular) and furthermore by dark-septate endophytes. The fungal root colonization is affected by the plant genotype, soil properties and their interactions. However, the impact of host diversity accomplished by mixing different Salix genotypes within the site on root-associated fungi and P-mobilization in the field is not known. It can be hypothesized that mixing of genotypes with strong eco-physiological differences changes the diversity and abundance of root-associated fungi and P-mobilization in the mycorrhizosphere based on different root characteristics. To test this hypothesis, we have studied the mixture of two fundamentally eco-physiologically different Salix genotypes (S. dasyclados cv. 'Loden' and S. schwerinii × S. viminalis cv. 'Tora') compared to plots with pure genotypes in a randomized block design in a field experiment in Northern Germany. We assessed the abundance of mycorrhizal colonization, fungal diversity, fine root density in the soil and activities of hydrolytic enzymes involved in P-mobilization in the mycorrhizosphere in autumn and following spring after three vegetation periods. Mycorrhizal and endophytic diversity was low under all Salix treatments with Laccaria tortilis being the dominating ectomyorrhizal fungal species, and Cadophora and Paraphaeosphaeria spp. being the most common endophytic fungi. Interspecific root competition increased richness and root colonization by endophytic fungi (four taxa in the mixture vs. one found in the pure host genotype cultures) more than by ectomycorrhizal fungi and increased the activities of hydrolytic soil enzymes involved in the P-mineralization (acid phosphatase and β-glucosidase) in mixed stands. The data suggest selective promotion of endophytic root colonization and changed competition for nutrients by mixture of Salix genotypes.

Keywords: arbuscular mycorrhizal fungi, ectomycorrhiza, soil enzymes, phosphorus, fine root density, short rotation coppice, willows

## INTRODUCTION

fmicb-09-01012 May 16, 2018 Time: 19:24 # 2

Mycorrhizal fungi are central to soil fertility and can affect both crop productivity and cropping security (Rooney et al., 2009). Both arbuscular mycorrhizal (AM) and ectomycorrhizal (EM) fungi are known to increase the uptake of nutrients like phosphorus (P) and nitrogen (N) by the host plants, especially in infertile soils. However, their functions and benefits for the host plants might not be equivalent (Jones et al., 1998). For this reason, a change in the proportions of mycorrhizal colonization, or type of mycorrhizal association when the host plant can form dual mycorrhiza, might affect the mycorrhizal benefit of the host plant. The mycorrhizal colonization of plants can be affected by soil nutrient level and water content, site management as well as the diversity of the vegetation composition. An increased host plant diversity is known to influence the nutrient fluxes and the interactions with soil microorganisms (Courty et al., 2011). Endophytic fungi and mycorrhizal fungi can interact in their impact on the plant growth (Rillig et al., 2014). For example, AM fungi were revealed to be able to modulate the impact of endophytic fungi beneficially for the plant growth (W˛ezowicz ˙ et al., 2017).

Although the majority of higher plants form mycorrhizal associations, only a low number is dual mycorrhizal, forming both AM and EM associations (Smith and Read, 2008). Salix belongs to the dual mycorrhizal plant genera, and can also harbor diverse endophytic fungi, which may affect the hosts growth and ecophysiological traits (An et al., 2015). Although Salix spp. were mainly shown to be pre-dominantly EM (Püttsepp et al., 2004; Hrynkiewicz et al., 2010b), the opposite has also been shown with a dominance of AM colonization, e.g., in response to flooding (Lodge, 1989). The ability to change their mycorrhizal association in response to the environmental conditions makes them well-suitable model plants to investigate the changes in the mycorrhizosphere in response to a changed host diversity.

Fast growing Salix genotypes have been cultivated successfully in short rotation coppice (SRC) for a long time, mainly for biomass production for energy purposes in boreal climates (Weih, 2004). They are viewed as a sustainable source of biomass with a positive greenhouse gas balance due to their potential to fix and accumulate carbon (Cunniff et al., 2015). Furthermore, compared to annual systems grown on arable land, the management of perennial Salix stands in SRC leads to decreased mechanical disturbance of the soil and changed biochemical soil properties; along with changes in the abundance and diversity of soil organisms including mycorrhizal fungi (Baum et al., 2009a; Hrynkiewicz et al., 2010a; Fransson et al., 2013; Toju et al., 2013a; Goldmann et al., 2015). In addition, Salix genotype identity can significantly affect the soil enzyme activities at the same site (Baum and Hrynkiewicz, 2006), and Salix stands grown in a floodplain revealed higher activities of the soil enzyme β-glucosidase compared to perennial grassland (Zimmer et al., 2012). Changed soil enzyme activities are considered to be a direct expression of the soil community to metabolic requirements and available nutrients, and can therefore be used as indicators of soil functional diversity. Enzyme activities might be more indicative for ecosystem productivity and stability than the taxonomic diversity of soil microorganisms (Caldwell, 2005). Acid phosphatases and β-glucosidases are hydrolytic soil enzymes involved in P- and C-cycling of the soil and originate from plant roots and microorganisms. Mycorrhizal fungi can significantly increase the activities of these enzymes in the soil (Burke et al., 2011).

In SRC, Salix is usually planted with single genotypes, but this could possibly make them less resistant to pest organisms, diseases and abiotic stress. This is a major concern as a SRC must remain viable for up to 20 years to be profitable (Aravanopoulos et al., 1999). For this reason, an increased host plant diversity within Salix stands would be advantageous, but it will also change the below-ground plant–plant interactions within the stand. The soil ecological consequences of mixed Salix stands on former arable land are scarcely known so far (Hoeber et al., 2017).

We hypothesize that the mixture of genotypes with strong eco-physiological differences changes the mycorrhiza formation and activities in the mycorrhizosphere, based on different root characteristics of the genotypes involved. To test this hypothesis we have studied the mixture of two fundamental eco-physiological different Salix genotypes (S. schwerinii × S. viminalis cv. 'Tora' and S. dasyclados cv. 'Loden') grown in pure and mixed cultures in a field experiment in Northern Germany. Specifically, we expected (1) a higher diversity of root-associated fungi under mixed culture than in culture with the pure genotypes; and (2) changed fine root growth, abundance of root-associated fungi and enzymatic activities in the mycorrhizosphere under host genotype mixtures resulting from changed competitive conditions for the individual genotypes. Such insights into the overall structure of belowground plant–fungal associations will help to understand mechanisms that regulate the coexistence of eco-physiological diverse dual mycorrhizal plant species.

#### MATERIALS AND METHODS

#### Study Site and Test Plants

The field site Rostock (54◦ 040 12<sup>00</sup> N, 12◦ 040 58<sup>00</sup> E) has been established in spring 2014 on former arable soil and is one out of three experimental field sites within the ECOLINK-Salix project (Hoeber et al., 2018), which is part of a global tree diversity network (TreeDivNet, Verheyen et al., 2016; Paquette et al., 2018). The plantation is a SRC stand with two willow genotypes ['Tora,' Svalöf-Weibull (SW) Cultivar No. 910007, S. schwerinii E. Wolf × S. viminalis L.; and 'Loden' SW 890129, S. dasyclados] grown in plots with one genotype being present (pure) and the combination (mixture) of the two. 'Tora' and 'Loden' in pure culture ('Tora P' and 'Loden P') were monocultures. In the mixture the two genotypes ('Tora M' and 'Loden M') were planted alternating one by one within the rows. Therefore, each genotype was directly associated with plants of the other genotype in the mixture.

Shoot biomass was harvested for the first time after 3 years of growth (in winter 2016/2017). The overall goal of the ECOLINK-Salix project is to assess the effects of genotype identity and diversity in willow SRC on various ecosystem functions. The

experiment has a randomized block design, with three replicates (blocks, plot size: 92.16 m<sup>2</sup> ). Planting density was c. 15600 plants ha−<sup>1</sup> . Each plot contains nine subplots (3.2 m × 3.2 m, 16 plants each).

The dominating soil type at the test site is a Stagnic Cambisol (FAO classification). The annual mean temperature is 8.5◦C; the annual mean precipitation 592 mm. The texture class of the topsoil is predominantly loamy sand. General soil chemical properties (0–10 cm soil depth) were: pH (CaCl2) 6.2, Corg 6.89 g kg−<sup>1</sup> , Ntotal 0.86 g kg−<sup>1</sup> , C/N 8.01, Ptotal 0.11 g kg−<sup>1</sup> , and Stotal 0.09 g kg−<sup>1</sup> .

#### Soil and Fine Root Sample Collection

Soil and fine root sampling was done in November 2016 (at the end of the first rotation period after 3 years of growth) and in April 2017 (after the first shoot biomass harvest during the winter rest in February 2017). Nine soil cores (diameter 30 mm, depth 0–100 mm) per treatment were collected per genotype (2), sampling date (2), and planting layout (2: mono- [pure] and di-clonal [mix]) in the center of the plots for the determination of the fine root density, and of soil chemical and biochemical analyses (72 soil samples in total). The same number of samples and the same sampling design was used to collect the samples for the mycorrhizal analyses, but using 10 cm × 10 cm × 10 cm soil cubes taken with a sharp knife. The samples were collected from the uppermost 0–10 cm of soil, c. 20 cm from the stem base of willow plants and kept at 4◦C until analyses. The majority of fine roots were found in 0–10 cm under diverse Salix species in SRC (Cunniff et al., 2015). For this reason the upper part of the topsoil seems to be most relevant for investigations of the mycorrhizosphere.

#### Soil Properties

The soil analyses were done using air-dried < 2 mm soil. Determination of soil pH was performed electrometrically using a glass electrode in 0.01 M CaCl<sup>2</sup> with a soil:solution ratio of 1:2.5. Total carbon (Ctotal), total nitrogen (Ntotal), and total sulfur (Stotal) contents of the soils were determined with a Vario EL elemental analyzer (Elementar Analysensysteme GmbH, Hanau, Germany). All C was organic C (Corg) due to the acidic pH. The total phosphorus (Ptotal) content was extracted from 0.5 g dry soil material by microwave-assisted digestion with aqua regia solution (3:1 hydrochloric acid–nitric acid) (Chen and Ma, 2001). The P concentrations in the digests were measured by inductively coupled plasma optical emission spectroscopy (ICP-OES) (Optima 8300, PerkinElmer LAS GmbH, Rodgau, Germany).

#### Soil Enzyme Activities

The activities of acid phosphatases (EC 3.1.3.2) and β-glucosidases (EC 3.2.1.3) in the soil were determined colorimetrically according to Tabatabai (1982) and Eivazi and Tabatabai (1988). The enzyme activities were expressed as µg p-nitrophenol (pNP) g−<sup>1</sup> soil h−<sup>1</sup> released from the pre-given substrate solution (p-nitrophenyl-phosphate for acid phosphatases and p-nitrophenyl-β-D-glucosid for β-glucosidase) within 1 h of incubation.

#### Fine Root Density

Soil cores with plant roots were soaked in tap water for 1 h in a bowl and then carefully separated with forceps and washed with the help of a sieve. Roots were dried at 65◦C for 48 h and the dry weights were calculated per soil volume and m<sup>2</sup> within the upper 0–10 cm soil depth.

#### Fungal Colonization of Fine Roots

For analyses of AM and other endophytic colonization roots were rinsed and cut into 10 mm segments. The segments were cleared with 10% (w/v) KOH for 15 min at 90◦C, bleached in 10% H2O<sup>2</sup> for 1 h, acidified with 1% HCl (van der Heijden, 2001) and stained with 0.05% (w/v) chlorazol black E for 90 min at 80◦C. The AM colonization was quantified microscopically using the intersection method of McGonigle et al. (1990), considering arbuscle formation as indicator of AM fungi vs. formation of sclerotia as indicator of dark septate endophytes (DSE). A minimum of 200 fine root intersections for AM colonization and a minimum of 200 fine root tips for DSE colonization were observed per plot.

For EM colonization and morphological characterization, 10 sub-samples of root fragments were randomly chosen on a grid for microscopical quantification of EM colonization. The numbers of living non-colonized root tips vs. obviously colonized EM root tips were counted using the method of Agerer (1991). In total c. 4800 root tips were scanned. A minimum of 200 root tips per genotype and treatment was investigated for each sampling date.

A minimum of two root tips per sub-sample was selected for subsequent molecular identification of the fungal partners. The morphological EM types were distinguished by macroscopical characteristics of the fungal mantle, such as color, surface appearance, presence of emanating hyphae and hyphal strands (Agerer, 1987–2002). Two to five root tips per sub-sample were separately frozen in Eppendorf-tubes and stored at −20◦C for molecular analysis.

#### Molecular and Phylogenetic Identification of Root Colonizing Fungi

The fungal taxa from the root samples were identified using analysis of DNA sequences of the internal transcribed spacer (ITS) region. DNA was isolated from root tips using the Plant & Fungi Purification Kit (EurX, Poland) according to the manufacturer's instructions. The ITS region within the ribosomal RNA genes was amplified using the primer pair ITS1F and ITS4 (White et al., 1990; Gardes and Bruns, 1993). Amplification of the ITS region was performed in a 25 µl final volume containing 50 ng of DNA, 0.125 µl of each primer (100 pM/µl), 10 µl of Master Mix (Qiagen) and 13 µl of nuclease-free water. The amplification protocol was as follows: 5 min at 94◦C, 40 cycles of 30 s at 94◦C,

1 min at 50◦C, and 1 min at 72◦C, and a final step at 72◦C for 5 min. The PCR products obtained were purified using GeneMATRIX PCR/DNA Clean-Up Purification Kit protocol (EurX, Poland) and sequenced using primers ITS1F and ITS4. Sequence editing was performed using Sequencher TW Version 5.1 (Gene Codes, Ann Arbor, MI, United States). BLAST searching with ITS-sequences was performed on the GenBank (Altschul et al., 1990). If the sequence of the fungus showed 98% identities over the whole length of the sequence (about 600 to 700 bp) with a known fungus, this fungus was assumed to be the fungal root colonizing taxa. Contigs of ITS sequences were edited with EditSeq and aligned using Clustal W (DNASTAR <sup>R</sup> ). DNA sequences were submitted to GenBank and accession numbers are presented in **Table 1**.

#### Statistical Analyses

The effect of the genotype-, the growth design (pure vs. mixed) and the interaction of both on biological and biochemical properties were analyzed by two-way ANOVA. The Detrended correspondence analysis (DCA) was used to disclose the effects of different host plant diversity (pure vs. mixed culture) on the root fungal colonization and soil enzyme activities in the mycorrhizosphere of Salix. Statistical analyses were computed using the software PAST (Hammer et al., 2001).

### RESULTS

#### Fine Root Biomass and Fungal Colonization

The fine root biomass was significantly affected by the interaction of genotype × growth design (pure vs. mixture) (**Table 2**). In pure culture 'Loden P' revealed higher fine root biomass than 'Tora P,' but under the mixture no significant differences between the two genotypes were observed (**Figure 1**). The means of the fine root biomass decreased in all treatments from autumn 2016 to spring 2017.

The AM colonization was relatively small (<12% of the total fine root length) in all treatments, but higher under 'Tora M and P,' than under 'Loden M and P' (**Figure 2**). The AM colonization of the fine roots was not significantly affected by the growth design (pure or mixed culture; **Table 2**). It was higher in spring than in autumn (**Figure 2**).

The EM colonization was significantly larger under 'Loden M and P' than under 'Tora M and P,' and significantly

TABLE 1 | Molecular identification of fungal partners in fine roots of Salix genotypes 'Tora' and 'Loden' in pure (P) and mixed (M) growth design in 0–10 cm soil depth at the short rotation coppice Rostock in autumn 2016 and spring 2017.


<sup>∗</sup>NCBI, ∗∗UNITE.


TABLE 2 | Results of two-way analysis of variance (ANOVA) on the effect of the genotype, the growth design (with different host plant diversity; pure vs. mixture) and their interactions (genotype × growth design) on root and soil properties in the mycorrhizosphere of Salix from autumn 2016 (autumn) and spring 2017 (spring).

p- Significance value (in bold are p < 0.05).

higher under the mixture in autumn 2016. In spring 2017 no significant impact of the growth design on the EM colonization was observed was (**Figure 3A** and **Table 2**). The EM colonization was in the mean slightly higher in autumn than in spring.

The DSE colonization was significantly larger in the mixture of genotypes ('Loden M' and 'Tora M') than in their pure cultures ('Loden P' and 'Tora P') (**Figure 3B**).

#### Fungal Diversity on the Fine Roots

With more than 95% of the total EM colonization (data not shown), Laccaria tortilis dominated as the single fungal partner in EM formation in all plots and was isolated from the roots of both genotypes (**Table 1**). Sporocarps of Laccaria tortilis were observed in all plots during the autumn sampling 2016. Other rarely occurring EM morphotypes observed were morphologically–anatomically determined to

belong to Inocybe and Russula spp. and were found under both clones.

One endophytic fungal taxa was identified from the genotype 'Tora' in pure culture (Pucciniomycotina) and four endophytic fungal taxa from the genotypes in mixed culture (Cadophora sp., Paraphaeosphaeria sp., Rhodotorula mucilaginosa, Pleosporales) (**Table 1** and Supplementary Figure 1).

#### Soil Enzyme Activities

The activity of acid phosphatase in the soil was significantly higher under the mixed growth design, but not significantly affected by the genotype and the interaction of genotype x growth design (**Figure 4A** and **Table 2**). The activities of β-glucosidase in the soil were higher under the genotype 'Loden P' in pure culture than under 'Tora P' in pure culture (**Figure 4B** and **Table 2**).

#### Comparison Between Plots With Pure Host Genotypes and Genotype Mixtures

The mixed growth design of the genotypes increased significantly the colonization of fine roots by DSE, and the activities of acid phosphatases and β-glucosidases in the soil (**Table 2**).

The DCA of all fungal parameters and the two soil enzyme activities (acid phosphatase and β-glucosidase) clearly differentiated the genotypes (L for 'Loden,' T for 'Tora') and the host genotype mixture (M) from the two pure cultures (P) in autumn 2016 (**Figure 5A**) and spring 2017 (**Figure 5B**). The difference between the pure and the mixed culture was larger in spring 2017 than in autumn 2016, since the data differentiate the treatments along axis 2 (explaining about one quarter of the variation) in autumn, but along axis 1 in spring (explaining about half of the variation). The pure culture 'Tora P' was always closer to the mixture than the pure culture 'Loden P.'

#### DISCUSSION

The present results confirm the significant impact of the Salix genotype on its phenotypical traits (Cunniff et al., 2015) including the mycorrhizal (see **Figures 2**, **3A**) and endophytic (see **Figure 3B**) fungal colonization and host-specific differences

in the fungal root colonization within a plant community as described by Toju et al. (2013b). The soil properties of the test site indicated P-deficiency (see section "Study Site and Test Plants"), which increases the importance of mycorrhizal fungi for the P-supply of the host plants (Smith et al., 2011). Assuming that plant neighbors with a faster growth rate can increase the competition for plants with slower growth, in the present mixture 'Tora M' had an advantage over 'Loden M.' Both, decreased fine root biomass under 'Loden M' grown in a mixture (see **Figure 1**) and greater similarity of the root traits and soil enzyme activities of the mixture compared with pure cultures 'Tora P' or 'Loden P' support this assumption (see **Figure 5**). These results are in line with our hypothesis that fine root growth, fungal abundance and activities are changed under host genotype mixtures, caused by changed competitive conditions for the individual genotypes and thereby changed interactions between them. However, the genotype-effect on the mycorrhizal colonization exceeds the impact of the plant design (pure or mixture of genotypes) on the mycorrhizal abundance (see **Figures 2**, **3A** and **Table 2**).

We predicted that the increased diversity of host plants in the genotype mixture will increase the diversity of root associated fungi too, since aboveground and belowground diversity can be linked (De Deyn and Van der Putten, 2005). However, we found increased fungal diversity only in endophytic fungi, not in mycorrhizal ones. This increased diversity was associated with increased colonization of the roots with DSE. Generally, low mycorrhizal diversity on Salix found here by only one

dominating EM fungal species (Laccaria tortilis; see **Table 1**) causes problems when interpreting the results. However, the low EM diversity is in line with the results on other genotypes of Salix (Püttsepp et al., 2004; Hrynkiewicz et al., 2012). It might be caused by the genotypic-specificity of this plant genus and additionally by the former arable site-conditions with lack of EM host plants. Therefore, we have focused rather on the ratios of root associations than on total mycorrhizal and endophytic diversity.

For the first time, we disclosed that increased host diversity of Salix genotypes can promote endophytic root colonization and increase soil enzyme activities involved in P-mobilization (see **Figures 2**–**4**). These results of a Continental SRC support the assumption of Pérez-Izquierdo et al. (2017) taken for Mediterranean forests, who highlighted the importance of structural shifts in fungal communities for their possible functional consequences. Functional consequences were revealed in an increased enzymatic P mobilization (see **Figure 4**) in the mycorrhizosphere with increased host plant diversity (mixture). These increased enzymatic activities can be caused directly by fungal impact, but also by the plant impact or most probably by a combination of both.

Dark septate endophytes were previously described to be the dominating root associated fungi under S. caprea on heavy metal contaminated sites (Likar and Regvar, 2013). In agreement with the results of these authors, we also found Cadophora spp. as a very common DSE in the roots of the two tested Salix genotypes at the non-contaminated arable site. This endophyte revealed plant growth promotion in heavy metal contaminated soils (Likar and Regvar, 2013). Also the second most common DSE in the present study (Paraphaeosphaeria sp., see **Table 1**), revealed plant growth promoting traits on another Salix sp. in heavy metal contaminated soil (An et al., 2015). The endophytic yeast Rhodotorula mucilaginosa, found under mixed Salix growth in the present study, was revealed to be plant growth promoting in Populus (Xin et al., 2009). However, from the present data no statement can be made on the growth response of Salix on the different fungal root colonization and the impact of DSE can vary from growth promotion to growth suppression of host plants (Aguilar-Trigueros and Rillig, 2016).

Increased colonization of the roots with DSE combined with consistent colonization with AM fungi in the mixture (see **Figures 1**, **3B**) were accompanied by increased soil enzyme activities involved in P-mobilization (see **Figure 4**). Della Monica et al. (2015) assumed, that DSE can be more effective to mobilize organic P, than AM fungi. However, in the present study increased enzyme activities might be stronger affected directly by the changed nutrient competition of the host plants in the mixture. Increased abundance of DSE without suppression of mycorrhiza formation supports that these fungal groups can interact beneficially in a joint host plant (W˛ezowicz et al., 2017 ˙ ). This might be caused by exudates of DSE, which can even stimulate the growth of mycorrhizal fungi (Scervino et al., 2009). Oppositely to the promotion of DSE in the mixture of different Salix genotypes, Becklin et al. (2012) revealed promoted of EM fungi on Salix in increased plant diversity by combination with herbaceous plants in an alpine meadow. However, these results have in common that host plant diversity changes the fungal root colonization of Salix.

The present data on dual mycorrhizal colonization of AM and EM fungi on Salix clearly confirmed the assumption of Lodge and Wentworth (1990), that these types of mycorrhizal fungi colonize

variance explained: 26.1%).

roots with an antagonistic behavior. Interestingly, Chen et al. (2000) found that antagonistic behavior of AM and EM fungi was most severe when the EM fungus Laccaria was present, which was the dominating EM fungus in the present study. Low fungal diversity was observed also in the sporocarp production in other SRCs (Baum et al., 2002).

The slow-growing genotype 'Loden' seems to be rather EM dominated, whereas the fast-growing genotype 'Tora' seems to promote AM formation (see **Figures 2**, **3**). This might be affected by the different litter quality of these genotypes, since the litter quality affects mycorrhizal communities (Lambers et al., 1998). In a pot experiment, leaves of 'Loden' revealed higher phenolic concentrations in the later growth than 'Tora' under the same environmental conditions (Baum et al., 2009b). Leaf litter with higher phenolic concentrations was described to support EM formation, whereas low phenolic concentrations were found to promote AM colonization of the host plants (Lambers et al., 1998).

Interestingly, the colonization of the roots with DSE was stronger affected by the host plant diversity than by the genotype-specific differences, which was revealed by an increased colonization of the roots (see **Figure 3B**) combined with increased species richness (see **Table 1**) in the mixture. Increased stress tolerance of Salix spp. was previously linked to colonization by several DSE (An et al., 2015). However, root endophytes can also include potential pathogens and each plant species and combination with different fungal strains might respond differently in plant growth (Aguilar-Trigueros and Rillig, 2016). Furthermore, the colonization of roots with DSE is also affected by edaphic properties of the site (Xie et al., 2017). The impact of edaphic differences within the present test site was initially balanced between the treatments by the randomized plot design in the present study. However, the increased colonization by DSE might also lead back to changed nutrient mobilization caused by the changed root growth of at least one of the mixed genotypes (see **Figure 1**) and indicated by changed enzymatic P mobilization under the mixture (see **Figure 4**).

Based on our results, it can be concluded that increased genotype diversity in Salix cultivation can lead to changed root competition, fungal association and enzymatic P-mobilization in the mycorrhizosphere. Subsequent investigations should assess the generality of promotion of DSE and P-mobilization in increased host diversity.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

MW conceived and designed the field experiment within the frame of ECOLINK-Salix. CB managed the field site Rostock, collected the samples, did soil and root analyses and statistical analyses and wrote the first draft of the manuscript; did all soil chemical and biochemical analyses and wrote part of Section "Materials and Methods." NV and KH isolated the ectomycorrhizal fungi, created figures and tables and did statistical analyses. KH and SS identified the fungal species. NV analyzed the colonization density of mycorrhizal fungi and analyzed these parts of the results. SH and PF evaluated and discussed the mycorrhizal and endophytic results. All authors edited and revised the manuscript and approved the publication.

#### FUNDING

The authors gratefully acknowledge the German Federal Ministry of Education and Research (BMBF) for funding the BonaRes project InnoSoilPhos (Project No. 031A558). The work of MW and PF was partly funded by the Swedish Energy Agency (Project Nos. 36654-1 and 36654-2). Partial support to SH was provided by the Swedish Research Council Formas (Project No. 2016- 00998).

#### ACKNOWLEDGMENTS

We are grateful to Ms. E. Heilmann (University of Rostock, Germany) for valuable technical assistance during the chemical soil analyses. This research was performed within the scope of the Leibniz ScienceCampus Phosphorus Research Rostock. We are grateful for the very valuable comments of the reviewers.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.01012/full#supplementary-material

Franch. in Three Gorges Reservoir Region. China. Microbiol. Res. 176, 29–37. doi: 10.1016/j.micres.2015.03.013



phosphorus uptake efficiency and external hyphal production. New Phytol. 140, 125–134.



**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 © 2018 Baum, Hrynkiewicz, Szymanska, Vitow, Hoeber, Fransson and ´ Weih. 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 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.

# Piriformospora indica Reprograms Gene Expression in Arabidopsis Phosphate Metabolism Mutants But Does Not Compensate for Phosphate Limitation

Madhunita Bakshi<sup>1</sup> , Irena Sherameti<sup>1</sup> , Doreen Meichsner<sup>1</sup> , Johannes Thürich<sup>1</sup> , Ajit Varma<sup>2</sup> , Atul K. Johri<sup>3</sup> , Kai-Wun Yeh<sup>4</sup> and Ralf Oelmüller<sup>1</sup> \*

1 Institute of General Botany and Plant Physiology, Friedrich-Schiller-University Jena, Jena, Germany, <sup>2</sup> Amity Institute of Microbial Technology, Amity University, Noida, India, <sup>3</sup> School of Life Sciences, Jawaharlal Nehru University, New Delhi, India, 4 Institute of Plant Biology, Taiwan National University, Taipei, Taiwan

#### Edited by:

Katarzyna Turnau, Jagiellonian University, Poland

#### Reviewed by:

Vito Valiante, Leibniz-Institute for Natural Product Research and Infection Biology - Hans Knöll Institute, Germany Susana Rodriguez-Couto, Ikerbasque, Spain

> \*Correspondence: Ralf Oelmüller b7oera@uni-jena.de

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 24 February 2017 Accepted: 23 June 2017 Published: 12 July 2017

#### Citation:

Bakshi M, Sherameti I, Meichsner D, Thürich J, Varma A, Johri AK, Yeh K-W and Oelmüller R (2017) Piriformospora indica Reprograms Gene Expression in Arabidopsis Phosphate Metabolism Mutants But Does Not Compensate for Phosphate Limitation. Front. Microbiol. 8:1262. doi: 10.3389/fmicb.2017.01262 Piriformospora indica is an endophytic fungus of Sebacinaceae which colonizes the roots of many plant species and confers benefits to the hosts. We demonstrate that approximately 75% of the genes, which respond to P. indica in Arabidopsis roots, differ among seedlings grown on normal phosphate (Pi) or Pi limitation conditions, and among wild-type and the wrky6 mutant impaired in the regulation of the Pi metabolism. Mapman analyses suggest that the fungus activates different signaling, transport, metabolic and developmental programs in the roots of wild-type and wrky6 seedlings under normal and low Pi conditions. Under low Pi, P. indica promotes growth and Pi uptake of wild-type seedlings, and the stimulatory effects are identical for mutants impaired in the PHOSPHATE TRANSPORTERS1;1, -1;2 and -1;4. The data suggest that the fungus does not stimulate Pi uptake, but adapts the expression profiles to Pi limitation in Pi metabolism mutants.

Keywords: root expression profiles, PHT1, WRKY6, Piriformospora indica, phosphate starvation

## INTRODUCTION

Piriformospora indica, an endophytic fungus of Sebacinaceae, colonizes the roots of many plant species and promotes their growth and performance (Peškan-Berghöfer et al., 2004; Waller et al., 2005, 2008; Shahollari et al., 2007; Oelmüller et al., 2009; Camehl et al., 2010; Nongbri et al., 2012; Varma et al., 2012; Jogawat et al., 2013; Ye et al., 2014). The fungus improves nutrition uptake from the soil to the host roots (Sherameti et al., 2005; Shahollari et al., 2005; Kumar et al., 2011) suggesting a strong fungal influence on the plant transport processes and metabolism. Mycorrhizal and beneficial root-colonizing fungi deliver phosphate (Pi) to the roots. Pi is taken up by the plants through Pi transporters, and low- and high-affinity Pi transporters of the PHOSPHATE TRANSPORTER1 (PHT1) family have been reported to be involved in the mycorrhizal pathway (Harrison et al., 2002; Garcia et al., 2016). While mycorrhizal plants contain fungus-inducible PHT1 genes (Bucher, 2007; Chen et al., 2007; Walder et al., 2015), plants which do not interact

**Abbreviations:** LP, low phosphate concentration; NP, normal phosphate concentration; Pi, phosphate; WT, wild-type.

with mycorrhizal fungi, such as Arabidopsis, lack inducible PHT1 genes. The expression of their Pi transporter genes is independent of root colonization by endophytic root-colonizing microbes, but some members of the PHT1 gene family respond to Pi deficiency (e.g., Chiou et al., 2001; Ai et al., 2009; Ayadi et al., 2015). In return, up to 50% of the carbon fixed by photosynthesis can be delivered to mycorrhizal fungi associated with the roots in natural ecosystems (Nehls et al., 2010).

Root colonization and nutrient exchange by P. indica results in substantial alterations in the root architecture, an effect which is highly host specific (cf. Lee et al., 2011; Dong et al., 2013). In Arabidopsis, P. indica stimulates root growth: while the primary root length is slightly reduced, lateral root growth and root hair development is promoted (e.g., Varma et al., 1999; Barazani et al., 2005; Vadassery et al., 2008, 2009a; Johnson and Oelmüller, 2009; Fakhro et al., 2010; Das et al., 2012, 2014; Lahrmann and Zuccaro, 2012; Lahrmann et al., 2013; Prasad et al., 2013; Venus and Oelmüller, 2013; Bakshi et al., 2015; Vahabi et al., 2015). Secondary metabolites, such as indole-3-acetaldoxime derivatives, hormones, and different defense compounds/pathways control the colonization of roots during Arabidopsis/P. indica interaction and thus influence root development (Sherameti et al., 2008; Camehl et al., 2011; Hilbert et al., 2012; Nongbri et al., 2012; Bakshi et al., 2015; Matsuo et al., 2015). P. indica also protects the roots by stimulating the antioxidant system under stress, which again influences developmental programs in response to environmental cues and the redox state of the roots (Baltruschat et al., 2008; Vadassery et al., 2009b; Harrach et al., 2013). To what extend root developmental and genetic programs are altered by endophytes such as P. indica is not completely understood. The quite different root responses of various plant species to P. indica demonstrate the important role of the host genetic programs (discussed in Lee et al., 2011; Dong et al., 2013).

Here, we investigate the role of Pi availability and the Arabidopsis transcription factor WRKY6 involved in the regulation of the plant Pi metabolism in the P. indica/Arabidopsis symbiosis. The Arabidopsis mutant lacking the WRKY6 transcription factor shows an altered response to LP stress (Chen et al., 2009; Bakshi et al., 2015), since WRKY6 is involved in regulating PHOSPHATE1 (PHO1) expression. PHO1 is a Pi exporter and required for the transfer of Pi from root epidermal and cortical cells to the xylem. The pho1 mutant has low shoot Pi and shows Pi deficiency symptoms, including poor shoot growth and overexpression of numerous Pi-deficiency responsive genes (cf. Wege et al., 2016). LP treatment reduced WRKY6 binding to the PHO1 promoter and thus allows PHO1 expression (Chen et al., 2009). Therefore, deletion of WRKY6 alters the root development of WT seedlings in a Pi-dependent manner (Bakshi et al., 2015). We compared the expression profiles of WT and wrky6 seedlings grown under normal Pi (NP) and low Pi (LP) conditions and found that the response to P. indica is quite different in the roots of WT and wrky6 seedlings, both under NP and LP conditions. This suggests that the genetic response of Arabidopsis to P. indica is highly dependent on the nutrient availability and genotype of the host. We also investigated the role of Arabidopsis Pi transporters in the symbiosis and demonstrate that Pi limitation, due to inactivation of the plant Pi transporters, cannot be compensated by P. indica colonization.

#### MATERIALS AND METHODS

#### Growth Conditions of Plant and Fungus

WT, wrky6, pht1;1 (At5g43350), pht1;2 (At5g43370), and pht1;4 (At2g38940) seeds and seeds of the pht1;1 pht1;4 double knockout line were surface sterilized and placed on Petri dishes containing MS (Murashige and Skoog, 1962) nutrient medium (with 13,7 g/l sucrose). After cold treatment for 48 h at 4◦C, the plates were incubated for 10 days at 22◦C under continuous illumination (100 µmol m−<sup>2</sup> s −1 ). P. indica was cultured as described previously on Aspergillus minimal medium (Johnson et al., 2011). After 10 days of growth, all seedlings were either transferred to fresh plates (MS or plant nutrient medium (PNM) as indicated in the figure legends, or to soil, cf. **Figure 6**) for the different treatments. The procedure has been described in details in Johnson et al. (2011)<sup>1</sup> . For the arsenate assay, they were transferred to fresh MS medium containing 1 mM Pi (KH2PO4) and 200 µM sodium arsenate (V) for 3 weeks (cf. **Table 2**). Alternatively, they were transferred to PNM medium (without sucrose) for 2 weeks. The PNM medium contained a fungal P. indica plaque in the middle (co-cultivation with P. indica) or an agar plaque without fungal hyphae (control, without P. indica), as described previously (Vadassery et al., 2009a). The fungus was allowed to grow on the PNM medium for 1 week before the seedlings were transferred to the fungal lawn. Control plates were treated in the same way. To test whether simultaneous inactivation of PHT1;1 and PHT1;4 has a long-term effect on plant performance, the 10-day old seedlings were transferred from MS plants (cf. above) to soil and kept in a greenhouse for 4 weeks. For the plate experiments shown in **Figure 6** (TOP), the seedlings were transferred to PNM medium with NP for 3 weeks.

#### Generation of the Homozygous Knock-Out Lines

The following SALK insertion lines were used for the PHT1 genes: pht1;1 (At5g43350; SALK\_088586C/N666665), pht1;2 (At5g43370; SALK\_110194) and pht1;4 (At2g38940; SALK\_103881). Homozygocity was tested with gene-specific primer pairs (cf. Supplementary Table S1) and primers given on the SALK homepage<sup>2</sup> . The pht1;1 pht1;4 double knock-out line was generated by crossing the two homozygote single knock-out lines. After confirmation of homozygocity of the two inactivated genes, the lack of PHT transporters was also confirmed with the arsenate resistance assay. The wrky6 line was described in Bakshi et al. (2015).

<sup>1</sup>http://pubman.mpdl.mpg.de/pubman/item/escidoc:1587455/component/ escidoc:1674163/IMPRS056.pdf <sup>2</sup>www.arabidopsis.org

#### Arsenate Resistance Assay

fmicb-08-01262 July 10, 2017 Time: 18:13 # 3

Seeds were germinated on full MS medium. After 10 days they were transferred to MS medium with 1 mM Pi (KH2PO4) and 200 µM sodium arsenate (V) for 3 weeks. Growth occurred in continuous white light (80 µmol m−<sup>2</sup> s −1 ).

#### Co-cultivation Experiments

Co-cultivation of A. thaliana (WT, wrky6, as well as pht1 single and double knock-out lines) with P. indica was performed under in vitro culture conditions on a nylon membrane placed on top of solidified PNM media (Johnson et al., 2011). For Pi stress treatment PNM media with two different Pi concentrations [2.5 mM (NP, control) and 0.25 mM (LP, Pi stress)] were used. For expression profiling, square Petri dishes were divided into two equal parts. On one plate two P. indica disks on Aspergillus medium, one in each part, and on another plate two disks without the fungus, were placed on each part and kept for 7 days. The disks were used for mock treatment. After 48 h of cold treatment and 10 days of growth as described above on MS medium, seedlings of equal sizes were used for the co-cultivation assays or mock treatment, using the pre-prepared plates. For each Pi concentration, 4 treatments were compared: WT, WT + P. indica, wrky6 and wrky6 + P. indica. Seedlings were maintained under two different Pi concentrations with and without P. indica as mentioned above for 3 days at 22◦C and 70–80% humidity in a 16-h light/8-h dark cycle. Roots were harvested and frozen in liquid nitrogen for total RNA extraction. They were used for gene expression analyses.

For analysis of the pht1 mutants, the same treatment was performed except that the seedlings were maintained in the two different Pi concentrations in normal Petri dishes for 14 days, before harvest for further analysis.

