# MOLECULAR AND CELLULAR MECHANISMS OF THE LEGUME-RHIZOBIA SYMBIOSIS

EDITED BY : Jianping Wang, Stig Uggerhøj Andersen and Pascal Ratet PUBLISHED IN : Frontiers in Plant Science and 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 978-2-88945-757-1 DOI 10.3389/978-2-88945-757-1

### 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

# MOLECULAR AND CELLULAR MECHANISMS OF THE LEGUME-RHIZOBIA SYMBIOSIS

Topic Editors:

Jianping Wang, University of Florida, United States; HIST, Fujian Agriculture and Forestry University, China Stig Uggerhøj Andersen, Aarhus University, Denmark Pascal Ratet, Institute of Plant Sciences Paris-Saclay IPS2, CNRS, INRA, Université Paris-Sud, Université Evry, Université Paris-Saclay, Paris Diderot, Sorbonne Paris-Cité, France

Image: Worachat Tokaew/Shutterstock.com

The legume-rhizobia symbiosis is a remarkable biological phenomena, which is critically important for sustainable agriculture. In the past decades, significant progress has been made to understand the mechanisms of the symbiotic processes. In this eBook, we present the most recent researches focusing on the molecular mechanisms of legume-rhizobia symbiosis including rhizobium characters, plant responses to different types of bacteria, phytohormones involved in the symbiosis, SYM pathway signals, and R genes and specificity of rhizobia infection. This eBook will be a great reference book highlighting the research frontiers in legume-rhizobia symbiosis.

Citation: Wang, J., Andersen, S. U., Ratet, P., eds. (2019). Molecular and Cellular Mechanisms of the Legume-Rhizobia Symbiosis. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-757-1

# Table of Contents

*05 Editorial: Molecular and Cellular Mechanisms of the Legume-Rhizobia Symbiosis*

Jianping Wang, Stig Uggerhøj Andersen and Pascal Ratet

### SECTION I

### RHIZOBIUM CHARACTERS

*08 The Very Long Chain Fatty Acid (C26:25OH) Linked to the Lipid A is Important for the Fitness of the Photosynthetic* Bradyrhizobium *Strain ORS278 and the Establishment of a Successful Symbiosis With*  Aeschynomene *Legumes*

Nicolas Busset, Flaviana Di Lorenzo, Angelo Palmigiano, Luisa Sturiale, Frederic Gressent, Joël Fardoux, Djamel Gully, Clémence Chaintreuil, Antonio Molinaro, Alba Silipo and Eric Giraud

*21 Type 3 Secretion System (T3SS) of* Bradyrhizobium *sp. DOA9 and its Roles in Legume Symbiosis and Rice Endophytic Association* Pongpan Songwattana, Rujirek Noisangiam, Kamonluck Teamtisong, Janpen Prakamhang, Albin Teulet, Panlada Tittabutr, Pongdet Piromyou, Nantakorn Boonkerd, Eric Giraud and Neung Teaumroong

### SECTION II

### PLANT RESPONSES TO DIFFERENT TYPES OF BACTERIA

*33 Distinct* Lotus japonicus *Transcriptomic Responses to a Spectrum of Bacteria Ranging From Symbiotic to Pathogenic* Simon Kelly, Terry Mun, Jens Stougaard, Cécile Ben and Stig U. Andersen

### SECTION III

### PHYTOHORMONES INVOLVED IN SYMBIOSIS


Celine Mens, Dongxue Li, Laura E. Haaima, Peter M. Gresshoff and Brett J. Ferguson

*69 Acropetal Auxin Transport Inhibition is Involved in Indeterminate but not Determinate Nodule Formation*

Jason L. P. Ng and Ulrike Mathesius

### SECTION IV

### SYM PATHWAY SIGNALS


Edgard Jauregui, Liqun Du, Cynthia Gleason and B. W. Poovaiah

### SECTION V

### R GENES AND SYMBIOTIC SPECIFICITY

*105 The Soybean* Rfg1 *Gene Restricts Nodulation by Sinorhizobium*  fredii *USDA193*

Yinglun Fan, Jinge Liu, Shanhua Lyu, Qi Wang, Shengming Yang and Hongyan Zhu

# Editorial: Molecular and Cellular Mechanisms of the Legume-Rhizobia Symbiosis

#### Jianping Wang1,2 \*, Stig Uggerhøj Andersen<sup>3</sup> and Pascal Ratet 4,5

<sup>1</sup> Key Laboratory of Genetics, Breeding and Multiple Utilization of Crops, Ministry of Education, Fujian Provincial Key Laboratory of Haixia Applied Plant Systems Biology, Center for Genomics and Biotechnology, Fujian Agriculture and Forestry University, Fuzhou, China, <sup>2</sup> Agronomy Department, University of Florida, Gainesville, FL, United States, <sup>3</sup> Department of Molecular Biology and Genetics, Aarhus University, Aarhus, Denmark, <sup>4</sup> Institute of Plant Sciences Paris-Saclay IPS2, CNRS, INRA, Université Paris-Sud, Université Evry, Université Paris-Saclay, Orsay, France, <sup>5</sup> Institute of Plant Sciences Paris-Saclay, Paris Diderot, Sorbonne Paris-Cité, Orsay, France

Keywords: legume, rhizobium, nodule, symbiosis, SYM pathway, biological nitrogen fixation