#### Microarray Analyses

Total RNA from roots of colonized/uncolonized WT and wrky6 mutants from three independent biological experiments grown under NP and LP conditions were harvested 3 days after transfer to the fresh plates. RNA from roots of mock-treated WT and wrky6 mutants (agar plaques instead of P. indica plaques) were used as control. The 3-day time point was chosen because in preliminary experiments, we observed a strong regulation of a selected number of genes. For each treatment, identical amounts of RNA from three independent biological replicates were labeled and hybridized according to Agilent's One-Color Microarray-Based Gene Expression Analysis (OAK Lab GmBH, Hennigdorf, Germany). Quality of RNA samples were checked by photometrical measurements with the Nanodrop 2000 spectrophotometer (Thermo Scientific) and then analyzed on agarose gels (2%) as well as by using the 2100 Bioanalyser (Agilent Technologies, CA, United States) for determining the RNA integrity and the exclusion of potential contaminants. After verifying the quality of RNA, the Low Input Quick Amp Labeling Kit (Agilent Technologies) was used for generation of fluorescent complementary RNA (cRNA). Default cRNAs were amplified by using oligo-dT primers labeled with cyanine 3-CTP (Cye-3) according to the manufacturer's protocol. Cye-3-labeled probes were hybridized to 8 ×60 k customdesigned Agilent microarray chips. For hybridization, the Agilent Gene Expression Hybridization Kit (Agilent Technologies) was used. The hybridized slides were washed and scanned using the SureScan Microarray Scanner (Agilent Technologies) at a resolution of 3 micron generating a 20 bit TIFF file, respectively.

### Microarray Data Analysis

Data extractions from Images were performed using the Agilent's Feature Extraction software version 11. Feature extracted data were analyzed using the DirectArray Version 2.1 software from Agilent. Normalization of the data was performed with DirectArray using the ranked median quantiles according to Bolstad et al. (2003). To identify significantly differentially expressed genes log2-fold changes are calculated and Student's t-test was performed. In summary, raw data were normalized by rank median quantiles, intensity values from replicate probes were averaged, log2-ratios between the treatments were calculated and Student's t-statistics applied to test for significance. Genes with log2-fold change <–1 or >1 and p-value < 0.05 were considered to be significantly different. All data show expression levels of genes regulated by P. indica relative to the control levels without P. indica. Differentially expressed genes were then assigned using the A. thaliana Gene Ontology software (TAIR's GO annotations) (Berardini et al., 2004) and transcript abundance were classified based on their functional categories and pathways using the MapMan<sup>3</sup> software.

The microarray data have been submitted to NCBI (GEO) under the accession number GSE63500.

#### Real Time PCR Analyses

RNA was isolated from root tissues of WT and mutant seedlings at the time points indicated in the Sections "Result" and "Figure Legends" using the RNeasy Plant Mini Kit (Qiagen), and reversetranscribed for quantitative real-time PCR (qRT- PCR) analyses, using an iCycler iQ real-time PCR detection system and iCycler software (version 2.2; Bio-Rad). cDNA was synthesized using the Omniscript cDNA synthesis kit (Qiagen, Hilden, Germany) with 1 µg of RNA. For the amplification of the reverse-transcription PCR products, iQ SYBR Green Supermix (Bio-Rad) was used according to the manufacturer's protocol in a final volume of 20 µl. The iCycler was programmed to 95◦C for 3 min; 40 x (95◦C 30 sec, 57◦C 15 s, 72◦C 30 sec), 72◦C for 10 min, followed by a melting curve program from 55◦C to 95◦C in increasing steps of 0.5◦C. All reactions were performed from three biological and three technical replicates. The mRNA levels for each cDNA probe were normalized with respect to the plant ACTIN2 mRNA level. Fold-induction values of target genes were calculated with the 11CP equation of Pfaffl (2001) and related to the mRNA level of target genes as indicated in the Result section. Primer pairs used in this study are given in Supplementary Table S2. They were designed using the CLC Main Workbench program<sup>4</sup> .

<sup>3</sup>http://mapman.gabipd.org/web/guest/mapman

<sup>4</sup>http://www.clcbio.com/products/clc-main-workbench

#### Pi Content Analysis

fmicb-08-01262 July 10, 2017 Time: 18:13 # 4

For Pi content analyses, the samples were dried in an oven at 105◦C overnight. The samples were mixed with 2 ml of 65% HNO<sup>3</sup> and kept for one hour at 160◦C. The final volume was adjusted to 10 ml and the pH to 3.0–4.0. Finally, samples were mixed with ascorbic acid reagent and ammonium molybdate reagent (DIN 38405) and the Pi content was analyzed by the phosphomolybdenum blue reaction using the UV-160A spectrophotometer. Total Pi concentration was determined for the complete seedlings and expressed in nmol/g dry weight. Experiments were repeated three times with different biological replicas.

#### Chlorophyll Fluorescence Measurements

Plant performance was measured for WT and the pht1;1 pht1;4 double knock-out line using chlorophyll fluorescence measurements. After germination and 10 days on MS medium (cf. above), the seedlings were transferred to fresh PNM plates with LP or NP concentrations under high light intensity (300 µmol m−<sup>2</sup> s −1 ) for 1 week. This high light intensity confers stress to the seedlings. The efficiency of the photosynthetic electron transfer describing the fitness of the plants was measured as Fv/Fm described by Maxwell and Johnson (2000) after dark adaptation of the seedlings for 20 min. The fluorescence parameters were measured with a FluorCam 700MF instrument and analyzed with the Flucam 5.0 software. The data are averages for 30 seedlings and three independent biological experiments.

### RESULTS

#### P. indica Regulates Different Genes in WT and wrky6 Roots Under LP and NP Conditions

The root architecture of WT and wrky6 seedlings differs substantially and the differences become stronger under Pi limitation conditions (Chen et al., 2009; Bakshi et al., 2015). This is reflected by different expression profiles in the roots. Here we analyze how the expression profiles of the roots of WT and wrky6 seedlings grown on either NP or LP respond to P. indica colonization.

**Figure 1** shows that P. indica affects the expression of ∼ 3000 genes (regulation > 2-fold) in WT and wrky6 roots under NP or LP conditions. The vast majority of the responsive genes are specific for a given genotype and Pi condition, since the number of genes which are equally regulated in any comparison of the four conditions is less than 25% (**Figures 1–3**). Interestingly, only 13 genes are common in the four datasets. (for details on individual genes, cf. accession number GSE63500 at NCBI, GEO). This shows an enormous flexibility of the roots to respond to the fungus, and the response is strongly dependent on the genotype (WT vs. wrky6 mutant) and the Pi availability (LP vs. NP) (**Figure 2**).

Functional categorization of the identified gene products using the A. thaliana Gene Ontology program (TAIR's GO annotations; Berardini et al., 2004) demonstrates that the overall number of genes belonging to one of the four categories codes for proteins with similar functions. The strongest differences among the four categories were found for genes involved in DNA/RNA metabolism and extracellular functions, or for genes which code for mitochondrial, plastid and plasma membrane proteins. Less than 20% of these genes are common in all four categories (**Figure 3**, and GSE63500 at NCBI). Among them are hydrolases and transferases, transcription factors and numerous transporters (**Figure 3**). In each of the four datasets, different genes for proteins involved in the perception of internal and external signals, abiotic and biotic stress responses, primary protein metabolism, transcription, developmental processes and cell organization respond to P. indica (**Figure 3**). This is further supported by the PageMan analysis for genes differentially responding to P. indica in WT and wrky6 roots under NP and LP conditions (Supplementary Figures S1–S4). For example, many DNA-related genes are down-regulated in P. indica-colonized WT roots under NP conditions and P. indica-colonized wrky6 roots exposed to LP, but are up-regulated in colonized WT roots exposed to LP. Many genes involved in diverse signaling processes are down-regulated by P. indica in WT under Pi limitation, while signaling-related genes in WT roots grown under NP are up-regulated. Stress-related genes are downregulated by P. indica in WT roots under Pi limitation, but up-regulated under the other three conditions (Supplementary Figures S1–S4). Categorization of the gene products according to enzyme families also demonstrates enormous differences under the four conditions, and often, the genes for one enzyme family are up-regulated under one condition and down-regulated or not regulated under other conditions (Supplementary Figure S1). Big differences can be observed for the large cytochrome P<sup>450</sup> enzyme family, but also for peroxidases, phosphatases and glutathione-Stransferases. The identification of genes for GDSL lipases (Dong et al., 2016) suggests that P. indica also affects lipid metabolism. Finally, P. indica targets different members of the glucosidase gene family under NP and LP conditions and this is particularly striking for the wrky6 mutant.

It is also obvious from Supplementary Figures S1–S4 that interfering with the Pi metabolism by either inactivating WRKY6 or growth under Pi limitation conditions has severe consequences on many genes involved in transport processes (Supplementary Figure S2), cellular responses (Supplementary Figure S3), and regulatory functions (Supplementary Figure S4). Besides the expected effects on Pi transporters, genes for nitrate, ammonium and sulfate transporters are differentially regulated under the four conditions. We also observe big differences on genes for sugar, potassium and amino acid transporters, P- and V-ATPases and lipid transfer proteins. Genes related to calcium transport processes are up-regulated under all conditions, although to different extents (Supplementary Figure S2). Overall, the data indicate that inactivation of WRKY6 activates biotic stress response genes under NP conditions (Supplementary Figures S3, S4). Furthermore, growth under LP conditions results in the down-regulation of many biotic-stress-related genes which are up-regulated under NP conditions, and this is observed for both WT and wrky6 roots. The latter observation holds also true for abiotic stress-related genes (Supplementary Figure S3)

and those with diverse functions (categorized as "miscellaneous") in WT roots. Peroxiredoxin genes are preferentially downregulated by P. indica in LP, and osmotic stress related genes (categorized as "drought/salt", Supplementary Figure S3) are downregulated in the wrky6 mutant under NP relative to the WT control. Finally, many genes related to the cell cycle are downregulated by P. indica in LP-grown WT seedlings when compared to those grown under NP conditions. As expected, the changes in the gene expression patterns for cellular responses (Supplementary Figure S3) are reflected by corresponding changes in regulatory pathways (Supplementary Figure S4). It is particularly striking that members of the receptor kinase gene family are downregulated by P. indica in LP-grown WT roots relative to roots grown under NP conditions. Taken together, the fungus activates quite different signaling pathways, as well as metabolic and developmental programs under the four conditions tested.

### PHT1 and Pi Regulator Genes in the Arabidopsis/P. indica Interaction

Pi uptake from the soil and distribution of Pi within the Arabidopsis plant is mediated by nine PHT1 family members (Ayadi et al., 2015). While some PHT1 genes are regulated in response to beneficial fungi in mycorrhizal plants, it is believed that the non-mycorrhizal plant Arabidopsis does not contain fungus-inducible PHT1 genes (cf. Tamura et al., 2012; Sisaphaithong et al., 2012). This is consistent with our observations that most of the PHT1 transporter genes do not respond to P. indica (>2-fold) under NP or LP conditions, or are only mildly regulated (PHT1;5; PHT1;6 and PHT1;8; cf. Discussion) (**Table 1**). Furthermore, WRKY42 modulates Pi homeostasis through regulating Pi translocation and acquisition (Su et al., 2015) and the transcription factor also regulates PHO1 expression (Su et al., 2015). WRKY45 activates PHT1;1 expression under Pi starvation (Wang et al., 2014). Both WRKY genes do not respond to P. indica (**Table 1**). Recently, the importance of posttranscriptional processes for PHT1 proteins has been shown (Cardona-López et al., 2015), and NITROGEN LIMITATION ADAPTATION (NLA) targets Pht1;4 for degradation during the regulation of Pi homeostasis (Lin et al., 2013; Park et al., 2014). Furthermore, ESCRT-III-associated protein ALIX mediates high-affinity Pi transporter trafficking to maintain Pi homeostasis in Arabidopsis (Cardona-López et al., 2015). **Table 1** demonstrates that only PHO1 responds to P. indica ( > 2-fold) in both WT and wrky6 roots under LP, but not NP conditions. This demonstrates that PHO1 is not only regulated under Pi limitation conditions, but also a mild target of signals from P. indica (cf. Discussion).

### P. indica Does Not Compensate for Pi Limitation in PHT1 Mutants

Although PHT1;1, PHT1;2 and PHT1;4 do not respond (>2 fold) to P. indica under all tested conditions, they show the highest expression level in roots (Shin et al., 2004; Ayadi et al.,

P. indica in WT or wrky6 roots grown on either NP or LP. Scale depicts level of expression with blue being high and red being low. Only significantly regulated (p-value <0.05) categories are shown. Numbers on the right refer to the categories defined by the PageMan program, which are presented in Supplementary Table S1. For specific genes, see data submitted to NCBI. For more detailed information on the program, cf. the MapMan software at http://mapman.gabipd.org/web/guest/ mapman.

2015). Therefore, we generate homozygote knock-out lines for these three transporters, and confirmed their homozygosity using standard techniques, gene-specific primer pairs as well as a primer combination with the T-DNA insertion (Supplementary Table S2). Furthermore, pht1;1 and pht1;4 were crossed to generate a double knock-out line. In addition to the molecular analyses (**Figure 4**), confirmation of homozygosity for all lines can easily be tested by growth of the seedlings in the presence of arsenate (V), because this heavy metal is transported into the roots via PHT1 transporters (DiTusa et al., 2016, and references therein). After 3 weeks on MS medium with 200 µM arsenate, the fresh weight of WT seedlings was reduced by 86% compared to seedlings grown without arsenate (**Table 2**). The fresh weights of the single mutants pht1;1 and pht1;2 were similarly reduced (79%). This demonstrates that inactivation of these Pi transporter genes has little effect on arsenate (and probably also Pi) uptake. The performance of pht1;4 is significantly better, suggesting that PHT1;4 is more important for arsenate (and probably Pi) uptake than PHT1;1 and PHT1;2. The fresh weight of the double knock-out line is reduced by 43% in the presence of arsenate, compared to growth without arsenate. This suggests that simultaneous inactivation of the two Pi transporters is more effective in restricting arsenate (and Pi) uptake than the effects observed for the single knockout lines. The degree of resistance can be taken as indication for the contribution of the transporter to the Pi/arsenate uptake. Although different growth conditions and arsenate concentrations were used, our results resemble those described by Shin et al. (2004).

WT seedlings and all mutants and were grown on NP and LP medium in the absence or presence of P. indica for 2 weeks. Shoots and roots were harvested and weighed (**Figures 5A,B**). The fresh weights of the shoots and roots of the pht1;1 and pht1;2 seedlings did not differ significantly from those of the WT, the weights of the shoots and roots of the pht1;4 seedlings were slightly reduced and those of the double knock-out line ∼ half of the weights of the WT. Only little differences can be observed for seedlings grown on NP or LP medium. This might be due to the pre-cultivation of the seedlings on full MS medium: the Pi that is taken up during this period might be sufficient for the seedling's growth during the next 2 weeks. In all cases, we observed an increase in the shoot and root fresh weights for seedlings grown in the presence of P. indica. However, the relative increases (% increase) among the different mutant and WT seedlings are comparable. Thus, P. indica does not compensate for the absence of specific PHT1s by transferring more Pi from the soil to the roots. This is particularly striking for the double knock-out mutant. Although the weight of the mutant is approximately half of the weight of a WT seedling, the % increases in their fresh weights induced by P. indica are approximately the same (∼ 20%). This indicates that P. indica promotes growth of all seedlings grown under LP and NP conditions, and this is independent of the presence or absence of the tested PHT1 transporters.

To test whether P. indica participates in Pi allocation, we used the same growth conditions and measured the total Pi content in the seedlings (**Figure 5C**). Similar to previous reports (Shahollari et al., 2005), we observed that all P. indica-exposed seedlings contained more Pi than the seedlings not exposed to the fungus. Closer inspection of the data shows that the % increase in the Pi content is not different between WT and mutant seedlings. This again is consistent with the idea that P. indica does not compensate for the lower Pi uptake of the Pi mutants. The higher Pi content in colonized seedlings might be caused by better access to the nutrient in the presence of the hyphal mycelium, and/or because Pi from the fungal hyphae is delivered to the host.

TABLE 1 | Regulation of the 9 PHOSPHATE TRANSPORTER1 (PHT1) genes and genes for regulatory proteins involved in PHT1 gene and PHT1 protein regulation by Piriformospora indica in WT and wrky6 roots under NP and LP conditions.


The numbers represent log<sup>2</sup> values of the mRNA levels in colonized roots vs. uncolonized roots, the numbers in brackets show fold-induction (colonized/uncolonized roots). Conditions with >2-fold regulation are in bold. The data from the microarray analysis (NCBI under the accession number GSE63500) are compared with qRT-PCR data shown in square brackets. Only those data are shown where a more than twofold regulation was observed. The primer pairs are given in Supplementary Table S1. Microarray and qRT-PCR data are based on three independent experiments, errors are SEs.



The fresh weights of 3-week-old Arabidopsis WT, pht1;1, pht1;2, pht1;4 and pht1;1 pht1;4 single and double mutants are given. % fresh weight is calculated relative to the fresh weight of the seedlings grown without arsenate. Based on three independent experiments with 30 seedlings each, errors are SEs.

Since the double knock-out line showed the strongest phenotype, we grew the seedlings under LP conditions in the absence or presence of P. indica and tested whether the mutant respond to Pi limitation by activating Pi starvation-response and transporter genes and whether P. indica has an effect on the regulation of these genes in the mutant. **Table 3** shows that the expression of several unrelated Pi starvation-response and transporter genes are up-regulated in the double mutant under LP and that their expression is not significantly affected by P. indica. This includes PHOSPHATE STARVATION RESPONSE1 (PSR1) which codes for a MYB transcription factor involved in the Pi starvation response (Rubio et al., 2001), PHOSPHATE STARVATION RESPONSE1 (PHR1), a master regulator of Pi homeostasis (e.g., Linn et al., 2017), which balances between nutrition and immunity in plants (Motte and Beeckman, 2017), the gene for the zinc-finger transcription factor ZAT6, which controls Pi homeostasis in roots (Devaiah et al., 2007a), for MYB62 involved in Pi homeostasis and hormone actions (Devaiah et al., 2009) and for WRKY45 which activates PHT1;1 expression in response to Pi starvation (Wang et al., 2014). Thus, the fungus does not interfere with the regulation of these genes in the LP-grown pht1;1 pht1;4 mutant.

The double mutant is impaired in long-term growth experiments on natural soil (**Figure 6**). After germination and initial growth in Petri dishes for 10 days, the seedlings were transferred to soil for additional 3 weeks. We observed an approximately 50% reduction in the fresh weight compared to WT control plants (**Figure 6** lower panel). This suggests the combination of the two transporters is important for growth on natural soil.

To test whether P. indica promotes the performance of the double knock-out line, we transferred the seedlings to PNM medium and high light intensity (300 µmol m−<sup>2</sup> s −1 ). The chlorophyll fluorescence parameters Fv/Fm showing the efficiency of the photosynthetic electron transport were measured daily under LP and NP conditions over a period of 7 days (**Figure 7**). Under NP conditions, no difference between colonized and uncolonized WT and double knock-out seedlings can be detected (data not shown). However, under LP conditions, a decline in the Fv/Fm values can be detected for uncolonized

pht1;1 pht1;4 seedlings, and this stress response is partially compensated in the presence of the fungus. Thus, P. indica partially promotes the performance of the double knock-out line under Pi limitation conditions.

is inserted into the respective gene close to the RP. The PCR samples were run on a 1% agarose gel next to a size ladder and the predicted sizes of the

#### DISCUSSION

fragments were confirmed.

#### The Genotype and Pi Availability Have a Strong Influence on P. indica-Targeted Genes

Previous interaction studies of root-colonizing microbes with plants have shown that the mutualistic interaction and the benefits for the two partners is strongly dependent on environmental conditions (e.g., Pánková et al., 2011; Moeller

et al., 2014) and differ between plant species when they are colonized by the same microbe (Lee et al., 2011; Lahrmann et al., 2013). Here, we compare a WT Arabidopsis line with a mutant impaired in the WRKY6 transcription factor which is a central regulator of Pi metabolism in Arabidopsis (Hamburger et al., 2002; Liu et al., 2012). By growing these seedlings under NP and LP conditions, we show that P. indica targets quite different genes in both genotypes and under the two Pi conditions. Genotype-dependent alterations in gene expression profiles in response to various biotic and abiotic stresses have been described



WT and double knock-out seedlings were grown in LP for 2 weeks in the presence or absence of P. indica before RNA extraction for qRT-PCR. Primer pairs for the analyzed genes are given in Supplementary Table S1. The mRNA level of a given gene from WT seedlings grown under LP without P. indica was set as 1.0 and the other values are expressed relative to it. For additional information about the genes not discussed in this paper: PHR1 (Rubio et al., 2001), ZAT6 (Devaiah et al., 2007a), and MYB62 (Devaiah et al., 2009). Based on three independent RNA extraction and two technical repetitions per RNA. Errors are SEs.

for many ecotypes, varieties, lines and mutants impaired in regulatory loci. Genotype-specific expression of miRNAs might explain distinct cold (Zhang et al., 2014) or salt sensitivities (Ding et al., 2009; Beritognolo et al., 2011; Yin et al., 2012). Genotypespecific defense gene expression programs were reported for two cultivars of Glycine max (Klink et al., 2011). In Sorghum bicolor a genotype-specific expression atlas for vegetative tissues was published (Shakoor et al., 2014). Ashraf et al. (2009) described comparative analyses of genotype-dependent expressed sequence tags and stress-responsive transcriptome for chickpea wilt. Here, we show that a mutation in the WRKY6 gene which strongly affects plant performance under Pi limitation conditions, results in a severe reprogramming of the root transcriptome after P. indica infection. The altered gene expression profile affects many biochemical, signaling and metabolic processes, which are not related to the Pi metabolism (**Figures 2**, **3** and Supplementary Figures S1–S4). To our knowledge this is the first report on a comprehensive analysis of gene reprogramming in response to P. indica in roots of two genotypes, the WT and the wrky6 mutant. The huge difference in the expression profiles clearly indicates that the fungus targets different genes and consequently induces completely different physiological responses in the roots of WT and wrky6 seedlings under the two different Pi conditions. This might have important implication for the application of the fungus in agriculture, since its interaction with different cultivars of a crop plant might differ and this again is dependent on environmental and soil conditions.

Analysis of genes involved in Pi starvation have been analyzed in many plant species (e.g., Nilsson et al., 2010). Interesting in the study performed here the number of P. indica stimulated genes in roots of both WT and wrky6 seedlings is quite different in LP compared to NP conditions Not only Pi-related signaling and metabolic pathways, but also developmental programs, defense strategy and transport processes are affected in WT and wrky6 roots. Under our standardized conditions, ∼70–80% of the plant genes are differentially expressed in response to P. indica by inactivation of WRKY6 and/or change in the amount of available Pi (**Figure 2**). We assume that similar phenomena could be observed for agriculturally important crops interacting with beneficial root-colonizing fungi. Therefore, the interplay between root-colonizing microbes with different cultivars, the soil quality and fertilizations may have a strong influence on the root metabolism.

### PHT1 Genes and P. indica

The members of the AtPHT1 protein family share a high degree of similarity with overlapping expression patterns (Ayadi et al., 2015). PHT1;1 and PHT1;4 form homomeric and heteromeric complexes (Fontenot et al., 2015) and both transporters play a major role in Pi acquisition from low- and high-Pi environments (Shin et al., 2004). Their genes show the highest expression of all PHT1 genes in Arabidopsis roots (Shin et al., 2004), but PHT1;1 also plays an important role in Pi translocation from roots to leaves in high Pi conditions (Ayadi et al., 2015). PHT1;2 cooperates with PHT1;1 and PHT1;4 in Pi uptake from the soil (Ayadi et al., 2015) and PHT1;6 is mainly expressed in flower tissue (Karandashov et al., 2004). Members of the PHT1 gene family are up-regulated by mycorrhizal fungi in mycorrhizaforming plant species (e.g., Ceasar et al., 2014; Walder et al., 2015; Kariman et al., 2016; for some recent publications), however the Arabidopsis homologs (pht1;5, pht1;6, and pht1;8) do not or barely respond to P. indica. The slight response of pht1;5 to P. indica might be related to the role of PHT1;5 as mobilizer of Pi between source and sink organs (Nagarajan et al., 2011). Under low Pi conditions, pht1;5 mutants show reduced Pi allocation to the shoots and elevated transcript levels for several Pi starvationresponse genes (Nagarajan et al., 2011). P. indica might interfere with the translocation and regulation of Pi under Pi limitation. Karandashov et al. (2004) identified six cis-regulatory elements which are present in different combinations and numbers in mycorrhiza-inducible PHT1 genes from different plant species. None of these elements were found in the P. indica-responsive AtPHT1;5 and AtPHT1;6 promoters, but four of them are present in the root-specific AtPHT1;8 promoter which shows a low response to P. indica (**Table 1**). Whether these elements are involved in P. indica -mediated PHT1;8 expression, is unclear.

Co-cultivation of pht1 mutants with P. indica demonstrates that the fungus does not compensate for Pi shortage by stimulating the expression of specific PHT1 family members (**Table 1** and **Figure 5**) or promoting Pi uptake via other ways. The % growth promotion and Pi uptake in the presence of the fungus is comparable for WT and mutant seedlings. Even for the double knock-out line, the positive effects of the fungus are comparable to the WT (**Figure 5**). Together with the observation that the PHT1 genes are not or barely regulated by P. indica, it is conceivable that the fungus targets additional genetic programs, which are not directly related to Pi availability.

### PHO1 Respond to P. indica

Besides PHT1 genes (Wasaki et al., 2003; Wu et al., 2003; Misson et al., 2005), the mRNA levels for several transcription factors involved in controlling Pi homeostasis, such as AtPHR1 (Rubio et al., 2001), the rice Pi starvation-induced transcription factor1 (Yi et al., 2005), AtWRKY75 (Devaiah et al., 2007a), the Arabidopsis zinc finger family member 6 (AtZAT6; Devaiah et al., 2007b), AtMYB62 (Devaiah et al., 2009), and AtWRKY6

FIGURE 6 | Phenotype of WT and pht1;1 pht1;4 seedlings/plants. Top: After 10 days, WT and pht1;1 pht1;4 seedlings were transferred to PNM medium with NP for 3 weeks. Bottom: Arabidopsis WT (upper 4 plants) and pht1;1 pht1;4 double knock-out lines (lower 4 plants) were grown in Petri dishes for 10 days before transfer to soil for additional 4 weeks. Growth occurred in a greenhouse.

(Chen et al., 2009) are upregulated under Pi limitation in different plant species (Wang et al., 2014). Also, PHO1 has been shown to be an important regulator in controlling Pi homeostasis (Chen et al., 2007). We observed a mild, but significant upregulation of the PHO1 mRNA level by P. indica in LP-exposed Arabidopsis roots (**Table 1**). PHO1 plays an important role in Pi translocation from roots to shoots (Poirier et al., 1991; Wang et al., 2014), is located primarily in root stellar cells and controls Pi xylem loading from root stellar cells (Hamburger et al., 2002). Thus, the protein is mainly involved in long distance Pi transport in Arabidopsis (Su et al., 2015). The pho1 mutant is impaired in loading Pi to the xylem vessel in the roots (Poirier et al., 1991). Taken together, our data suggest that P. indica interferes primarily with the Pi distribution and metabolism in Arabidopsis under Pi limitations rather than by promoting Pi uptake from the soil.

### AUTHOR CONTRIBUTIONS

MB: performed most of the experiments, except those described for others authors. IS: generated the ko lines. DM:

### REFERENCES


analysed microarray data, performed low Pi experiments with Arabidopsis. JT: performed Pi experiments with the double mutant. AJ: supervision of Ph.D. students together with RO (common India-Germany DAAD project). AV: supervision of Ph.D. students together with RO (common India-Germany DAAD project). K-WY: supervision of microarray analyses. RO: supervision of Ph.D. students together with RO (common India-Germany DAAD project)

#### ACKNOWLEDGMENT

This work was supported by CRC 1127 to RO and collaborative DAAD grants to RO, Germany and AV, India and K-WY, Taiwan.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2017.01262/full#supplementary-material




genes of soybean. Biosci. Biotechnol. Biochem. 76, 309–313. doi: 10.1271/bbb. 110684


Zhang, Y., Zhu, X., Chen, X., Song, C., Zou, Z., Wang, Y., et al. (2014). Identification and characterization of cold-responsive microRNAs in tea plant (Camellia sinensis) and their targets using high-throughput sequencing and degradome analysis. BMC Plant Biol. 14:271. doi: 10.1186/s12870-014-0271-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 © 2017 Bakshi, Sherameti, Meichsner, Thürich, Varma, Johri, Yeh and Oelmüller. 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) or licensor 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.

# Genetic Diversity Studies Based on Morphological Variability, Pathogenicity and Molecular Phylogeny of the *Sclerotinia sclerotiorum* Population From Indian Mustard (*Brassica juncea*)

Pankaj Sharma<sup>1</sup> , Amos Samkumar <sup>2</sup> , Mahesh Rao<sup>2</sup> , Vijay V. Singh<sup>1</sup> , Lakshman Prasad<sup>3</sup> , Dwijesh C. Mishra<sup>4</sup> , Ramcharan Bhattacharya<sup>2</sup> and Navin C. Gupta<sup>2</sup> \*

<sup>1</sup> Sclerotinia Lab, ICAR, Directorate of Rapeseed and Mustard Research, Bharatpur, India, <sup>2</sup> Brassica Lab, ICAR, National Research Centre on Plant Biotechnology, New Delhi, India, <sup>3</sup> ICAR, Indian Agricultural Research Institute, New Delhi, India, 4 ICAR, Indian Agricultural Statistics Research Institute, New Delhi, India

#### *Edited by:*

Katarzyna Turnau, Jagiellonian University, Poland

#### *Reviewed by:*

Zbigniew Miszalski, The Franciszek Górski Institute of Plant Physiology (PAS), Poland Sylwia Rózalska, ˙ University of Łódz, Poland ´

> *\*Correspondence:* Navin C. Gupta navinbtc@gmail.com

#### *Specialty section:*

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

*Received:* 14 September 2017 *Accepted:* 14 May 2018 *Published:* 05 June 2018

#### *Citation:*

Sharma P, Samkumar A, Rao M, Singh VV, Prasad L, Mishra DC, Bhattacharya R and Gupta NC (2018) Genetic Diversity Studies Based on Morphological Variability, Pathogenicity and Molecular Phylogeny of the Sclerotinia sclerotiorum Population From Indian Mustard (Brassica juncea). Front. Microbiol. 9:1169. doi: 10.3389/fmicb.2018.01169 White mold or stem rot disease are ubiquitously distributed throughout the world and the causal organism of this disease Sclerotinia sclerotiorum (Lib.) de Bary, is known to infect over 400 plant species. Sclerotinia stem rot is one of the most devastating fungal diseases and poses a serious threat to the worldwide cultivation of oilseed Brassica including India. S. sclerotiorum pathogen usually infects the stem but in severe cases leaves and pods also affected at different developmental stages that deteriorate not only the oil quality but also causing the seed and oil yield losses up to 90% depending on the severity of the disease infestation. This study investigated the morphological and molecular characterization of pathogenic S. sclerotiorum (Lib) de Bary geographical isolates from oilseed Brassica including Brassica juncea (Indian mustard). The aim of this study was to compare isolates of S. sclerotiorum originated from different agro-climatic conditions and to analyse similarity or differences between them as well as to examine the virulence of this pathogen specifically in Brassica for the first time. The collection of S. sclerotiorum isolates from symptomatic Brassica plants was done and analyzed for morphological features, and molecular characterization. The virulence evaluation test of 65 isolates on four Brassica cultivars has shown 5 of them were highly virulent, 46 were virulent and 14 were moderately virulent. Phylogenetic analysis encompassing all the morphological features, SSR polymorphism, and ITS sequencing has shown the existence of high genetic diversity among the isolates that categorized all the isolates in three evolutionary lineages in the derived dendrogram. Further, genetic variability analysis based on sequences variation in ITS region of all the isolates has shown the existence of either insertions or deletions of the nucleotides in the ITS region has led to the interspecies variability and observed the variation were in a clade-specific manner. Together this analysis observed the existence of higher heterogeneity and genetic variability in S. sclerotiorum isolates collection and indicates the presence of clonal and sexual progenies of the pathogen in the mustard growing regions of India surveyed in this study. With a higher level of genetic variability and diversity among the S. sclerotiorum population needs robust screening approaches to identify the donor parent and utilize them in resistance breeding program for effectively counter the menace of stem rot disease in Brassica.

Keywords: morphological, molecular, phylogeny, *Sclerotinia sclerotiorum*, stem rot, *Brassica*, diversity

### INTRODUCTION

Globally India continues to be at a 3rd position after Canada and China in acreage (19.3%) and after China and Canada in production (11.1%) of rapeseed-mustard. In India, among nine edible oilseed crops, the share of rapeseed-mustard is about one-fourth of total area and one-third of total oil production in the country. During 2015–2016, production (6.82 mt) and productivity (1184 kg/ha) was achieved (Anonymous, 2016). Rapeseed-mustard is the major source of income especially for the marginal and small farmers in rainfed areas which are about 25% of the total cultivated area. In spite of its increase in demand for the year the production of oilseed Brassica remains to stagnate over the year and most of the demands are being met through import from outside the India. The main reason behind productivity stagnation in Indian Brassica is its susceptibility and damages caused to the crop by various insect pests and disease infestation in addition to the other yield-limiting factors. Out of thirty diseases known to infest the Brassica crops in India, stem rot has been found one of the most devastating diseases that heavily damages the crops during the flowering stage of development. The stem rot disease which is caused by fungal pathogens, Sclerotinia sclerotiorum (Lib) de Bary, ubiquitously found throughout the world is a polyphagous, soil-borne plant pathogen that infects more than 400 plant species of diverse phylogenetic origin (Boland and Hall, 1994; Saharan and Mehta, 2008; Sharma et al., 2015). In India, during the eighties and nineties, the stem rot (SR) disease in rapeseed-mustard was of a minor importance, because of its seldom appearance over the ground level of the isolated plants after mycelial infection. A widely adopted monocropping practices and cultivation of rapeseed-mustard under irrigated condition has significantly increased the sclerotial population in the soil that has made SR very serious disease of oilseed Brassica crops in states including Rajasthan, Haryana, Punjab, Uttar Pradesh, Bihar, Assam, West Bengal and Madhya Pradesh (Sharma et al., 2015). This fungus has been long considered as prototypical necrotrophs as it begins highly pathogenic phase by releasing oxalic acids and cellulolytic enzymes immediately upon host cuticle penetration followed by mycelial proliferation inside the host cell followed by a saprophytic phase that supports the sclerotia formation (Hegedus and Rimmer, 2005). However, the recent studies decipher the fact of evidence for the occurrence of a brief biotrophic phase just within the apoplastic space next after the establishment of the host-pathogen connection and hence based on these it is more appropriately classified as a hemibiotroph (Kabbage et al., 2015). The information related to the genetic diversity of the pathogen and their effective virulence over the target crop is the foremost requirement for taking the breeding program for development of pathogen resistance in the release of the regionspecific cultivars. Various diversity analysis tools based on the molecular methods like microsatellite haplotype (Aldrich-Wolfe et al., 2015), SSR (simple sequence repeat or microsatellitebased marker; Meinhardt et al., 2002), AFLP (amplified fragment length polymorphism; Cubeta et al., 1997), and SRAP (sequencerelated amplified polymorphism technique; Li et al., 2009) have been used in analyzing the genetic diversity of the pathogen S. sclerotiorum from different host species. Very limited variability was observed in ITS (internal transcribed spacer) sequences in S. sclerotiorum isolates from various host species (Njambere et al., 2008) and thus a universal barcode markers were developed from the nearly conserved nature of the nuclear ribosomal internal transcribed spacer (ITS) region for imparting the individual identity to the fungus up to genus level (Schoch et al., 2012). Furthermore, MCGs is another diversity analysis method based on the mycelial compatibility grouping (MCG) has been used in establishing the kinship among S. sclerotiorum isolates from chickpea (Kull et al., 2004; Li et al., 2008). In addition to it, the diversity based on the morphological appearance of sclerotia, mycelial growth, and ascospores formation have also been reported in analyzing the genetic diversity of S. sclerotiorum isolates in previous studies (Li et al., 2008; Sharma et al., 2013). However, polymorphism and genetic diversity of the S. sclerotiorum isolates at the morphological and DNA sequence level has not been comprehensively studied so far especially for the isolates from Brassica species of India.

Being a polyphagous nature, S. sclerotiorum pathogen is usually infecting not only the majority of the economically important dicotyledonous species but also serve as the major pathogen for several monocotyledonous plant species (Boland and Hall, 1994). The yield loss estimated with the S. sclerotiorum infestation has exceeded hundred million dollars annually because of the lack of resistance cultivar of the crop species and also because of lack of the effective management practices (Tok et al., 2016). In general, the management of S. sclerotiorum borne disease is not much easy because of its widespread existence, irregular incidence, and the long-term survival by producing huge numbers of sclerotia in the soil. Although several control measures like chemical and cultural methods have been devised and adopted for countering the Sclerotinia stem rot menace (Rousseau et al., 2007) none of them were found fully effective in preventing either the process of disease infection or pathogenesis progression after infection. The extent of genetic diversity of the pathogen and their widespread distribution among host species across the growing regions play an important role in determining and devising the control strategies to efficiently control the diseases in the more effective way. Hence, the availability of the genetic variability information related to the target pathogen is the foremost requirement for designing the management means to counter the disease incidence more effectively. In plant-pathogen interactions, development of new pathogenic races, and the breakdown of host resistance are the limiting factors in resistance deployment against plant diseases. The pathogen's life history characteristics and evolutionary potential are major factors leading to the pathogen overcoming host resistance. Therefore, major efforts should be focused not only in understanding the genetic structure of the fungal populations but also to determine how populations will evolve in response to different control strategies (McDonald and Linde, 2002).

In recent past, India observed the frequent incidence of the S. sclerotiorum infestation on cereals and horticultural species that draws the wide attention of the researcher over this fungal pathogen. In pursuance of basic understanding about pathogenicity, diversity and distribution pattern of the S. sclerotiorum were extensively studied on isolates collected from various host species like chickpea (Mandal and Dubey, 2012), vegetable crops (Choudhary and Prasad, 2012), cumin (Prasad et al., 2017), carnation (Kumar et al., 2015), oilseed Brassica (Sharma et al., 2013), and their identity were established on the basis of morphological features and cultural conditions. However, major variation in the growth characteristics has been reported in the growing collection of S. sclerotiorum isolates even as they belong to the same Sclerotinia species. Indeed, as projected with diversity analysis S. sclerotiorum isolates has been reported of possessing the variation in morpho-physiological, biochemical properties, molecular features and pathogenicity in terms of virulence because of the presence of clonal and sexual progenies together even in the same crop and in same region (Atallah et al., 2004; Sexton and Howlett, 2004; Irani et al., 2011). For establishing the console features of the pathogen especially in Brassica growing regions of India, the present study was aimed at determining the genetic diversity within S. sclerotiorum population from various Brassica species based on morphological characteristics, genotyping with simple sequence repeats (SSR) markers and molecular phylogeny by ITS sequence analysis.

### MATERIALS AND METHODS

#### *Sclerotinia sclerotiorum* Isolates

Brassica growing areas in 10 states of India (Rajasthan, Haryana, Punjab, Delhi, U.P., Bihar, Uttarakhand, Himachal Pradesh, Jammu & Kashmir, and Jharkhand) were surveyed and stem rot disease infected plants were collected from 65 different locations (**Table 1**). The sclerotia obtained from the stem rot infested Indian mustard plant samples were first washed with sterile water than surface sterilized with 70% ethanol for 2 min and again washed two times with sterile water. The drained sclerotia over pre-sterilized filter papers were placed on nutrient media (Potato Dextrose Agar; PDA) plates supplemented with 50µg/ml tetracycline antibiotics to prevent the growth of bacterial contamination. The samples were wrapped in a brown envelope and kept for incubation at 20◦ ± 2 ◦C in dark for 4–5 days. After the development of the white fluffy mass of mycelial growth of S. sclerotiorum, the mycelial plaques were used to subculture the isolates on PDA slants and pure culture of them was stored at 4◦C for future use. Morphological identification of the isolates was based on cultural characteristics of the S. sclerotiorum and morphology of the mycelial mat and sclerotia formation traits.

### Morphological Characterization of *S. sclerotiorum* Isolates

Freshly grown 3–4 days old cultures of Sclerotinia isolates were used for analyzing the morphological features like the radial spread of mycelial growth (mm) at 72 and 96 h after inoculation. Whereas, the number of sclerotia produced from each inoculum per Petri plates (90 mm), length and diameter of the sclerotia (mm) and weight of individual sclerotia (mg) was measured 10–15 days after inoculation.

#### Pathogenicity of Isolates

An experiment to study pathogenicity and pathogenic variability among 65 geographical isolates of S. sclerotiorum were conducted during 2016–2017 post-rainy season at ICAR-Directorate of Rapeseed-Mustard Research, Bharatpur, India. Seven different oilseed Brassica species were selected for this study. The crop was destroyed after the experiment by cutting and collecting the debris followed by autoclaving before disposing of them.

#### Inoculation

The inoculum was mass multiplied in the laboratory on autoclaved sorghum grains in glass jars and incorporated into the soil prior to sowing. Further, the plants were inoculated at 60 days after sowing on the stem with the pathogen growing on agar blocks. Stem inoculation procedure was followed as described by Buchwaldt et al. (2003). A single 5 mm mycelial bit cut from S. sclerotiorum colony of 4–5 days old culture growing on potato dextrose agar was used to inoculate each plant. The mycelial bit along with cotton swab soaked in sterilized distilled water was placed on a small piece of parafilm (5–7 cm). Mycelial bit touching the stem at 15 cm height was then secured by wrapping the parafilm strip around the stem. Wet cotton swab maintained high humidity during the infection period (Sharma et al., 2013).

#### Disease

Disease incidence was assessed by recording the size of stem lesion length (cm) and disease severity (%) 15–21 days after inoculation. This has been shown to be an ideal time to demonstrate the host response to the pathogen (Li et al., 2007). The disease parameters recorded were statistically verified by data analysis using Duncan's test (P < 0.05) and analysis of variance (ANOVA).

#### DNA Extraction

Sclerotinia sclerotiorum isolates were grown on PDA medium at 20◦ ± 2 ◦C in BOD incubator for 5 days. Mycelial mat was harvested by scraping with a sterile spatula and ∼2 g of the mycelial mat was used for DNA extraction using modified CTAB method. The fine powder of the grounded mycelial mat in liquid nitrogen was transferred in preheated 10 ml DNA extraction TABLE 1 | Sclerotinia sclerotiorum (Lib) de Bary population used for studying the virulence and genetic diversity within isolates based on morphological features, SSR profiling, and ITS sequence analysis.


(Continued)

#### TABLE 1 | Continued


The severity of disease assessed on Brassica cultivars was measured in lesion length (cm). Disease index for the strains listed, accessed on length of the lesion developed on stem (in cm). Lesion size 0.0 cm: Non-virulent; 0.1–3.0 cm: less virulent; 3.1–10.0: Moderately virulent; 10.1–20.0 cm: virulent; >20.0 cm: highly virulent. NRCDR-02 stands for National Research Centre Disease Resistance-02, YS stands for yellow sarson

buffer (1% Cetyltrimethylammonium bromide; 250 mM NaCl; 100 mM Tris-HCl, pH 8.0; 100 mM EDTA, pH 8.0; along with 4 mM Spermidine) and incubated at 65◦C with intermittent mixing at regular interval for 60 min. Denatured proteins were removed by extracting once with an equal volume of Trissaturated phenol: chloroform: isoamyl alcohol (25:24:1, v/v/v), followed by repeated extractions with an equal volume of Trissaturated chloroform: isoamyl alcohol (24:1, v/v). The aqueous phase obtained after centrifugation was added with 0.6 volume of chilled isopropanol and stored overnight at −20◦C for DNA precipitation. DNA was pelleted, dried and resuspended in 500 µl of Tris–EDTA buffer [10 mM Tris-HCl, pH 7.5; 1 mM EDTA (ethylenediaminetetraacetic acid)]. The purity and concentration of the DNA were determined through NanoDrop (Thermo Scientific, USA) and stored in aliquots at −20◦C.

#### Species-Specific PCR Assay

Sclerotinia-specific primers as described by Freeman et al. (2002) (**Table 2**) were used to carry out PCR amplification in all the isolates. The PCR reaction of 15 µL contained 1.5 µL of 10x PCR buffer with 15 mM MgCl2, 50µM dNTPs, 10µM forward and reverse primers, 1 U Taq DNA polymerase (Bangalore Genei, India) and 50 ng of template DNA was performed in a thermal cycler (Biometra, ILS, USA) and the program made with initial denaturation at 95◦C for 4 min followed by 35 cycles of denaturation at 94◦C for 45 s, annealing at 57◦C for 45 s and TABLE 2 | Details of the primers used for identification of Sclerotinia species, rDNA conserved sequence, and ITS analysis.


extension at 72◦C for 1 min, and a final extension at 72◦C for 10 min. The PCR amplicons were resolved on 1% agarose gel along with 1 kb standard DNA ladder and visualized in a UV transilluminator. Gels were photographed using Alpha image (Alpha innotech, USA) Gel Doc system.

#### Genetic Diversity

#### Simple Sequence Repeat (SSR) Marker Analysis

The pathogenicity tested 65 S. sclerotiorum isolates were subjected to SSR fingerprinting. A total of 25 SSR primers (Sirjusingh and Kohn, 2001; **Table 3**) were synthesized from IDT, Agrigenome, Bengaluru, India and used for amplification of microsatellite loci. PCR reactions were performed using a BioRad C1000 thermocycler (BioRad, USA). The PCR reaction was set up in a 25 µL reaction volume with 2.5 µL of 10x Taq buffer (with 15 mM MgCl2), 0.25 mM dNTPs (10 mM dNTPs mix), 10 nM of forward and reverse primers, 1 U of Taq polymerase and 200 ng template DNA. The reproducibility of the amplification was confirmed by repeated PCR with a similar set of ingredients and condition. For each experiment, negative controls were taken with sterile water in place of the template. The thermal cycler program consisted of an initial denaturation for 4 min at 95◦C, followed by 35 cycles of denaturation at 94◦C for 45 s, optimized annealing temperature for 45 s, and extension at 72◦C for 1 min with a final extension of 72◦C for 10 min. The PCR amplicons were resolved on 3.5% agarose gel in 1x TAE buffer by electrophoresis at 60 V cm−<sup>1</sup> .

#### SSR Profiling and Data Analysis:

The standard binomial matrices were followed in SSR profiling and the presence and absence of the allele at a particular locus was scored as 1 and 0, respectively. Though Jaccard's coefficient pairwise distance were calculated and the resultant distance matrices were further used in NTSYSpc2 software version 2.0 (Rohlf, 2000) for clustering the isolates under investigation based on the Unweighted Pair Group method using arithmetic means (UPGMA) method. Allele frequency and Polymorphism information content (PIC) value for each marker was also calculated based on the formula: Hn = 1–Σpi2, where pi is the allele frequency of the ith allele (Nei, 1973).

#### Internal Transcribed Spacer (ITS) Amplification

The universal primers pair ITS4 and ITS5 (White et al., 1990; **Table 2**) were used for amplifying the ITS region of the rDNA. The PCR was carried out in a total volume of 50 µL with 1xPCR buffer (with 15 mM MgCl2), 250µM each dNTPs, 1 µL of each primer (10 nmol), 1.5 U Taq DNA polymerase, and 200 ng of genomic DNA. The thermal cycler program consisted of an initial denaturation at 95◦C for 4 min followed by cycling conditions included denaturation at 94◦C for 45 s, annealing at 49◦C for 45 s, and elongation at 72◦C for 1 min (35 cycles), followed by a final extension at 72◦C for 10 min. All the PCR reactions were set up along with a negative control (without template DNA). The amplification product was analyzed on 1% agarose gels by running a standard DNA molecular weight 1 kb marker parallel to it.

#### rDNA Sequence Analysis

The PCR amplified internal transcribed (ITS) rDNA region from each of the isolates was purified using Favorgen PCR purification kit (FAVORGEN Biotech Corp, Taiwan), as per the manufacturer's instructions. The purified ITS amplicons were sequenced by Agrigenome, Bengaluru, India and then the sequences were analyzed in GenBank (http://www.ncbi.nlm.nih. gov/) by using the Mega BLAST sequence analysis tool. The full length ITS sequences were used as queries for establishing kinship with the published sequences and search result with maximum homology and highest score were marked for further analysis. All the ITS sequences obtained from the isolates used in the present study were submitted to GenBank. ClustalW program of the Bioedit sequence alignment editor (Thompson et al., 1994; Hall, 1999) was used to align all the obtained sequences. The resulting multiple-alignment file was used for phylogenetic analyses which were performed using Molecular Evolutionary Genetics Analysis (MEGA 7.0) with Neighbour-Joining method (Kumar et al., 2016).

## RESULTS AND DISCUSSION

#### Morphological Identification

All the 65 S. sclerotiorum isolates collected from the major Brassica growing regions of India were observed exhibited morphological characteristics specific to Sclerotinia sclerotiorum. The mycelia of S. sclerotiorum isolates produced abundantly and appeared white to off-white, fluffy, delicate and generally with a brownish to blackish tinge. Sclerotia produced by the fungus were often dark brown to black and appeared concentrically in some case and at the periphery in others. The sclerotia were formed in abundant at the terminal stage of the growth and vary in its shape and size viz. oval-ellipsoid, straight to curved, 1.8–2.4 × 2.9–6.2 mm in size (**Table 4**). Usually, the sclerotia were produced solitary in the chain of ring-shaped either at the periphery or in the center and occasionally also formed in pairs randomly. Substantial variability in culture and morphology were exhibited by the S. sclerotiorum isolates during their growth in a controlled environment (**Figure 1**).

## Species-Specific PCR Assay

For ascertaining the originality, the S. sclerotiorum isolates was tested by using the species-specific PCR assay. For this, the species-specific primers designed and verified by Freeman et al. (2002) were used and a single 278 bp DNA fragment amplified by PCR was specific to S. sclerotiorum, which confirms the speciesspecific identity of the isolates (Figure S10).

### Morphological Variability

The growth rates of mycelia in all the 65 isolates of S. sclerotiorum were varied irrespective of their collection from the similar host species and they were prominently distinct in terms of the morphological characteristics. Depending on the growth characteristics of mycelium after 72 h of incubation the isolates were categorized into 3 groups: (i) Fast growing (ESR-01, 02, 03, 05, 06, 07, 09, 12, 13, 15, 17–22, 25, 27, 31, 32, 35, 36 and 37); (ii) Intermediate (ESR-08, 11, 14, 16, 23, 26, 28, 29, 34, 38, 40, 42, 43, 46, 48, 50, 53, 55–58, and 64); and (iii) Slow growing (ESR-04, 10, 24, 30, 33, 39, 41, 44, 47, 49, 51, 52, 54, 59, 60–63, and 65). Although most of the S. sclerotiorum isolates has attained full mycelial growth and filled up the 90 mm Petri plates within 4–5 days whereas other isolates namely ESR-04, 10, 24, 26, 28, 30, 49, 52, and 62 were taken 7–9 days for attaining the full growth and in filling the 90 mm Petri plate with the mycelial mat. Among these S. sclerotiorum isolates collection, most of them were had whitish mycelial growth with a smooth texture but the remaining isolates ESR-02, 06, 08, 12, 14, 17, 21, 31, 42, and 48 had an off-white color.

TABLE 3 | Optimized annealing temperature (Tm), Polymorphism information content (PIC), product size and a total number of bands observed for the SSR markers used to check the presence genetic diversity among S. sclerotiorum isolates.


\*Locus names; #Repeat motifs as mentioned by Sirjusingh and Kohn (2001).


#### TABLE 4 | Variability in mycelial growth and sclerotia formation (on PDA) in a different geographical isolate of Sclerotinia sclerotiorum.

(Continued)

#### Sharma et al. Genetic Diversity of S. sclerotiorum in B. juncea

#### TABLE 4 | Continued


Scl. Ini, Sclerotia initiation; Dia, Diameter; dou. Ring, double ring.

All the distinct Sclerotinia isolates were categorized based on the number and rate of sclerotia produced by them at a given time point into three groups: (i) High producer (ESR-20, 23, 26, 39, 46, 51, 58, and 65), (ii) Intermediate (ESR-01–19, 21, 22, 24, 25, 27– 29, 31–38, 40–45, 47–49, 52–57, 59–64), and (iii) Low producer (ESR-30 and 50). In the majority of the isolates, sclerotia were observed produced within 6–9 days after growth. The sclerotia of isolates ESR-26, 27, 44, 49, 55 and 65 were larger in size (2.4–2.5 mm in diameter), while those of ESR-01, 02, 04, 09, 13, 18 were smallest (1.8–1.9 mm in diameter). The isolates varied in their sclerotial length (2.9–6.6 mm), however, the sclerotia of isolate ESR-65 had a maximum length (6.6 mm) (Table S1).

The morphological features among S. sclerotiorum collections from Brassica field of different agronomic regions in India (**Figure 1**), were used in genetic diversity analysis. The dendrogram generated (Jaccard's coefficient) have three distinct clusters at 65% of distance (**Figure 2**). Cluster I, with the highest degree of genetic variation, comprises 60 isolates from different regions; cluster II has four isolates, collected from distinct geographical locations or provinces, and cluster III contains one isolate, from Haryana. The robustness of the tree was estimated by bootstrap resampling, and the bootstrap values for clusters I, II, and III were very high, 100, 99, and 100%, respectively. Out of three clusters, cluster I was the largest and differentiated into seven distinct sub-clusters: IA-IG. Almost 99% of the isolates from different provinces fell under these sub-clusters. The majority of the S. sclerotiorum isolates from Rajasthan were grouped in the sub-cluster IC to IG (83%) and rest were distributed across the dendrogram. Cluster II had 4 isolates were grouped into two sub-clusters IIA and IIB. All the four isolates one in IIA and three in IIB were from four different states whereas single isolate present in cluster III was from Haryana region. Individual isolates from Jharkhand and J&K distributed in sub-cluster IIB. Interestingly out of 4 isolates in cluster II, 2 of the isolates one each from Rajasthan and Jharkhand was moderately virulent. One notable thing in the dendrogram was the distribution of the highly virulent isolates, one from each states UP, Punjab and Rajasthan had clustered into sub-clad ID whereas another highly virulent isolates one each from UP and Rajasthan falls under IA and IB, respectively.

FIGURE 2 | Dendrogram derived from the morphological characters of 65 different S. sclerotiorum isolates, based on growth, color, texture, shape and size of the mycelia and sclerotia grown on PDA medium and their respective pathogenicity over Brassica species.

#### Pathogenicity

Plants inoculated with S. sclerotiorum developed necrotic and bleached lesions 6–9 days after inoculation, and white cottony mycelial growth appeared on stem surface after successful infestation. Pathogenic variability test conducted on 4 Brassica differentials has shown that all the isolates collected were pathogenic and able to develop stem rot disease after infection. However, the aggressiveness of the isolates was observed varies contrarily on all the host differentials and thus diverse groups were identified for each character by comparing the pathologically quantitative characteristics i.e., stem lesion length and percentage of disease incidence on different host differentials. All the 65 S. sclerotiorum isolates from different geographical regions showed variations in pathogenicity in respect of stem lesion length. After artificial inoculation of S. sclerotiorum, the typical water-soaked lesion was observed on the stems of host differentials. On the basis of longest and shortest stem lesion formed by different geographical isolates of four Brassica species, following 4 groups were made: Group I: (Brassica juncea var. NRCDR-02), Group II: (B. rapa var. toria), Group III: (B. rapa var. yellow sarson), and Group IV: (B. rapa var. brown sarson). Group I, composed of 9 isolates viz., ESR−12, 15, 17– 19, 25, 35, 36, 65 produced longer stem lesion and among them ESR-36 produced the longest lesion i.e., 49.0 cm, while ESR-26, 40 and 64 isolates produced shortest stem lesion i.e., 1.0 cm (range:1.0–49.0 cm). Similarly, other 6 groups also showed pathogenic variability. Based on the stem lesion formation, out of 7 groups, ESR-36 and ESR-60 isolates were found producing longest and shortest stem lesion on the inoculated stem, respectively.

The results of pathogenicity tests conducted by using the 65 isolates of S. sclerotiorum over the 4 different Brassica differential has shown that the isolates had differences in their pathogenicity, infectivity and disease severity on tested Brassica cultivars and are presented in **Table 1**. The disease severity was measured in terms of Mean Disease Severity (MDS) by measuring the lesion length appeared on infected stem and that provides the basis for recording the virulence of individual isolate as low (MDS: < 3 cm), medium (MDS: < 3–10 cm) and high (MDS: > 10 cm). Thus, based on the disease severity index of the isolates S. sclerotiorum collection was distributed into three groups viz., highly virulent (5 isolates), virulent (46 isolates), and moderately virulent (14 isolates) based on the variations observed by analyzing the mean disease severity index in the treated mustard variety (**Table 5**). The un-inoculated and control plants of test variety inoculated with plain agar media showed no symptoms. The Sclerotinia isolates were further re-isolated and cultured from the infected Brassica plants following Koch's postulates.

### SSR Genotyping

The population dynamics study of the pathogen plays a vital role in deciphering the mechanism behind behavioral pattern as well as in understanding the distribution pattern of the pathogen in distinct geographical areas where host species envisioned frequently by the infective pathogen. In this direction, the genetic diversity of S. sclerotiorum isolates were analyzed by simple sequence repeat (SSR) based fingerprinting. All the 25 SSR primers were analyzed after optimizing the melting temperature (Tm) for polymorphism assay on collective set of 65 isolates comprising representatives from the major Brassica growing states of India included in the study: Rajasthan (31), Uttar Pradesh (16), Punjab (8), Haryana (4), Delhi (1), Bihar (1), Madhya Pradesh (1), Uttarakhand (1), Jharkhand (1), and Jammu & Kashmir (1). Among the genome-wide distributed primers, only four primers (9I, 12L, 14N, and 17Q) has generated multiple bands whereas remaining primers were produced a single polymorphic band (**Table 3**). Two primers 9I and 17Q were observed produced 2 polymorphic bands of 100–500 bp size range per isolate. Subsequently other two primers 12L and 14N have produced 4 polymorphic bands within a range of 200– 1,000 bp (Figures S1–S9). The differences in genetic diversity among the S. sclerotiorum pathogen population obtained through SSR fingerprinting in this study were found reproducible. Subsequently, UPGMA dendrogram drawn based on presence and absence of the banding patterns showed substantial diversity among the isolates. Based on the topology and similarity indices of the dendrogram, the isolates were grouped into three major clades and each clade was signified by a roman numeral (group I to III; **Figure 3**).

The detailed analysis of the dendrogram revealed, maximum numbers of isolates grouped in clade I and it was the largest and based on the similarity indices the clade I was further separated into seven divergent sub-clades: IA-IG. Among these sub-clades, IA, IF and IG harbors nearly 60% of the S. sclerotiorum isolates from Rajasthan whereas, remaining 40% were distributed in other clades of the dendrogram. The other two major clades, clade II and III had 2 and 4 isolates and in which 1 isolate in clade II and 2 isolates in clade III were from Rajasthan. Almost all the isolates from UP were clustered in clade I under sub-clades IA, IB, IC, IE, and IG. But, isolates from Punjab has a different trend of distribution across the dendrogram. All the isolates from Haryana were clustered in sub-clades IA and IC. Individual isolates from Delhi and J&K distributed in sub-clade IA, from Bihar, Jharkhand, and MP distributed in IC, and one isolate from Uttarakhand distributed in the major Clade II. The distribution of highly virulent isolates in dendrogram was found interesting were two from UP and one isolates from each Rajasthan and Punjab had clustered into IA sub-clade whereas another highly virulent isolate from Rajasthan was clustered in clade III. The prominent genetic diversity feature in clustering was observed in sub-clade IE, IF, IG and clade II, III those included exclusively virulent or highly virulent isolates and more interestingly these clades had at least one isolate from individual state included in this study.

### ITS Sequence Analysis

The fungal-specific ITS4-ITS5 universal primers pair was used in amplifying the internal transcribed spacer (ITS) region from the DNA of all the 65 S. sclerotiorum isolates collected from infected Brassica crop grown in the diverse genetic background. The amplicon lengths and purity was estimated by gel electrophoresis and found about 600 bp in size (Figure S11). The sequence analysis result of the rRNA data by NCBI BLAST tool observed connoted the morphological variability and supported their individual existence. Further, in NCBI GenBank database the TABLE 5 | Virulence grade for Sclerotinia sclerotiorum population assigned based on disease severity assessment and GenBank accession numbers for the ITS sequences.


(Continued)

#### TABLE 5 | Continued


GenBank accession IDs of each isolate based on ITS sequence information.

99–100% closest match was found with S. sclerotiorum. The sequence data information for the ITS rDNA region of all the 65 S. sclerotiorum isolates obtained in this study were submitted in the GenBank database of the NCBI (GenBank Accession numbers MF408233-MF408297; **Table 5**).

Based on the ITS sequence information a dendrogram was constructed by using the ClustalW and Mega7.0 software with the Neighbor-Joining method (**Figure 4**). The resulted phylogenetic tree was clustered in total 11 clades and six of them were formed major clades, four clades with two members each and one clade with a single member that clearly indicates the distribution of the isolates across dendrogram irrespective of their properties of pathogenicity or geographic origin. Isolates from Rajasthan and UP were found distributed in nine and six different clades, respectively, whereas isolates from Punjab were found distributed in five different clades. Clade I was found to be the largest and consisted of 21 divergent isolates from Rajasthan, UP, Punjab, Bihar, and Delhi with varied pathogenicity (highly virulent, virulent and moderately virulent). Clade II had virulent and moderately virulent isolates from Rajasthan and UP whereas isolates from Haryana were all virulent and distributed in clades IV, V and VI. Clade VII had only two isolates and both of them were from Rajasthan and virulent. Interestingly, the clade IV and V were also found heterogeneous as of clade I in terms of the isolates of different region and their virulence features. The isolate from Jharkhand was found distinctly in the clade VIII and was moderately virulent. Remarkably clade X had four isolates from the four different states. Two pathogenic isolate each from Rajasthan and Punjab were formed clade III, from UP and Punjab have formed clade IX and from UP and Rajasthan were formed clade X. One moderately pathogenic isolate from Jharkhand formed the single member clade VIII. The highly pathogenic isolates from UP, Rajasthan, and Punjab were distributed in Clades I, V, and VI.

The ITS sequence analysis of all the isolates in the study depicted on an average 95–100% of sequence similarity that entails conspecific nature of the isolates and their host from which the isolates were collected irrespective of different geographic regions.

### DISCUSSION

The incidence of stem rot disease in India, especially in oilseed Brassica, has been known to damage the crops since from the mid-1990 but the genetic diversity of the pathogen especially in oilseed Brassica has not been comprehensively studied so far which is primarily required for understanding the mechanism of the pathogenesis and virulence of the pathogen. From the last decade, the stem rot disease in Brassica poses a serious threat to the crop every year in India and causes significant damage to the crop in terms of yield and economic losses to the farmers due to lack of effective management practices and other control measures. From the frequent disease occurring reports and extent of its severity over the oilseed crops, this pathogen has drawn attention and interest of the researchers to analyze the host-pathogen interaction in a broader spectrum as the mustard growing farmers experiencing this problem almost every years during the cropping season. The present investigation thoroughly reporting the existence of S. sclerotiorum pathogen in mustard fields and the genetic diversity among the population distributed throughout the Brassica growing regions of India.

The field-based crop-specific disease survey in different states of the India has shown the high occurrence and well-dispersed existence of the stem rot disease that commonly associated with the scorched lesion on the stem of the infected plants and sclerotia infested soil samples in the severely affected crop field. The regular appearance of the stem rot disease in Brassica indicates it is a recurrent menace for the crop and the lack of promising resistance source to the pathogen in existing cultivated varieties. Although the isolates in the present investigation were collected from Brassica field only but even though they were found genetically diverse in terms of their culture conditions and in morphological appearance as well. This might be due to the divergent geographical location from where the isolates have been collected. The evaluation of the pathogenicity of different Brassica cultivars has revealed that there was differential susceptibility in host species against different isolates of the S. sclerotiorum and thus based on the pathogenicity assay and virulence properties the pathogen population was grouped into three subgroups highly virulent, virulent and moderately virulent isolates. Depending on their virulence all the isolates resulted in the development of lesions of white patches of a different length over the infected stem of the Brassica cultivars that indicates the existence of genetically diverse and variable population of S. sclerotiorum. The pathogenicity assay and virulence variability test have shown that out of 65 S. sclerotiorum isolates 5 of them were highly virulent, 46 were virulent and 14 isolates were falls under moderately virulent category and they disseminated through either contaminated soil or seed across Brassica growing regions in India. The genetic diversity analysis by all the means followed in this study has clearly shown there were not many differences exists in terms of virulence and in groups that comprised from the isolates of different virulence nature belongs to different localities. Our results are in accordance with the observations made in previous studies where aggressiveness of the isolates are being independent of their geographic distribution (Atallah et al., 2004; Xie et al., 2014). Hence, the nature of the virulence of the S. sclerotiorum pathogen possesses a potential threat to the planting of Brassica cultivars susceptible to the stem rot disease.

Sclerotinia sclerotiorum apart from reproducing asexually by myceliogenic germination of sclerotia it is also able to produce sexually through apothecia and ascospores formation. The diversity analysis and genetic architecture of S. sclerotiorum have shown the existence of homogeneity across the isolates with little variation in 18 and 28 s rDNA regions (Kohli et al., 1995). The genetic diversity of the pathogen, severity of the disease and frequency of its occurrence are the important factors that directly correlates with the epidemics of the disease at the regional level and hence their comprehensive studies are the foremost requirement for understanding the behavioral pattern of the pathogens for effectively design the management practices for the control of this disease more efficiently. The polymorphism assay with the genome-wide distributed SSR markers has been proven effective in studying the existence of genetic diversity within the pathogen population. The SSR is the co-dominant molecular markers and it is a PCR-based robust technique that amplifies the putative repeat regions of the DNA and therefore it is able to identify a higher level of polymorphism than PCRbased another less robust markers like RAPD and hence it has been extensively used for analyzing the fungal population. The genetic diversity among S. sclerotiorum isolates based on the morphological features and SSR fingerprinting had observed

FIGURE 4 | Phylogenetic analysis of S. sclerotiorum isolates based on rONA ITS sequences by Neighbor-Joining method and respective pathogenicity over Brassica species.

quite diverse and reveals that some of the isolates grouped together belonging to the same region while others from the same region distributed across the dendrogram. The morphological features of the isolates and their SSR fingerprinting confirms the existence of a highly variable population of S. sclerotiorum even within the same crop and the region. However, among some isolates of S. sclerotiorum those belonging to the same geographic regions has been found having lower genetic diversity and that might be due to the locally adapted cropping system where introduction of newer varieties are usually not preferred over the existing one and hence diversity of the pathogen remained to stagnate due to lack of introduction of the newer genotypes of the pathogen in the area under cultivation.

The result of the present investigation depicting the genetic diversity of S. sclerotiorum population based on the morphological parameters and SSR analysis has been observed in coherence with earlier findings of the studies made on the same pathogen but in different host species. In eggplant, the genetic diversity analysis of S. sclerotiorum population by using the SSR and RAPD molecular markers, Tok et al. (2016) had shown the existence of high level of heterogeneity in the pathogen populations and presence of different isolates of S. sclerotiorum pathogen in the same regions. The presence of high heterogeneity in genetic composition of the S. sclerotiorum isolates in different regions of Turkey has reflected by their distribution across the dendrogram in different groups made from the RAPD and SSR profiling. In a similar study, Barari et al. (2013) have characterized the genetic diversity within the population of S. sclerotiorum isolates from Iran and the variations among them were projected by SSR profiling. The depiction of variability in closely related isolates by SSR markers demonstrated its effectiveness in analyzing the genetic diversity among S. sclerotiorum pathogen. In chickpea, the use of RAPD, ITS-RFLP, ITS sequencing, and mycelial compatibility groupings (MCG) has shown the limited variability among the distinct S. sclerotiorum isolates possess the higher genetic homogeneity and partially correlated with their geographical origin (Mandal and Dubey, 2012). The robustness and reproducibility features of the SSR markers make it more effective and reliable marker for deciphering the genetic relationship among population studies and thus the phylogenetic analysis of S. sclerotiorum based on the SSR genotyping revealed the existence of wider genetic diversity among its population. The present findings are in connotation with the earlier reports that molecular typing by SSR is an efficient technique for establishing the genetic relationship among the diverse population and it is being preferred over other molecular tools for diversity analysis.

Further, the ITS (internal transcribed spacer) sequence based molecular phylogeny was established by using the ITS region as a genetic marker to investigate the genetic association within the S. sclerotiorum isolates. This phylogenetic study has facilitated the revelation of the evolutionary relationship within S. sclerotiorum species collected from a single host Brassica species from different geographical regions of India. Based on the ITS phylogeny all the 65 isolates investigated in the present study were observed grouped into six major and five minor evolutionary lineages of S. sclerotiorum isolates collected from different mustard growing states of India. In all the lineages deviation in virulence was not so evident and this suggests that the isolates from single host species may not differ ominously in their pathogenicity. The least variation in virulence of the pathogen has been observed in the case where isolates from defined geographical regions were taken into the investigation (Alvarez and Molina, 2000; Atallah et al., 2004; Sexton and Howlett, 2004). The differences in virulence can be more evident and conclusive when isolates from the widely distant geographical area will be taken for comparison. Kull et al. (2004) has found the conclusive evidence for the host specialization among S. sclerotiorum isolates.

Although, there is no report of any toxin is usually produced by the S. sclerotiorum pathogen it has been observed the production of oxalic acid (Cessna et al., 2000) and extracellular lytic enzymes production (Riou et al., 1991) invariably by all the isolates during pathogenesis. However, extensive biochemical and physiological studies are required to investigate the production of any natural mycotoxins by the pathogen if any like other fungus and further needs to find the effect of those toxins on the host plants and in herbivores consuming those plants. In this study, we attempted to verify the authenticity of S. sclerotiorum isolates by using the Sclerotinia-specific PCR primers and found all the geographical isolates (65) investigated were genetically diverse.

Development of the Genetic fingerprinting database for distinguishing the pathogen diversity between and within the fungal species is one of the foremost requirement for determining the origin and evaluation of the pathogen. Unequivocally, environmental factors are also influencing the properties of the pathogen and play crucial roles in determining the genetic differentiation within the pathogen population (Weller, 1988). The existence of a mixed population of S. sclerotiorum within a region could be due to the possible human interventions like the movement of infested soil through seeds, packaging materials or agricultural equipment that might carry the pathogen from one region to another. Results of the present study, as well as evidence from the previous studies, indicates the minute changes in physiological properties of the pathogen like changes in virulence due to evolutionary changes happening in due course of time that brings changes among genetically distinct populations of S. sclerotiorum isolates. Such changes may happen because of the varying force of selection over pathogen imposed by the selected set of genes of the host germplasm being tolerant to the invasive pathogen. The genes involved in host defense system and coding for the virulence factors in pathogen have been observed comparatively much more vulnerable for accumulating the mutation under positive selection pressure than the genes those regulating the basic physiological processes in the organism (Matute et al., 2008). As a consequence, in most of the hostpathogen interaction pathogen often adapts themselves to the evolutionary changes arises in response to the variation in the host genotype.

### CONCLUSION

The current investigation reports the studies on morphological features, pathogenicity variance over the Brassica cultivars, and genetic diversity based on molecular characteristics of the S. sclerotiorum isolates population collected from 65 locations of the 10 major Brassica growing states of India. This is the first comprehensively studied report on S. sclerotiorum isolates especially from Brassica field of India that indicates the existence of broader genetic variations among the pathogen and this would facilitate the plant breeders to use the information in Brassica breeding program for stem rot disease resistance development. The SSR-based markers utilized in present investigation for diversity analysis can be effectively used for identification and analysis of S. sclerotiorum population from various host species. The present study offers the basis for increasing the basic understanding of host-pathogen interaction and also for deciphering the genes and associated pathways involved in combating the pathogen invasion. The evidence of the present study may provide a broader understanding of the pathogen virulence and that will facilitate the development of the effective disease management strategies based on molecular breeding and other advanced approaches.

### AUTHOR CONTRIBUTIONS

PS, NG, and LP conceived and obtained funding from ICAR-EMR project for Sclerotinia work in Indian mustard. PS and VS collected isolates from diseased plants and along with LP done the morphological characterization and pathogenicity evaluation. AS and MR obtained isolates from PS, carried out microsatellite genotyping. NG and RB do the ITS sequencing of S. sclerotiorum isolates and wrote the paper. NG and DM carried out clustering and phylogenetic analyses of microsatellite data and ITS sequences. RB and VS edited the manuscript.

#### FUNDING

The authors are grateful to ICAR (Indian Council for Agricultural Research), New Delhi, India for providing

#### REFERENCES


the financial support under ICAR-Extramural research Project F. No.: CS/18(15)/2015-O&P and Director, IARI, New Delhi, Director, DRMR, Bharatpur; Project Director, NRCPB, New Delhi for providing facilities to conduct these experiments.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.01169/full#supplementary-material

plant pathogenic fungus, Sclerotinia sclerotiorum. Mol. Ecol. 4, 69–77. doi: 10.1111/j.1365294x.1995.tb00193.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 © 2018 Sharma, Samkumar, Rao, Singh, Prasad, Mishra, Bhattacharya and Gupta. 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 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.

# Integrated Translatome and Proteome: Approach for Accurate Portraying of Widespread Multifunctional Aspects of Trichoderma

#### Vivek Sharma<sup>1</sup> \*, Richa Salwan<sup>2</sup> , P. N. Sharma<sup>1</sup> and Arvind Gulati<sup>3</sup>

<sup>1</sup> Department of Plant Pathology, Choudhary Sarwan Kumar Himachal Pradesh Agricultural University, Palampur, India, <sup>2</sup> Department of Veterinary Microbiology, Choudhary Sarwan Kumar Himachal Pradesh Agricultural University, Palampur, India, <sup>3</sup> Institute of Himalayan Bioresource Technology, Palampur, India

#### Edited by:

Katarzyna Turnau, Jagiellonian University, Poland

#### Reviewed by:

Somayeh Dolatabadi, Westerdijk Fungal Biodiversity Institute, Netherlands Ravindra Nath Kharwar, Banaras Hindu University, India

> \*Correspondence: Vivek Sharma ankvivek@gmail.com

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 21 April 2017 Accepted: 07 August 2017 Published: 29 August 2017

#### Citation:

Sharma V, Salwan R, Sharma PN and Gulati A (2017) Integrated Translatome and Proteome: Approach for Accurate Portraying of Widespread Multifunctional Aspects of Trichoderma. Front. Microbiol. 8:1602. doi: 10.3389/fmicb.2017.01602 Genome-wide studies of transcripts expression help in systematic monitoring of genes and allow targeting of candidate genes for future research. In contrast to relatively stable genomic data, the expression of genes is dynamic and regulated both at time and space level at different level in. The variation in the rate of translation is specific for each protein. Both the inherent nature of an mRNA molecule to be translated and the external environmental stimuli can affect the efficiency of the translation process. In biocontrol agents (BCAs), the molecular response at translational level may represents noise-like response of absolute transcript level and an adaptive response to physiological and pathological situations representing subset of mRNAs population actively translated in a cell. The molecular responses of biocontrol are complex and involve multistage regulation of number of genes. The use of high-throughput techniques has led to rapid increase in volume of transcriptomics data of Trichoderma. In general, almost half of the variations of transcriptome and protein level are due to translational control. Thus, studies are required to integrate raw information from different "omics" approaches for accurate depiction of translational response of BCAs in interaction with plants and plant pathogens. The studies on translational status of only active mRNAs bridging with proteome data will help in accurate characterization of only a subset of mRNAs actively engaged in translation. This review highlights the associated bottlenecks and use of state-of-the-art procedures in addressing the gap to accelerate future accomplishment of biocontrol mechanisms.