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

### **Molecular and Cellular Mechanisms of the Legume-Rhizobia Symbiosis**

### Edited by:

Andrea Genre, Università degli Studi di Torino, Italy

#### Reviewed by:

Beatrice Lace, University of Freiburg, Germany Joëlle Fournier, UMR2594 Plant Interactions Laboratory Microorganisms (LIPM), France

> \*Correspondence: Jianping Wang wangjp@ufl.edu

#### Specialty section:

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

Received: 14 September 2018 Accepted: 27 November 2018 Published: 12 December 2018

#### Citation:

Wang J, Andersen SU and Ratet P (2018) Editorial: Molecular and Cellular Mechanisms of the Legume-Rhizobia Symbiosis. Front. Plant Sci. 9:1839. doi: 10.3389/fpls.2018.01839 Legume-rhizobia symbiosis is a remarkable and mutually beneficial association between higher plants and microbes, which is extremely important for sustainable agriculture and ecology. During this association, biological nitrogen fixation occurs in the nodule, which is a specialized accessory legume organ, generally formed on roots. In mature nodules, rhizobia convert inert atmospheric N<sup>2</sup> into ammonia (NH3), essential for plant growth. In return, bacteria obtain photosynthetic carbon from the plant. The biologically fixed nitrogen during symbiosis accounts for approximately 65% of nitrogen use in agriculture (Burris and Roberts, 1993).

Due to its agricultural importance and interesting biological features, the legume-rhizobia symbiosis fascinated many researchers in the past decades. It has been discovered that the legume-rhizobia symbiosis is initiated through mutual chemical communication. The plant root releases flavonoids, which induce rhizobia to produce specific lipo-chitooligosaccharides (LCOs), called nodulation factors (Nod factors, or NFs) (Dénarié et al., 1996). These NFs are specifically recognized by the host plant and trigger a sophisticated symbiotic signaling cascade in the host root cells to coordinate rhizobial infection, nodule organogenesis and later on nitrogen fixation (Oldroyd et al., 2011). The main aim of this research topic was to assemble papers addressing and discussing the fundamental science of the molecular mechanisms of legume-rhizobia symbiosis. A total of nine articles were published under this research topic related to rhizobium characters (2), plant response to different type of bacteria (1), phytohormones (3), SYM pathway signals (2), and R genes determining specificity of rhizobia infection (1).

Rhizobia have two lifestyles, a free-living state in soil and a symbiotic state in plant cells when engaged in symbiosis (Gibson et al., 2008). Rhizobia therefore have to adapt to changing conditions outside and inside the host plants. Besides LCOs, a few other bacterial components are important for a successful symbiotic interaction. Lipopolysaccharides (LPSs) are major components of the outer membrane of gram negative bacteria including rhizobia and play a central role in rhizobia infection, in adaptation to the host environment, and in facilitating symbiosis (Lerouge and Vanderleyden, 2002; Raetz and Whitfield, 2002). Busset et al. characterized a cluster of five genes presumably involved in LPS biosynthesis and investigated the specific roles of one gene (lpxXL) in Bradyrhizobium free-living and symbiotic states.

Some rhizobia can also use a type 3 secretion system (T3SS) to translocate bacterial effector proteins into host cells to trigger and facilitate symbiosis in certain legume species (Jiménez-Guerrero et al., 2015; Okazaki et al., 2016). Songwattana et al. examined the role of a T3SS present on a Bradyrhizobium symbiotic plasmid. The symbiotic performance of a T3SS mutant Bradyrhizobium strain was tested on 9 different legume species. In addition, the role of the R3SS in Bradyrhizobium endophytic colonization of rice roots was examined. This work confirmed that the T3SS plays significant roles in symbiosis for certain legume species.

Legume plants, when perceiving compatible NFs, respond very rapidly to rhizobial infection at the physiological and molecular levels (Desbrosses and Stougaard, 2011; Oldroyd, 2013). However, how legumes differentiate compatible from incompatible rhizobia or pathogens was not clear. Kelly et al. investigated the transcriptome profiles of Lotus japonicus in response to a spectrum of interacting bacteria, including compatible symbiotic Mesorhizobium loti, semi-compatible symbiotic Sinorhizobium fredii, incompatible symbiotic Bradyrhizobium elkanii, and pathogenic Pseudomonas syringae and Ralstonia solanacearum. This comparative analysis revealed distinct transcriptional responses of the legume plant to different types of bacteria and challenged the concept that an early defense-like response is evoked by compatible rhizobia in L. japonicus during the establishment of symbiosis.

Rhizobial infection and organogenesis events during legumerhizobia symbiosis are tightly coordinated, and defects in the infection process often result in defective nodule organogenesis. The tight coordination with signaling across root layers strongly suggests the involvement of hormonal pathways. The mini review presented by Buhian and Bensmihen summarized how phytohormones, as key regulators of cellular and developmental plasticity in plants, control the symbiosis process, specifically the early symbiotic events involving NFs. New findings illustrating NF and phytohormone signaling crosstalk in controlling symbiosis establishment were summarized and synergized comprehensively in this mini review.

Cytokinins are essential phytohormones for rhizobia infection and nodule development, particularly for initiation of root cortical cell divisions and induction of critical gene expression in the NF-signaling pathway. To further investigate the role of cytokinin in controlling nodule numbers in soybean. Mens et al. evaluated and characterized the IPT gene family involved in cytokinin biosynthesis in soybean with respect to gene number, sequences, and expression in response to nodulation and soil nitrogen. The results illustrated the exact function of various IPT genes in soybean nodulation and indicated that in general low levels of cytokinins promote nodulation, whereas high levels inhibit it possibly due to toxic effects, and interestingly IPT genes may not play a role in the autoregulation of nodulation pathway.