Keywords: transcripts, active mRNA, regulation, integrated omic, translatome

## INTRODUCTION

Trichoderma is a cosmopolitan and cardinal representative soil microflora of various climatic conditions (Herrera-Estrella, 2014). The biocontrol role of Trichoderma spp. have emerged as an attractive choice in agriculture sector due to their environmentally friendly nature over synthetic pesticides (Mukherjee et al., 2012, 2013). Among different biocontrol agents (BCAs),

the genus Hypocrea/Trichoderma containing Trichoderma harzianum, Trichoderma atroviride, Hypocrea virens are probably the most explored BCAs (Schuster and Schmoll, 2010; Sharma and Shanmugam, 2012; Sharma and Salwan, 2017) and occupies over 60% of all registered biopesticides (Mukherjee et al., 2013). The continuous efforts on the evaluation of biocontrol potential of Trichoderma have led to the identification of several promising species/strains including T. harzianum (Yedidia et al., 1999; Cloyd et al., 2007), Trichoderma virens (Hermosa et al., 2000; Howell, 2006), Trichoderma viride (Papavizas, 1985), T. atroviride (Longa et al., 2009), Trichoderma polysporum (Zhang et al., 2015), and Trichoderma asperellum GDFS1009 (Wu et al., 2017). In recent studies, another potential strains of Trichoderma saturnisporum has been identified for its biocontrol potential (Sharma and Shanmugam, 2012; Diánez Martínez et al., 2016). In addition to primary application in agriculture, Hypocrea jecorina/Trichoderma reesei strains are molecular factory for cellulolytic enzymes (Merino and Cherry, 2007; Singh et al., 2015). The natural potential to secrete lytic enzymes, antibiotics, and defeating opponent for space and nutrition are largely considered responsible for its success against plant pathogenic fungi (Viterbo et al., 2002; Benítez et al., 2004). The root colonization and intimate association of Trichoderma spp. with plant roots are known to promote plant growth and boost immune response against a number of plant pathogens (Contreras-Cornejo et al., 2011; Brotman et al., 2012; Mukherjee, 2012). Biocontrol strains of Trichoderma are used worldwide for the management of various plant pathogens like vascular wilt caused Fusarium (Al-Ani et al., 2013), Botrytis blight or gray mold caused by Botrytis (Elad and Kapat, 1999), anthracnose caused by Colletotrichum spp., and several other plant fungal diseases (Sharma et al., 2016a,b, 2017a). The improvement of Trichoderma species as BCAs for various agricultural applications required, detailed understanding of its active biological repertoire involved in mycoparasitism antibiosis as well as others components (**Table 1**). Genome sequencing and its annotation in mycoparasitic species have depicted genome sizes of 38.8 and 36.1 Mb for T. viride and of T. atroviride for biocontrol strains, respectively, compared to 34 Mb of T. reesei an industrial strain. Annotation of complete genome depicted a gene pool of 11,800 genes for T. atroviride and 12,400 genes for T. viride, compared to 9,143 genes in saprophytic strain T. reesei. The abundance of gene pool in mycoparasitic strains of Trichoderma genome (Lin et al., 2012; Atanasova et al., 2013) and expression of over 60% of the encoding transcripts during interaction of T. virens and T. atroviride against Rhizoctonia solani have revealed a complex nature of biocontrol mechanisms (Atanasova et al., 2013). Liu and Yang (2005) using simulated mycoparasitic conditions and cDNA libraries identified a total of 3,298 expressed sequence tags (ESTs) which corresponds to 1,740 transcripts. Using inducible conditions for T. harzianum CECT 2413, Vizcaíno et al. (2006) characterized, nearly 8,710 ESTs whereas Yao et al. (2013) identified 1,386 ESTs for T. harzianum 88. Among different ESTs, significant differential expression was observed only for limited transcripts. These EST represents a fragment of mRNA have several biotechnological applications and are being explored for either complementing the sequenced genome projects or cost effective alternatives for identification of genes as well as elucidation of functional genomics of plant– microbe interactions (Vieira et al., 2013). Advancement in molecular tools such as transcriptome profiling using RNA-seq and quantitative real-time PCR (RT-qPCR) technologies also predicted a large number of genes (14,095) for T. harzianum during augmentation on plant pathogen such as Sclerotinia sclerotiorum cell wall and only 297 were found differentially expressed among them (Steindorff et al., 2012, 2014). In addition to plant diseases management potential of biocontrol strains of Trichoderma, its growth promotion abilities in plants have been identified which are significantly enhanced during their antagonistic interactions with pathogens in soil. The molecular action of its biocontrol arsenal is mediated through adaptive recruitment and reprogramming of unique reservoir of several transcripts (Shaw et al., 2016). A comparative account using bioinformatic approaches such as BLAST analysis has revealed a very low overlap for different ESTs libraries (Yao et al., 2013). Therefore, the microarrays set designed based on genome coverage and ESTs may not provide accurate information.

The comprehensive analyses using different molecular approaches including ESTs (Chambergo et al., 2002), subtractive hybridization (Carpenter et al., 2005; Scherm et al., 2009; Vieira et al., 2013), microarray (Chambergo et al., 2002; Breakspear and Momany, 2007; Samolski et al., 2009), and transcriptomes (Atanasova et al., 2013) have established the complex response of Trichoderma species in biocontrol process which induces numerous genes having morphogenetic or other functions as well (Mehrabi-Koushki et al., 2012; Puglisi et al., 2012; Cacciola et al., 2015; Cetz-Chel et al., 2016). The complexity in different attributes may not be related to a particular stress and hence can lead to either imperfect transcriptional representation or a complex response. The continuous development in molecular technologies and advent of cloning free libraries using genome sequencing, deep RNA sequencing and proteomics has played vital role in the accurate identification and enhancing our capabilities of cataloging mRNA and protein populations exclusive to Trichoderma strains in response to changing environmental conditions (Shentu et al., 2014; Xie et al., 2015; Schmoll et al., 2016).

The Trichoderma–plant–pathogen interaction can produce significant amount of noise. Therefore one can speculate that the substantial amount of response at the gene expression level represents noise and that only a few changes are adaptive. Also, the microbes in the environment are continuously subjected to challenges and respond simultaneously to these factors in a complex way. Understanding the regulatory interactions necessitates an approach that can encompasses simultaneous both the transcriptome and proteome to observe and systematically view the adaptive expression at RNA and protein level. The integrated studies based on translatome and proteome level can provide a better state of these adaptive responses during biocontrol interaction. The regulation of mRNA at transcriptional and post-transcriptional levels contributes to reprogramming the behavior of BCAs through protein and secondary bioactive metabolites secretion to counter the pathogen associated challenges.


TABLE 1 | List of a few selected glycosyl hydrolases, secondary metabolites, and different transcripts of biocontrol strains/species of Trichoderma characterized for their role in biocontrol.

So far studies on Trichoderma have been conducted extensively using ESTs and transcriptome approach revealed the expression of several genes related to mycoparasitism of BCAs directly (Reithner et al., 2011; Sharma et al., 2017b) or indirectly through the modulation of host transcriptome (Morán-Diez et al., 2012; Perazzolli et al., 2012). In our previous studies, attempts were made to identify the role of different transcripts related to lytic enzymes, transporter system, and other gene related to metabolites of T. harzianum (Sharma et al., 2016a,b, 2017b) and characterization of extracellular proteins from T. saturnisporum (Sharma and Shanmugam, 2012) using autoclaved mycelium of different plant pathogenic fungi. These studies revealed only a limited number of proteins compared to transcripts. The approaches used for cDNA cloning and other array technologies have also created artifacts in accurate identification of candidate transcripts. Therefore, the integrated translatome and proteome based studies can help in a better and accurate depiction of key regulators involved in Trichoderma– plant–pathogen interaction (**Figure 1**). Recent studies showed that the gene expression of mycoparasitic T. harzianum and T. atroviride strains changes not only to plant-pathogenic fungi (Sharma et al., 2016a, 2017b) but also to itself (Reithner et al., 2011). Thus translational response is a key determinant contributing to adaptation under such interaction stress (Picard

et al., 2013). Therefore, present review emphasizes the role of translatome based approach in accurate determination of active mRNA population in a complex dialog coupled to proteome data in a three way interaction of Trichoderma–plant–pathogen.

## MECHANISMS OF Trichoderma

Trichoderma strains are used as BCAs in agriculture largely due to their abilities to directly antagonize plant-pathogenic fungi through the production of hydrolases (Benítez et al., 2004; Gruber and Seidl-Seiboth, 2012), antibiotics (Rubio et al., 2009; Vinale et al., 2014) and their tolerance to toxin produced by plant pathogens (Sharma et al., 2013) (**Table 1**). The interaction of Trichoderma with host plants reprograms not only the gene expression of biocontrol strains but also of its associated host plant (Harman, 2011; **Figure 2**). For example, strains of Trichoderma are explored for growth promotion and boosting immune responses, root development, and activation of seed germination or amelioration of abiotic stresses (Harman et al., 2004; Lorito et al., 2010; Shoresh et al., 2010; Hermosa et al., 2012). The immune responses in host plant are primed through systemic resistance (Tucci et al., 2011), involving a complex signaling of jasmonic acid/ethylene-induced systemic resistance and/or salicylic acid-dependent pathways which may behave differently in plant–Trichoderma interactions (Shoresh et al., 2010). The three way interaction between biocontrol, host plant, and pathogen from initial root colonization is known to change both the transcripts and proteome of host plants (Alfano et al., 2007; Segarra et al., 2007; Shoresh and Harman, 2008; Palmieri et al., 2012; Gomes et al., 2017; Martínez-Medina et al., 2017a,b; Pelagio-Flores et al., 2017). The availability of microarrays, next generation DNA sequencing, RNA-seq, and genome annotation have provided a global insight into the transcriptome response of plant–Trichoderma and Trichoderma– plant pathogen interaction.

### Omics APPROACHES IN UNCOUPLING GENOME AND TRANSCRIPTOME PROFILE

The characterization of candidate transcripts involved in various biological functions using transcriptome is one of the best approach. In comparison to stable nature of the genome, transcriptome is more dynamic and vary in response to different stimuli. The massive transcriptome response to various factors can be tentatively identified, quantified, and correlated to a biological process using ESTs, subtractive libraries, and DNA microarrays (Herrera-Estrella, 2014). A number of studies have been done at genome-wide and transcriptional level to understand the molecular behavior of different Trichoderma strains under contrasting conditions ranging from mycoparasitism of plant pathogens to imparting direct beneficial aspects to plants under stress conditions (Arvas et al., 2006; Vizcaíno et al., 2007; Seidl et al., 2009). The transcriptome analysis of T. atroviride IMI206040 at different stages of

interaction with R. solani identified 7,797 out of 11,863 estimated genes which represented over 65% of total gene of the organism genome whereas only 1.47% of total gene (175) transcripts were found significantly differentially expressed in mycoparasitic interactions. The differentially expressed transcripts were also investigated during pathogenic attack on Phytophthora capsici, Botrytis cinerea, and R. solani (Reithner et al., 2011). In comparison to a large number of transcribed genes predicted for T. atroviride based on genomic data, only 38.4% of genes involved in interaction with R. solani, were expressed before contact whereas 52.8% were found responsible for Trichoderma confrontation with itself (Reithner et al., 2011).

The use of EST (Vizcaíno et al., 2006, 2007), subtractive cDNA libraries and DNA array (Rosales-Saavedra et al., 2006; Alfano et al., 2007; Mathys et al., 2012) based studies carried under environmental conditions have helped dramatically to the global-scale identification of active genes of Trichoderma which are not directly linked to plant pathogens but are required for colonization and imparting other beneficial aspects to the host plant. For example, hydrophobins, aspartyl proteases, expansinlike protein of Trichoderma origin have been explored for their involvement in the mycoparasitism mediated biocontrol of these microbes (Brotman et al., 2008; Samolski et al., 2009). Subsequently, the sequencing of complete genome and highthroughput transcriptome using 454 sequencing (Barakat et al., 2009) has enhanced our understanding on investigation of mechanisms at global cellular level under different conditions in better way (Reithner et al., 2011). The transcriptome based approach is far more robust, dynamic, and refined technique compared to genome sequencing which is stable as described below.

#### Trichoderma Genome Organization

Since the genome sequencing of T. reesei industrial strain nine years back (Martinez et al., 2008), presently the genome of a number of strains representing T. virens, T. harzianum, T. atroviride, T. asperellum, Trichoderma longibrachiatum, and Trichoderma citrinoviride have been sequenced and revised (http://genome.jgi.doe.gov/). A comparative account of genome revealed presence of seven chromosomes in industrial strain T. reesei (Carter et al., 1992; Mantyla et al., 1992; Herrera-Estrella et al., 1993) whereas six chromosomes in biocontrol strains T. harzianum and T. viride (Gómez et al., 1997; Martinez et al., 2008). The genomic annotation of T. virens, T. atroviride, and T. reesei also unveiled lack of transposons and remarkable similarity of genes up to 78–96% among them. In the genome of T. virens and T. atroviride no true orthodox were reported for 2,756 and 2,510 genes, respectively in other species. The genome of T. virens and T. atroviride share 1,273 exclusive orthologs and 26 expanded families which were missing in T. reesei genome that may be a probable answer to mycoparasitic nature of T. atroviride and T. virens (Kubicek et al., 2011; Herrera-Estrella, 2014). A comparative study of genome organization of two Trichoderma species has revealed the expansion of considerable expansion genes involved in mycoparasitic T. virensstrain which are missing in T. reesei (Kubicek et al., 2011).

#### Transcriptome

fmicb-08-01602 August 25, 2017 Time: 17:45 # 6

The development of modern sophisticated omics technologies has played a vital role in developing better system-level understanding of gene expression. In particular, transcriptome based studies have proved a yardstick in the investigation of global cellular mechanisms and identification of several key genes involved in mycoparasitism and imparting other benefits to the host by Trichoderma strains. The measurement of the entire set of RNAs through transcriptome coupled with DNA microarrays or high-throughput RNA sequencing is a reliable and reproducible tool for wide analysis of transcripts. A number of transcriptome studies have been done on Trichoderma–plant– pathogen interaction (Marra et al., 2006; Chacon et al., 2007; Samolski et al., 2009; Mehrabi-Koushki et al., 2012; Rubio et al., 2012).

Stating from initial use of EST for the determination of glucose metabolism in T. reesei (Chambergo et al., 2002) and TrichoEST project (Vizcaíno et al., 2006), ESTs based studies have been done in T. harzianum (Liu and Yang, 2005; Vizcaíno et al., 2006; Suárez et al., 2007; Yao et al., 2013), T. atroviride, T. asperellum (Vizcaíno et al., 2007; Liu et al., 2010), T. virens (Vizcaíno et al., 2007; Morán-Diez et al., 2010), Trichoderma aggressivum, T. viride, and T. longibrachiatum (Vizcaíno et al., 2007) for the identification of transcripts induced during mycoparasitism and other environmental conditions. From a total of unique sequences (3,478), in T. harzianum CECT2413, 23% were found related to secretory chitinases, glucanases, and proteases. A large number of transcripts expressed (9478 ESTs containing 2,734 unique sequences) during the early interaction of T. atroviride with B. cinerea and R. solani were identified (Seidl et al., 2009) whereas 66 genes covering 442 ESTs were induced under mycoparasitic interaction (Herrera-Estrella, 2014).

Similarly, the analysis of transcriptomics changes in T. harzianum, T. virens, and T. hamatum during interactions with tomato plants revealed expression of 1,077 genes and only six of them being common to all three. The majority of genes encoding enzymes belong to chitin degradation during early interactions with tomato plants whereas genes encoding other secreted proteins were likely to involve in the signaling between Trichoderma and plants. Transcriptome based studies have led to the identification of new candidate genes having role in redox reaction, possible elicitors, transporters (Sharma et al., 2017b), lipid metabolism and detoxification (Chacon et al., 2007; Sharma et al., 2013), small secreted proteins (Ruocco et al., 2009; Samolski et al., 2009; Rubio et al., 2014). The de novo sequencing of T. atroviride IMI206040 transcriptome obtained during mycoparasitic interaction in presence of plant-pathogenic fungus R. solani revealed thousands of high-quality reads. An account of transcripts expressed during interaction to the total number of genes predicted in the genome of T. atroviride revealed that almost 45% were induced during interaction with R. solani and only 175 of them were host responsive (Reithner et al., 2011; Gupta et al., 2016).

Microarray analysis of T. harzianum T34 strain interaction with Arabidopsis identified approximately 24,000 transcripts of the host plant which were modulated by the BCA. The significance and global impact of this beneficial microbe in reprogramming the molecular physiology of host plant to stress responses through the regulation of transcription, signal transduction pathways has been reported in different studies (Morán-Diez et al., 2012; Lamdan et al., 2015). Further host specific response of Trichoderma strain with plants representing monocot and dicot hosts under the same conditions have also been explored to identify signature transcriptome repertoires and answer the widely prevalent questions of specificity of responses and role of secreted proteins in mutualistic interaction, root colonization, and induction of immune responses (Morán-Diez et al., 2015; Ho et al., 2016; Sharma et al., 2017b). These studies indicate the limitations of transcriptome based studies in precise estimation of ribosome loaded active mRNA population involved in complex mycoparasitic behavior of Trichoderma species as BCAs.

#### Translatomes

The mRNA and protein levels do not perfectly correlate in native or engineered systems (Tian et al., 2004; Jayapal et al., 2008; Vogel and Marcotte, 2012; Payne, 2015). The post-transcriptional regulation of transcripts is a complex process and may not be compared with transcription level regulation of genes. Therefore, the post-transcriptional regulation is of great significance for better characterization of functional role of genes (Picard et al., 2013). Although ESTs and transcriptome based experimental studies have provided valuable information in mining genes incited by various stress responses in Trichoderma interaction with plants and plant pathogens, its application is limited because the levels of the proteins and their encoding mRNA are not correlated to each other. Therefore considering the use of the cutoff standards in transcriptome based studies and appearance of artifacts in the differential expression of genes, translatome based studies offers potential choice and a better alternative involving only active mRNA populations (Picard et al., 2013; Yanguez et al., 2013; Piccirillo Ciriaco et al., 2014; King and Gerber, 2016; Meteignier et al., 2017).

Studies involving translational regulation of gene expression are emerging as a prominent tool for the understanding the regulation of protein abundance in adaptive responses of the host (Halbeisen and Gerber, 2009; Spriggs et al., 2010). In the genetic flow of information, the translational regulation reprograms the cell activities by protein synthesis. In last decade due to rapid advancements in technology, efforts on understanding the modulatory role of translation in gene expression have increased significantly. The translatome referring to the active mRNAs population associated with ribosomes has facilitated the removal of background noise and useful for the accurate determination of active mRNA. Originally used in oocytes and embryos (Terman, 1970; Gurdon et al., 1971), translational control has emerged as a key point of eukaryotes. The process is executed by loading of ribosomes on mRNA followed by translation elongation (Groppo and Richter, 2009; Jackson et al.,

2010). Since, the translatome based studies are focused only on the pools of genome-wide translated mRNA and therefore have helped in identification of key regulatory factors that are under translational control (Zupanic et al., 2013). This technique offers immense potential in the targeting key regulators which are active during interaction and play important role for the host plant in combating various stress responses. Translatome studies also help in determination of the ribosome number on active mRNA molecule in response to stress in the cellular genes (Koritzinsky and Wouters, 2007; Thomas and Johannes, 2007; Picard et al., 2013; **Figure 3**).

Presently, there are three methods used for translatome analysis; (a) polysomal profiling, (b) ribosomal profiling, and (c) ribosome affinity purification (RAP) (**Figure 3**). Polysomal profiling discovered in 1960s involves the separation of actively translated mRNAs bound by several ribosomes from free RNA by sucrose gradient centrifugation and then mRNAs can be coupled to northern blot or RT-qPCR or cDNA microarrays, or RNA-seq on a global level (Karginov and Hannon, 2013; Spangenberg et al., 2013). The second method known as ribosomal profiling was developed by Weissman group in Saccharomyces cerevisiae, determines the location of ribosomes at codon or nucleotide scale (Ingolia et al., 2009). The advantage of this technique is acquisition of information at global scale with respect to the position of the ribosomes on translated mRNA.

The deep nucleotide sequencing of ribosome protected RNA fragments obtained after RNase I treatment of cell lysate helps in accurate determination of ribosome position and its densities along RNA (Ingolia et al., 2012). Both polysome and ribosome based profiling need relatively large sample size to obtain enough RNA for microarray/RNA-seq analysis. The third method known as RAP developed by Inada et al. (2002) in S. cerevisiae capture monosomes and polysomes by using antiFLAG affinity resin. The RAP also known as translating RAP provides a better approximation of the translated mRNA population if coupled with transcriptome analysis (Halbeisen and Gerber, 2009; Jiao and Meyerowitz, 2010).

#### Integrating Translatome and Proteomic Study

The post-transcriptional events such as translation regulation and protein stability are the principle causes of weak correlations and variations in proteomic, transcriptomic, and genomic data. The associated errors in transcriptome analysis are subjected to arise from the suppression by microarrays which can further impede the identification of active candidate transcripts. On the other side, methods opted for protein staining, limitations associated in visualizing low-abundant and co-migrating proteins seriously hampers proteomic based study. The recent developments in proteomics methods such as use of mass spectrometric (MS) and liquid chromatography (LC) techniques have made quantitative proteomic profiling, currently a driving force for identification of proteins. The highly stable and reproducible performance of mass spectrometers such as Q Exactive hybrid quadrupole-Orbitrap mass spectrometer MS and Triple TOF 5600 MS is capable of identification of both proteomics (Chang et al., 2014) and characterization of bioactive metabolites. Integrated analyses of active mRNAs coupled with protein expression are available for bacteria, yeast, mice, and humans. Similar to transcriptome, the translatome based studies are focused only on transcripts level which are intracellular in nature. The coupling of multiomic approaches based on active mRNA, proteomes,

and protein turnover of both intra as well extracellular proteins and biologically active metabolites under different environmental conditions will provide a better answer of reprogramming biocontrol to various plant beneficial attribute and its resiliencies to combat different environmental conditions (**Figure 2**).

#### CONCLUSION

The availability of the fully sequenced genomes of Trichoderma spp. has accelerated our research on understanding of the behavior of different species of this genus and how the information on their gene pool determines their capabilities and limitations. The genomes of Trichoderma which is known to contain thousands of genes encoding different glycosyl hydrolases, secondary metabolites, antibiotics, lectins with insecticidal properties, and transporters with potential in bioremediation involved in antibiotics biosynthesis, and several other candidate genes (Druzhinina et al., 2012; Atanasova et al., 2013). Exploration of genes and their encoding proteins involved in developing tolerance against various stresses such as cold, below-average precipitation, salty conditions, pH, herbicide resistance as well biotic factor are an active field of research. The predicted genome of Trichoderma strains are known to encode a large number genes therefore coupling of translatome studies with proteomics of both extracellular and intracellular proteins offers a wide scope for better understanding the complex behaviors of Trichoderma as BCA.

The genomic comparison of mycoparasitic species of T. harzianum with non-mycoparasitic strains of T. reesei already provides evidences of the expansion of several genes in biocontrol strains. The secretion of a large number of cell wall targeting enzymes and bioactive secondary metabolites require adaptive molecular reprogramming of Trichoderma transcriptome. The variation at genomic, transcriptomics, and proteomic levels is a challenging task and difficult to correlate due to complex and non-systematic post-transcriptional and limitation of proteomic techniques. Further, the translational control is a widespread phenomenon with intense effect; nevertheless it is underestimated for its regulatory roles. In general, extensive uncoupling of both RNA movements and inferred cell activities has been observed for 19 different transcriptome and translatome. Therefore, coupled quantitative transcript and protein abundance studies can serve as a gold standard for proper and accurate depiction of interaction involving Trichoderma–plant–plant pathogens. Although

#### REFERENCES


detecting changes in the transcriptome level (total mRNAs), translatome level (ribosome loaded mRNAs) and the proteome is experimentally feasible in a high-throughput way, the integration of these omic technologies is still far away. Systematic global analyses aims at integrating transcriptome, translatome, and proteome level can provide accurate view of widespread adaptive mechanisms of interaction between Trichoderma– plant–pathogen.

In future, integrated efforts will help us to better understand, identify, and then explore the molecular behavior of Trichoderma arsenal involved in its success as BCAs as well as industrial sectors. In such instances, the integration of the translatome using ribosomal profiling and coupling it with proteomic approaches such as liquid chromatography-tandem mass spectrometry (LC-MS/MS) for both extracellular and intracellular proteins offers a lot of scope for accurate characterization of active molecular components involved in biocontrol and then subsequently their utilization of various applications.

### FUTURE DIRECTIONS

A comparative multiomic coupled insights of Trichoderma– plant–plant pathogens in three way interaction will play vital role in accurate characterization of transcripts responsible for cosmopolitan nature of Trichoderma and then targeting the promising one for agricultural based applications. The latest advancements and complete genome sequencing have already provided a platform of gene pool. Further integration with latest functional techniques such as translatome will lead another step close to identification of targets in the form of active transcripts involved in a complex interaction of plant–BCA–plant pathogens.

### AUTHOR CONTRIBUTIONS

VS and RS prepared the manuscript. PS and AG edited the manuscript.

### ACKNOWLEDGMENT

The authors are thankful to SERB, Department of Science and Technology-New Delhi India for providing funding under DST-FAST Track young scientist scheme (award letter NO. SB/FT/LS-365/2012).






brevicompactum under different culture conditions. PLoS ONE 9:e94203. doi: 10.1371/journal.pone.0094203



**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 © 2017 Sharma, Salwan, Sharma and Gulati. 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) or licensor 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.

# *Verticillium dahliae-Arabidopsis* Interaction Causes Changes in Gene Expression Profiles and Jasmonate Levels on Different Time Scales

Sandra S. Scholz <sup>1</sup> , Wolfgang Schmidt-Heck <sup>2</sup> , Reinhard Guthke<sup>2</sup> , Alexandra C. U. Furch<sup>1</sup> , Michael Reichelt <sup>3</sup> , Jonathan Gershenzon<sup>3</sup> and Ralf Oelmüller <sup>1</sup> \*

<sup>1</sup> Department of Plant Physiology, Matthias Schleiden Institute of Genetics, Bioinformatics and Molecular Botany, Friedrich-Schiller-University Jena, Jena, Germany, <sup>2</sup> Systems Biology and Bioinformatics Group, Leibniz Institute for Natural Product Research and Infection Biology—Hans-Knöll-Institute, Jena, Germany, <sup>3</sup> Department of Biochemistry, Max-Planck Institute for Chemical Ecology, Jena, Germany

#### *Edited by:*

Katarzyna Turnau, Jagiellonian University, Poland

#### *Reviewed by:*

Sotiris Tjamos, Agricultural University of Athens, Greece Sergio Casas-Flores, Institute for Scientific and Technological Research, Mexico

*\*Correspondence:*

Ralf Oelmüller ralf.oelmueller@uni-jena.de

#### *Specialty section:*

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

*Received:* 30 September 2017 *Accepted:* 30 January 2018 *Published:* 13 February 2018

#### *Citation:*

Scholz SS, Schmidt-Heck W, Guthke R, Furch ACU, Reichelt M, Gershenzon J and Oelmüller R (2018) Verticillium dahliae-Arabidopsis Interaction Causes Changes in Gene Expression Profiles and Jasmonate Levels on Different Time Scales. Front. Microbiol. 9:217. doi: 10.3389/fmicb.2018.00217 Verticillium dahliae is a soil-borne vascular pathogen that causes severe wilt symptoms in a wide range of plants. Co-culture of the fungus with Arabidopsis roots for 24 h induces many changes in the gene expression profiles of both partners, even before defense-related phytohormone levels are induced in the plant. Both partners reprogram sugar and amino acid metabolism, activate genes for signal perception and transduction, and induce defense- and stress-responsive genes. Furthermore, analysis of Arabidopsis expression profiles suggests a redirection from growth to defense. After 3 weeks, severe disease symptoms can be detected for wild-type plants while mutants impaired in jasmonate synthesis and perception perform much better. Thus, plant jasmonates have an important influence on the interaction, which is already visible at the mRNA level before hormone changes occur. The plant and fungal genes that rapidly respond to the presence of the partner might be crucial for early recognition steps and the future development of the interaction. Thus they are potential targets for the control of V. dahliae-induced wilt diseases.

#### Keywords: *Arabidopsis*, calcium, JA, defense, *Verticillium dahliae*

## INTRODUCTION

Vascular wilts caused by members of the genus Verticillium are among the most devastating fungal diseases worldwide, and these soil-borne ascomycete fungi attack a large variety of plant hosts in many parts of the world (Decetelaere et al., 2017) which leads to massive yield losses (Pegg and Brady, 2002). Among the 10 species within the Verticillium genus, Verticillium dahliae has the broadest host range with the ability to infect >200 plant species worldwide (Agrios, 1997; Inderbitzin et al., 2011; Inderbitzin and Subbarao, 2014). Verticillium species produce microsclerotia, which can survive in soil or dead plant material for more than 10 years, but they also form resting mycelia which survive in dead plant material. Newly growing hyphae rapidly penetrate the roots of their hosts, reach the vascular tissue and ultimately propagate in the xylem (Puhalla and Bell, 1981; Schnathorst, 1981). Initially, these hemibiotrophic fungi show biotrophic behavior that does not lead to severe reductions in plant performance. However, at later stages, they shift to a necrotrophic interaction characterized by the reprogramming of phytohormone metabolism (Veronese et al., 2003; Thaler et al., 2004; Tjamos et al., 2005), synthesis of hydrogen peroxide (H2O2) and nitric oxide (NO, Yao et al., 2011, 2012), and defense gene activation which ultimately induces host cell death (Reusche et al., 2012; Zhang et al., 2016). A typical symptom of these pathogens is wilting, which occurs as a consequence of impaired vascular transportation (Reusche et al., 2012).

Arabidopsis is an ideal model plant to study the Verticilllium infection process at the molecular level. Sun et al. (2014) showed that jasmonate phytohormones are highly induced upon V. dahliae infection, including the jasmonic acid (JA) pre-cursor cis-(+)-12-oxo-phytodienoic acid (cis-OPDA) and the jasmonic acid-isoleucine conjugate (JA-Ile). JA-Ile, synthesized by the enzyme JASMONATE RESISTANT 1 (JAR1), is the active signaling compound that binds to the JA receptor CORONATINE INSENSITIVE 1 (COI1) to initiate the downstream signaling cascade (Xie et al., 1998; Staswick and Tiryaki, 2004; Fonseca et al., 2009). Hydroxylation of JA-Ile by a P450 enzyme leads to the deactivation of the molecule (Koo et al., 2011). Previous observations on JA-deficient def1 tomato plants showed decreased fitness after V. dahliae treatment (Thaler et al., 2004), while external application of methyl JA (MeJA) to cotton and tomato plants partially blocked disease development due to the reduced growth of the fungus (Li et al., 1996). Recent RNAseq analysis in cotton showed that in addition to the biosynthesis of JA, also other components of the JA signaling cascade are targets of Verticillium. For instance, the gene of a repressor protein of COI1, GhJAZ10, was significantly upregulated in a V. dahliae-resistant cotton cultivar (Chini et al., 2007; Thines et al., 2007; Zhang W. W. et al., 2017). On the other hand, Verticillium itself needs an activated COI1 signaling pathway in the plant to cross the root-shoot barrier and to induce disease symptoms in the leaf. Consequently, V. dahliae-infected coi1 mutant plants showed less severe wilt symptoms (Ralhan et al., 2012).

Besides jasmonates, Verticillium-infested plants also accumulate other phytohormones like salicylic acid (SA) and ethylene (ET), (Fradin and Thomma, 2006; Ratzinger et al., 2009; Sun et al., 2014). SA plays an important role in plant defense against pathogens by activation of systemic acquired resistance (SAR) (Metraux, 2001). In Brassica napus the concentration of SA increased after infection with Verticillium longisporum in the xylem sap of the plants (Ratzinger et al., 2009). Also in Arabidopsis, a significant increase of SA could be detected in V. dahliae-infested plants (Sun et al., 2014). Exogenous application of SA protected cotton callus cells against a V. dahliae (VD)-toxin preparation by increase of chitinase and β-1,3-glucanase activities (Li et al., 2003; Zhen and Li, 2004). Interestingly, several Arabidopsis genotypes affected in different steps of SA signaling did not show altered resistance against V. dahliae infection (Veronese et al., 2003).

The role of ET in pathogen responses is still controversial although it was shown that ET can increase resistance and control symptom expression in some hosts. In soybean, tobacco and Aarabidopsis, ET is involved in host resistance against particular classes of pathogens (Knoester et al., 1998; Hoffman et al., 1999; Thomma et al., 1999), while a tomato mutant impaired in ET perception exhibited a significant reduction in disease symptoms after inoculations with different bacterial and fungal pathogens (Lund et al., 1998). Similarly, a tomato mutant with decreased ET biosynthesis showed significantly reduced symptom development (Robinson et al., 2001). The involvement of ET in response to Verticillium infection and VDtoxins has been shown repeatedly (Pegg and Cronshaw, 1976; Mansoori and Smith, 2005; Sun et al., 2014). Also the ET-resistant Arabidopsis mutant etr1-1 was found to display less chlorosis upon Verticillium inoculation (Veronese et al., 2003; Tjamos et al., 2005; Pantelides et al., 2010).

A growing number of other chemical compounds of both plant and fungal origin have been identified that participate in disease development of infected host plants. For Verticillium, a range of peptides, hormones like cytokinins, and other metabolites were shown to confer partial resistance to infection (Reusche et al., 2013; Bu et al., 2014; Gaspar et al., 2014; Roos et al., 2014). In addition, the Ve gene provides resistance against isolates of V. dahliae in tomato and Arabidopsis (Kawchuk et al., 2001; Fradin et al., 2009, 2011) and the VET1 gene was able to convey increased tolerance with milder chlorosis symptoms (Veronese et al., 2003).

Due to the vascular location of the growing fungus, Verticillium wilt of agricultural plants is difficult to control by chemicals or molecular tools. Even after removal of infected plants from agricultural fields, the resting structures (microsclerotia) remain in soil making them a hazard for future plantings (Fradin and Thomma, 2006). Biocontrol studies with bacterial and fungal isolates like Pseudomonas putida B E2, Pseudomonas chlororaphis K15, Serratia plymuthica R12 or Paenibacillus alvei K165 on Solanaceae, Malvaceae, and Brassicaceae have been shown to be an alternative strategy to restrict Verticillium-induced wilts (Berg et al., 2001; Tjamos et al., 2005; Li et al., 2012). Sun et al. (2014) showed that Piriformospora indica, a beneficial endophytic fungus of Sebacinales which colonizes the roots of many plant species, is an efficient biocontrol agent that restricts V. dahliae growth in the model plant Arabidopsis thaliana.

V. dahliae rapidly and efficiently colonizes Arabidopsis roots within 24 h post germination. In order to identify plant and fungal genes involved in these early recognition processes we performed an RNA-seq analysis under standardized cocultivation conditions. We found a massive reprogramming of both the plant and fungal expression profiles that occurred before pathogen induced, defense-related phytohormone levels changed in the host. This alteration in expression profiles clearly indicates that both plant and pathogen respond very rapidly to the presence of the other.

## MATERIALS AND METHODS

### Growth Conditions of Seedlings

A. thaliana wild-type (ecotype Columbia-0) seeds, and seeds of the jar1, coi1-16, and cyp94B3 mutants (kindly provided by Axel Mithöfer, MPI-CE, Jena) were surface-sterilized and placed on Petri dishes with MS media supplemented with 0.3% gelrite (Murashige and Skoog, 1962). After cold treatment at 4◦C for 48 h, plates were incubated vertically for 9 or 14 days at 22◦C under long day conditions (16 h light/8 h dark; 80 µmol m−<sup>2</sup> s −1 , for experimental setup, cf. Figure S1).

### Growth Conditions of Fungi and Preparation of Spore Solutions

V. dahliae wild-type (FSU-343, Jena Microbial Resource Center, Germany) and a GFP-labeled strain (IPP0860, kindly provided by Prof. Tiedemann, University of Göttingen) were grown for 1–2 weeks on Potato-Dextrose-Agar (PDA) medium (Bains and Tewari, 1987) at 23◦C in the dark. To obtain high spore production, the still white mycelia (with low amount of microsclerotia) were transferred to liquid KM medium (Hill and Kaefer, 2001) and incubated for 4–5 days at room temperature (RT) in the dark and 110 rpm. The cultures were filtered through two layers of a nylon membrane (75µm pore size), pelleted and washed with water. The spore concentration was determined with a hemocytometer and adjusted to 5 ×10<sup>6</sup> per ml.

#### Co-cultivation Assays

For short time co-cultivation assays, 9 day-old A. thaliana seedlings of equal sizes were used. Co-cultivation of A. thaliana and the fungus was performed under in vitro culture conditions on a nylon membrane on PNM medium (Johnson et al., 2011). Two days prior to use, 100 µl of the generated spore solutions (in water) or an equal amount of water were plated on six sterile membrane stripes (4 × 1 cm) placed on PDA plates and incubated at room temperature (Figure S1). For the co-cultivation the membrane stripes with fungus or water were placed on the PNM plates together with two plants per membrane. The plants were placed in the way that the roots were in contact with the fungus while the leaves were not (Figure S1). Plates were sealed with 3MTM Micropore tape and incubated for 24 h at 80 µmol m−<sup>2</sup> s <sup>−</sup><sup>1</sup> with light from one side (leaves directed to light). Roots and leaves were harvested separately for further analysis. All experiments were performed three times independently (≈120 seedlings per replicate).

For long term co-cultivation [10 and 20 days post infection (dpi)], 14 day-old A. thaliana seedlings of equal size were used. Plants were grown and infected as described above and transferred to Magenta boxes (Sigma-Aldrich, Germany) after 24 h. One plant was added per box which contained 30 g sterile vermiculite mixed with 100 ml liquid PNM medium. The boxes were incubated at 23◦C under short day conditions (9 h light/15 h dark; 80 µmol m−<sup>2</sup> s −1 ). After 10 or 20 dpi, plants were visually examined and photographed. The performance of the plants [n = 6 (control) and 9 (VD-infected)] was quantified based on the efficiency of the photosynthetic electron transfer measured with the Fluorcam as described before (Matsuo et al., 2015).

### RNA Isolation and RNA-Seq

RNA of both plant and fungus was isolated from 100 mg root material (≈ 120 seedlings per replicate, n = 3). The frozen roots were finely ground with mortar and pistil and weighed. The RNA from the samples was extracted with peqGOLD TriFastTM FL (VWR, Darmstadt, Germany) according to the manufacturer's protocol. The RNA was further processed by use of the PureLinkTM RNA Mini Kit (Thermo Fisher Scientific, Dreieich, Germany) with on-column DNAse treatment. RNA was dissolved in water and checked for quality. The isolated RNA was shipped to GeneCore (Heidelberg, Germany) where library construction was performed using the mRNA sequencing Sample Preparation Guide (Illumina, Cat#RS-930-1001), followed by a validation of the library and RNA sequencing (Hi Seq 2000, single-end 75 bp, ∼50 Mill. reads/sample). The removal of low quality reads and Illumina adapters was performed using Trimmomatic (Bolger et al., 2014). The remaining reads were then aligned to the A. thaliana (TAIR10.33) and V. dahliae (ASM15067v2.37) reference genomes using the RNA-seq aligner STAR (Dobin et al., 2013). Differential gene expression analysis was performed using DESeq2 (Love et al., 2014) with the raw counts obtained from FeatureCounts (Liao et al., 2014). Differentially Expressed Genes (DEGs) of both interaction partners were analyzed [FoldChange ≥ 2 and p-Value (FDR) ≤ 0.05]. The function of the DEGs was analyzed with the TAIR (www.arabidopsis.org) and the Fungi Ensembl (http://fungi.ensembl.org/Verticillium\_dahliae) databases. Pathway analysis for the plant and the fungus was executed with the KEGG (Kyoto Encyclopedia of Genes and Genomes) Mapper tool (http://www.genome.jp/kegg/tool/map\_ pathway2.html; RRID:SCR\_012773).

### Analysis of Gene Expression and Fungal Colonization

The RNA (n = 3) was isolated as described above. One µg of RNA was transcribed to cDNA using the Omniscript RT Kit (Qiagen, Hilden, Germany). Fifty nanograms of synthesized cDNA was used as template for RT-qPCR in a CFX ConnectTM Real-Time PCR Detection System (Bio-Rad, Munich, Germany) with the Brilliant II SYBR <sup>R</sup> Mastermix (Agilent, Böblingen, Germany). The mRNA levels for each cDNA probe were normalized with respect to the RPS18B (plant) or VD\_Actin2 (fungus, e.g., Klimes and Dobinson, 2006; Yang et al., 2013) mRNA levels. The normalized fold expression of GOIs was calculated according to 11CP (Pfaffl, 2001). The primer pairs are given in Table S5. To analyze fungal colonization of different mutant plants, normalized fungal housekeeping gene expression was compared to that in WT plants.

#### Confocal Microscopy

Samples for confocal laser scanning microscopy were prepared according to the method for the short time co-cultivation described above. The GFP-labeled V. dahliae strain on Arabidopsis seedlings was imaged using a LSM 880 (Zeiss Microscopy GmbH, Jena, Germany) with the 488 nm laser line of an argon multiline laser (11.5 mW). Images were taken with a 40x objective (Plan-Apochromat 40x/0.8). Lambda stacks were created using the 32 channel GaAsP detector followed by Linear Unmixing with ZEN software (Zeiss, Jena, Germany). Z-stacks were taken from specific areas of the sample and Maximum Intensity Projections were produced with ZEN software. Crosssections of the roots with a width of 14µm were done with a Microm HM560 Cryostar (Southeast Pathology Instrument Service, Charleston, USA).

#### Quantification of Phytohormones

Phytohormones were extracted from the green parts of co-cultivated seedlings (1 sample = seedlings from 1 plate, ≈ 50–100 mg, n = 10). Frozen samples were homogenized for 30 s at 1000 rpm in a 2010 Geno/Grinder <sup>R</sup> (Spex SamplePrep, Stanmore, UK) and mixed with 1 ml methanol containing 40 ng/ml of D6-JA, D6-ABA, D4-SA, and 8 ng/ml of D6-JA-Ile (Scholz et al., 2017). All samples were shaken for 30 min at 4◦C and centrifuged at 13,000 rpm for 20 min at 4◦C. The supernatants were collected and the sample re-extracted with 500 µl methanol. The combined supernatants were evaporated to dryness at 30◦C using a vacuum concentrator. Residues were resuspended in 200 µl methanol and centrifuged at 13,000 rpm for 10 min. The supernatants were collected and measured with the API 5000 LC-MS/MS system (Applied Biosystems, Framingham, USA) as previously described (Vadassery et al., 2012). Since it was observed that both the D6-labeled JA and JA-Ile contained 40% of the corresponding D5-labeled compounds, both peaks were combined for analysis.

#### Statistics

Experiments were repeated three times to ensure reproducibility and 120–150 seedlings were used in each treatment for each mutant. Data of all independent experiments were pooled and analyzed. For comparison of two groups, the Mann Whitney U-test was applied. For statistical analyses of multiple groups, one-way analysis of variance (one-way ANOVA) was used as indicated in the figure legends. Different letters indicate significant differences between treatments. SigmaPlot13 and Origin Pro were used for data analysis and graph composition.

## RESULTS

#### Co-cultivation of *Arabidopsis* Seedlings with *V. dahliae* for 24 H

Verticillium species are considered as hemibiotrophs, where a biotrophic phase—within the root xylem without a visible disease phenotype—is followed by a necrotrophic phase in the aerial parts of the plant (Reusche et al., 2012). We focused on the very early phase of infection, the pre-vascular growth phase, and analyzed the plant and fungal expression profiles during the first 24 h of fungal root colonization in Arabidopsis. By developing a stable co-cultivation method (scheme in Figure S1), a reproducible colonization of the seedlings by the fungus was achieved. Confocal microscopic pictures taken 24 h after coculture demonstrate that Arabidopsis roots are already heavily colonized by the GFP-labeled V. dahliae (**Figure 1**). The hyphae of the fungus form a net over the root surface and root tip while the first spores can be observed at the site of lateral root formation (**Figures 1G–I**). In cross-sections of the colonized root (**Figures 1J–O**) it can be observed that the fungus penetrates the root surface, but did not yet invade the vascular tissue after 24 h of co-culture.

#### Dual-RNA-Seq of *Arabidopsis-Verticillium* Co-culture Reveals High Number of Differentially-Expressed Genes (DEGs)

Recent studies of plant-Verticillium co-cultures focus on analysis of the plant transcriptome by RNA-seq and indicate major changes in nearly 19% of pathways in early phases of infection (1 or 4 dpi, e.g., Faino et al., 2012; Zhang W. W. et al., 2017). The RNA-seq data generated in this study 24 h after co-cultivation was analyzed for changes in both the plant and fungal transcriptomes and revealed a total number of 4432 DEGs for both organisms (**Table 1**). Compared to Arabidopsis seedlings grown alone, co-cultivation with Verticillium results in 1143 DEGs, 903 of them are significantly up-regulated and 240 down-regulated (Tables S1, S2). For the fungus, 3289 DEGs were detected with 1695 significantly up-regulated (Table S3) and 1594 significantly down-regulated genes (Table S4). This suggests that the fungal expression profile responds more strongly to the presence of a host than the plant genome to a pathogen. The results for selected genes were verified by RT-qPCR (**Figures 7**, **8**, WT).

#### Multiple *Arabidopsis* Pathways Are Affected by *V. dahliae* Infection

Previous studies during longer co-cultivation times have shown that infection of Arabidopsis plants with Verticillium species results in disruption of water transport and massive accumulation of phytohormones like jasmonates (JAs), ethylene (ET), salicylic acid (SA), or abscisic acid (ABA), as well as stomata closure (Pegg and Brady, 2002; Fradin and Thomma, 2006; Klosterman et al., 2009; Ralhan et al., 2012; Sun et al., 2014). To identify early targets of the fungus in Arabidopsis and to analyze and classify the plant DEGs, we chose the 24 h cocultivation time point and mapped the identified plant DEGs to distinct pathways in the KEGG database (Kanehisa, 1997). In total, 78 pathways were affected by the fungus (**Table 2** and **Figure 2A**). Metabolic pathways involved in carbon and amino acid metabolism seemed to be the major targets, such as SWEET genes and sugar efflux transporters, which often respond to pathogens and symbionts for nutritional gain (Chen et al., 2010). In our study, SWEET11, −3, and −12 were down-regulated. Genes involved in plant development were also affected. So was ROOT CAP POLYGALACTURONASE28, a crucial player in root tip growth, which was also down-regulated at the mRNA level (Kamiya et al., 2016). Defense-related genes, genes involved in the synthesis and propagation of defense signals like JA, SA, NO, and reactive oxygen species (ROS), membrane-associated receptor kinases are up-regulated while pathways involved in or related to photosynthesis, carotenoid and flavonoid biosynthesis are down-regulated. We observed the activation of genes for various defense-related WRKY transcription factors (WRKY18, −28, −30, −33, −41, −45, −53, −55, −71). It has been shown that WRKY71 promotes shoot branching, acceleration of flowering and cell death (cf. references in TAIR, www.Arabidopsis.org); WRKY33 is a target of Botrytis to repress defense in Arabidopsis (Liu et al., 2017), WRKY53 is involved in disease resistance against Verticillium longisporium (Reusche et al., 2013), and WRKY30 confers general abiotic stress response to the plant (Scarpeci et al., 2013). Upregulation of these genes suggest that the plant reprograms its metabolism for defense. Also candidate genes for the perception of pathogen-associated molecular patterns, and enzymes as well as signaling components which might lead to the activation


TABLE 2 | KEGG pathway classification of genes differentially expressed in Arabidopsis alone vs. co-culture with V. dahliae.


of defense compounds in the host were upregulated in the presence of the fungus. This includes six genes for receptorlike protein kinases (RLP7, −15, −19, −20, −35, −38) and seven genes for proteins with TIR-NBS-LRR domains involved in disease resistance responses (At1g57630; At1g66090; At5g45240; At5g41750; At1g63750; At1g72900, At1g72920). For the RLP genes, no clear function has been described yet. The mRNAs for the Toll and Interleukin-1 Receptor (TIR)-Nucleotide-Binding Site (NBS)-Leucine-Rich Repeat (LRR) proteins (TIR-NBS-LRR), At1g66090 and At1g72920, have been shown to travel to distant tissue upon stress and thus might be involved in systemic signal propagation (Thieme et al., 2015). Furthermore, numerous genes for cytochrome P450 enzymes responded to the fungus. Of these, 17 are up-regulated and 11 have a described role in plant defense (cf. TAIR). Two of the three downregulated genes (CYP87A2 and CYP705A12) are involved in cytokinin signaling (cf. TAIR). Many of the cyp mRNAs are also known to be mobile within the plant. Finally, defense gene activation is often mediated via Ca2<sup>+</sup> signaling and Ca2+ binding proteins. The majority of genes that code for proteins with Ca2+-related functions are involved in signaling, ion uptake and distribution. Many of these are also up-regulated in Arabidopsis upon V. dahliae infection. At an early phase of interaction with the pathogen, the plant appears to shift its resources from primary metabolism to defense processes (cf. Discussion).

Interestingly, two genes for SNARE proteins (soluble Nethylmaleimide-sensitive-factor attachment receptor, At5g39630, and At1g16225) also responded to the fungus. SNARE proteins are involved in vesicle trafficking and membrane fusion and deliver defense products to infection sites during exocytosisassociated immune responses (Wang et al., 2017, and references cited therein). Stimulation of EXO70H4 and EXO70A3 for EXOCYST subunits is consistent with the stimulation of exocytosis-mediated callose deposition and cell wall maturation by the fungus (Li et al., 2010; Kulich et al., 2015).

#### *V. dahliae* Pathways Respond to the Interaction with *Arabidopsis*

Many genes annotated in the 117 fungal pathways that were affected by co-cultivation with Arabidopsis (**Table 3** and **Figure 2B**), are still uncharacterized or with unknown function, which limits our analysis on the fungal side. However, like on the plant side, the majority of the affected pathways involved in primary metabolism, e.g., genes for the amino acid metabolism, were highly down-regulated, while those for sugar metabolism and sugar transport processes were up-regulated. This suggests that the fungus starts quite early to reprogram its primary


(Continued)

TABLE 2 | Continued


metabolism and adapts amino acid and sugar metabolism to being inside a host. mRNA for several enzymes employed in degradation of the plant cell wall, such as the cell wall glycosyl-hydrolase YteR (Moore et al., 2016), were up-regulated as shown previously. Genes involved in processes such as oxidative phosphorylation, ribosome formation, RNA transport and degradation or proteasome functions were preferentially down-regulated in the co-cultivated fungus. Closer inspection of the data supports the idea that the fungus down-regulates genes for essential processes (primary sucrose, N and P metabolism, ion homeostasis, redox processes, defense, and secondary metabolites) perhaps because of its increasing reliance on the plant. Reduced defense gene activation may indicate that the fungus prevents the synthesis of compounds that restrict its growth and propagation in the host.

### *V. dahliae* Growth Is Reduced in *Arabidopsis* JA Mutants

Former studies indicate that the accumulation and perception of jasmonates are key events for both plant and fungal responses to the interaction. The plant defense machinery is activated by an elevation of jasmonates and activation of the receptor COI1, while the fungus activates the plant COI1-dependent JA pathway to induce disease symptom development in the host (Feys et al., 1994; Xie et al., 1998; Ralhan et al., 2012). Upregulation of genes involved in JA biosynthesis and responses within the first 24 h of co-cultivation demonstrates that the course is already set even before significant changes in the plant hormone levels can be detected (cf. below). We observed several DEGs involved in the α-linolenic acid pathway as well as in plant hormone signal transduction pathways (**Table 2** and **Figure 2A**). While the expression of growth-associated genes like the auxin-responsive genes ARF5 (Krogan et al., 2016) and several members of the GH3-family (e.g., GH3.17, GH3.4, DFL2, WES1, Staswick et al., 2005) was decreased, an upregulation of the JA biosynthetic genes LIPOXYGENASE 3 and 4 (LOX3 and 4, Acosta and Farmer, 2010; Umate, 2011) and OXOPHYTODIENOATE-REDUCTASE 3 (OPR3, Müssig et al., 2000) as well as of an ET response factor (ERF1, Fujimoto et al., 2000) was observed.

To further investigate the role of JA in the Arabidopsis-V. dahliae interaction, we analyzed the phenotype and vitality of Verticillium-infected JA mutants jar1 (Staswick et al., 1992; Staswick and Tiryaki, 2004), coi1-16 (Ellis and Turner, 2002), and cyp94B3 (Koo et al., 2011). Although there was no visual difference between the mutants and the WT plants 10 dpi, the WT plants were dead after 20 dpi, while the mutant plants showed disease symptoms in leaves but were still vital and alive (**Figure 3**). This result was confirmed by analysis of the efficiency of the electron transfer during photosynthesis using chlorophyll fluorescence measurements, a sensitive marker for plant vitality (**Figure 4**). At 10 dpi, there were only small differences between the mutants and the WT, and chlorophyll fluorescence values around 0.81 indicate that the plants were capable of photosynthesis. At 20 dpi, the JA mutants still possessed a fluorescence value between 0.81 and 0.83, while WT plants were at 0.24 corresponding to very low levels of electron transport and photosynthetic efficiency. Thus, these non-invasive measurements provide an efficient tool to determine and quantify disease development in Verticilliuminfected hosts. To obtain further insight into the colonization efficiency of V. dahliae, the content of fungal RNA—in the plants previously analyzed for chlorophyll fluorescence—was determined in roots and shoots separately, and the levels

were compared to those in WT plants (Figure S2). The colonization pattern at the two tested time points was very different. While coi1-16 and cyp94B3 roots showed a significantly lower colonization at 10 dpi compared to WT (38 and 30%, respectively), jar1 roots showed an intermediate level with 83% (Figure S2A). The respective shoots of the plants show the same trend (Figure S2B). At 20 dpi, no difference in colonization of the mutant shoots was detectable; all of them showed significantly lower fungal RNA levels compared to the WT (Figure S2D). The levels of fungal RNA in the roots of coi1- 16 plants was significantly lower compared to those of all other genotypes.

TABLE 3 | KEGG pathway classification of genes differentially expressed in Verticillium alone vs. co-culture.

#### TABLE 3 | Continued


(Continued)

(Continued)


### Root Colonization and Plant Hormone Levels Are Not Altered after the First 24 h of Co-culture with *V. dahliae*

To compare the colonization of the mutant plants in the very early phase, the content of fungal RNA was also analyzed 24 h after co-cultivation (**Figure 5**). All plants showed a similar fungal colonization with no statistically significant differences (**Figure 5A**). Compared to the WT colonization level (set as 100%), jar1 plants showed a colonization of 90.0%, coi1-16, and cyp94B3 of 121.4 and 129.8%, respectively (**Figure 5B**). These results suggest that there are also no differences in plant JA levels during the first 24 h of co-culture.

In a previous study we demonstrated that V. dahliae infection leads to an elevation of JA, JA-Ile, and SA 21 dpi (Sun et al., 2014). To analyze whether the accumulation of these phytohormones was already induced in the first 24 h, leaf tissue of WT and the mutants was analyzed, but the content of SA (**Figure 6A**) and JA-Ile (**Figure 6C**) were found not to be significantly induced by the fungus. There was also no difference in the JA content in WT, coi1-16 and cyp94B3, while a slight difference was observed in jar1 plants. Accumulation of JA in this mutant is caused by the inactivation of the Ile-conjugating enzyme. These data confirm that the changes in the gene expression profiles described above occur before the fungus induces changes in the plant phytohormone levels. Thus, the identified genes respond most likely to fungal signals and not to fungus-induced phytohormone changes in the plant. JA signaling in the host is necessary for disease development only during later phases of the interaction, which can be strongly repressed or retarded when JA function is impaired.

#### Gene Expression in JA Mutants Is Different to That in the WT in the Early Phase of Colonization

To validate the RNA-seq results obtained, the expression of selected regulated genes in the data set was analyzed by RTqPCR for co-cultivated WT plants (**Figure 7**) and for the fungus (**Figure 8**). Highly regulated members of the SWEET family (Chen et al., 2010) and a Wall-Associated Kinase-like gene (WAKL10, Meier et al., 2010) were chosen for the plant and VDAG\_02979 (a glucose transporter) and VDAG\_06565 (poly-A-ribonuclease, PARN) for the fungus. In the RNA-seq analysis, the expression of WAKL10 was 56-fold induced in V. dahliae-infested plants compared to plants grown alone. A 10 fold induction was observed in the RT analysis (**Figure 7A**). SWEET11 and -3 were 20- and 10-fold downregulated in the RNA-seq data set, respectively, while a 9.7- and 6.8-fold decrease was detected by RT-qPCR (**Figures 7B,C**). Interestingly, the expression of these genes was differently regulated in the JA mutants. The expression of WAKL10 was induced ∼100-fold in jar1, i.e., significantly higher than in WT seedlings, while SWEET3, which is down-regulated in the WT, was induced 17-fold in the mutant (**Figure 7**). Additionally, SWEET11 was less repressed in the mutants: for jar1, the repression was 50% lower compared to WT plants. This clearly indicates that the expression of interaction-specific plant genes is influenced by JA content.

The RNA-seq results for the fungal genes were also validated by RT-qPCR analysis (**Figure 8**). The mRNA levels of VDAG\_02979 (RNA-seq: + 200-fold; RT-qPCR: + 120-fold) and VDAG\_06565 (RNA-seq: + 3-fold; RT-qPCR: + 27-fold) in response to the fungus showed the same trend. Again, when cocultivation was performed with the JA mutants, the expression of both genes was less induced compared to WT plants (**Figure 8**). Thus, both partners require plant JA for the typical regulation of genes that respond to the symbiotic interaction.

### DISCUSSION

#### *V. dahliae* Is in the Pre-vascular Growth Phase 24 h after Co-cultivation with *Arabidopsis*

The propagation of Verticillium species within their hosts is characterized by a biphasic interaction: initially a biotrophic phase allows rapid growth of the microbe in the xylem without major effects on plant performance followed by a necrotrophic phase in which the host is ultimately killed (Reusche et al., 2012). Analysis of the initial contact and early phase of the interaction between both partners could lead to the identification of target genes that will help understand further phases of the interaction and provide useful targets for pest control. Confocal analysis 24 h after co-cultivation demonstrated that the pathogen colonizes the root efficiently (**Figure 1**) by developing a dense hyphal network at the root surface and root tip. In areas of lateral

then transferred to the boxes.

root formation, an accumulation of conidia spores was detectable (**Figures 1G–I**). This matches previous observations that showed the lateral root to be a primary area of Verticillium infection in diverse host plants (Zhou et al., 2006; Vallad and Subbarao, 2008; Zhao et al., 2014). Further analyses of root cross-sections demonstrated that the fungus invades the plant tissue but did not reach the vascular system within the first 24 h of co-culture (**Figures 1J–O**). We conclude that the fungus was still growing in the pre-vascular phase during our experiments, optimal for the identification of genes regulated during the very early interaction phase. During this period we also did not observe significant differences in phytohormone levels, which is consistent with the microscopic information.

#### Initial Contact of *Arabidopsis* and *V. dahliae* Leads to Reprogramming of Primary Metabolism in Both Organisms

Plant-pathogen interactions involve an adaptation of both partners. While the plant typically recognizes the pathogen and induces appropriate defense responses, the fungus manipulates the biology of its host to gain sufficient nutrients for growth and reproduction (Boyd et al., 2013). To identify early targets in the Arabidopsis—V. dahliae interaction, an RNA-seq analysis after 24 h of co-cultivation was performed for both partners. This time point was chosen because no structural changes were detected in plant and fungal tissues under the microscope, and because the phytohormone levels in the plant were not yet altered, thereby avoiding secondary effects due to the response of the plant genome to an altered phytohormone environment. We detected 4432 DEGs for both organisms (**Table 1,** Tables S1–S4), 1143 DEGs in Arabidopsis belong mainly to primary metabolic pathways (21%, **Figure 2A**). This result is comparable to a recent study in cotton where 26% of the DEG reads could be assigned to metabolic pathways (Zhang W. W. et al., 2017). Comparably, in both studies, pathways associated with plant-pathogen interaction (3%) and plant hormone signal transduction (2%) were also regulated upon Verticillium stress.

Several members of the plant sucrose efflux transporters of the SWEET family were highly regulated in the analyzed interaction.

SWEET11 as well as –3, −12, −15, and –8 were downregulated in decreasing order (Table S2). SWEET11 and −12 are involved in phloem loading with sugars (Chen et al., 2010, 2012; Boyd et al., 2013). By silencing of OsSWEET11 in rice it could be demonstrated that the growth of Xanthomonas oryzae pv. oryzae (Xoo) is decreased which resulted in more resistant plants (Yang et al., 2006). Also SWEET3 was reported to be involved in defense and was downregulated after Geminivirus infection (Ascencio-Ibá-ez et al., 2008). These observations suggest that the plant is trying to prevent export of its sugars to restrict fungal growth.

Besides the defense genes discussed above, members of the WAK family are also highly represented in our data sets. WAK-L10 and WAK3, –5, −1, and −4 responded to Verticillium treatment (in declining order, Table S1). WAK1 is induced upon infection with Ps. maculicola ES4326 and the protein stimulated PR1 expression (He et al., 1998). Additionally, WAK1, −2, −3, and −5 are inducible by SA which accumulated during pathogen infection (He et al., 1999). Likewise, the WAK-like gene WAKL10 is SA-induced and involved in defense against bacteria and fungi (Meier et al., 2010). Among the genes involved in controlling Ca2<sup>+</sup> homeostasis, genes for in- and efflux transporters of the cyclic nucleotide-gated channel (CNGC), autoinhibited Ca2+-ATPase (ACA), and glutamate-like receptor (GLR) protein families responded to the fungus, indicating that many signaling events induced during the early phase of the interaction are Ca2+-dependent. Consistent with this idea, genes for Ca2+-binding proteins involved in signal perception and propagation showed increased expression, especially members of the calmodulin-like (CML) protein family. However, closer inspection of the genes did not allow any meaningful conclusion about which pathways are major targets of the fungus. Besides regulation of defense responses, many genes for Ca2+-binding proteins are involved in ion homeostasis, enzyme activity control, and biotrophic plant/microbe interactions (for details compare genes in Table S1 with the TAIR database). An induction of defense genes is always accompanied by a lower investment in plant growth (Huot et al., 2014). This is reflected by the down-regulation of ROOT CAP POLYGALACTURONASE28 and enzymes involved in cytokinin signaling. The phase of reprogramming of plant primary metabolism lasts for several days, as in Verticillium-infected tomato plants where it was shown that both gene expression and protein synthesis of metabolic pathways proteins are still down-regulated at 7 dpi (van Esse et al., 2009; Witzel et al., 2017). The global transcriptome analysis performed in this study provides significant insights into components involved in early phases of the Arabidopsis—V. dahliae interaction and suggests a number of genes and pathways that could be employed as markers in breeding for wilt tolerance.

In contrast to previous studies focusing on analysis of the plant's reaction to the fungus (van Esse et al., 2009; Faino et al., 2012; Witzel et al., 2017; Zhang W. W. et al., 2017), we present also DEGs from the fungus caused by response to the plant. The majority of the 3289 DEGs code for proteins with unknown

VD\_Actin2 in VD-infected WT, jar1, coi1-16 and cyp94B3 plants (A) and the relative expression in % (B). The expression level of VD\_Actin2 in VD-infected WT plants was used as control and set to 1.0. The mRNA levels for each cDNA probe were normalized with respect to the RPS18B mRNA level. Statistically significant differences between the mutants were analyzed by one-way ANOVA, p < 0.05 (Sidak). Different letters indicate a statistically significant difference.

functions (Tables S3, S4). The up-regulation of VDAG\_02979, which encodes for a glucose transporter, suggests that it might play an important role in the early interaction. Besides this initial observation, a huge number of identified genes belong to families which might become important targets after elucidation of their function in V. dahliae. Therefore, this list might provide useful information for genes with interaction-specific functions, in particular since the knowledge about the function of V. dahliae genes is strongly increasing. For example, very recent studies revealed that homeodomain and bZIP transcription factors (Fang et al., 2017; Sarmiento-Villamil et al., 2017), two less characterized transcription factors (Sarmiento-Villamil et al., 2010; Zhang W. Q. et al., 2017), polyketide synthases (Zhang

T. et al., 2017), endochitinases (Cheng et al., 2017), a novel V. dahliae protein that targets the plant nucleus (Zhang L. et al., 2017), LysM effectors (Kombrink et al., 2017), a isochorismatase hydrolase (Zhu et al., 2017), a factor involved in the fungal

FIGURE 7 | Expression of selected genes in Arabidopsis WT and JA mutant plants after 24 h of V. dahliae infection. Shown is the normalized fold expression (± SE, n = 3) of WAKL10 (A), SWEET11 (B), and SWEET3 (C) in VD-infected WT, jar1, coi1-16, and cyp94B3 plants. The expression level of genes of interest (GOIs) in water-treated plants was used as control and set to 1.0. The mRNA levels for each cDNA probe were normalized with respect to the RPS18B mRNA level. Statistically significant differences between the mutants were analyzed by one-way ANOVA, p < 0.05 (Sidak). Different letters indicate a statistically significant difference.

Arabidopsis WT and JA mutant plants for 24 h. Shown is the normalized fold expression (± SE, n = 3) of VDAG\_02979 (A) and VDAG\_06565 (PARN) (B) in VD-colonized WT, jar1, coi1-16, and cyp94B3 plants. The expression level of GOIs in VD alone was used as control and set to 1.0. The mRNA levels for each cDNA probe were normalized with respect to the VD\_Actin2 mRNA level. Statistically significant differences between the mutants were analyzed by one-way ANOVA, p < 0.05 (Sidak). Different letters indicate a statistically significant difference.

secretory pathway (Xie et al., 2017), a RACK1-like protein involved in root entry (Yuan et al., 2017), pathogenesis-related exudated proteins (Chen et al., 2016), and the mitogen-activated protein kinase 2 (Tian et al., 2016) are important components in controlling V. dahliae-induced disease development in various plant species. Members of all these protein families can be found in the list of V. dahliae genes up-regulated after infection of Arabidopsis.

#### JA Mutants Are Less Susceptible to *Verticillium* Infection

Various studies have shown the involvement of several classes of plant hormones in the control of Verticillium growth and propagation in Arabidopsis. While ET perception mutants are more susceptible to Verticillium infection, an elevation of cytokinins enhances plant resistance (Pantelides et al., 2010; Reusche et al., 2013; Sun et al., 2014). Downregulation of plant genes involved in cytokinin signaling might therefore be induced by the fungus. Interestingly, jasmonates are not only accumulated by the plant to induce defense, but the fungus also requires a JA-independent COI1 function in roots to elicit disease symptoms in Arabidopsis shoots (Ralhan et al., 2012). To further analyze the role of JA levels in the interaction of Arabidopsis and V. dahliae, JA biosynthesis (jar1), perception (coi1-16) and degradation (cyp94B3) mutants were studied. All of these mutants performed better, showed less severe disease symptom development in the leaves at 20 dpi compared to the WT control, which were already dead at this time point (**Figure 3**), and had a higher photosynthetic potential, as demonstrated by their QY\_max value above 0.80 (Fv/Fm, **Figure 4**) (e.g., Kim et al., 2009; Sztatelman et al., 2015). The observation, that jar1 plants performed better than WT plants, contradicts earlier findings where jar1 plants were as susceptible as WT plants (Fradin et al., 2011). To analyze this contradiction, the colonization of the mutants used was compared to WT plants (Figure S2). While there was no difference to the WT colonization level after 10 dpi in the roots and the shoots, there was a significant difference at 20 dpi. The colonization level of the root was similar to WT while the colonization in the shoots was significantly lower in jar1 (Figure S2D). This observation could explain the better performance of the aerial parts of the jar1 plants. Taken together, the altered expression of the interactionspecific genes of plant and fungal origin in the JA mutants confirms the important role of this phytohormone also during early phases. Apparently, the altered expression profile occurs before a significant change in phytohormone levels become detectable.

To gain insight into the growth behavior of V. dahliae on JA mutants during the first 24 h of co-cultivation, the colonization of the roots was analyzed. The detected differences in the colonization (**Figure 5**) were not significant at this time point, but may have a greater impact in later phases where a clear difference is obvious. After 24 h of co-cultivation, there was also no detectable difference in the levels of the phytohormones SA, JA and JA-Ile (**Figure 6**). Since changes in phytohormone levels upon pathogen attack are normally very rapid in plants, Arabidopsis might have not yet recognized the microbe as friend or foe, in spite of the already initiated reprogramming of its gene expression pattern. It is also conceivable that the penetration rate is still too low to induce the accumulation of JA and JA-Ile, since this is often associated with wounding or pathogeninduced cell disruption, which was not visible in our microscopic studies (e.g., Suza and Staswick, 2008; Koo et al., 2009). The biphasic interaction with an initial biotrophic period followed by a necrotrophic period may also leave the plant undecided whether it responds to the pathogen with SA- or JA-dependent defense strategies. Furthermore, both phytohormones cross-talk (Thaler et al., 2012; Proietti et al., 2013).

The low content of active JA-Ile in jar1 plants and the reduced perception in the receptor mutant coi1-16 lead to a decreased activation of the downstream signaling pathway by the receptor complex SCFCOI1 (Thines et al., 2007). Since Verticillium propagation from the roots to the leaves depends on an activated COI pathway (Ralhan et al., 2012), this could be a reason why Verticillium causes a reduced leaf growth rate of the JA mutant plants (**Figure 3**). The reduced spread in the green parts of the JA mutants is also reflected in the gene expression analysis of chosen genes for both the plant and the fungus (**Figures 7**, **8**). In the mutant plants, the SA-induced defense gene WAKL10 is more highly expressed than in WT plants and the downregulation of SWEET11 is less pronounced (**Figure 7**). In the fungus, both target genes VDAG\_02979 and VDAG\_06565 are more weakly expressed in the mutants than in the WT plants (**Figure 8**). VDAG\_02979 codes for a glucose transporter (http:// fungi.ensembl.org/Verticillium\_dahliae), which contributes to the nutrient supply of the fungus. The lower growth rate in the mutants may be a consequence of this transporter being less expressed.

In conclusion, biotrophic plant-microbe interactions are characterized by the stimulation of SA, but not JA levels, while the opposite hormone regulation occurs during nectrophophic interactions (reviewed in Chanclud and Morel, 2016). Within the first 24 h of interaction studied here, none of these phytohormone levels increase significantly, while hormone-synthesis related genes, as well as defense-related genes responding to both hormone types are already up-regulated in the host. This suggests that no clear decision has been taken yet about which strategy will follow initial contact. The numerous genes identified during early reprogramming of the fungal and plant development might be crucial for the initiation and propagation of the pest, and thus may be helpful for developing strategies which potentially restrict fungal development after infection. Considering our results, identification of crucial players which control the interaction at early stage is apparently difficult, because many metabolomic pathways are already re-adjusted within the first 24 h of the contact of the two partners.

The strong retardation of disease symptom development in host plants impaired in jasmonate mutants has been attributed to the fact that Verticillium stimulates the host JA functions in order to promote host cell death during the later necrotrophic phase. There is a substantial crosstalk between JA and SA signaling in which each hormone inhibits the accumulation and/or function of the other. Complete or strong inhibition of JA functions in the mutants may favor SA accumulation and/or SA signaling function which—in turn—may prolong the biotrophic phase and thus retard necrosis and disease development.

### AUTHOR CONTRIBUTIONS

SS: designed and performed the experiments; WS-H: analyzed the RNA-seq data and DEGs; AF: prepared, analyzed, and evaluated samples for confocal microscopy; MR: analyzed the phytohormones; RG, JG, and RO: supervised the projects; RO: coordinated the project; SS, JG, and RO: wrote the manuscript; All authors read and approved the final manuscript.

#### DATA ACCESS

Raw and calculated RNA-seq data was submitted to NCBIs Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) under the accession number GSE104590.

#### ACKNOWLEDGMENTS

We like to thank Anna-Sophie Enke for technical assistance and Julia Starke for pre-experiments and set-up of the

#### REFERENCES


cultivation method. Special thanks go to Dr. Vladimir Benes and co-workers at GeneCore (http://genecore3.genecore.embl. de/genecore3/index.cfm) for RNA sequencing. This work was supported by the DFG (CRC1127, projects A03 and INF).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00217/full#supplementary-material


WRKY33-mediated host immunity. Plant Cell Environ. 40, 2189–2206. doi: 10.1111/pce.13022


tomato expressing a bacterial ACC deaminase. Mol. Plant Pathol. 2, 135–145. doi: 10.1046/j.1364-3703.2001.00060.x


K-165 and pathogenesis-related proteins gene expression. Mol. Plant Microbe. Interact. 18, 555–561. doi: 10.1094/MPMI-18-0555


Zhu, X., Soliman, A., Islam, M. R., Adam, L. R., and Daayf, F. (2017). Verticillium dahliae's isochorismatase hydrolase is a virulence factor that contributes to interference with Potato's Salicylate and jasmonate defense signaling. Front. Plant Sci. 8:399. doi: 10.3389/fpls.2017.00399

**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 © 2018 Scholz, Schmidt-Heck, Guthke, Furch, Reichelt, Gershenzon and Oelmüller. 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 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.

# Metabolomics Investigation of an Association of Induced Features and Corresponding Fungus during the Co-culture of *Trametes versicolor* and *Ganoderma applanatum*

#### *Edited by:*

Erika Kothe, Friedrich Schiller Universität Jena, Germany

#### *Reviewed by:*

Nico Jehmlich, Helmholtz-Zentrum für Umweltforschung (UFZ), Germany Martin Taubert, Friedrich Schiller Universität Jena, Germany

#### *\*Correspondence:*

Song Yang yangsong1209@163.com

† These authors have contributed equally to this work.

#### *Specialty section:*

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

*Received:* 03 September 2017 *Accepted:* 19 December 2017 *Published:* 09 January 2018

#### *Citation:*

Xu X-Y, Shen X-T, Yuan X-J, Zhou Y-M, Fan H, Zhu L-P, Du F-Y, Sadilek M, Yang J, Qiao B and Yang S (2018) Metabolomics Investigation of an Association of Induced Features and Corresponding Fungus during the Co-culture of Trametes versicolor and Ganoderma applanatum. Front. Microbiol. 8:2647. doi: 10.3389/fmicb.2017.02647 Li-Ping Zhu<sup>1</sup> , Feng-Yu Du<sup>4</sup> , Martin Sadilek <sup>5</sup> , Jie Yang<sup>6</sup> , Bin Qiao<sup>7</sup> and Song Yang1, 8 \* <sup>1</sup> Shandong Province Key Laboratory of Applied Mycology, Qingdao International Center on Microbes Utilizing Biogas, School of Life Science, Qingdao Agricultural University, Qingdao, China, <sup>2</sup> Central Laboratory, Qingdao Agricultural University,

, Yuan-Ming Zhou<sup>2</sup>

, Huan Fan<sup>3</sup>

,

Xiao-Yan Xu1†, Xiao-Ting Shen1†, Xiao-Jie Yuan<sup>1</sup>

Qingdao, China, <sup>3</sup> Tianjin Animal Science and Veterinary Research Institute, Tianjin Academy of Agricultural Sciences, Tianjin, China, <sup>4</sup> School of Chemistry and Pharmacy, Qingdao Agricultural University, Qingdao, China, <sup>5</sup> Department of Chemistry, University of Washington, Seattle, WA, United States, <sup>6</sup> Department of Biochemistry and Molecular Biology, School of Basic Medical Sciences, Tianjin Medical University, Tianjin, China, <sup>7</sup> School of Chemical Engineering and Technology, Tianjin University, Tianjin, China, <sup>8</sup> Key Laboratory of Systems Bioengineering, Ministry of Education, Tianjin University, Tianjin, China

The co-culture of Trametes versicolor and Ganoderma applanatum is a model of intense basidiomycete interaction, which induces many newly synthesized or highly produced features. Currently, one of the major challenges is an identification of the origin of induced features during the co-culture. Herein, we report a <sup>13</sup>C-dynamic labeling analysis used to determine an association of induced features and corresponding fungus even if the identities of metabolites were not available or almost nothing was known of biochemical aspects. After the co-culture of T. versicolor and G. applanatum for 10 days, the mycelium pellets of T. versicolor and G. applanatum were sterilely harvested and then mono-cultured in the liquid medium containing half fresh medium with <sup>13</sup>C-labeled glucose as carbon source and half co-cultured supernatants collected on day 10. <sup>13</sup>C-labeled metabolome analyzed by LC-MS revealed that 31 induced features including 3-phenyllactic acid and orsellinic acid were isotopically labeled in the mono-culture after the co-culture stimulation. Twenty features were derived from T. versicolor, 6 from G. applanatum, and 5 features were synthesized by both T. versicolor and G. applanatum. <sup>13</sup>C-labeling further suggested that 12 features such as previously identified novel xyloside [N-(4-methoxyphenyl)formamide 2-O-beta-D-xyloside] were likely induced through the direct physical interaction of mycelia. Use of molecular network analysis combined with <sup>13</sup>C-labeling provided an insight into the link between the generation of structural analogs and producing fungus. Compound 1 with m/z 309.0757, increased 15.4-fold in the co-culture and observed <sup>13</sup>C incorporation in the mono-culture of both T. versicolor and G. applanatum, was purified and identified as a phenyl polyketide, 2,5,6-trihydroxy-4, 6-diphenylcyclohex-4-ene-1,3-dione. The biological activity study indicated that this compound has a potential to inhibit cell viability of leukemic cell line U937. The current work sets an important basis for further investigations including novel metabolites discovery and biosynthetic capacity improvement.

Keywords: basidiomycete, interaction, <sup>13</sup>C-labeling, origin of induced features, phenyl polyketide, metabolite identification, LC-MS

### INTRODUCTION

Secondary metabolites are an important source of valuable drug leads, of which compounds derived from various fungi, especially medicinal fungi of basidiomycetes, represent an important part. To explore chemical diversity, several approaches such as epigenetic modification or non-targeted metabolic pathway manipulation have recently been developed in the Aspergillus and Streptomyces species (Scherlach and Hertweck, 2009; Chiang et al., 2011). In the case of basidiomycetes, the induction of novel secondary metabolites or enhancement of secreting extracellular enzymes can be achieved by activating cryptic biosynthetic pathways through establishment of a fungi interaction in the co-culture (Peiris et al., 2008; Hiscox et al., 2010). This coculture strategy mimics natural ecosystem, in which interspecies interaction of basidiomycetes is very common (Hiscox et al., 2015, 2017). Recently, Zheng et al. demonstrated that co-culture of basidiomycetes Inonotus obliquus and Phellinus punctatus resulted in the accumulation of lanostane-type triterpenoids, polyphenols, and melanins, compounds capable of scavenging free radicals and inhibiting tumor cell proliferation (Zheng et al., 2011). The co-culture of Trichoderma Reesei with Coprinus comatus was an important approach for an on-site generation of lignocellulolytic enzyme leading to the increase of lignocellulose degradation rate as described by Ma and Ruan (2015). In our previous work, fifteen wood-decaying basidiomycetes and two straw-decaying basidiomycetes were used to establish 136 pairwise co-cultures on agar plates (Yao et al., 2016). The co-culture system of Trametes versicolor and Ganoderma applanatum showed an interaction zone, in which the accumulation of a series of known carboxylic acids as well as novel xylosides were observed.