Two different types of nodules, indeterminate and determinate, are generally formed by legumes. The model legume Medicago truncatula forms indeterminate nodules (typically elongate and maintaining an apical nodule meristem), while L. japonicus forms determinate nodules (typically round with no meristematic activity in mature nodules (Hirsch, 1992). What determines the difference between these two organogenesis programs is largely unknown. The phytohormone auxin is essential for cell divisions in plants and required for both indeterminate or determinate nodule development (Suzaki et al., 2012). Ng and Mathesius carried out a detailed analysis of auxin transport and localization during indeterminate and determinate nodule formation. Their results suggest that auxin response is observed in pericycle, endodermis, and inner cortex cells of the roots of indeterminate nodule-forming species**,** while increased auxin responses in outer cortex cells likely take place during determinate nodule formation. Specifically, the tissue difference correlates nicely with the fact that the initial cell divisions start in pericycle and endodermis in M. truncatula (indeterminate nodules) while the first cell divisions leading to determinate nodule formation initiate in the outer cortex for plants with determinate nodules.

Research on the legume-rhizobia symbiosis benefits from the advances of technologies at the cellular and molecular levels. Cellular calcium oscillations are critical for signal transduction. The analysis of Ca2<sup>+</sup> signaling in different cellular compartments requires tools adapted for monitoring Ca2<sup>+</sup> dynamics with high spatial and temporal resolution. Kelner et al. demonstrated the successful application of dual color sensors for simultaneously monitoring calcium signal dynamics in vivo in single root cell compartments during symbiosis, thus providing a powerful tool to study the signal transduction involving Ca2<sup>+</sup> signaling.

CCaMK is a calcium/calmodulin-dependent protein kinase localized in the nucleus and plays a critical role downstream of symbiotic calcium spiking. Activated by Ca2<sup>+</sup> and Ca2+ loaded calmodulin, CCaMK phosphorylates CYCLOPS or IPD3 in M. truncatula (Yano et al., 2008). Jauregui et al. demonstrated that mutations in the calmodulin-binding domain of CCaMK alter its calmodulin binding capacity, generating a constitutively activated kinase that confers an altered nodulation phenotype.

Particular rhizobial species or strains normally nodulate only a narrow group of legume species or genotypes (Wang et al., 2012, 2017). Understanding the genetic and molecular basis of this symbiotic specificity is important for developing strategies to improve symbiosis. Through genetic mapping Fan et al. identified a dominant gene in soybeans that restricts nodulation with S. fredii USDA193. Further validation indicated that an Rfg1 allele encoding a member of the Toll-interleukin receptor/nucleotidebinding site/leucine-rich repeat class of plant resistance proteins restricts nodulation by S. fredii strains USDA257 and USDA205. In combination with several previous studies, they showed that the Rfg1 allele likely provided broad-spectrum resistance to nodulation by many S. fredii and B. japonicum strains in soybean.

### AUTHOR CONTRIBUTIONS

JW prepared the first draft of the manuscript. SA and PR revised the draft critically and constructively. All authors reviewed and approved the final draft.

### REFERENCES


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

# The Very Long Chain Fatty Acid (C26:25OH) Linked to the Lipid A Is Important for the Fitness of the Photosynthetic Bradyrhizobium Strain ORS278 and the Establishment of a Successful Symbiosis with Aeschynomene Legumes

Nicolas Busset<sup>1</sup> , Flaviana Di Lorenzo<sup>2</sup> , Angelo Palmigiano<sup>3</sup> , Luisa Sturiale<sup>3</sup> , Frederic Gressent<sup>1</sup> , Joël Fardoux<sup>1</sup> , Djamel Gully<sup>1</sup> , Clémence Chaintreuil<sup>1</sup> , Antonio Molinaro<sup>2</sup> , Alba Silipo<sup>2</sup> and Eric Giraud<sup>1</sup> \*

#### Edited by:

Pascal Ratet, Centre National de la Recherche Scientifique (CNRS), France

#### Reviewed by:

Christian Staehelin, Sun Yat-sen University, China Oswaldo Valdes-Lopez, National Autonomous University of Mexico, Mexico

> \*Correspondence: Eric Giraud eric.giraud@ird.fr

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Microbiology

Received: 07 June 2017 Accepted: 06 September 2017 Published: 21 September 2017

### Citation:

Busset N, Di Lorenzo F, Palmigiano A, Sturiale L, Gressent F, Fardoux J, Gully D, Chaintreuil C, Molinaro A, Silipo A and Giraud E (2017) The Very Long Chain Fatty Acid (C26:25OH) Linked to the Lipid A Is Important for the Fitness of the Photosynthetic Bradyrhizobium Strain ORS278 and the Establishment of a Successful Symbiosis with Aeschynomene Legumes. Front. Microbiol. 8:1821. doi: 10.3389/fmicb.2017.01821 1 Institut de Recherche pour le Développement, LSTM, UMR IRD, SupAgro, INRA, Université de Montpellier, CIRAD, Montpellier, France, <sup>2</sup> Dipartimento di Scienze Chimiche, Complesso Universitario Monte Sant'Angelo, Università di Napoli Federico II, Naples, Italy, <sup>3</sup> Istituto per i Polimeri, Compositi e Biomateriali IPCB, Consiglio Nazionale delle Ricerche, Catania, Italy