The application of co-culture has enhanced the number of discovered novel secondary metabolites but also raises several challenges. One of the major challenges is how to timely and accurately identify the species responsible for the newly induced metabolites during the microorganism interaction. Typically, the structure of compounds and related biochemical information were required to verify the unique secondary metabolite and its particular producer (Riedlinger et al., 2006). Recently, a combined technique of nanospray desorption electrospray ionization and matrix-assisted laser desorption/ionization-time of flight mass spectrometry has been developed as a platform to reveal many unknown metabolites produced by Streptomyces coelicolor when paired with five other actinomycetes (Traxler et al., 2013). However, this approach was limited to associate the novel metabolites with corresponding producer grown on agar plates and was not suitable to the co-culture in the liquid medium. The metabolic incorporation of stable isotopes <sup>13</sup>C or <sup>15</sup>N is a powerful approach used for quantitative proteome and metabolome studies (Julka and Regnier, 2004; Yang et al., 2010). Moreover, <sup>13</sup>C-dynamic labeling analysis has been applied for analyzing the metabolite turnover rates, distinguishing the flux distribution between two pathways starting from the same metabolic point, and interpreting the synthetic process of novel metabolites (Yang et al., 2012; Shlomi et al., 2014; Hammerl et al., 2017). With the advantage of a <sup>13</sup>C-labeling approach, the current study reports on the identification of the producer of novel induced metabolites in liquid medium after the two interactive mycelia were sterilely separated and mono-cultured after the induction by the co-culture.

The research of the interspecies crosstalk expands our possibilities to discover novel metabolites, and also increases our understanding of how these metabolites are induced in microbial consortia. A chemical warfare in the fungal-fungal communication is often described as diffusion of harmful and chemically complex metabolites from one partner to the other (Bertrand et al., 2014). For instance, the interaction between the Aspergillus niger and Aspergillus flavus led to the inhibition of aflatoxin B1 produced by A. flavus through signal molecules downregulating expression of major biosynthetic genes (Xing et al., 2017). Under some conditions, co-cultivation appears to trigger production and accumulation of novel metabolites without the involvement of released signaling molecules. For instance, co-cultivation with Corynebacteria glutamicum or Tsukamurella pulmonis was indicated to stimulate a novel pathway in S. endus, contributing to a new heterocyclic chromophore-containing antibiotics alchivemycin A (Onaka et al., 2011). Notably, the mono-culture of S. endus did not generate the same compound with or without the addition of filter sterilized supernatants from bacterial culture. The production of alchivemycin A therefore appeared to require a direct physical interaction between S. endus and the coryneform bacteria cells. For this research, <sup>13</sup>C dynamic labeling was further utilized to suggest the potential mechanism of the induction of increased and newly synthesized features.

In this work, we built upon the characteristic of mycelial pellets of basidiomycetes and <sup>13</sup>C-labeling analysis, to analyze and catalog a broad range of <sup>13</sup>C-labeled features, which were highly accumulated in the co-culture of T. versicolor and G. applanatum. <sup>13</sup>C dynamic labeling further suggested two potential mechanism of the induction of these features. Ultimately, compound **1** with M+H<sup>+</sup> m/z 309.0757, that displayed <sup>13</sup>C incorporation in the mono-culture of both T. versicolor and G. applanatum, was isolated and identified as novel phenyl polyketide, and its biological activity was evaluated.

## MATERIALS AND METHODS

### Chemicals

All chemicals including standards (ascorbic acid and phenolic antioxidant 2,6-di-tert-butyl-4-methoxyphenol) were purchased from either Sigma-Aldrich (St. Louis, MO, USA) or TCI (Kita-ku, Tokyo, Japan). Millipore water (Billerica, MA, USA) was used for the preparation of all the media and sample solutions.

#### Fungi Material and Culture Conditions

G. applanatum (CGMCC No. 5.249) and T. versicolor (CGMCC No. 12241) were deposited at the Shandong Province Key Lab of Applied Mycology in China. The culture medium was supplied at concentrations as follows: 2 g glucose, 0.2 g KH2PO4, 0.1 g MgSO4, 0.4 g peptone, and 4 g agar (only in solid medium) in 200 mL of sterilized water.

### Mono-culture of *T. versicolor* and *G. applanatum* on Agar Plate

The mono-culture procedure was adapted from our previous publication (Yao et al., 2016), and mono-culture medium was the same as above. Briefly, a 5 mm agar plug of each fungus scraped from agar slant culture-medium was cultured on a Petri dish (9 cm diameter), and were incubated at 28◦C for 10 days.

### Co-culture of *T. versicolor* and *G. applanatum* in Liquid Medium and Sample Preparation

Eight 5 mm agar plugs of T. versicolor and G. applanatum were separately pre-cultured in 500 mL shake flask containing 200 mL of culture medium at 28◦C for 4 days on orbital shakers at 180 rpm. Then 100 mL culture broth of G. applanatum was transferred into the culture of 100 mL T. versicolor, and co-cultivated up to 18 days. All the co-cultures had three independent biological replicates. At the harvest time, 10 mL of co -culture broth was filtered by using MILLEX-GP PES membrane filters (0.22µm, 33 mm, Merck Millipore, Germany) and the filtrate was dried in a freeze-dryer ALPHA 1-2LDplus (Christ, Osterode, Germany). Five milliliters of freshly prepared dichloromethane/methanol/water (64:36:8, v/v) solvent mixture was added to the dried samples (Yao et al., 2016). The sample extractions were carried out in a water bath sonicator (KQ-300GVDV, Kunshan, China) at 25◦C for 20 min, and were centrifuged at 12,000 rpm for 10 min. Finally, the extracts were dried on a rotational vacuum concentrator (Christ, Osterode, Germany) and stored in a −80◦C freezer.

### Measurement of the Metabolome

The extracts were dissolved in 200 µL methanol, and then were centrifuged at 12,000 rpm for 15 min. The supernatants were transferred into 250 µL Agilent autosampler vials. The samples were analyzed on an Agilent liquid chromatograph-quadrupole time-of-flight mass spectrometer (LC-QTOF-MS, Agilent 1290 Infinity-6530B, Agilent Technologies, Santa Clara, CA, USA) as previously described (Cui et al., 2016; Hu et al., 2016). Briefly, 10 µL of the samples was separated on an Acquity UPLC BEH C18 column (100 × 2.1 mm, 1.7µm, Waters, Milford, MA, USA). The mobile phase A was water with 0.1% formic acid. The mobile phase B was pure acetonitrile. The reversed-phase liquid chromatographic elution gradient was optimized in order to maximize the resolution of the induced features. The gradient was the following: 0–3 min, 5% B, 3–12 min, 5–40% B, 12–38 min, 40–95% B, 38–46 min, 95% B, 46–48 min, 95–5% B, 48–55 min, 5% B. This shallow gradient provided reproducible separation for many co-eluting compounds in a time window from 5.0 to 8.0 min. The TOF m/z range was set to 50–1,200 amu in centroid mode with a scan rate of 1.5 spectra/s. All the samples had three independent biological replicates. Each biological replicate had two analytical replicates.

### Data Pre-processing and Principal Component Analysis

LC-QTOF-MS data were converted into mzML format using MS Convert software (Holman et al., 2014). Data pre-processing and statistical analysis were performed with MZmine 2 (Version 2.11) (Pluskal et al., 2010) and SIMCA-P 11.5. For the MZmine 2, the peak detection threshold for MS signal intensity was set to 1.0 × 10<sup>3</sup> . The chromatogram building was realized using a minimum time span of 0.01 min, minimum height of 2.5 × 10<sup>3</sup> , and m/z tolerance of 0.005 (or 10 ppm). Chromatograms were deconvoluted with the following settings: search minimum in absolute retention time (RT) range 0.1 min, minimum relative height 10%, minimum absolute height 2.5 × 10<sup>3</sup> and baseline level 1.2. The chromatogram isotopic peaks grouper algorithm was set as m/z tolerance of 0.005 (or 10 ppm) and absolute RT tolerance of 0.10 min. Chromatograms were peak aligned with m/z tolerance at 0.008 (or 15 ppm) and absolute RT tolerance 1 min. The peak list was eventually gap-filled with m/z tolerance at 0.008 (or 15 ppm), and absolute RT tolerance of 0.20 min. To classify m/z in the peak list, principal component analysis (PCA) was carried out by using SIMCA-P (version 11.5). Two steps were required to perform PCA. The first step was to set the variables m/z and RT as Primary ID and Secondary ID. The secondary step was to do the normalization of the variables with Pareto scaling. This normalized method was embedded in SIMPCA-P, which did not require further parameter tuning. After that, PCA were displayed by a scores plot, mainly observing the overall cluster of the different treatments as well as the presence of outliers. The correlation coefficient loading plot was used to identify the variables responsible for the clustering or separation of the treatments.

#### Molecular Network Analysis

MS/MS data for molecular network analysis were acquired in targeted MS/MS mode on the same LC-QTOF-MS system (the precursor ions are listed in **Table 1**). The collision energy and m/z range for different precursor ions were optimized based on their own characteristics as our previous publication (Yao et al., 2016). MS/MS data were converted to mzML format, and then were subjected to the Molecular Networking workflow of Global Natural Products Social Molecular (GNPS at gnps.ucsd.edu) using the Group Mapping feature (Watrous et al., 2012; Wang et al., 2016). The subnetworks were generated with settings of minimum pairs cosine 0.65, parent mass tolerance 1.0 Dalton, TABLE 1 | List of induced features in the co-culture of T. versicolor and G. applanatum analyzed by LC-MS in the positive/negative mode and corresponding production fungus.


(Continued)

#### TABLE 1 | Continued


Yellow label, the features were observed by LC-MS in both negative and positive modes. These features were only counted once. Plus mark, the features discovered in this work were tagged with plus mark. NA, the features had no <sup>13</sup>C incorporation. ND, the features had weak MS signal intensity or no signal in the supernatants from the co-culture on day 10. Fold increase, fold value was the ratio of MS signal intensity of increased feature in the co-culture to that in the control mono-culture. Data show the mean with standard deviation calculated from three independent biological replicates.

ion tolerance 0.5 Dalton, maximum connected components 50, minimum matched peaks 6, minimum cluster size 2. The results were then visualized using Cytoscape (Version 3.1.1) (Su et al., 2014).

### <sup>13</sup>C-Labeling Analysis

T. versicolor and G. applanatum were co-cultivated up to 10 days as described above. Then mycelium pellets of T. versicolor and G. applanatum were respectively harvested using sterilized tweezers based on the difference of diameter size and color. The mycelium pellets of T. versicolor had the size ranging from 6 to 8 mm with the color of faint yellow and those of G. applanatum had the size ranging from 2 to 4 mm with red color. The harvested mycelium pellets were washed with sterile water three times and then mono-cultured in 50 ml shake flask containing 10 mL fresh medium with <sup>13</sup>C-labeled glucose at the final concentration of 5 g L−<sup>1</sup> and 10 mL of co-cultured supernatants collected on day 10. For the control experiment no co-culture supernatants were added, only 20 mL of fresh medium with <sup>13</sup>C-labeled glucose as carbon source. After the mono-culture for 10 or 20 days, 10 mL of T. versicolor and G. applanatum culture broths were harvested for analysis of <sup>13</sup>C-labeling. All samples had three independent biological replicates and two analytical replicates. Samples were extracted and analyzed by LC-QTOF-MS as described above. The mass isotopomer distributions were corrected for the contribution from natural isotopes by a matrixbased method (Jennings and Matthews, 2005). The total <sup>13</sup>Cincorporation for each feature was obtained by normalizing to its total carbon number as in our previous publication (Yang et al., 2013). Relative isotopic abundance (Mi) for a feature in which i <sup>13</sup>C atoms were incorporated was calculated by the Equation (1):

$$\text{Mi(\%)} = \frac{m\_i}{\sum\_{j=0}^{n} m\_j} \tag{1}$$

where m<sup>i</sup> representsthe isotopic abundance for a feature in which i <sup>13</sup>C atoms were incorporated and n represents the maximum number of <sup>13</sup>C atoms incorporated.

Total <sup>13</sup>C-incorporation of a feature with N carbon atoms was obtained by normalizing to its total carbon number according to the Equation (2):

$$\text{Total } ^{13}\text{C}-\text{incororption (\%)} = \frac{\sum\_{i=1}^{N} i \times \text{Mi}}{\text{N}} \tag{2}$$

The significance of difference of <sup>13</sup>C-incorporation between experimental data points was determined by t-tests (Origin 8.0). A P-value <0.05 was considered to be statistically significant.

#### Isolation and Purification of Compound 1

Twenty liters of the co-cultured supernatant was extracted three times with ethyl acetate (EtOAc). 2.5 g of crude EtOAc extract was concentrated under reduced pressure and purified on a silica gel (200–300 mesh) column and then eluted with petroleum ether/ethyl acetate and chloroform/methanol system to yield ten fractions (Song et al., 2014). Eight hundred and fifty milligrams of fraction 6 and 7 from the chloroform/methanol elution (20:1 and 10:1, v/v) was used to a medium pressure liquid chromatography (Flash CO140080-0, Agela Technologies, China) and eluted with methanol/water (the elution gradient was 10–90% methanol in 65 min) to generate a mixture of 216 mg. This mixture was purified on a silica gel (200–300 mesh, Qingdao Haiyang Chem. Ind. Co. Ltd. China) column by eluting with petroleum ether/ethyl acetate/ methanol (3:3:1, v/v) and then separated on a preparative column (Venusil XBP C18, Agela Technologies, China) by eluting with methanol/water (the elution gradient was 10–50% methanol in 40 min and flow rate was 8 mL/min) to obtain 10 mg of compound **1**.

#### NMR Analysis of Purified Compound 1

<sup>1</sup>H, <sup>13</sup>C, and 2D NMR spectra of the purified compound **1** were all performed by using a Bruker Avance 600 MHz spectrometer (Karlsruhe, Germany) at 25◦C. The compound **1** was accurately weighted, and dissolved in 0.5 mL of deuterated methanol as the internal lock. The resulting spectra were manually phased and baseline corrected and calibrated to methanol, using TOPSPIN (Version 2.1, Bruker).

#### Cell Viability Assay

The human leukemic cell line U937 and lung cancer cell line A549 were purchased from American Type Culture Collection (Manassas, VA, USA). They were cultured in RPMI-1640 (HyClone, USA) with 10% FBS (fetal bovine serum, Gibco). Cells were inoculated in 96-well plates at a density of 3,000 cells per well overnight and then treated to different concentrations of compound **1** for 48 h. The compound **1** was solved in ethanol. At a concentration of 300µM of compound **1**, the ethanol was 0.8% (v/v) in the cell culture medium, showing minimal effect on the viability of both cell lines. The effect of compound **1** on the viability of U937 and A549 cells was evaluated with CellTiter 96 <sup>R</sup> AQueous One Solution (Promega, Madison, WI, USA) (Soman et al., 2009). The absorbance was read at 490 nm with a microplate spectrophotometer (Multiskan FC, Thermo scientific, USA). Three independent experiments were performed and each one had six replicates. Data show the mean with error bar indicating standard deviation calculated from biological replicates by Origin 8.0. IC<sup>50</sup> defined as the concentration with the inhibition of 50% cells was calculated by using SPSS (Statistical Package for Social Sciences) package 6, version 15.0.

#### Antioxidant Activity Assay

The 2,2′ -azinobis-(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) method used is based on the reduction of the ABTS•+ radical action by the antioxidants present in the sample (Rigano et al., 2014). Three biological replicates were performed. A solution of 7.4 mM ABTS•+ (5 mL) mixed with 140 mM K2S2O<sup>8</sup> (88 µL) was prepared and stabilized for 12 h at 4◦C in the dark. This mixture was then diluted by mixing ABTS•+ solution with ethanol (1:88) to obtain an absorbance of 0.70 ± 0.10 unit at 734 nm using a spectrophotometer (Multiskan FC, Thermo scientific, USA). Compound **1** (100 µL) and the standard controls (ascorbic acid and phenolic antioxidant 2,6-di-tert-butyl-4-methoxyphenol) reacted with 1 mL of diluted ABTS•+ solution for 2.5 min, and then the absorbance was taken at 734 nm against a blank constituted by ABTS•+ solution added with 100 µL of ethanol. ABTS•+ scavenging activity was calculated by the Equation (3):

$$\text{Scavenging effect (\%)} = \left(1 - \frac{Abs.sample}{Abs.blank}\right) \times 100\qquad(3)$$

where Abs blank = 100 µL of ethanol+1 mL of diluted ABTS•+ solution.

#### RESULTS

#### Induced Feature Discovery in Fungal Interaction

Unsupervised PCA is well-suited for comparing different biological samples and identifying statistically significant differences (Chen et al., 2007). As shown in **Figure 1** (left figure), examination of the scores plot showed that the co-culture treatment is clearly separated from the two mono-cultures. The variable features responsible for discriminating these three groups are shown in the loading plots (**Figure 1**, right). More

than 4,000 features were recorded, out of which 58 features were detected only in the co-culture and 16 features were at least 5-fold more abundant in the co-culture than those in the control mono-culture (**Table 1**, **Figure 2**). Compared to our previous work (Yao et al., 2016), 28 additional features were discovered mainly due to the optimized chromatography (**Table 1**, **Figure 2**).

### <sup>13</sup>C-Labeling Analysis to Associate Induced Features with Corresponding Fungus

To associate induced features with corresponding fungus, we initially harvested individual mycelium pellets of T. versicolor and G. applanatum after the co-culture for 10 days and then detected the abundance of in vivo induced features. Many of the induced features were observed for both samples of T. versicolor and G. applanatum (data not shown). This made it impossible to distinguish if individual or both fungi were induced to generate the compounds in the co-culture. To overcome this issue, we designed a <sup>13</sup>C-labeling approach in the liquid co-cultivation. The workflow is shown in **Figure 3**. First, T. versicolor and G. applanatum were co-cultivated for 10 days to activate the cryptic genes. Next, their mycelia were sterilely separated and mono-cultivated in the liquid medium which contained half of fresh medium with <sup>13</sup>C-labeled glucose as carbon source and half of co-cultured supernatants collected on day 10. Then, the samples were harvested in the mono-cultures of T. versicolor and G. applanatum on days 10 and 20 and analyzed by LC-QTOF-MS. In the preliminary experiments, the samples were also harvested after 5 days of mono-culture with the addition of the supernatants. However, many of induced features were only slightly labeled. It was likely due to a relatively long lag phase and low growth rate in the mono-culture of T. versicolor and G. applanatum after the stimulation of the co-culture. In addition, for the unlabeled features, <sup>13</sup>C-labeling was not detected on day 5 either. Therefore, the samples were harvested later in order to obtain the strong MS signal. As the incorporation of <sup>13</sup>C-labeled carbons from glucose increases the molecular weight of metabolites, the mass shift determined from the mass spectra

provides then the information about which fungi generated the induced features.

Total 31 induced features were found to have <sup>13</sup>C incorporation in the mono-culture, and 43 features did not incorporate any label (**Figure 2**). Among the labeled features activated by the co-culture, 20 originated from T. versicolor and 6 features were derived from G. applanatum. Five features were induced by both T. versicolor and G. applanatum (**Figure 2**). Several representative <sup>13</sup>C labeling results are shown in **Figure 4**. In **Figure 4A**, the m/z 490 (experimental m/z 490.3025, predicted elemental component C24H45NO9, error 0.3 ppm) was a newly synthesized feature in the co-culture. As <sup>13</sup>C-labeled carbon was incorporated into the feature m/z 490, intensity of m/z 491 (m/z + 1) up to m/z 506 (m/z + 16) increased. This change

was observed only in the mono-culture of T. versicolor and indicates that m/z 490 was particularly induced in this species. The calculation of total <sup>13</sup>C incorporation indicated that about 19 and 25% carbon was replaced with <sup>13</sup>C on days 10 and 20, respectively (bottom part of **Figure 4A**). The m/z 167.0348 in **Figure 4B** was highly produced during the co-culture and identified as orsellinic acid in the previous publication (Yao et al., 2016). In contrast to the above example, the m/z 167 incorporated <sup>13</sup>C-labels in the same time frame only in the mono-culture of G. applanatum, resulting in the parent ions shifted from m/z 167 to m/z 174 in the <sup>13</sup>C-labeled mass spectra and total <sup>13</sup>C incorporation level reached 14% on day 20 (**Figure 4B**). The m/z 165.0554 was previously identified as 3-phenyllactic acid (Yao et al., 2016), and <sup>13</sup>C-labeled mass spectra ranging from m/z 165 to m/z 174 were observed in both mono-cultures of T. versicolor and G. applanatum (**Figure 4C**). Similar example, feature with m/z 309 (experimental m/z 309.0756, C18H14O5, error 0.37 ppm) increased 15.4-fold during the co-culture in comparison with MS signal in the control mono-culture and <sup>13</sup>C-labeling was observed for both fungi (**Figure 4D**). It is worth mentioning that a contamination from the undesired fungus cannot be absolutely excluded during the transfer from the co-culture to the mono-culture, but that it does not affect the identification of the fungus producing a feature only found in the mono-culture of either T. versicolor or G. applanatum based on the <sup>13</sup>C-labeling analysis.

<sup>13</sup>C incorporation could be due to the induction of features which have been released into the medium during the coculture. To confirm whether diffusible features were involved in triggering the silent gene expression in this study, we treated T. versicolor and G. applanatum with fresh medium containing <sup>13</sup>C labeled glucose as carbon source but without the addition of the supernatant of co-culture on day 10. As shown in **Figure 5** as an example, <sup>13</sup>C labeling of m/z 490.3025 and 136.0403

was not observed up to 20 days but when the supernatant was added the features were significantly labeled. Notably, for 12 induced features with high signal intensities, including newly synthesized xyloside [N-(4-methoxyphenyl)formamide 2- O-beta-D-xyloside; Yao et al., 2016], we did not detected any <sup>13</sup>C incorporation in the mono-culture of either fungus (**Table 1**, **Figure 4E**, Supplementary Figure 1D). Therefore, it was not possible to assign their origin to a specific fungus.

### Dereplication of Newly Discovered Features by Molecular Network Analysis

To investigate the structural similarities occurring for 28 newly discovered features in this study, MS/MS fragmentation spectra of the induced features in **Table 1** were processed and organized as molecular network with the previously identified features. **Figure 6** (upper part) shows a constructed subnetwork to dereplicate m/z 196.0944 and 216.1021. These two new features likely possessed similar backbone structure with the previously identified N-(2-hydroxy-4-methoxyphenyl) formamide (m/z 168.0653) and its analogous features (m/z 140.0708, 150.0548, 230.1177, 287.1030) due to their close fragmentation patterns (Supplementary Figure 2). This data was also in agreement with <sup>13</sup>C-labeling analysis, in which most of features involved in this subnetwork were not labeled either. In another example, five newly identified features and three previous features (m/z 251.0712, 281.0808, and 334.0733) were clustered together with high scores (**Figure 6**, bottom part). Comparison of the fragment ions of m/z 306.0775, 334.0733, 581.1208, 629.1419, and 279.0656 showed some common ions of m/z 77.04, 117.03, 235.08 which were likely derived from fragmentation of m/z 279.0656, suggesting that these features probably had the same backbone structure and belonged to a series of structural derivatives (Supplementary Figure 2). Moreover, <sup>13</sup>C incorporation of m/z 279.0656, 251.0712, and 281.0808 were clearly detected in the mono-culture of T. versicolor with the addition of the supernatant of co-culture (Supplementary Figure 1), suggesting that the derivatives of m/z 279.0656 (i.e., m/z 306.0775, 334.0733, 581.1208, and 629.1419, which did not show <sup>13</sup>C incorporation due to the weak signals in the mono-culture) were also likely biosynthesized by T. versicolor under the co-cultured condition.

### Identification of Induced Compound 1 and Analysis of Its Biological Activity

Since compound **1** was derived from both T. versicolor and G. applanatum (**Figure 4D**), and 15.4-fold more abundant in the co-culture, we isolated and purified sufficient amount for detailed characterization. Compound **1** was a pale yellow powder with the molecular formula of C18H14O5. The <sup>1</sup>H, <sup>13</sup>C-NMR and HSQC spectrum showed the presence of two carbonyl carbon, one oxygen connected CH, one oxygen connected to quaternary carbon, two single benzene rings and two olefinic carbon atoms (one with oxygen attached to it) (Supplementary Table 1 and Supplementary Figure 3). All the information suggested that the basic skeleton of the compound **1** is a terphenyl derivative with two mono-substituted benzene rings. The COSY spectrum showed the <sup>1</sup>H–1H spin systems of H-2′ /H-3′ /H-4′ /H-5 ′ /H-6′ and H-2′′/H-3′′/H-4′′/H-5′′/H-6′′, assigned two monosubstituted benzene rings (A and B). The HMBC correlations from H-2 (δ<sup>H</sup> 4.49) to C1 (δ<sup>C</sup> 203.4), C3 (δ<sup>C</sup> 197.4), C4 (δ<sup>C</sup> 113.5), and C6 (δ<sup>C</sup> 90.9), and from OH-6 (δ<sup>H</sup> 5.44, dimethylsulfoxide (DMSO-d6) to C1 (δ<sup>C</sup> 202.45), C5 (δ<sup>C</sup> 191.28) and C6 (δ<sup>C</sup> 90.9) established the ring C. The fragment ring A was linked to C-6 of ring C supported by the HMBC correlations from H-6' (δ<sup>H</sup> 7.98) to the carbonyl C1 (δ<sup>C</sup> 203.4). Likewise, the linkage of the ring B to the ring C at C-4 was confirmed by HMBC correlations from H-2′′ (δ<sup>H</sup> 7.88) to C4 (δ<sup>C</sup> 113.5) and C-1′′(δ<sup>C</sup> 135.4). Therefore, the structure of compound **1** was identified as a phenyl polyketide, 2,5,6-trihydroxy-4,6-diphenylcyclohex-4 -ene-1,3-dione (**Figure 7**).

We further tested biological activity of compound **1**. The human lung cancer cell lines A549 and leukemic cell lines U937 were treated with compound **1** at various concentrations for 48 h. As shown in **Figure 8A**, no visible changes in cell viability were detected for human lung cancer cell line A549 when the concentrations were increased to 300µM. In contrast, compound **1** inhibited the viability of leukemic cells in a dose-dependent manner. The IC<sup>50</sup> at 48 h was determined

to be 276 ± 5µM (equal to 85 mg/L). In addition, based on the structural characteristics of compound **1**, we also studied whether compound **1** had antioxidant properties using ABTS assay (**Figure 8B**). The highest percentage of antioxidant capacities (82.65 ± 1.25%) was observed for compound **1** at the concentration of 200µg/mL. This is comparable with the report of crude extracts from berries (Abu-Bakar et al., 2016). The comparison of the antioxidant activity with ascorbic acid (VC) and phenolic antioxidant 2,6-di-tert-butyl-4 methoxyphenol (BHT) revealed that ascorbic acid> compound **1** > BHT.

### DISCUSSION

Determining the origin of induced features in the co-culture remains a large challenge, because it generally requires the chemical structures with their biological information. Previously Bertrand et al. discovered five de novo induced compounds from the co-culture of Trichophyton rubrum and Bionectria ochroleuca, and elucidated the origin of one of them based on its nonsulfated form detected in the mono-culture of B. ochroleuca (Bertrand et al., 2013). However, other four compounds could not be associated with corresponding fungi due to the lack of their structures. In another study Ola et al. isolated nine compounds and speculated that four of them detected only in the co-culture originated from Fusarium tricinctum in terms of structural analogies with the known fungal products from the Xylariaceae family, but the producer of remaining five compounds was still unclear because of insufficient biochemical evidences (Ola et al., 2013). In this work, among 74 induced features, 31 features were shown to be produced specifically by either T. versicolor or G. applanatum, or by both fungi using a <sup>13</sup>C-based labeling analysis. This methodology was able to distinguish the origin even if the identities of compounds were not available or almost nothing was known about biochemical aspects. In more detail, the total <sup>13</sup>C incorporation in the same fungal culture varied noticeably among features. Isotopic steady state is the state that <sup>13</sup>C-labeling signatures in metabolites become time invariant (Antoniewicz et al., 2007). In current hydrogen connectivity.

work, <sup>13</sup>C incorporation into 20 features (e.g., orsellinic acid) had the similar abundance levels between days 10 and 20 and these features had the similar <sup>13</sup>C-labeling pattern (data not shown), indicating that they likely reached isotopic steady state around 10 days (**Figure 4B**, Supplementary Figure 1). In contrast, <sup>13</sup>C incorporation level of 11 features including novel phenyl polyketide (compound **1**) increased from day 10 to day 20 (**Figures 4A,C,D**, Supplementary Figure 1), suggesting either the fluxes via their synthetic pathways were low or their metabolite pools were relatively large (Zamboni et al., 2009).

By comparison of <sup>13</sup>C-labeling patterns between with and without the addition of the supernatant of co-culture, it was demonstrated that the induction of gene expression and synthesis of corresponding metabolites during the co-culture of T. versicolor and G. applanatum depended indeed on the signaling molecules released into the medium. Forty-three induced features in the supernatant of co-culture on day 10 were detected with signal intensity ranging from 10<sup>3</sup> to 10<sup>5</sup> (**Table 1**), but in the current work we were not able to determine which features were responsible for signaling to activate gene expression in T. versicolor or G. applanatum. Additional studies will be needed to confirm this link as well as to elucidate the structure and function of signaling molecules. Notably, the previous report also demonstrated that an intimate physical interaction of the actinomycete S. hygroscopicus and fungal mycelia of Aspergillus nidulans was required to induce specific stimulation of the silent polyketide synthases and non-ribosomal peptide synthetases gene clusters (Schroeckh et al., 2009). In our research, we found 12 highly produced features that did not incorporate <sup>13</sup>C labeling in the mono-culture. One potential explanation is that these features were also produced in the co-culture via mycelium physical interaction to elicit the specific response. In addition, compound **1** had over 15-fold higher MS signal intensity in the co-culture than in the control mono-culture of T. versicolor where the mycelia were not pre-induced by the co-culture. It was not detected in the control mono-culture of G. applanatum, yet <sup>13</sup>C incorporation in G. applanatum was 1.3-fold higher than that in T. versicolor on day 20 (**Figure 4D**). This finding suggests that the co-culture could significantly activate the encoding genes or gene clusters related to the synthesis of Compound **1** in G. applanatum. This provides an important insight into possible manipulation of G. applanatum to enhance biosynthesis of novel metabolites.

Comparing MS fragmentation similarity including common losses, molecular network analysis is able to obtain a simultaneous visual investigation of identical molecules, analogs, or metabolite families, thereby assisting the structural analysis (Winnikoff et al., 2014; Cabral et al., 2016). In our previous research, this method was utilized to show that the common neutral loss of 132 Dalton resulted from the deglycosylation reaction, which helped to find a series of novel xylosides generated during the co-culture (Yao et al., 2016). Here, combined molecular network analysis and <sup>13</sup>C-labeling analysis suggested further that some newly discovered features were not only structurally analogous but also had similar induction mechanism and were likely produced by the same fungus. Thus, combination of network analysis and <sup>13</sup>C labeling shows promise to accelerate the elucidation of biosynthetic pathways of novel metabolites.

Several type II polyketides have been reported to be highvalue medicals (Zhang W. et al., 2012). In antioxidant assay, compound **1** had better activity than BHT. More interestingly, compound **1** at micromolar concentration was able to inhibit the viability of leukemic cells. For comparison, Zhang et al. reported that matrine extracted from Sophora flavescens inhibited the proliferation of acute myeloid leukemia cell lines U937 in a dose- and time-dependent manner with the IC<sup>50</sup> of 590 mg/L at 24 h, and resulted in the maximal apoptosis rates with 37.2% for 24 h (Zhang S. et al., 2012). Wang et al. also demonstrated that 100µg/mL of Ganoderma lucidum extracts could greatly suppress leukemic cell growth with the inhibition rate of 75% (Wang et al., 1997). In our case, the IC<sup>50</sup> of compound **1** was at the same level with matrine and crude extracts from G. lucidum. Thus, it can become a starting point for development of lead compounds to cure leukemia or other cancers.

#### CONCLUSION

The application of <sup>13</sup>C-labeling analysis produced valuable insight into the role of individual partners in the co-culture in production of known or unknown induced metabolites. Moreover, <sup>13</sup>C-labeling approach combined with molecular network analysis can reveal whether certain structural analogs were produced by the same fungus or through the similar activation mechanism. The current <sup>13</sup>C-labeling information sets an important foundation for further studies in the basidiomycetes, including but not limited to novel metabolites discovery and biosynthetic capacity improvement.

ascorbic acid; BHT, 2,6-di-tert-butyl-4-methoxyphenol.

#### AUTHOR CONTRIBUTIONS

X-YX, BQ, and SY: conceived and designed the project; X-YX, X-TS, X-JY, Y-MZ, HF, and JY: performed the experiments; X-YX, X-TS, L-PZ, F-YD, MS, and SY: interpreted the data. All authors contributed to the preparation of the manuscript, read and approved the final manuscript.

#### FUNDING

This work was supported by a grant from the Petrochemical Joint Fund of National Natural Science Foundation of China (No. U1462109), a grant from Shandong Province Natural Science Foundation (No. ZR2013CM024), a grant from National Natural Science Foundation of China (No. 31600028), a grant from

#### REFERENCES


Qingdao Applied Basic Research Program (No. 16-5-1-76-jch), a grant from Key Laboratory for Industrial Biocatalysis (Tsinghua University) Ministry of Education (No. 2015102) and a grant from Shandong provincial key research and development plan (Grant No. 2016GSF121010).

#### ACKNOWLEDGMENTS

We thank Dr. Yan Li at the University of Texas Health Science Center for his assistance with NMR analysis.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2017.02647/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.

The reviewer MT and handling Editor declared their shared affiliation.

Copyright © 2018 Xu, Shen, Yuan, Zhou, Fan, Zhu, Du, Sadilek, Yang, Qiao and Yang. 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) or licensor 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.

# How Does Salinity Shape Bacterial and Fungal Microbiomes of Alnus glutinosa Roots?

Dominika Thiem1,2, Marcin Goł ˛ebiewski2,3, Piotr Hulisz<sup>4</sup> , Agnieszka Piernik<sup>5</sup> and Katarzyna Hrynkiewicz1,2 \*

<sup>1</sup> Department of Microbiology, Faculty of Biology and Environmental Protection, Nicolaus Copernicus University in Torun,´ Torun, Poland, ´ <sup>2</sup> Centre for Modern Interdisciplinary Technologies, Nicolaus Copernicus University in Torun, Toru ´ n, Poland, ´ <sup>3</sup> Chair of Plant Physiology and Biotechnology, Faculty of Biology and Environmental Protection, Nicolaus Copernicus University in Torun, Toru ´ n, Poland, ´ <sup>4</sup> Department of Soil Science and Landscape Management, Faculty of Earth Sciences, Nicolaus Copernicus University in Torun, Toru ´ n, Poland, ´ <sup>5</sup> Chair of Geobotany and Landscape Planning, Faculty of Biology and Environmental Protection, Nicolaus Copernicus University in Torun, Toru ´ n, Poland ´

#### Edited by:

Katarzyna Turnau, Jagiellonian University, Poland

#### Reviewed by:

Christopher Blackwood, Kent State University, United States Christel Baum, University of Rostock, Germany

\*Correspondence: Katarzyna Hrynkiewicz hrynk@umk.pl; hrynkiewicz.katarzyna@gmail.com

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 09 October 2017 Accepted: 20 March 2018 Published: 18 April 2018

#### Citation:

Thiem D, Goł ˛ebiewski M, Hulisz P, Piernik A and Hrynkiewicz K (2018) How Does Salinity Shape Bacterial and Fungal Microbiomes of Alnus glutinosa Roots? Front. Microbiol. 9:651. doi: 10.3389/fmicb.2018.00651 Black alder (Alnus glutinosa Gaertn.) belongs to dual mycorrhizal trees, forming ectomycorrhizal (EM) and arbuscular (AM) root structures, as well as represents actinorrhizal plants that associate with nitrogen-fixing actinomycete Frankia sp. We hypothesized that the unique ternary structure of symbionts can influence community structure of other plant-associated microorganisms (bacterial and fungal endophytes), particularly under seasonally changing salinity in A. glutinosa roots. In our study we analyzed black alder root bacterial and fungal microbiome present at two forest test sites (saline and non-saline) in two different seasons (spring and fall). The dominant type of root microsymbionts of alder were ectomycorrhizal fungi, whose distribution depended on site (salinity): Tomentella, Lactarius, and Phialocephala were more abundant at the saline site. Mortierella and Naucoria (representatives of saprotrophs or endophytes) displayed the opposite tendency. Arbuscular mycorrhizal fungi belonged to Glomeromycota (orders Paraglomales and Glomales), however, they represented less than 1% of all identified fungi. Bacterial community structure depended on test site but not on season. Sequences affiliated with Rhodanobacter, Granulicella, and Sphingomonas dominated at the saline site, while Bradyrhizobium and Rhizobium were more abundant at the non-saline site. Moreover, genus Frankia was observed only at the saline site. In conclusion, bacterial and fungal community structure of alder root microsymbionts and endophytes depends on five soil chemical parameters: salinity, phosphorus, pH, saturation percentage (SP) as well as total organic carbon (TOC), and seasonality does not appear to be an important factor shaping microbial communities. Ectomycorrhizal fungi are the most abundant symbionts of mature alders growing in saline soils. However, specific distribution of nitrogen-fixing Frankia (forming root nodules) and association of arbuscular fungi at early stages of plant development should be taken into account in further studies.

Keywords: salinity, ectomycorrhiza, arbuscular mycorrhiza, Frankia sp., metagenomics, Illumina MiSeq

### INTRODUCTION

fmicb-09-00651 April 16, 2018 Time: 17:59 # 2

The global scarcity of water resources, environmental pollution and increased salinization of soil and water are the most pressing problems of the 21st century. Salinity affects more than 7% of global land surface and 70% of all irrigated agricultural soils worldwide (Bencherif et al., 2015). Current predictions indicate that salinity is expected to be responsible for 30% land loss within the next 25 years, and up to 50% within the next 35 years (Chandrasekaran et al., 2014). Afforestation may contribute to solving the problem of soil reclamation, as it is a sustainable land use system that serves as an alternative to agriculture. In general, trees are more tolerant to salt stress than herbaceous crops (Chen et al., 2014). Among them, representatives of Betulaceae family are the most widely planted around the world for the rehabilitation of salinity affected lands (Diagne et al., 2013).