In rhizobium strains, the lipid A is modified by the addition of a very long-chain fatty acid (VLCFA) shown to play an important role in rigidification of the outer membrane, thereby facilitating their dual life cycle, outside and inside the plant. In Bradyrhizobium strains, the lipid A is more complex with the presence of at least two VLCFAs, one covalently linked to a hopanoid molecule, but the importance of these modifications is not wellunderstood. In this study, we identified a cluster of VLCFA genes in the photosynthetic Bradyrhizobium strain ORS278, which nodulates Aeschynomene plants in a Nod factorindependent process. We tried to mutate the different genes of the VLCFA gene cluster to prevent the synthesis of the VLCFAs, but only one mutant in the lpxXL gene encoding an acyltransferase was obtained. Structural analysis of the lipid A showed that LpxXL is involved in the transfer of the C26:25OH VLCFA to the lipid A but not in the one of the C30:29OH VLCFA which harbors the hopanoid molecule. Despite maintaining the second VLCFA, the ability of the mutant to cope with various stresses (low pH, high temperature, high osmolarity, and antimicrobial peptides) and to establish an efficient nitrogen-fixing symbiosis was drastically reduced. In parallel, we investigated whether the BRADO0045 gene, which encodes a putative acyltransferase displaying a weak identity with the apo-lipoprotein N-acyltransferase Lnt, could be involved in the transfer of the C30:29OH VLCFA to the lipid A. Although the mutant exhibited phenotypes similar to the lpxXL mutant, no difference in the lipid A structure was observed from that in the wild-type strain, indicating that this gene is not involved in the modification of lipid A. Our results advance our knowledge of the biosynthesis pathway and the role of VLCFAs-modified lipid A in free-living and symbiotic states of Bradyrhizobium strains.

Keywords: VLCFA, lipid A, Bradyrhizobium, acyltransferase, symbiosis, Aeschynomene

### INTRODUCTION

fmicb-08-01821 September 19, 2017 Time: 16:32 # 2

Rhizobia are Gram-negative bacteria with two life styles, one in a free-living state in the soil where they have to cope with changing environmental conditions (hydric, acid, saline stresses, nutrient starvation, etc.) and the other in symbiosis with plants, inside an organ called a nodule, in which they reduce atmospheric nitrogen to ammonium for the benefit of the host plant. A simplistic view would be that inside the nodule, the bacteria benefit from a safe ecological niche with unlimited carbon and energy provided by the plant in exchange for ammonium. In fact, inside the host cells, bacteria also encounter stressful conditions imposed by the plant (low oxygen tension, low pH, hyperosmosis, and various oxidative stresses) (Gibson et al., 2008; Haag et al., 2013). In addition, in some host species, including in inverted repeatlacking clade (IRLC) legumes or the Aeschynomene species, the bacteria have to cope with antimicrobial peptides called NCR, that are used by plants to control bacterial metabolism and can led to a marked change in the shape of the bacteria (Van de Velde et al., 2010; Czernic et al., 2015; Montiel et al., 2017; Wang et al., 2017). The life of the rhizobia therefore does not resemble that of a long peaceful river since they have to adapt to changing stressful conditions outside and inside the host plants.

The first barrier used by the bacteria against biotic and abiotic stresses is its outer membrane (OM). Lipopolysaccharides (LPSs) are major components of the OM of Gram-negative bacteria. These compounds have three components (i) the O-antigen side chain that is in direct contact with the host plant, (ii) the core oligosaccharide, and (iii) a glycolipid moiety named the lipid A that anchors the LPS to the OM. LPSs are known to play a central role in bacterial invasion and adaptation to the host environment (Lerouge and Vanderleyden, 2002; Raetz and Whitfield, 2002). The variability of the O-antigen region observed in most rhizobia species is assumed to be a strategy to modulate or suppress plant defense responses, thereby facilitating the establishment of the symbiosis (Kannenberg and Carlson, 2001; Gourion et al., 2015). In addition, the lipid A of rhizobia LPSs is characterized by the presence of a C<sup>26</sup> to C<sup>30</sup> very long-chain fatty acid (VLCFA) (Bhat et al., 1994; Raetz et al., 2007). A lipid A-linked VLCFA is also encountered in pathogenic or intracellular bacteria such as Brucella or Legionella, suggesting that this structure promotes intracellular life by increasing the stability of the membrane (Bhat et al., 1991; Zähringer et al., 1995; Lerouge and Vanderleyden, 2002; Becker et al., 2005). The biosynthesis of the VLCFA and its addition to the lipid A require a cluster of five genes found in all the bacteria that synthesize a VLCFA-modified lipid A (Ardissone et al., 2011). This region is composed of genes encoding an acyl carrier protein (acpXL), fatty acid elongation proteins (fabF1XL and fabF2XL), a beta-hydroxyacyl-acyl carrier protein (ACP) dehydratase (fabZXL) and an acyltransferase (lpxXL) (Ardissone et al., 2011; Brown et al., 2011).

Analysis of acpXL and lpxXL mutants of Rhizobium leguminosarum and Sinorhizobium meliloti showed that the VLCFA plays a major role in the resistance to various abiotic stresses (high osmolarity, detergents, and dessication) but also in the establishment of a functional symbiosis (Ferguson et al., 2005; Haag et al., 2009; Ardissone et al., 2011; Bourassa et al., 2017). The plants inoculated with the VLCFA mutants are less efficient than the WT to fix nitrogen and present a nodulation delay. In addition, the nodules elicited by the VLCFA mutants present various alterations such as a white color, a smaller size, a disorganization of the infected zone and an early senescence. Furthermore, some bacteroids of the VLCFA mutants strains are abnormally large and have aberrant forms indicating also an alteration of the bacteroid differentiation process (Ferguson et al., 2005; Haag et al., 2009; Ardissone et al., 2011; Bourassa et al., 2017). Interestingly, while mutations in acpXL and lpxXL completely abolished VLCFA attachment to lipid A, it is observed that the acpXL mutants of R. leguminosarum and S. meliloti are able to substitute the VLCFA with a C16:0 or C18:0; which is not the case for lpxXL mutants (Ferguson et al., 2005; Bourassa et al., 2017). This suggests that in the absence of AcpXL, LpxXL could transfer shorter chain to the lipid A.

The OM of Bradyrhizobium strains has several peculiarities. First, unlike other rhizobia, hopanoid molecules are present in the membranes of all the Bradyrhizobium strains analyzed (Kannenberg et al., 1996). This family of compounds displays structural and functional similarities with eukaryotic sterols, such as cholesterol, and form an important class of membrane lipids that are widely distributed in diverse bacteria that reinforce the rigidity and stability of the OM (Ourisson et al., 1987; Kannenberg et al., 1996; Welander et al., 2009; Saenz et al., 2012). Second, structural analysis of the lipid A in various Bradyrhizobium strains revealed the occurrence of up to four VLCFAs that differ in their length and decoration (Choma and Komaniecka, 2011). Third, it has been reported that a hopanoid molecule can be covalently linked to the VLCFA(s) of the lipid A (**Figure 1**); this unusual lipid A structure was described for the first time in the photosynthetic Bradyrhizobium BTAi1 strain, named HoLA for Hopanoid-Lipid-A (Komaniecka et al., 2014; Silipo et al., 2014). Two recent studies reported that hopanoids play an important role in Bradyrhizobium strains by helping them to cope with various stresses in their two life styles (Silipo et al., 2014; Kulkarni et al., 2015). A hopanoid deficient mutant of the photosynthetic Bradyrhizobium BTAi1 strain, lacking a squalene hopene cyclase (1shc), displays increased sensitivity to stressful conditions and is unable to maintain chronic intracellular infection in Aeschynomene species (Silipo et al., 2014). Similarly, hpnP and hpnH mutants of B. diazoefficiens USDA110 affected in the synthesis of respectively methylated or extended (C35) hopanoids, displayed several disorders in both free-living and symbiotic states (Kulkarni et al., 2015). However, no study has been conducted on Bradyrhizobium strains to understand how the different VLCFAs linked to the lipid A are synthetized and to describe their respective roles.

Interestingly, only one VLCFA gene cluster has been identified in the bradyrhizobial genomes (**Figure 2**). In this study, we investigated whether this cluster is responsible for the synthesis of the different VLCFAs linked to the lipid A using the photosynthetic Bradyrhizobium ORS278 strain as model. This strain has the unusual property of inducing the formation of nitrogen-fixing nodules on some tropical legumes of the Aeschynomene genus in the absence of the synthesis of Nod factor

(Giraud et al., 2007). In addition, inside the host cell, bacteria undergo a drastic morphological change, suggesting important modifications of the bacterial cell wall (Bonaldi et al., 2011). This study focused on the effects of the mutations of the VLCFA gene cluster on the structure of the lipid A and on the free and symbiotic state of the bacteria.

### MATERIALS AND METHODS

### Bacterial Strains and Growth Conditions

The bacteria used in this study are indicated in Supplementary Table S1. The Bradyrhizobium strain ORS278 and its derivatives mutants were grown in yeast extract mannitol (YM) medium at 34◦C or in minimal medium buffered nodulation medium (BNM) (Podlešáková et al., 2013). Escherichia coli strains were grown in Luria-Bertani medium (LB) at 37◦C. When required, the media were supplemented with kanamycin (100 µg ml−<sup>1</sup> ) or spectinomycin (20 µg ml−<sup>1</sup> ) or a mixture of the two (100 and 20 µg ml−<sup>1</sup> ).

### Construction of ORS278 VLCFA Mutants

Standard molecular biology techniques were used for all cloning work. All the primers and plasmids used for cloning of DNA fragments are listed respectively in Supplementary Tables S1, S2. For the construction of Bradyrhizobium strain ORS278 insertional mutants, 300–400 base pair (bp) internal fragments were amplified by PCR and cloned into the plasmid pVO155 npt2-GFP (Okazaki et al., 2016). The constructions were first transferred into the E. coli S17-1 strain and then into ORS278 by mating as previously described (Giraud et al., 2010). For the construction of mutants by deletion mutagenesis, flanking regions of the genes were first amplified by PCR and then merged by overlap extension PCR. The overlap PCR fragment obtained was then digested by restriction enzymes corresponding to the restrictions sites of the primers (Supplementary Table S1) and was subsequently cloned in the sacB suicide pNPTS129 plasmid by ligation (Tsai and Alley, 2000). The spectinomycin resistance gene cassette from the pHRP315 vector (Parales and Harwood, 1993) was liberated by BamHI digestion and was introduced between the upstream and downstream regions previously cloned in the pNPTS129 plasmid. The resulting plasmid was then transferred into ORS278 by biparental conjugation, as previously described (Giraud et al., 2010). Antibiotic selection was used to select single recombinant which were verified by PCR. Double recombinant clones were then obtained by growth on sucrose in the presence of spectinomycin but without the addition of kanamycin. Candidate clones were checked for the loss of kanamycin resistance from the pNPTS129 plasmid, and the deletion of the gene was verified by PCR.