Alnus glutinosa Gaertn., commonly known as black alder, belongs to this plant family and is an actinorrhizal plant that forms symbiosis with nitrogen-fixing actinomycete Frankia sp. (McEwan et al., 2017; Roy et al., 2017). Additionally, A. glutinosa can be colonized by both ectomycorrhizal (EM) and arbuscular (AM) fungi at the same time (Pritsch et al., 1997). Alnus sp. has high potential in forestry, land reclamation and biomass production (Roy et al., 2007). Alder trees belong to pioneer tree species and can grow in poor or disturbed soils and are well adapted to abiotic stresses, e.g., drought, salinity, and flooding (Diagne et al., 2013).

The ability of alders to form multifactorial symbiotic association can be a key factor improving their tolerance to saline stress conditions. The symbiotic relationship with Frankia sp. increases the soil fertility and enhances the performance of tree during their plantation under unfavorable conditions (Diagne et al., 2013). Alnus plants inoculated with Frankia sp. in general display improved plant growth, total biomass, nitrogen supply, and chlorophyll content in leaves tissues under saline conditions (Oliveira et al., 2005; Ngom et al., 2016). However, the selection of tolerant Frankia sp. strains is needed because this symbiosis is facultative for the actinobacteria (Põlme et al., 2013). Frankia strains vary in their sensitivity and response to salinity (Ngom et al., 2016) and the number of root nodules, vesicle production and nitrogenase activity is affected under saline stress conditions (Oshone et al., 2013).

Mycorrhizal fungi – ectomycorrhizal and arbuscular – play an important role in protection of plants against abiotic stress in the environment (Schützendübel and Polle, 2002; Meharg, 2003; Ruotsalainen et al., 2009). Mycorrhizal fungi prevent Na<sup>+</sup> and Cl<sup>−</sup> translocation to shoot and leaf tissues and enhance nutrients uptake, e.g., phosphorus (P) and nitrogen (N), in moderately salttolerant plants, e.g., black alder (Tang et al., 2009; Evelin et al., 2012), which indirectly increases plant growth and subsequent diminution of toxic ion effects (Tang et al., 2009). Mycorrhiza indirectly increase salinity tolerance by other mechanisms, e.g., by synthesis of phytohormones, amelioration of rhizosphere and bulk soil conditions (Asghari et al., 2005), enhancement of photosynthetic activity, water uptake (Hajiboland et al., 2010), accumulation of compatible solutes (Evelin and Kapoor, 2013), and production of higher levels of antioxidant enzymes (Manchanda and Garg, 2011). However, saline stress can affect mycorrhizal association (Hrynkiewicz et al., 2015) by decreasing fungal colonization capacity, the growth of fungal hyphae and germination of fungal spores in soil (Hameed et al., 2014). A broad taxonomic range of EM fungi can grow at ∼170 mM NaCl, although at a lower rate than in the absence of salt (Chen et al., 2001). This fact can suggest high salt resistance of many EM fungi (e.g., Dixon et al., 1993). Many reports describe successful application of salt-tolerant mycorrhizal fungi in plants protection against saline stress (Langenfeld-Heyser et al., 2007; Estrada et al., 2013; Talaat and Shawky, 2014; Sarwat et al., 2016). High concentrations of soluble salts have negative effect on microorganisms, e.g., decrease microbial activity and biomass and affect microbial community structure (Yan et al., 2015). However, this environmental factor can also predominantly stimulate occurrence of some salt tolerant bacterial and fungal strains, and the knowledge of changes in microbial community under salt stress conditions can be useful in development of new technologies used in forestry.

The aim of our study was to assess the effect of the unique ternary structure of alder symbionts on community of other rootassociated microorganisms (bacterial and fungal endophytes), particularly under seasonally changing salinity. Specifically, we hypothesize that: (i) one type of symbiosis will dominate, (ii) salinity and seasonality will affect diversity and species richness of plant-associated bacteria and fungi, and (iii) salinity can preferentially promote occurrence of certain groups of symbionts or halotolerant bacterial and/or fungal taxa. Broadening of the knowledge on alder's endophytes can open new horizons in plant-microbial interactions occurring in forests.

### MATERIALS AND METHODS

#### Site Description and Sampling

The study was carried out in mid-northern Poland at two test sites: a non-saline one in Pszczółczyn (53◦ 000 22.9<sup>00</sup> N, 17◦ 540 57.2<sup>00</sup> E; NS) and a saline one in Słonawy (53◦ 010 26.6<sup>00</sup> N, 17◦ 370 47.3<sup>00</sup> E; S) (Thiem et al., 2017). Salinity of soils in this area results from the impact of saline springs contacting with the Zechstein (Permian) salt deposits (Dadlez and Jaroszewski, 1994). The investigated saline site is located in the close vicinity of salt marshes (Solniska Szubinskie) belonging to the European ´ Ecological Network Nature 2000. Analyzed sites belong to the State Forests, outside of any protected areas. The control site Pszczółczyn (NS) is located 18 km away from the saline test site (S). Both sites are located in Odra drainage basin, elevation is 100 m over the sea level and the sites are flat (<1 ◦ slope). They are characterized by similar climatic, hydrological and pedological conditions and harbor tree stands of similar age [about 20 years old, planted in 1995 (S) and 1996 (NS), respectively]. Climate is temperate, mean annual temperature is 8.4◦C, and precipitation 520 mm. Groundwater table usually lies at lower than 1 m depth, the sites are located ∼1.5 km from the nearest river, and as such are not flooded by river waters. Soils are mineral-mucky developed on sands. Groundcover at

the NS site is typical for riparian forest and consists mainly of Urtica dioica, Stachys sylvatica, and Ranunculus lanuginosus, but also Rubus occidentalis could be found. At the S site the groundcover was even less diverse consisting almost exclusively of U. dioica.

Roots and soil samples (20 cm × 20 cm, 20 cm depth, litter layer was first removed) were collected from two test sites (NS and S) in two seasons of 2015 (spring – April and fall – September). In each test site three plots (10 m × 10 m) were selected, and within them three randomly selected trees were analyzed (9 trees per site). In total, 36 samples were analyzed (9 × 4 variants of experiment: NS-spring, NS-fall, S-spring, S-fall) (Supplementary Figure 1).

#### Soil Description and Analysis

The air-dried soil samples were passed through a 2 mm mesh and analyzed using the following methods: soil moisture content – gravimetrically, total organic carbon (TOC) and total nitrogen (TN) content using a CNS Vario MAX analyzer, pH (in H2O and 1 M KCl) by potentiometric method, and phosphorus in 1% citric acid solution (Pca) by colorimetric method. The saturation paste extracts were prepared to evaluate the soil salinity level. The electrical conductivity (ECe) was measured conductometrically and saturation percentage (SP) – gravimetrically (van Reeuwijk, 2006). Soil moisture content (M) was determined by drying to constant weight at 105◦C.

#### Metagenomic Analysis

#### Roots Cleaning and Preparation

First, soil was gently separated from roots to obtain rhizosphere soil for analysis of physico-chemical properties. Next, the pre-cleaned roots were thoroughly washed with sterile distilled water. Residual soil was separated from the roots with a sterile dissecting needle under magnifying glass and the roots were washed again with sterile distilled water. All steps were performed under sterile conditions. Ca. 500 mg samples were randomly collected from pools of cleaned roots and lyophilized.

#### Isolation of Metagenomic DNA

Total DNA was extracted from 50 mg of lyophilized black alder roots with the use of Plant & Fungi DNA Purification Kit (EURx, Poland) according to the manufacturer's protocol with the number of washing steps increased to four. Three technical replicates (independent DNA isolations) were prepared for each sample. Tubes with sterile glass beads (Mo Bio Laboratories) were used for plant material homogenization. The amount of isolated DNA was quantified fluorometrically (Qubit 2.0) and the quality was assessed spectrophotometrically (NanoDrop 2000) and the preparations were diluted to 1 ng/µl.

#### PCR Amplification of 16S rRNA Gene as Well as ITS Fragments and Sequencing

Bacterial 16S rRNA and fungal ITS amplicon libraries were generated in two-step PCR, first with the specific primers bearing M13/M13R overhangs (Goł˛ebiewski et al., 2014) (bacteria: u357f and u786r; fungi: uITS1 and uITS2) then with M13 and M13R primers with P5/P7 adapters and barcodes (different MID sequences for each sample) (**Table 1**). Mock community sample with known composition of bacterial species (Human Microbiome Project Mock Community A, BEI Resources) was also processed. Negative control (without DNA) and positive control (Escherichia coli K12 DH10B [NC\_00930.1] and Paxillus involutus [GQ389624.1] purified DNA) were included in each PCR round.

The first PCR reaction mix consisted of: 1 ng of DNA, 5 pmol of each primer, 4 nmol of each dNTP, 0.4 U of Phusion polymerase (Thermo Scientific), 100 µg of BSA and 1× concentrated buffer with 1.5 mM MgCl<sup>2</sup> in 20 µl. The cycling conditions were as follows: 98◦C – 30 s; 30 cycles of: 98◦C – 10 s, either 55◦C (bacteria) or 53◦C (fungi) – 15 s, 72◦C for 20 s; then 5 min at 72◦C. The PCR products were checked on 1.5% agarose gels in TBE and then they were purified using DNA Clean-Up Purification Kit (EURx) according to the manufacturer's protocol. Next, PCR products were quantified on Qubit 2.0 and diluted to 1 ng/µl.

The second PCR round was performed using Taq PCR Master Mix Kit (Qiagen) according to the manufacturer's protocol. The cycling conditions were as follows: 95◦C – 5 min, 14 cycles of: 95◦C – 30 s, 54◦C –15 s, 72◦C – 30 s; and finally 72◦C for 1 min. The products were checked again on 1.5% agarose gel, quantified with Qubit 2.0 (Thermo Scientific) and pooled in equimolar amounts.

Libraries were purified twice with Agencourt AMPure XP (Beckman Coulter) according to the manufacturer's protocol. The quality of the pooled libraries was assessed on a Bioanalyzer chip (Agilent) and they were quantified with KAPA Library Quantification Kit for Illumina Platform using LightCycler 480 (Roche) according to the manufacturers' protocols. The final pool was diluted to 4 nM, denaturated, mixed with 5% of PhiX control library and sequenced with the use of 2 × 300 cycles kit v.3 on a MiSeq machine (Illumina). Sequencing was performed using HPLC-purified versions of forward and reverse primers as well as reverse-complement of the reverse primer (**Table 1**).

#### Bioinformatic and Statistical Analyzes

The resulting read pairs were quality filtered with Sickle (Joshi and Fass, 2011), merged with Pandaseq (Masella et al., 2012) and denoised with BayesHammer (Nikolenko et al., 2013). The sequences were classified with naive Bayesian classifier (Wang et al., 2003) with SILVA Seed v.123 database then bacterial and non-bacterial ones were separated.

In case of the bacterial 16S rRNA sequences, the processing was performed essentially as described in Goł˛ebiewski et al. (2014). In brief: the sequences set was dereplicated, aligned to a template alignment (SILVA v. 123) and screened for the sequences covering the desired region of the alignment. Gap-only and terminal gap-containing columns were filtered out of the alignment, the set was pre-clustered to reduce the error rate and putative chimeras were identified with UCHIME (Edgar et al., 2011). The fungal sequences were processed with ITSx (Bengtsson-Palme et al., 2013), and all fungal ITS1 sequences were used in the downstream analyzes. The reads were dereplicated and OTUs were constructed using vsearch (Rognes

#### TABLE 1 | Primer sequences.

fmicb-09-00651 April 16, 2018 Time: 17:59 # 4


1 – u denotes M13/M13R tagged sequences; 2 – M13 and M13R sequences are given in boldface font; 3 – key sequence in italics; 4 – X denotes barcode (MID sequence).

et al., 2016) at 0.03 dissimilarity level, then singletons as well as doubletons (OTUs consisting of one or two sequences only) were removed. The sequences were classified with naive Bayesian classifier (minimum 80% bootstrap support was required; Wang et al., 2003) using SILVA v.123 (bacteria) and ITS1 extracted from UNITE v.7 (fungi), and the non-bacterial and non-fungal sequences were removed from the respective sets. The final data were subsampled to 500 (bacteria) and 300 (fungi) sequences per sample twenty times, sequences names were mangled to reflect the iteration, the sets were pooled, dereplicated, and OTUs were constructed as described earlier. OTU tables were then averaged over the twenty subsamples and the entries were rounded to the nearest integer with a Perl script to yield the final tables. Bray–Curtis distance matrices based on Wisconsin double-standardized OTU tables were calculated with vegdist in R. Non-metric multidimensional scaling (NMDS) and canonical correspondence analysis (CCA) analyzes were performed within R with vegan's metaMDS and cca functions. In case of CCA, forward selection procedure implemented in ordistep was used for model building. Significance of differences between sample clusters was assessed with ANOSIM and PERMANOVA in vegan's anosim and adonis functions, respectively. p-value < 0.05 was considered significant. Variance partitioning was performed with the varpart function.

Differences between soil parameters were analyzed by the non-parametric Kruskal–Wallis test and the Dunn test for post hoc comparison (Statistica ver. 7.1, StatSoft et al., 2006). Significance of differences in means (number of observed OTUs, Shannon's H', Shannon's E, taxa and functional groups distribution) was assessed with ANOVA with post hoc Tukey's HSD analysis, unless assumptions of normality of data and/or homogeneity of variance were violated, in which case Robust ANOVA implemented in raov of the Rfit package was used to check for general p-value. All figures were plotted with standard R graphic functions.

Bacterial sequences were classified to functional groups (nitrogen-fixing, nitrifying, denitrifying, halotolerant/halophilic) based on a BLAST analysis. Matches were considered significant when the alignment spanned over 99% of query length and identity was over 99% (details in Supplementary Material).

Fungal sequences classified down to the species level were assigned to functional groups [saprophytic (S), parasitic/pathogenic (P), endophytic (E), mycorrhizal (M)] based on information contained in NCBI's databases including source of isolation, host and data contained in references pertaining to a given sequence (details in Supplementary Material).

Possible metagenomes were imputed based on 16S rRNA fragments data with PICRUSt v.1.1.3 (Langille et al., 2013). Briefly, sequences were reclassified using Greengenes v.13\_8\_99 (DeSantis et al., 2006), then a BIOM file was produced with MOTHUR and made compatible with PICRUSt using BIOM tools<sup>1</sup> . Then, the BIOM file was normalized with normalize\_by\_copy\_number.py and metagenomes were predicted as well as NSTI scores calculated with predict\_metagenomes.py, using KEGG Orthology<sup>2</sup> as the base for predictions. Finally, an.spf file was prepared with biom\_to\_stamp.py. Predicted metagenomes were analyzed at the KEGG Orthologs (KO) level with STAMP v.2.1.3 (Parks et al., 2014), using the following parameters: remove unclassified reads, analysis of two groups, two sided Welch's test with Benjamini-Hochberg FDR correction. Q-value threshold was set to 0.01 and minimum ratio of proportions of five was used as a filter. PCA plot as well as barplot for the most significantly different categories were prepared.

#### RESULTS

#### Soil Physico-Chemical Parameters Differ Between Test Sites and Seasons

Physico-chemical soil parameters of samples are presented in **Table 2**. The EC<sup>e</sup> values at the S site ranged between 2.4 and 5 dS·m−<sup>1</sup> in spring, and they were significantly lower (∼1 dS·m−<sup>1</sup> ) in fall. At the NS site the values ranged between 0.55 to 1.13 dS·m−<sup>1</sup> . We supposed that the low EC<sup>e</sup> values at the saline site in fall were due to heavy rainfalls occurring right before sampling, which caused increase of soil water content and decreased salinity. It was confirmed by the higher level of soil moisture (M) at the S site, close to the values of SP. Similar phenomenon was observed at the NS site (**Table 2**). The soil samples from the S site were more acidic (pH-H2O 6.1 and 5.6 in spring and fall, respectively) than the NS ones (pH-H2O 6.7 and 6.5 in spring and fall, respectively). Soils at both sites were mineral, but the nutrients content (TOC, TN, and Pca) was consistently higher at the NS site (**Table 2**).

<sup>1</sup>https://github.com/rprops/PICRUSt\_from\_mothur <sup>2</sup>www.genome.jp


TABLE 2 | Physicochemical and chemical parameters of the studied soils (mean ± standard deviation) and Kruskal–Wallis test with the Dunn post hoc comparisons for analyzed variants (site: NS, non-saline; S, saline; season – fall, spring).

Variants labeled with the same letters are not significantly different (p ≤ 0.05). ECe, electrical conductivity of the saturated extract; pH-H2O – pH in water; pH-KCl, pH in 1M KCl; TOC, total organic carbon; TN, total nitrogen; Pca, phosphorus in 1% citric acid solution, M, soil moisture, SP, saturation percentage.

#### Sequencing Statistics

3 158 772 read pairs were obtained, of which 2 317 906 met the quality criteria and were merged. Of these, 546 961 sequences were classified as bacterial, and the rest (1 770 945) was regarded to be ITS sequences. 319 355 putative chimeras and 43 184 doubletons and singletons were removed from the bacterial dataset, leaving 184 422 sequences in the analysis (1–29 056 per sample).

ITSx correctly identified fungal ITS1 region in 863 931 putative ITS sequences. After classification with naive Bayesian classifier using the UNITE database, 588 198 non-fungal sequences were culled. They originated mainly from the host (Alnus glutinosa), demonstrating that the primers amplify not only fungal sequences. Finally, after removal of 11 264 singletons and doubletons 264 499 sequences were included in downstream analyses (2–18 341 per sample).

The reads, bacterial and fungal separately, were deposited in NCBI's SRA under accession no. SRP119174.

#### Microbial Community – Species Richness and Diversity Indices

Bacterial diversity, species richness as well as evenness for OTUs constructed at 0.03 dissimilarity threshold were lower at the S site both in spring and fall (**Figures 1A–C**). The opposite tendency was observed for fungal community, however, in fall the values were comparable at both sites, and they decreased at the S site and increased at the NS site (**Figures 1D–F**). Parameters measured for the bacterial community were always higher than for the fungal one: diversity ∼fourfold (bacteria: 4.3–5.0, fungi: 0.6–1.2), species richness ∼100 times (bacteria: 180–350, fungi: 18–34), and evenness ∼threefold (bacteria: 0.84–0.86, fungi: 0.21–0.34) (**Figure 1**).

#### Bacterial and Fungal Community Structure

Bacterial community was dominated by bacteria belonging to proteobacterial classes Alphaproteobacteria (NS: 36.4–38.7%, S: 27.0–28.3%) and Gammaproteobacteria (NS: 9.1–9.9%, S: 14.4–17.1%), Actinobacteria (NS: 18.0–19.4%, S: 16.0–16.6%) and Acidobacteria (NS:1.7–2.1%, S: 9.4–11.7%) (**Figure 2A**), however, the observed differences between the test sites were not significant. To the contrary, significant differences were observed at the genus level, where the most abundant genera differed between the NS and S sites: Bradyrhizobium and Rhizobium were more frequent at the NS site, while an unknown actinobacterial genus, Rhodanobacter, Granulicella, and Sphingomonas were more abundant at the S site (**Figure 2B**). At the level of order, sites differed significantly in abundance of Frankiales, comprising bacteria of genus Frankia, and greater abundance was observed at the S site (Supplementary Figure 2B). No significant differences between seasons were observed for all taxonomic levels.

Fungal community was dominated by Agaricomycetes (NS: 34.5–60.9%, S: 50.6–57.2%), Leotiomycetes (NS: 3.3–10.9, S: 13.7–17.4%), Mortierellomycotina Incertae Sedis (NS: 3.2–8.5%, S: 0.6–3.7%), and Sordariomycetes (NS: 06–2.1%, S: 0.9–1.7%) (**Figure 3A**). Similarly to bacteria, differences between sites and seasons at the class level were not significant, however, in general, greater abundance of Agaricomycetes was observed in fall, and Leotiomycetes were more frequent at the S site, and at the NS site in fall, while Mortierellomycotina Incertae Sedis and Sordariomycetes were most abundant at the NS site in fall, and at the S site in spring. At the genus level, Tomentella, Lactarius, Phialocephala, and Mortierella were the most frequent taxa, however, their share depended on site. Tomentella, Lactarius, and Phialocephala were more abundant at the S site, while Mortierella and Naucoria displayed the opposite tendency (**Figure 3B**). Tomentella, Lactarius, Thelephora, and Naucoria were the most frequently identified ectomycorrhizal fungi in all variants of the experiment (**Figure 3B**). Smaller number of sequences was noted for other EM fungal genera, such as Cortinarius or Amanita that were confounded within 'other fungi' group due to the low number of reads. Arbuscular mycorrhizal fungi were found in all samples and represented less than 1% of all identified fungi in each of them. They belonged to Glomeromycota (orders Paraglomales and Glomales, data not shown).

#### Microbial Communities Ordinations

Unconstrained ordination of bacterial community matrix (Bray– Curtis distance-based NMDS) demonstrated that the samples

were divided according to site and this grouping appeared to be significant according to ANOSIM and PERMANOVA analyses (p < 0.001) (**Figure 4A**). Notably, grouping according to season was not significant (p > 0.05). CCA, a method of constrained ordination (i.e., showing which environmental variables explain observed differences in community composition), showed that site together with Pca, pH, and TOC were significant environmental factors shaping the structure of bacterial communities in analyzed samples (**Figure 4B**). Percent variance explained by the variables found in the CCA analysis was assessed with variance partitioning. Total variance explained by the variable in question is followed by variance explained solely by this variable: site – 25.00% (7.00%), Pca – 15.60% (0.10%), pH – 14.40% (3.20%), TOC – 9.30% (1.00%).

Fungal communities, like bacterial ones, were grouped according to test site, and the grouping was significant (ANOSIM and PERMANOVA analyses, p < 0.001) (**Figure 4C**). Similarly to bacterial communities, CCA analysis revealed significant effect of site, Pca and pH (**Figure 4D**). Variance partitioning showed that season site explained 7.90% (3.70%), SP – 6.10% (3.90%), Pca – 5.80% (0.70%), pH – 5.10% (4.30%) of variance.

### Functional Analysis of Bacterial and Fungal Communities

Bacterial sequences were classified to functional groups potentially involved in nitrogen cycling based on their similarity to 16S rRNA genes of organisms of known function (**Figure 5A**). Sequences similar to those coming from denitrifiers were more frequent at the S site, while nitrifying and nitrogen-fixing ones were more abundant at the NS site. Counter intuitively, the potentially halotolerant taxa were more frequent at the NS site in spring (data not shown).

Fungal sequences were classified based on descriptions of species found in NCBI's databases (**Figure 5B**). The percentage of saprophytic and parasitic/pathogenic taxa appeared to be consistently low. Occurence of endophytes and mycorrhizal fungi turned out to be negatively correlated, although the correlation was not significant (Spearman's ρ = −0.14, p-value = 0.2351). Endophytes were more frequent in fall and at the NS site, while the opposite was true for mycorrhizal fungi.

#### PICRUSt Analysis

PCA analysis demonstrated that sets of KOs were different in genomes of organisms coming from the two investigated sites (Supplementary Figure 4).

FIGURE 4 | Analysis of log-transformed and Wisconsin double-standardized Bray–Curtis dissimilarity matrix for bacterial and fungal communities, respectively: (A) stress = 0.1430 and (C) stress = 0.1814 – NMDS (non-metric multidimensional scaling analysis), (B,D) – CCA (canonical correspondence analysis). Circles represent OTUs, their fill color denotes consensus taxonomy at the phylum level. Fifty most abundant OTUs were plotted in each case. Squares and triangles represent sites, their fill color means season. Area of figures is proportional to square root of EC<sup>e</sup> (electrical conductivity) in soil samples.

Altogether, 5101 KO categories were found by PICRUSt to be encoded in genomes of organisms thriving in alder root samples. Functional diversity (measured as the number of categories) did not differ significantly between samples groups, nor between sites (Kruskal–Wallis test, p > 0.05, Supplementary Table 1). Sixtysix KO categories were significant (**Figure 6**), five of them were represented more frequently in genomes of organisms coming from the S site. They were responsible for DNA replication and repair, biosynthesis and metabolism of pyrimidines as well as CoA biosynthesis. The remaining 61 categories were more frequent at the NS site. They were engaged in a plethora of processes, such as amino acids, peptides, carbohydrates, opines, and aromatic hydrocarbons metabolism, as well as type IV secretion, resistance to antibiotics and their biosynthesis (Supplementary Table 3).

### DISCUSSION

Each plant species hosts a genotype-specific microbiome (i.e., endophytic microbiome) that dynamically responds to the environment, e.g., soil quality (Podolich et al., 2015). A recent meta-analysis of soil microbial communities revealed that the global microbial composition in saline soils is affected more by salinity than by extremes of any other abiotic factor, e.g., pH or temperature (Lozupone and Knight, 2007). There is a direct relationship between soil and root salinity levels, which may significantly influence endophytic microbial community structure (Yaish et al., 2016b). However, the effect of salinity on endophytic communities is largely unexplored. Most of the studies on the endophytes were focused on nonwoody crops, moreover most of them used culture-dependent

methods (Shen and Fulthorpe, 2015; Szymanska et al., 2016 ´ ). Data on bacterial and fungal endophytes in roots of forest trees under saline stress are very limited (Ju et al., 2014; Hrynkiewicz et al., 2015; Thiem et al., 2017), and our work is the first report describing application of metagenomics to such a community with particular emphasis on root symbionts.

In general, we have noted more bacterial than fungal OTUs in alder roots, which may be caused by the 10-fold greater number of bacterial than fungal species observed in most soils (Larsen et al., 2017). Consistently bacteria showed higher diversity (measured as Shannon's H') than fungi. Moreover, our findings demonstrated significant differences in the endophytic microbial community composition due to the level of salt in the soil. Bacterial diversity, species richness, and evenness of black alder roots was decreased at the saline site (S), an effect similar to that observed by Yaish et al. (2016a) in Phoenix dactylifera. The lower average number of OTUs observed at the S site can be due to the fact that only a fraction of endophytes can thrive under conditions of increased salinity.

Black alder roots endophytic bacterial community was dominated by three phyla: Proteobacteria, Actinobacteria, and Acidobacteria that are the most common groups found in studies concerning soils (e.g., Goł˛ebiewski et al., 2014; Canfora et al., 2017), including saline ones (Valenzuela-Encinas et al., 2009; Ma and Gong, 2013). Our results are also in agreement with studies of Shakya et al. (2013) concerning root endophytes of Populus deltoides. Moreover, our studies showed that Alphaproteobacteria, Actinobacteria, and Betaproteobacteria were found more frequently at the NS site, which is consistent with report of Yaish et al. (2016b). Decrease of Alphaproteobacteria and Actinobacteria level in response to environmental stress is a common observation in soil studies (e.g., Chodak et al., 2013; Goł˛ebiewski et al., 2014). In our study, Acidobacteria and Gammaproteobacteria were more frequently found at the S site. It can be related to the higher abundance of this bacteria in saline soils (e.g., Yang et al., 2016; Yaish et al., 2016b).

Bacteria belonging to Rhodanobacter, Granulicella, and Sphingomonas genera were significantly more abundant in libraries coming from black alder roots from the S site. Rhodanobacter is a Gram-negative, aerobic bacterium currently not regarded as halotolerant. However, this bacterial species was found as a relatively abundant organism in A. glutinosa nodules (McEwan et al., 2017). Sphingomonas sp. belongs to a group of Gram-negative, chemoheterotrophic, aerobic bacteria that are widely distributed in nature, having been isolated from many different land and water habitats as well as from plant roots (Yang et al., 2016). Representatives of this genus make up a significant proportion of endophytic community in many trees (Izumi et al., 2008). They have the capacity to survive at low nutrient concentrations, as well as to metabolize a wide variety of carbon sources and some of them showed characteristics of nitrogen fixation and denitrification (Yang et al., 2016). Moreover, different species of Sphingomonas were shown to be endophytes effective in protection of crops against salts stress (Khan et al., 2014, 2017). Members of the Granulicella genus were found in the nodules of alder, similarly to Rhodanobacter (McEwan et al., 2017). Interestingly, according to Marupakula (2016), Rhodanobacter and Granulicella belong to bacteria colonizing ectomycorrhiza in boreal forest, which may suggest that they are mycorrhiza helper bacteria.

Frankia was found to be scarce in our study, and was found exclusively at the saline site. This might have been caused by a specific spatial distribution of these bacteria in plant roots, namely they are found mainly in nodules (Diagne et al., 2013). Roy et al. (2017) found 12 Frankia OTUs in A. glutinosa nodules,


FIGURE 6 | PICRUSt analysis of bacterial sequences. Barplots for most significantly different KO categories. Error bars represent standard error of the mean of given category abundance in different sample types. Categories belonging to DNA replication and repair supercategory and are shown in green boxes, while antibiotic synthesis and resistance is displayed in black boxes. Q-values (False Discovery Rates, based on two-sided Welche's test with Benjamini-Hochberg correction) are given to the right of each category plot.

however, the bacteria might constitute <1% of all sequences (McEwan et al., 2017), which is similar to our results. This scarcity might have been caused by bias imposed by universal primers used. The universal primers were used to obtain broad picture of the bacterial community, at the expense of underrepresentation of certain groups.

Bradyrhizobium and Rhizobium were more abundant at the NS site. Members of both of these bacterial genera are common microsymbionts of nodulating legumes (Guimarães et al., 2015). However, many data indicate that non-legumes also react to the presence of bradyrhizobia and rhizobia in the rhizosphere. Root hair curling induced by these symbiotic bacteria was observed on maize, rice and oat plants (Antoun et al., 1998). To date, Bradyrhizobium was found to be endosymbiont of only one tree species – Parasponia (Trinick and Hadobas, 1988). Bradyrhizobium sp. is relatively sensitive to unfavorable environmental conditions, e.g., salinity (it tolerates up to 0.5–1% NaCl) (Guimarães et al., 2015) or soil pH (Rascovan et al., 2016). A member of the Rhizobium genus (R. metallidurans) was isolated from roots of silver birch and alder growing on heavy metalscontaminated sites (Złoch et al., 2016). However, representative of these genera can be also robust heterotrophs that can persist in bulk soil (Gano-Cohen et al., 2016).

Classification of bacterial OTUs into functional groups indicated that sequences from those potentially involved in nitrogen cycling comprised together ∼20% of their total number. This high percentage is in accordance with the key role of bacteria in nitrogen cycling in soil (Boyle et al., 2008). However, it must be noted that such a kind of analysis is intrinsically limited, as there may exist organisms with closely related 16S rRNA sequences, yet differing in genome content and not performing the same ecological roles. Therefore, these results should be treated with caution. Nitrogen-fixing bacteria were significantly more frequent at the NS site, and these organisms belonged to Alphaproteobacteria (Rhizobium and Bradyrhizobium genera). It might be connected to possible greater abundance of these taxa in non-saline soils, as it is known that environmental stress decreases level of Alphaproteobacteria in soils (e.g., Chodak et al., 2013).

To learn more on possible functions contained in genomes of organisms found in our samples, PICRUSt analysis was performed. Possible gene content was imputed based on Greengenes classification of OTUs. This approach allows to get insight into metagenomes without performing shotgun sequencing, which is much more expensive than amplicon sequencing. However, it has its intrinsic limits, due to the lack of certain sequences in the database and the fact that even organisms closely related in terms of 16S rRNA sequences may significantly differ when it comes to the gene content. Moreover, one has to bear in mind that this kind of analysis yields potential functions that does not need to be expressed. Nevertheless, when carefully interpreted, PICRUSt may yield interesting results at a very low cost. Our results suggest that bacterial functional diversity does not differ between sites, and, as antibiotic biosynthesis and resistance genes were more frequent at the site, that competition is more intensive. This is supported by the greater diversity observed at the NS site. On the other hand, higher salinity at the S site might have selected organisms more adapted to extreme environments requiring greater capabilities of DNA repair.

The ITS region sequencing indicated a positive relationship between higher salinity and biodiversity. We suggest that this phenomenon can be caused by two mechanisms: (i) salt-sensitive fungi may more readily enter plant roots where the environment is more stable, and (ii) plants may attract endophytes under stress conditions, as they might increase their tolerance to the stressors.

These mechanisms are connected with plant reaction to abiotic stress. Differences in the outside environment put the plant metabolism out of homeostasis and impose the necessity to harbor specific genetic and metabolic traits within its cellular system (Meena et al., 2017). Plant microbiome often facilitates reduction of abiotic stress (Singh et al., 2011; Meena et al., 2017), and endophytes from different habitats confer habitatspecific stress tolerance to plants (Rodriguez et al., 2004). It is also known that plant physiology together with environmental physico-chemical parameters determine endophytic community structure (Hardoim et al., 2015).

Alder's endophytic fungal community was dominated by members of Agaricomycetes, Leotiomycetes, and Mortierellomycotina Incertae Sedis. In spite of large differences in taxa abundances between sites and seasons, statistical analysis indicated that the differences were not significant. This effect was caused by large variability among analyzed replicates, which might have been caused by specific spatial distribution of fungal mycelia, among them the mycorrhizal ones. We observed lower abundance of Agaricomycetes and Leotiomycetes at the NS site in spring. These groups comprise many ectomycorrhizal representatives that may be less abundant under conditions of high soil phosphorus content (Balzergue et al., 2013; Nouri et al., 2014). Our analysis revealed mainly ectomycorrhizal fungi, while arbuscular ones turned out to be rare in our dataset. This is probably due to two factors: (i) primers' specificity and (ii) the age of the tree stands under study. Arbuscular mycorrhizal structures usually are formed only at early stages of seedling development (van der Heijden, 2001; Põlme et al., 2013).

Tomentella, Lactarius, and Phialocephala members were more abundant at the S site. Tomentella was commonly found as an ectomycorrhizal partner of many trees, such as willow, birch, or alder, with tendency to dominate at unfavorable environmental conditions (Ishida et al., 2009; Regvar et al., 2010; Hrynkiewicz et al., 2015). Our results suggest that members of this genus can be tolerant to salinity. Mechanisms involved in Tomentella tolerance to abiotic environmental stressors still remain to be elucidated. Lactarius, belonging to russula-lactarius fungal linkages (Tedersoo et al., 2009), was frequently identified in alder roots (Kennedy and Hill, 2010; Rochet et al., 2011), but never in saline environment. Phialocephala spp. belong to dark septate endophytes (DSE) and dominate the endophytic mycobiota in roots of conifers and members of the Ericaceae family in heathlands, forests, and alpine ecosystems (Grünig et al., 2009). Phialocephala spp. may form mycorrhizal structures or live as endophytes (Lukešová et al., 2015).

Contrary to the above-mentioned fungal taxa, Mortierella and Naucoria were found more frequently at the NS site. Members of the Mortierella genus can grow on a wide variety of substrates,

including chitin (Leake and Read, 1990), and are commonly found as soil-inhabiting saprophytes (Kirk et al., 2008), however, some species were found to adopt endophytic lifestyle (Jankowiak et al., 2016). They were also found to increase effectiveness of mycorrhiza formation and to solubilize phosphorus (Zhang et al., 2011). This fungal genus is known to prefer high organic matter content (Wagner et al., 2013) and its abundance was greater at the non-saline site. It is thus possible that the higher level of extractable phosphorus at the NS site is related to the incidence of Mortierella. Genus Naucoria (also known as Alnicola) comprises many closely related endophytic species that colonize Alnus roots (e.g., Moreau et al., 2006; Rochet et al., 2011).

As salinity and other soil physico-chemical parameters differed significantly between the test sites, we expected that the identity of sites would be the key driver of microbial diversity. Indeed, unconstrained ordinations showed that the samples coming from the same site clustered together, moreover, this grouping appeared to be significant. This was corroborated by the CCA analysis, wherein site TOC, pH, parasitic/pathogenic (P), and SP were found to be significant factors explaining variance in microbial communities. Seasonality is commonly regarded as an important parameter influencing plant-associated microbial communities (Hrynkiewicz et al., 2015; Shen and Fulthorpe, 2015). This is due to the differences in nutrients levels and temperature, both factors can change microorganisms' abundance and community structure (Carrero-Colo et al., 2006), however, we did not see significant influence of season on the communities under study. This effect might be explained by relative stability of environment within plants. Moreover, as we analyzed the communities by amplification of rDNA, we were unable to capture changes in microorganismal activity that probably fluctuated seasonally.

#### CONCLUSION

Salinity affects both bacterial and fungal diversity, albeit in different manners. Bacterial diversity decreases with salinity, while the response of fungi to this parameter is more complex. Bacterial and fungal community structure depends on salinity, and seasonality does not appear to be an important factor explaining variance in communities of root alder endophytes. There are taxa that are more abundant at

#### REFERENCES


the saline site: Rhodanobacter, Sphingomonas, and Granulicella (bacteria); Tomentella, Lactarius, Phialocephala (fungi). We found a low number of sequences coming from obvious halotolerant/halophilic bacteria and fungi in our dataset. Ectomycorrhizal fungi (Tomentella, Lactarius, and Thelephora) are an important component of endophytic community of alder roots at the saline site. The lasting challenge would be to evaluate microbiome influence on nutrients cycling and plant physiology of alder at a greater number of sites differing in salinity. This could be done by means of shotgun metatranscriptomic analysis coupled with marker amplicons sequencing both at the DNA and RNA levels.

### AUTHOR CONTRIBUTIONS

DT did all the laboratory analyses and wrote the first version of the manuscript. MG designed the sequencing system, managed the lab experiments, performed the bioinformatic and statistical analyses, and participated in the preparation of the manuscript. PH did the soil chemical analyses and interpreted the results. AP selected the areas for field experiments, analyzed the M level in soils, and interpreted the results of soil analysis. KH designed and managed the field and lab experiments as well as participated in the preparation of the manuscript. All authors revised the manuscript and approved the publication.

### FUNDING

This investigation was conducted under the framework of COST Action (European Cooperation in Science and Technology) FP1305 Linking Belowground Biodiversity and Ecosystem Function in European Forests, and financially supported by the National Science Centre (Poland) PRELUDIUM 2016/23/N/NZ8/00294.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00651/full#supplementary-material



tomato (Solanum lycopersicum L.) plants. Plant Soil 331, 313–327. doi: 10.1007/ s11104-009-0255-z




**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 © 2018 Thiem, Goł˛ebiewski, Hulisz, Piernik and Hrynkiewicz. 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 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.