### Lipid A Structure Analysis – MALDI MS and MS/MS

Lipopolysaccharide was isolated and purified from the ORS278 WT strain and the lpxXl, BRADO4680 and BRADO0045 mutants, as previously described, and the lipid A isolated through a mild acid hydrolysis (Silipo et al., 2014). Like for MS data, the lipid A was analyzed using a 4800 MALDI MS/MS (ABSciex) mass spectrometer equipped with a Nd:YAG laser at a wavelength (λ) of 355 nm with a 500-ps pulse and a 200 Hz firing rate. External calibrations were performed using an ABSciex calibration mixture, allowing mass accuracies close to 55 ppm. An aliquot of bacterial pellets was also subjected to a micro-extraction procedure to obtain the lipid A moiety, as described by El Hamidi et al. (2005), and analyzed in an ABSCIEX TOF/TOFTM 5800 Applied Biosystems mass spectrometer equipped with an Nd:YLF laser with a λ of 345 nm, a <500-ps pulse length and a repetition rate of up to 1000 Hz. All mass spectra were acquired either in negative (not shown) and positive polarity. In the MS experiments, each spectrum resulted from the accumulation of 1,500 laser shots, whereas 5,000–7,000 shots were summed for the MS/MS data acquisitions (Spina et al., 2000; Sturiale et al., 2011). Samples were dissolved in CHCl3/CH3OH (50:50, v/v) at a concentration of 1 mg ml−<sup>1</sup> . Matrix solution was

prepared by dissolving 2,4,6 trihydroxyacetophenone (THAP) in CH3OH/0.1% trifluoroacetic acid/CH3CN (7:2:1, by volume) at a concentration of 75 mg ml−<sup>1</sup> . One microliter of the sample/matrix solution (1:1, v/v) was deposited on the well plate and allowed to dry at room temperature (Leone et al., 2007).

### Growth Curves at Different Temperatures

To monitor growth in rich (YM) and minimal (BNM) media, cultures were inoculated at 10−<sup>2</sup> dilution using YM-grown logphase (OD<sup>600</sup> = 0.5–0.7) WT or mutant strains. Growth was measured at OD<sup>600</sup> using a Cary 50 Scan (Varian). Different incubation temperatures were tested (28, 34, and 37◦C). Growth curves were performed in duplicate.

### Disk Diffusion Assays and NaCl Resistance Assay

The sensitivity to various abiotic stresses (SDS, H2O2, HCl, and NaCl) was assayed as previously described (Silipo et al., 2014). The experiments were conducted in triplicate for the lpxXL and BRADO4680 mutants and only once for the two BRADO0045 mutants.

### Resistance to Antibiotics

The minimum inhibitory concentration (MIC) of polymyxin B was determined as previously described (Kulkarni et al., 2015) by the Etest method using the disk diffusion assay (Biomérieux, Marcy-l'Etoile, France). The experiment was performed in triplicate.

### Plant Cultivation and Symbiotic Analysis

Aeschynomene indica and A. evenia seeds were surface sterilized, cultivated, and inoculated as previously described (Gully et al., 2016). For the nodulation and nitrogen fixation assay, 10 plants per condition were taken at 14 days post infection (dpi) to count the number of nodules on the roots and to analyze nitrogenase activity using an acetylene reduction assay (ARA) as previously described (Bonaldi et al., 2010).

### Cytological Analyses and Microscopy

Cytological analyses were conducted on 5–10 nodules originating from three different plants for each condition using the protocol described by Bonaldi et al. (2011). Confocal microscopy observations were carried out using a confocal laser-scanning microscope (Carl Zeiss LSM 700; Jena, Germany). Sections of ORS278 WT and BRADO4680 nodules were incubated for 20 min in live/dead staining solution [5 µM SYTO 9 and 30 µM propidium iodide (PI) in 50 mM Tris pH 7.0 buffer; Live/Dead BacLight, Invitrogen]. All the nodule sections were incubated for 15 min in 10 mM phosphate saline buffer (PBS) containing calcofluor white M2R (Sigma, Munich) at a final concentration of 0.01% (w/v) to stain the plant cell wall (Nagata and Takebe, 1970). Calcofluor was excited at 405 nm and emission signals were collected from 405 to 470 nm. For SYTO 9 or GFP and PI, an excitation wavelength of 488 and 555 nm was used with emission signal collected at 490–522 nm and 555–700 nm, respectively. Images were obtained using the ZEN 2008 software (Zeiss).

### Statistical Analysis

Statistical analysis was performed using XLSTAT version 2016.6 software. Differences between groups of samples were evaluated with the Tukey's range test. Differences were considered statistically significant at a P-value < 0.01. The box plots were made in R 3.2.2 software. Results are shown as box plots. Each graph contains median, quartiles, and whiskers which show the last sample in [1st quartile −1.5<sup>∗</sup> IQR, 3rd quartile + 1.5<sup>∗</sup> IQR] range. The free points on the graph represent the outliers samples.

### Accession Numbers

fmicb-08-01821 September 19, 2017 Time: 16:32 # 5

The GenBank accession numbers of BRADO4675 to BRADO4680 are respectively CAL78405 to CAL78410 and the one of BRADO0045 is CAL74017.

### RESULTS

### Bradyrhizobium Strains Displayed Only One VLCFA Gene Cluster

A BLAST search of the ORS278 genome led to the identification of a single gene cluster containing several homologs of genes shown to be involved in VLCFA biosynthesis in the R. leguminosarum bv. viciae 3841 strain (**Figure 2**) (Bourassa et al., 2017). This gene cluster is composed of the CDS BRADO4675 to BRADO4679, which are homologs of acpXL, fabZXL, fabF1XL, fabF2XL, and lpxXL genes, respectively (**Figure 2**). Besides sharing a high level of identity (>40%) with the corresponding R. leguminosarum proteins, except for the absence of the adhA2XL gene in ORS278, the organization of the genes is perfectly conserved between the two strains (**Figure 2**).

We also analyzed the distribution and the organization of the VLCFA genes in other photosynthetic (ORS285, BTAi1, B. oligotrophicum S58) and non-photosynthetic (B. diazoefficiens USDA110, B. japonicum USDA124, and B. elkanii USDA76) Bradyrhizobium strains. In all cases, only one homolog region was identified in which the genes displayed similar organization to that in ORS278 notably with the absence of the adhA2XL homolog. However, the photosynthetic Bradyrhizobium strains differed from the non-photosynthetic ones by the presence, downstream of lpxXL, of a gene (BRADO4680) that codes for a protein of unknown function (**Figure 2**).

The identification of only one VLCFA gene cluster among the Bradyrhizobium genomes suggests that the genes present in this region might be sufficient for the synthesis and the attachment of the various VLCFAs that are linked to the lipid A in these bacteria.

### Lipid A of the lpxXL Mutant Lost the C26:25OH VLCFA, but Still Contains the C30:29OH VLCFA

To confirm that the identified region in the ORS278 strain is involved in the synthesis of VLCFAs and in their transfer to the lipid A, we tried to mutate the different genes present in this region, including BRADO4680. Different mutagenesis strategies were tested, deletion by double crossing over in acpXL and BRADO4680 and disruption of fabZXL, fabF1XL, fabF2XL, or lpxXL by insertion of the non-replicative plasmid pVO155 npt2-GFP in the corresponding coding region. Despite several attempts, only two mutants were obtained, one in the lpxXL homolog (BRADO4679) and the second in BRADO4680. The repeated failure to select mutants in acpXL, fabZXL, fabF1XL, or fabF2XL suggests that these four genes are essential because they are required for the synthesis of VLCFAs without which the bacteria cannot survive or grow under the conditions used to select the mutants. Conversely, the selection of mutants in BRADO4680 and lpxXL could result in functional redundancy, the absence of role, or in a specific role in the synthesis or the attachment of only one type of VLCFA to the lipid A.

To investigate the impact of the BRADO4679 and BRADO4680 mutations on the synthesis of the VLCFAs, MALDI MS analyses were performed on the lipid A of the ORS278 WT and mutants strains. The mass spectrometry analysis of the WT lipid A showed a mixture of ion species which essentially varied by the acylation pattern, i.e., penta- and hexa-acylated species. All the lipid A species were stoichiometrically built up of a pentasaccharide backbone formed by a β-(1-6)-linked 2,3 diamino-2,3-di-deoxy-D-glucopyranose (DAG) carrying a GalpA residue on the vicinal DAG and an α-(1-6)-Manp disaccharide on the distal DAG residue. Indeed, the mass differences among neighboring ion species were due to acyl number and length of ester linked fatty acids (**Figure 3A**). The ion peak at m/z 2128.52 was established to be a penta-acylated sodiated lipid A carrying two 14:0 (3-OH), one 12:0 (3-OH), one 12:1 and one acetylated VLCFA, 26:0 (25-OAc). The most abundant sodiated hexa-acylated lipid A species (m/z 2594.81) further carried a secondary VLCFA, the 30:0 (2,29-2OH), in turn esterified, in the hepta-acylated species centered around m/z 3107.22, by the hopanoid moiety (**Figures 3A,D**). The ORS278 lipid A structure was therefore very similar to the one previously described for the BTAi1 strain with the presence of two VLCFAs, the longest of which harbored a hopanoid molecule (Silipo et al., 2014). The lipid A of the BRADO4680 mutant had a similar structure to that of the WT-strain (**Figures 3B,D**), showing that BRADO4680 is not involved in the modification of the lipid A.

In contrast, the lipid A of the lpxXl mutant showed differences in the acylation pattern. In fact, it was composed of a mixture of tetra-, penta-, and hexa-acylated species in which the 26:0 (25-OAc) VLCFA was absent (**Figures 3C,E** and Supplementary Figure S2), indicating that LpxXL is exclusively involved in the transfer of the C26:25OH VLCFA to the lipid A of ORS278.