# Two P1B−1-ATPases of Amanita strobiliformis With Distinct Properties in Cu/Ag Transport

#### Vojtech Beneš, Tereza Leonhardt, Jan Sácký and Pavel Kotrba ˇ \*

Department of Biochemistry and Microbiology, University of Chemistry and Technology Prague, Prague, Czechia

As we have shown previously, the Cu and Ag concentrations in the sporocarps of Ag-hyperaccumulating Amanita strobiliformis are correlated, and both metals share the same uptake system and are sequestered by the same metallothioneins intracellularly. To further improve our knowledge of the Cu and Ag handling in A. strobiliformis cells, we searched its transcriptome for the P1B−1-ATPases, recognizing Cu<sup>+</sup> and Ag<sup>+</sup> for transport. We identified transcripts encoding 1097-amino acid (AA) AsCRD1 and 978- AA AsCCC2, which were further subjected to functional studies in metal sensitive Saccharomyces cerevisiae. The expression of AsCRD1 conferred highly increased Cu and Ag tolerance to metal sensitive yeasts in which the functional AsCRD1:GFP (green fluorescent protein) fusion localized exclusively to the tonoplast, indicating that the AsCRD1-mediated Cu and Ag tolerance was a result of vacuolar sequestration of the metals. Increased accumulation of AsCRD1 transcripts observed in A. strobiliformis mycelium upon the treatments with Cu and Ag (8.7- and 4.5-fold in the presence of 5 µM metal, respectively) supported the notion that AsCRD1 can be involved in protection of the A. strobiliformis cells against the toxicity of both metals. Neither Cu nor Ag affected the levels of AsCCC2 transcripts. Heterologous expression of AsCCC2 in mutant yeasts did not contribute to Cu tolerance, but complemented the mutant genotype of the S. cerevisiae ccc21 strain. Consistent with the role of the yeast Ccc2 in the trafficking of Cu from cytoplasm to nascent proteins via post-Golgi, the GFP fluorescence in AsCCC2-expressing ccc21 yeasts localized among Golgilike punctate foci within the cells. The AsCRD1- and AsCCC2-associated phenotypes were lost in yeasts expressing mutant transporter variants in which a conserved phosphorylation/dephosphorylation site was altered. Altogether, the data support the roles of AsCRD1 and AsCCC2 as genuine P1B−1-ATPases, and indicate their important functions in the removal of toxic excess of Cu and Ag from the cytoplasm and charging the endomembrane system with Cu, respectively.

Keywords: ectomycorrhizal fungi, P1-type ATPase, copper transporter, silver transporter, metal homeostasis, Amanita strobiliformis

#### INTRODUCTION

Studies have revealed that ectomycorrhizal (EM) fungi effectively mobilize heavy metals from soils and minerals (Gadd et al., 2012) and that ectomycorrhizae improve plant fitness in metal polluted environments also because metal tolerant mycobionts function as a barrier for the entry of metals into plant tissues (Colpaert et al., 2011; Reddy et al., 2016). High concentrations of heavy metals and

#### Edited by:

Erika Kothe, Friedrich-Schiller-Universität Jena, Germany

#### Reviewed by:

Christopher Rensing, Fujian Agriculture and Forestry University, China Michael Bölker, Philipps University of Marburg, Germany

> \*Correspondence: Pavel Kotrba pavel.kotrba@vscht.cz

#### Specialty section:

This article was submitted to Fungi and Their Interactions, a section of the journal Frontiers in Microbiology

Received: 28 September 2017 Accepted: 03 April 2018 Published: 23 April 2018

#### Citation:

Beneš V, Leonhardt T, Sácký J and Kotrba P (2018) Two P1B−1-ATPases of Amanita strobiliformis With Distinct Properties in Cu/Ag Transport. Front. Microbiol. 9:747. doi: 10.3389/fmicb.2018.00747

**186**

metalloids accumulated in the sporocarps further support the notion that EM fungi substantially contribute to the environmental cycling of these elements, including Cu and Ag (Falandysz and Borovicka, 2013 ˇ ). It is noteworthy that studies indicate that macrofungi could be considered the most effective Ag accumulators among eukaryotes with two known outstanding EM species, Amanita strobiliformis and Amanita solitaria (Borovicka et al., 2007, 2010 ˇ ). The concentrations of Ag in their sporocarps collected from unpolluted sites range from 200 to 1200 mg kg−<sup>1</sup> . We have documented that the intracellular detoxification of Cu and Ag in A. strobiliformis largely relies upon binding with cysteinyl-rich, cytosolic metallothionein (MT) peptides, AsMT1a, 1b, and 1c (Osobová et al., 2011; Beneš et al., 2016; Hložková et al., 2016); and that two A. strobiliformis transporters of the copper transporter family (CTR; specifically AsCTR2 and AsCTR3) can recognize not only Cu, but also Ag for uptake (Beneš et al., 2016).

Studies in eukaryotes have revealed that while CTRs transport Cu ions into the cytoplasm, the members of P1B−<sup>1</sup> subgroup of P1B-type ATPases (also called heavy metal ATPases, HMA) contribute to the homeostasis and redistribution of essential Cu by exporting the metal ion from the cytoplasm into the subcellular compartments or out of the cell (Nevitt et al., 2012; Bashir et al., 2016). The homology of P1B-ATPases and their characteristic sequence features suggest a division into seven subgroups (Smith et al., 2014). While the roles of the members of the P1B−<sup>5</sup> to P1B−<sup>7</sup> subgroups (predicted so far only in prokaryotes) remain elusive, the transporters belonging to P1B−1, P1B−2, prokaryote P1B−3, and P1B−<sup>4</sup> subgroups are known for distinct preferences for their substrate heavy metal ion(s). The transporters highly specific for monovalent Cu ions (the dominant intracellular Cu species in eukaryotes; Nevitt et al., 2012) comprise P1B−1-subgroup, while P1B−2, P1B−3, and P1B−<sup>4</sup> transport Cd2+/Zn2+/Pb2+, Cu+/Cu2+, and Co2+, respectively.

The intracellular handling of Cu involves in Saccharomyces cerevisiae Ccc2 protein (Bleackley and MacGillivray, 2011), and in mammals the Menkes protein ATP7A and Wilson protein ATP7B (La Fontaine and Mercer, 2007; Nevitt et al., 2012). These P1B−1-ATPases are responsible for the transport of the physiological Cu into the post-Golgi. Unlike with Ccc2 in S. cerevisiae, the Cu overload in mammalian cells triggers trafficking of ATP7A to the plasma membrane and ATP7B to the excretory vesicles, and both transporters then facilitate the efflux of the excess metal to rescue the cell from Cu toxicity. Similar trafficking [from the endoplasmic reticulum (ER) to the plasma membrane] stimulated by Cu overload has been documented in Arabidopsis thaliana for its AtHMA5 and heterologously expressed SvHMA5I from Silene vulgaris (Li et al., 2017). It is noteworthy that several P1B−1-ATPases have been shown to also recognize Ag for transport (Argüello et al., 2007; Smith et al., 2014; Migocka et al., 2015). Among fungi, the plasma membrane Cu+- and Ag+-efflux CaCRD1 of Candida albicans provides the primary source of cellular resistance against both metals (Riggle and Kumamoto, 2000; Weissman et al., 2000). Recently, the P1B−1-ATPase CrpA that also localizes to the plasma membrane has been shown to confer substantial Cu- but not Ag-tolerance in filamentous fungus Aspergillus nidulans (Antsotegi-Uskola et al., 2017).

Since our previous studies revealed certain overlap in the cell biology of Ag and Cu in A. strobiliformis, we investigated whether or not this species may employ P1B−1-ATPases in the intracellular handling of both Cu and Ag. We searched its transcriptome for the homologs of P1B−1-ATPases and describe here the isolation and functional characterization of cDNA coding the Cu- and Ag-inducible AsCRD1 that can protect metal-sensitive yeasts against the toxicity of both metals. We also describe the second isolated P1B−1-ATPase of A. strobiliformis, the homolog of yeast Ccc2 named AsCCC2. To our knowledge, these are the first P1B−1-ATPases characterized in mycorrhizal fungi.

#### MATERIALS AND METHODS

### Amplification of AsCRD1 and AsCCC2 Genes and Sequence Analyses

Partial sequences of AsCRD1 and AsCCC2 transcripts were obtained from tBLASTn analysis (Altschul et al., 1990) of the transcriptome of A. strobiliformis (Paulet ex Vittad.) isolate PRM 857486 (Hložková et al., 2016) by using characterized fungal P1B−1-type ATPases as queries. The entire coding sequence information was established by 5<sup>0</sup> and 3<sup>0</sup> RACE, using a SMARTer RACE cDNA Amplification Kit (Clontech Labs) with 1 µg of total RNA to produce the population of the first cDNA strand; the Q5 High-Fidelity DNA polymerase (New England Biolabs) was used to obtain double-stranded cDNAs. The total RNA was isolated by using an RNeasy Plant Mini Kit and RNase free DNase set (Qiagen) from 50 mg of freeze-dried tissue of the A. strobiliformis PRM 857486 sporocarp. Transcript-specific primers were 5rCRD1\_R1 to R5 for AsCRD1 5 <sup>0</sup> RACE, and 5rCRD2\_R1 to R3 or 3rCRD2R1 and R2 for AsCCC2 5 <sup>0</sup> or 3<sup>0</sup> RACE, respectively (for primer sequences see **Supplementary Table S1**), and the amplicons were subjected to 3<sup>0</sup> -A tailing with GoTaq DNA polymerase (Promega). Genomic fragments harboring AsCRD1 and AsCCC2 genes were amplified from 200 ng of chromosomal DNA template by PCR using Q5 DNA polymerase and pairs of gene-specific primers designed based on 5<sup>0</sup> and 3<sup>0</sup> untranslated regions of the corresponding cDNAs; the primers were CRD1\_F/R for AsCRD1 and CRD2\_F/R for AsCCC2 (**Supplementary Table S1**). The chromosomal DNA was isolated from 50 mg of freeze-dried tissue of A. strobiliformis PRM 857486 by using a NucleoSpin Plant II Kit (Macherey-Nagel). The amplicons were inserted to a pGEM-T vector (Promega) and then amplified in E. coli DH5α according to standard protocols. The recombinant DNAs were subjected to custom DNA sequencing on both strands with the vector-specific primers. The sequences of AsCRD1 and AsCCC2 cDNAs were deposited in GenBank under the accession numbers MF317930 and MF317931, respectively.

#### Sequence Analyses

The protein sequences deduced from the cDNAs were subjected to a transmembrane domain and signal peptide predictions

in silico at the CCTOP web server (Dobson et al., 2015). The signal peptide prediction was also done by submitting the sequences to SignalP 4.1 server (Pettersen et al., 2004). The homology modeling of transporter 3D structure used the Phyre2 protein homology/analogy recognition engine (Kelley et al., 2015), the Modeller (Webb and Sali, 2014), and UCSF Chimera (Pettersen et al., 2004) programs. The closest AsCRD1 and AsCCC2 homologs among the RCSB Protein Data Bank (PDB) entries used for comparative modeling were 2EW9 (N-terminal domain of ATP7B, 23% and 40% identity, respectively) and 3J09 (P1B−1-ATPase of Archaeoglobus fulgidus; 34% and 41% identity, respectively). A MEGA 6.0 package (Tamura et al., 2013) incorporating ClustalW (Thompson et al., 1994) was used to align AsCRD1, AsCCC2, and related amino acid (AA) sequences (retrieved from UniProtKB datase by using BLASTp) and construct the corresponding unrooted phylogenetic tree using the Neighborjoining method with Poisson correction model and 10,000 bootstrap replications.

#### Functional Complementation in Yeasts

The S. cerevisiae strains used in complementation assays were cup11 strain DTY113 (MATα trp1-1 leu2-3,-112 gal1 ura3-50 cup1161; Tamai et al., 1993) and the Euroscarf<sup>1</sup> Y00569 (yap11; YML007w::kanMX4) and Y03629 (ccc21; YDR270w::kanMX4) mutant strains of BY4741 (MATa his311 leu210 met1510 ura310). To constitutively express AsCRD1 and AsCCC2 in yeasts, the entire coding sequences produced by Q5 DNA polymerase from cDNA using primer pairs eifCRD1\_F/R (AsCRD1) and eifCRD2\_F/R (AsCCC2) were inserted into the HindIII-treated and EcoRI-treated yeast expression vector p416GPD (Mumberg et al., 1995), respectively, by using an In-Fusion HD Cloning Kit (Clontech Labs) according to manufacturer's instructions. Site-directed mutagenesis of AsCRD1 and AsCCC2 in p416GPD was performed by the inverse PCR method (Füzik et al., 2014) with Phusion High-Fidelity DNA Polymerase (Thermo Scientific); the overlapping primers used were mCRD1\_F plus mCRD1\_R and mCRD2\_F plus mCRD2\_R, respectively. Primer sequences are listed in **Supplementary Table S1**. The yeasts transformed with p416GPD-based plasmids were routinely grown at 30◦C on URA<sup>−</sup> selective SD medium containing (w/v) 0.7% yeast nitrogen base (Difco), 0.005% adenine hemisulfate, 2% glucose, and 0.003% of each of the essential amino acids (Sigma-Aldrich).

For complementation plate assays, the mid-log cultures of transformed S. cerevisiae were adjusted to an optical density at 590 nm (OD590) of 0.1, and 5 µl of serial dilutions were spotted on agar medium. The metal (added as CuCl<sup>2</sup> or AgNO3) tolerance of cup11 and yap11 transformants was assayed on SD medium and non-fermentable YPEG medium [1% (w/v) yeast extract, 2% ([w/v) peptone, 2.5% (v/v) ethanol and 2.5% (v/v) glycerol], respectively. The growth tests of ccc21 transformants used non-fermentable YPEG medium.

### Fluorescence Microscopy of AsCRD1:GFP and AsCCC2:GFP-Expressing Yeasts

To construct the translational AsCRD1:GFP and AsCCC2:GFP fusions, the coding sequences without the termination codons were amplified from cDNA by using primer pairs gifCRD1\_F plus gifCRD2\_R for AsCRD1 and gifCRD2\_F plus gifCRD2\_R for AsCCC2 (**Supplementary Table S1**). The amplicons were inserted into a BamHI-digested plasmid p416GFP. The plasmid p416GFP is a p416GPD derivative, harboring GFP from plasmid pEGFP-C1 (Clontech Labs) inserted as a BamHI/HindIII DNA fragment (Hložková et al., 2016). The cells of AsCRD1:GFPexpressing cup11 and AsCCC2:GFP-expressing ccc21 yeasts were obtained from mid-log cultures grown in SD medium supplemented with 0.5 µg·ml−<sup>1</sup> DAPI (Invitrogen) when needed. Vacuoles were labeled at 30◦C for 4 h in SD medium with 400 µg·ml−<sup>1</sup> of the tonoplast-specific FM4-64 dye (Molecular Probes). The fluorescence microscopy was performed by using a BioSystems Imaging station CellˆR with a MT20 illumination and a DSU semiconfocal unit on a IX-81 microscope (Olympus BioSystems) equipped with the model C9100 EM-CCD camera (Hamamatsu Photonix). A GFP-deriving fluorescence was observed with the U-DM-DA-FI-Tx2 FITC filter (excitation band: 495/15 nm, emission band: 530/30 nm; Olympus) and nuclei stained with DAPI were visualized with the U-DM-DA-FI-Tx2 DAPI filter (excitation band: 400/15 nm, emission band: 460/20 nm). Vacuoles were observed with the U-DM-Cy5 filter (excitation band: 590–650 nm, emission band: 665–740 nm). The recorded black and white images were processed using the ImageJ software<sup>2</sup> .

#### Gene Expression Analysis in A. strobiliformis

The mycelium isolate from the PRM 857486 pileus (Osobová et al., 2011) was grown at 25◦C and routinely maintained on potato dextrose (PD) agar containing 4 g·l <sup>−</sup><sup>1</sup> potato extract (Sigma-Aldrich) and 10 g·l −1 glucose (0.5× PD). The metal dose-dependent growth was observed with mycelia grown for 8 weeks on 0.5× PD agar with CuCl<sup>2</sup> or AgNO<sup>3</sup> supplements. The expression of target genes was assessed in the mycelium propagated in liquid PD medium (basal Cu, Ag and Cd concentrations below the atomic absorption spectrometry detection limit of 0.21, 0.04, and 0.09 µM, respectively) for 16 weeks and then subjected to metal (added as CuCl2, AgNO3, or CdCl2) exposures for 24 h. The gene expression analysis was performed on independent biological samples from three replicate experiments in two technical replicates. The RNA extraction from freeze-dried mycelia and quantitative reversetranscribed PCR measurements including the quality/specificity controls were conducted essentially as described previously (Hložková et al., 2016). Briefly, the population of transcripts present in 1 µg of total RNA was reverse transcribed in a 20 µl reaction and 1.5 µl of the resulting cDNA product was used in a 12 µl quantitative PCR (qPCR) reaction for the measurements

<sup>1</sup>http://web.uni-frankfurt.de/fb15/mikro/euroscarf/

<sup>2</sup>http://imagej.nih.gov/ij/

with 0.35 µM gene-specific primers (**Supplementary Table S1**). The measurements used a DyNAmo Flash SYBR Green 2- Step qPCR Kit (Life Technologies) and a MiniOpticon Real Time PCR System (Bio-Rad). The primers were qF- plus qR-CRD1 for AsCRD1, qF- plus qR-CRD2 for AsCRD1, and qFtub-b plus qRtub-b for β-tubulin AsTUB-b gene (GenBank: JX463743), which was used for normalization of the qPCR data as internal reference, stably expressed under Ag and Cu exposures (Hložková et al., 2016). A Bio-Rad CFX Manager was used to calculate the baseline range and the experiment threshold cycle (Cte) values recorded during the elongation period of the qPCR. The levels of gene transcription as relative to the controls (unexposed mycelium) were calculated by using the 2−11Ct1 method (Livak and Schmittgen, 2001), where Ct1 = Cte × [log(1+E)/log2]. The amplification efficiency values (E) were calculated using the equation E = [10(−1/slope) ]−1; the slopes were determined from the standard quantification curves obtained with serial dilutions of first strand cDNA templates. The obtained E values for AsCRD1, AsCCC2 and AsTUBb genes were 102%, 98%, and 108%, respectively.

#### RESULTS

#### Identification and Sequence Analysis of AsCRD1 and AsCCC2

To obtain information about the sequences coding for P1B−1- ATPases in A. strobiliformis, the sporocarp transcriptome of A. strobiliformis was screened by using tBLASTn search with known P1B−1-ATPases as queries. The screening retrieved two partial transcript sequences: one 822 nucleotides long in which a termination codon was included (a part of mRNA named AsCRD1) and another 528 nucleotides long without a termination codon (a part of mRNA named AsCCC2). As the deduced protein fragments showed a substantial identity with the C-terminal sequences of known P1B−1-ATPases, the corresponding full-length coding sequences were established via the RACE method.

The predicted 1097-AA AsCRD1 and 978-AA AsCCC2 proteins showed the characteristic sequence features of P1B−1- ATPases described in other organisms (Argüello et al., 2007; Smith et al., 2014). These involve putative N-terminal Cu/Agbinding CxxC motifs (three in AsCRD1, two in AsCCC2) and two P1B−<sup>1</sup> subgroup signature sequences in predicted transmembrane domains (TMD), Nx6YNx4P (x represents any AA residue), and Px6MxxSSx5S, which are in P1B−1-ATPases conserved in TMD7 and TMD8, respectively (**Figure 1** and **Supplementary Figure S1**). Like other P1B-type ATPases, AsCRD1 and AsCCC2 contained eight predicted TMDs with CPCx6P sequence in TMD6 and HP locus between TMD6 and TMD7. In addition, both predicted proteins possess features typical for all the members of the P-ATPase superfamily (**Figure 1**), particularly the DKTGTxT motif in the predicted large cytoplasmic loop with an aspartyl residue whose phosphorylation from ATP and dephosphorylation is prerequisite for active metal ion transport (Palmgren and Nissen, 2011). Despite the identified regions of conservancy at the protein level, the corresponding genes showed different structure and appeared dissimilar. The cDNA and genomic sequences of AsCRD1 and AsCCC2 were clearly distinct, with coding sequences interrupted with nine and three introns, respectively (**Figure 1**).

The comparison of the predicted AsCRD1 and AsCCC2 proteins revealed that along the sequence, they show lower identity and similarity with each other (25% and 38%, respectively) than they individually showed to P1B−1-ATPases characterized from other species. Predicted AsCRD1 shares 38%, 36%, and 31% identity (54%, 50%, and 48% similarity) with A. nidulans CrpA, C. albicans CaCRD1, and cucumber (Cucumis sativus) CsHMA5.2, respectively, while AsCCC2 shows 35% identity and 51% similarity with both the S. cerevisiae Ccc2 and A. thaliana AtHMA5. As further indicated in the Neighbor-joining tree (**Supplementary Figure S2**), AsCRD1

Beneš et al. Two P1B−1-ATPases of Amanita strobiliformis

and AsCCC2 sort into two distinct clusters. The AsCRD1 containing cluster comprised the characterized CaCRD1 and a clade of predicted agaricomycete P1B−1-ATPases. The second cluster involved clearly separated clades of mammalian and plant P1B−1-ATPases together with the AsCCC2-containing agaricomycete clade, which was more closely related to plant than to mammalian or yeast transporters. It is noteworthy that among the characterized transporters from Ascomycetes and Basidiomycetes, the closest relatives of AsCCC2 were P1B−1- ATPases from plant pathogens Botrytis cinerea (Saitoh et al., 2010) and Colletotrichum lindemuthianum (Parisot et al., 2002), and human pathogen Cryptococcus neoformans (Walton et al., 2005).

#### Functional Expression of AsCRD1 and AsCCC2 in S. cerevisiae

The homology to known fungal P1B−1-ATPases suggested that AsCRD1 and AsCCC2 are P1B−1-ATPases, which could be involved in metal tolerance and delivery of Cu to metalloproteins, respectively. In order to gain information regarding the function of AsCRD1 and AsCCC2 in handling Cu and Ag, the corresponding coding sequences were constitutively expressed in mutant S. cerevisiae strains grown on agar media with or without metal supplements. To attest the importance of the DKTGTxT motif in which the conserved aspartyl is in P-ATPases, a target of phosphorylation/dephosphorylation during the transport reaction cycle, the corresponding mutant AsCRD1D742A and AsCCC2D555A variants were constructed, in which the codons for aspartyl 742 (in AsCRD1) and aspartyl 555 (in AsCCC2) were changed to encode alanyl residues.

The Cu tolerance assays were conducted in the cup11 strain carrying a deletion of its single-copy MT gene cup1, which renders the cells hypersensitive to Cu. Heterologous expression in yeasts grown on SD medium containing 50 or 100 µM Cu2<sup>+</sup> revealed that only AsCRD1, but not AsCCC2, protected the yeasts form Cu toxicity (**Figure 2A**). The protective effect of AsCRD1 became weaker when the cells were subjected to 200 µM Cu2<sup>+</sup> (**Figure 2A**). Considering that Ag+, particularly in respiratory conditions, acts as a potent inducer of oxidative stress (Mijnendonckx et al., 2013), and yeasts with defects in oxidative stress response proved useful in attributing Agdetoxification functions to heterologous proteins (Sácký et al., 2014; Migocka et al., 2015), the yap11 strain, deficient in a transcription factor upregulating genes involved in oxidative stress response (Rodrigues-Pousada et al., 2010), was used in Ag toxicity assays. As documented in **Figure 2B**, the yap11 cells grown on non-fermentable YPEG medium and expressing AsCRD1 grew much better in the presence of 5–30 µM Ag<sup>+</sup> than did the controls. The observation that the expression of AsCRD1D742A did not confer increased resistance against either Cu (**Figure 2A**) or Ag (**Figure 2B**) suggested that the Cu- and Ag-tolerance phenotypes associated in the model yeasts with wild-type AsCRD1 were indeed due to the metal-transport ability of the encoded protein.

The apparent lack of the Ag/Cu toxicity-related phenotype of AsCCC2 in cup11 and yap11 yeasts was congruent with the expected function of AsCCC2 as the transporter involved in

the same expression vector inserted with AsCRD1 or AsCCC2, their translational fusions with GFP, or mutant variants (AsCRD1D742A , AsCCC2D555A). Metal tolerance assays were performed using SD medium with or without indicated metal supplement and assays with ccc21 strain were conducted on non-fermentable YPEG medium.

handling of physiological Cu inside the cell. The properties of AsCCC2 were thus further tested in the ccc21 strain in which the absence of Ccc2 causes a severe growth defect on nonfermentable media because of the lack of sufficient mitochondrial iron (Fu et al., 1995; Yuan et al., 1997); note that high affinity iron uptake pathway in S. cerevisiae involves Fet1 permease that works together with Cu-dependent, plasma membrane ferroxidase Fet3 that receives its Cu ions (supplied by Ccc2) in Golgi. The growth tests on YPEG medium revealed that AsCCC2 was able to fully complement the respiratory deficiency of the ccc21 cells, whilst the control cells transformed with empty p416GPD and those expressing AsCCC2D555A (and AsCRD1; not shown) failed to grow under the same conditions (**Figure 2C**). The

controls, AsCRD1 (not shown), and AsCCC2D555A cells showed full growth on the YPEG medium supplemented with 1 mM Cu2+, respectively.

### Targeting of AsCRD1 and AsCCC2 in S. cerevisiae

Distinct phenotypes associated with AsCRD1 and AsCCC2 in yeasts suggested that the corresponding proteins localized to different membranes. To obtain information about the cellular localization of AsCRD1 and AsCCC2 using direct fluorescence microscopy, the proteins were translationally fused with GFP at their C-termini, and the recombinant AsCRD1:GFP and AsCCC2:GFP genes were expressed in cup11 and ccc21 yeasts grown in SD medium. Complementation assays revealed that the phenotypes conferred by the fusions upon the yeasts were essentially the same as those observed with the corresponding transporters without GFP (**Figure 2**), thereby indicating that AsCRD1 and AsCCC2 tagged with GFP at their C-termini remained functional.

The microscopy of AsCRD1:GFP-expressing cup11 yeasts revealed strong GFP fluorescence co-localizing exclusively with the tonoplast stained with the vacuole-specific fluorophore FM4-64 (**Figure 3A**). The expression of AsCCC2:GFP in the ccc21 strain resulted in a strong, punctuated GFP signal in vesicular bodies within the cell (**Figure 3B**). The absence of GFP fluorescence from the perinuclear region attributable to ER may suggest that AsCCC2:GFP localized to Golgi rather than ER. The localization of GFP fluorescence in AsCRD1:GFPand AsCCC2:GFP-transformed yeasts was not affected by the presence of subtoxic concentrations of Cu or Ag or the length of culture period (not shown).

#### Metal Responsiveness of AsCRD1 and AsCCC2 in A. strobiliformis

Considering the AsCRD1-associated, metal tolerance-related phenotypes in the model yeasts and the typically induced expression of metal tolerance genes during metal overload, the transcription rates of the studied P1B−1-ATPases genes were analyzed by using qRT-PCR, measuring mRNA levels in the mycelium of A. strobiliformis treated with 5 and 50 µM Cu2+, 5, 20, and 50 µM Ag+, or 5 µM Cd2<sup>+</sup> for 24 h. The Cu and Ag concentrations used in the 24-h exposures proved sublethal also in long-term exposures (**Figure 4A**), although the radial growth of mycelia was strongly reduced (by 70%) in the presence of 50 µM Ag. The mycelia always developed brown zones already at 5 µM of any of the metals, presumably due to the induced production of stress-related melanin (Gostincar et al., 2012 ˇ ).

As shown in **Figure 4B**, 24 h treatments of mycelia with Ag and Cu clearly elevated the expression of AsCRD1, but not AsCCC2, relative to the unexposed control. The average levels of AsCRD1 transcripts increased 4.5- and 8.7-fold in the presence of 5 µM Ag<sup>+</sup> and Cu2+, respectively, and they further nearly doubled when the concentration of the two metals was 50 µM. Neither AsCRD1 nor AsCCC2 showed significant response when the mycelia were treated with a 5 µM concentration of Cd2<sup>+</sup> (**Figure 4B**), which in A. strobiliformis induces the expression of

Zn2+/Cd2+-related MT gene AsMT3, but not Cu+/Ag+-related AsMT1s (Hložková et al., 2016).

### DISCUSSION

fluorescence, and GFP/DAPI merged image.

Our previous studies revealed a certain overlap in the cellular biology of Cu and Ag in the EM, Ag-hyperaccumulating fungus A. strobiliformis – both metals can enter the cells via AsCTR2 and AsCTR3 transporters (Beneš et al., 2016) and intracellular Cu and Ag are sequestered in the cytoplasm through binding with AsMT1s (Hložková et al., 2016). It is worth noting that MTs have been considered principal in the sequestration of Cu or Ag in many EM fungi, including Pisolithus albus (Reddy et al., 2016), Laccaria bicolor (Reddy et al., 2014), Hebeloma mesophaeum (Sácký et al., 2014), Hebeloma cylindrosporum (Ramesh et al., 2009), Amanita submembranacea (Borovicka et al., 2010 ˇ ), and Paxillus involutus (Bellion et al., 2007). The present study aimed to identify P1B−1-ATPases of A. strobiliformis and inspect their potential role in the handling of intracellular Cu and Ag in this species. Our search of the sporocarp transcriptome suggested the presence of several putative P1B-ATPases of which only two showed sequence features characteristic of the P1B−<sup>1</sup> subgroup.

Unlike for Zn or Cd, information about the deposition of the excess of the accumulated Cu in fungal vacuoles is scarce. In Aspergillus niger (Fomina et al., 2007) and in arbuscular mycorrhizal Rhizophagus intraradices

(González-Guerrero et al., 2008), the vacuolar sequestration of excess Cu was revealed by X-ray microanalyses, which further suggested the association of Cu with vacuolar polyphosphate in A. niger. The vacuole is an important organelle for Cu homeostasis in S. cerevisiae and the strains defective in vacuolar assembly are hypersensitive to Cu (Szczypka et al., 1997). While the transporters of the CTR family responsible for the mobilization of the vacuolar Cu back into the fungal cytoplasm are well characterized (e.g., Ctr2 in S. cerevisiae; Bleackley and MacGillivray, 2011), the Cu-specific, high-affinity transporters that can deliver Cu into the vacuoles remained elusive.

Besides the sequence features common in P-ATPases, in particular of the P1B-subtype, several lines of experimental evidence implicate that AsCRD1 can act as a detoxification P1B−1-ATPase that can transport Cu<sup>+</sup> and Ag<sup>+</sup> into vacuoles in A. strobiliformis. First, the expression of AsCRD1, but not AsCRD1D742A, protected the model yeasts from Cu and Ag toxicity. The observation that the replacement of aspartyl with alanyl in the DKTGTxT motif (to prevent the phosphorylation in AsCRD1D742A from ATP) abolished the AsCRD1-associated phenotype in both cup11 and yap11 yeast mutants further indicates that AsCRD1 can recognize both Cu and Ag for an active, ATP-dependent transport, and that it was the metal transport that increased the metal tolerance in the yeasts, not a mere immobilization of Cu<sup>+</sup> or Ag<sup>+</sup> through binding to cytoplasmic N-terminal metal binding motifs as it is the case, e.g., of Cu-binding to the Cd-transporting PCA1 in S. cerevisiae (Adle et al., 2007). Second, the functional GFP-tagged AsCRD1 was targeted to the tonoplast in model yeasts. Although vacuolar P1B−1-ATPases have not been described in fungi before, such localization is not without precedent. Recent studies in plants have identified the Cu-transporting S. vulgaris SvHMA5II (Li et al., 2017), and cucumber (Cucumis sativus) Cu- and also Agactivated CsHMA5.1 and CsHMA5.2 proteins (Migocka et al., 2015), as tonoplast-localizing P1B−1-ATPases facilitating metal detoxification in root cells. Third, the observation that the expression of AsCRD1 was in A. strobiliformis effectively induced by Cu and Ag makes it reasonable to assume that AsCRD1 is involved in the cellular biology of both metals and the fungus raises the levels of AsCRD1 to handle excess intracellular Cu and Ag. Considering that our previous metal speciation analyses using size exclusion chromatography revealed that the majority of the Ag and Cu accumulated in A. strobiliformis is stably bound in 6-kDa complexes with Ag- and Cu-inducible, cytosolic AsMT1s (Osobová et al., 2011; Beneš et al., 2016; Hložková et al., 2016), one may then ask the question of what role AsCRD1 would have in metal detoxification. We propose that vacuolar storage could provide the second line of defense against high intracellular Ag and Cu levels, perhaps during a temporal deficiency of Cu+- and Ag+-binding AsMT1s, akin to the function of zincosome vesicles acting as transient stores of the excess accumulated Zn in S. cerevisiae (Devirgiliis et al., 2004). However, considering the plasma membrane localization of the closely related CaCRD1 in C. albicans (Riggle and Kumamoto, 2000; Weissman et al., 2000) and CrpA in A. nidulans (Antsotegi-Uskola et al., 2017), the possibility that AsCRD1 mislocalizes in S. cerevisiae and in A. strobiliformis acts as a transporter that exports the excess Cu and Ag out of the cells should not be excluded.

The predicted AsCCC2 and its homologs from Agaricomycetes appeared phylogenetically associated with the Ccc2 protein from the unicellular basidiomycete C. neoformans and to a lesser extent with Ccc2s from ascomycetes B. cinerea, C. lindemuthianum and S. cerevisiae. Congruent with this observation, AsCCC2 functionally complemented the CCC2 gene in S. cerevisiae ccc21 that is unable to charge its multicopper oxidase Fet3 with Cu in Golgi to establish the Fet3-Ftr1-based iron uptake system (Bleackley and MacGillivray, 2011). The lack of the AsCCC2-associated phenotype resulting from the D-to-A substitution in the DKTGTxT motif of the encoded protein (in the ccc21 cells expressing AsCCC2D555A), and the GFP fluorescence localizing to the intracellular punctuate bodies resembling Golgi in yeasts expressing AsCCC2:GFP provides further support to the notion that AsCCC2 can mediate active transport of Cu into the Golgi. In C. neoformans, B. cinerea, and C. lindemuthianum, the corresponding functional CCC2 gene appeared critical for the biosynthesis of melanin; the lack of CCC2 in these species lead to a disruption in the delivery of Cu to extracellular multicopper oxidases

(laccases in particular) during their trafficking through Golgi (Parisot et al., 2002; Walton et al., 2005; Saitoh et al., 2010). Multiple copies of laccase genes have been predicted in both saprobic and EM species (Kohler et al., 2015); for example, the genomes of saprobic Amanita thiersii and EM Amanita muscaria contain 15 and 18 putative nonallelic laccase genes, respectively. Recent studies indicate that laccases expressed in EM fungi are, besides the pigmentation, involved in the sporocarp development or nutrient acquisition in extraradical mycelia (Courty et al., 2009; Kües and Rühl, 2011; Ellström et al., 2015; Shah et al., 2016). Considering this and the fact that Fet3-like ferroxidase genes have been found in most sequenced basidiomycetes, including Amanita species (Kües and Rühl, 2011; Kohler et al., 2015), it could be possible that A. strobiliformis benefits from AsCCC2 for both the Fe-uptake complex and laccase(s) assembly via Cu handling.

The results obtained in this study indicate that AsCRD1 and AsCCC2 belong to two separate protein clusters of the P1B−1-ATPase subgroup. The collected data strongly suggest that AsCRD1 is in A. strobiliformis, like AsMT1s and AsCTRs, involved in the handling of both Ag and Cu, specifically in supporting the detoxification of Ag and Cu, which is, besides efficient transport, the prerequisite for (hyper)accumulation. Our data further indicate that AsCCC2, identified as another P1B−1-ATPase of A. strobiliformis, is a functional homolog of yeast Ccc2, involved in the delivery of physiological Cu into organelles of endomembrane system for the biosynthesis of Cudependent proteins. It is worth noting that BLASTp returned putative P1B−1-type ATPases of Agaricomycetes species in which homologs of AsCRD1 and AsCCC2 were identified. These species belong to different orders (**Supplementary Figure S2**) of different lifestyles. It is thus tempting to speculate that the functional specialization and roles of P1B−1-type ATPases, which we here discussed for AsCRD1 and AsCCC2, are widespread among Agaricomycetes.

#### AUTHOR CONTRIBUTIONS

VB conducted the experimental work and analyzed and interpreted data. TL and JS jointly contributed to the conception and design of the study, the bioinformatic analyses, and helped with the interpretation of data. PK was responsible for the concept and design of the work and the interpretation of the results, ensured the scientific issue was appropriately investigated, and wrote the manuscript. All of the authors assisted in writing

#### REFERENCES


the manuscript, discussed the results, and commented on the manuscript.

### FUNDING

This work was supported by the Czech Science Foundation through grant no. 16-15065S. Publication fees of the work have been co-financed by the endowment from the Ministry of Education, Youth and Sports of Czechia for the institutional development plan at UCT Prague.

#### ACKNOWLEDGMENTS

We are grateful to Prof. Dennis J. Thiele (Duke University Medical Center) for the gift of DTY113 (cup11) strain. We thank Dr. Jan Borovicka (Institute of Geology and Nuclear ˇ Physic Institute, CAS) for the valuable discussions.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00747/full#supplementary-material

FIGURE S1 | (A) Comparative sequence analysis of the predicted AsCRD1 and AsCCC2 proteins. Sequences were aligned by using ClustalW and the identical, conservative, and semiconservative residues were marked with asterisks, double dots, and single dots, respectively. The predicted transmembrane domains (TMD, numbered) are highlighted with light blue background and the characteristic sequence motifs are boxed in yellow (metal binding CxxC motif), green (CPCx6P in TMD6, Nx6YNx4P in TMD7, Px6MxxSSx5S in TMD8, and HP locus), and red (DKTGTxT motif). The signal peptide predictions in AsCRD1 and AsCCC2 at CCTOP and Signal P 4.1 servers did not unveil any potential signal sequences. (B) 3D homology models of AsCRD1 and AsCCC2. The PDB entries used for the comparative modeling of AsCRD1 and AsCCC2 were 2EW9 (N-terminal domain of ATP7B, 23% and 40% identity, respectively) and 3J09 (P1B−1-ATPase CopA of Archaeoglobus fulgidus; 34% and 41% identity, respectively). The positions of characteristic sequence motifs are indicated with arrows.

FIGURE S2 | An unrooted, neighbor-joining-based tree of characterized and predicted P1B−1-ATPases. Species name and UniProt accession numbers of functionally characterized ascomycete, plant and mammalian, and predicted Agaricomycetes P1B−1-ATPases (30% to 40% identical with AsCRD1 or AsCCC2; UniProt expect values of 0 to 10−97) are indicated. The tree was generated by MEGA version 6.0 after the sequence alignment by using ClustalW. Bootstrap values (%; 1,000 replicates) are shown at nodes (values <40% are omitted for clarity) and branch lengths are proportional to phylogenetic distances.

TABLE S1 | Primers used in this study.

by the transcription factor AceA. Front. Microbiol. 8:912. doi: 10.3389/fmicb. 2017.00912


expression in Hebeloma cylindrosporum. New Phytol. 174, 151–158. doi: 10.1111/j.1469-8137.2007.01973.x


genes in mycorrhizal mutualists. Nat. Genet. 47, 410–415. doi: 10.1038/ng. 3223



**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 © 2018 Beneš, Leonhardt, Sácký and Kotrba. 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 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.