### The C26:25OH VLCFA Lipid A Is Essential for the ORS278 Strain to Deal with Stresses in Free-Living Conditions

Very long-chain fatty acids have been shown to contribute to stress tolerance in free-living and symbiotic states, in diverse rhizobia like R. leguminosarum or S. meliloti (Becker et al., 2005; Gibson et al., 2008; Bourassa et al., 2017). To test the hypothesis that the removal of the C26:25OH of ORS278 lipid A reduces the ability of the strain to resist stressful conditions, we challenged lpxXl and BRADO4680 mutants with a variety of stressors that occur during the initiation and progression of symbiosis. In addition, the growth kinetics of the lpxXl and BRADO4680 mutants at various temperatures and their resistance to membrane destabilizer were quantified to determine if these mutations have an effect on the stability of the membrane.

As shown in **Figures 4A,D**, the lpxXl mutant displayed similar growth to that of the WT strain at 28◦C in both rich and minimal media. In contrast, at higher temperatures (34 and 37◦C), the growth kinetics of the mutant were reduced compared with those of the WT strain (**Figures 4B,C,E,F**). These observations

suggest a reduction in the stability of the membrane. In addition, disk diffusion assays showed that the lpxXl mutant was more sensitive to H2O<sup>2</sup> and SDS than the WT strain (**Figure 4H**). The mutant was also more sensitive to osmotic stress than the WT strain, as evidenced by a reduction in its growth at a concentration of 100 mM of NaCl (**Figure 4G**). Because ORS278 is exposed to NCR-like peptides in Aeschynomene plants, we also tested the sensitivity of the mutants to polymyxin B which is a cationic peptide displaying similar effects than some NCRs on the alteration of the bacterial OM permeability (Mikuláss et al., 2016). Contrary to the lpxXL mutant of R. leguminosarum which is not affected in polymyxin B resistance (Bourassa et al., 2017), we observed that the lpxXl mutant of ORS278 displayed an eightfold lower MIC (8 µg/ml) for polymyxin B than the WT strain (64 µg/ml) (**Figure 4I**). The BRADO4680 mutant withstood all the stressors, as did the WT (**Figures 4A–I**). Together, these data suggest that the C26:25OH VLCFA linked to the lipid A of ORS278 plays an important role in the ability of this strain to cope with various stressful conditions by increasing the stability of the OM.

lpxXL (gray) and BRADO4680 (dark gray) mutants, as determined by disk diffusion assays using 5 µl of 5.5 M H2O2, 2N HCl or 10% of SDS (n = 9). (I) Box plots representation of the polymyxin B resistance of ORS278 and the lpxXL and BRADO4680 mutants, determined by Etest (Etest <sup>R</sup> bioMérieux) on YM medium (n = 3). (G–I) <sup>∗</sup>P < 0.01, by Tukey's honestly significant difference test.

### The C26:25OH VLCFA Lipid A Is Essential for an Effective Symbiosis between the Photosynthetic Bradyrhizobium ORS278 Strain and Aeschynomene Plants

To investigate if the lpxXl and BRADO4680 mutations have an effect on the symbiotic properties of ORS278, we inoculated the WT and the two mutants strains on two Aeschynomene plants, A. evenia and A. indica, that can be nodulated in a Nod-independent manner by this strain.

No effect of the BRADO4680 mutation was detected on either Aeschynomene species (Supplementary Figure S1). In contrast, a clear effect of the lpxXl mutation was observed in both Aeschynomene species. Plants inoculated with this mutant, particularly A. evenia, displayed typical symptoms of nitrogen starvation, including leaf chlorosis and reduced plant growth at 14-dpi (**Figures 5A,B**). These observations were correlated with reduced nitrogenase activity, as estimated by the ARA compared to the WT strain, despite the fact that the lpxXL mutant led to a higher number of nodules per plant (**Figures 5C,D**). In

addition, some nodules induced by the lpxXL mutant in both A. evenia and A. indica plants were yellowish instead of green, as observed for the WT nodules (**Figures 5E–G,L,M**). The absence of chloroplasts in the epidermal tissue of the Aeschynomene nodules is typically observed in mutants with altered nitrogen fixation, such as nif or hopanoid-minus mutants (Bonaldi et al., 2010; Silipo et al., 2014). Furthermore, some A. evenia nodules elicited by the lpxXl mutant were hollow suggesting a degradation of the tissue of the infection zone (**Figure 5H**), which is a sign of early senescence, as already reported in inefficient nitrogenfixing mutants (Delmotte et al., 2014; Silipo et al., 2014). In A. indica, no hollow nodules were found, but brownish compounds that autofluorescent in the red spectrum (excitation, 488 nm; emission, 600–660 nm) were observed in some lpxXl nodules (**Figure 5P**). This is indicative of the accumulation of polyphenol compounds generally associated with plant defense reactions, as previously described in other legumes (Bourcy et al., 2013). Finally, cytological analysis performed by confocal microscopy showed that the process of bacteroid differentiation was altered in the lpxXL mutant, in both A. evenia and A. indica plants. In fact, in the lpxXL mutant, both undifferentiated bacteria and bacteroids were observed, which were not perfectly spherical

and were of abnormal size, in contrast to the WT bacteroids (**Figures 5I–K,N–P**). As a WT reference, we used a tagged strain containing the pVO155-npt2-GFP plasmid inserted in the BRADO5083 gene which encodes a protein of unknown function. Previous studies did not reveal any particular symbiotic defect due to the plasmid insertion.

Taken together, these data indicate that the lpxXL mutation drastically alters the ability of the ORS278 strain to form an efficient symbiosis with Aeschynomene plants.

### Search for Another Acyltransferase Transferring the C30:29OH VLCFA to the Lipid A of ORS278

The fact that the lipid A of the lpxXL mutant still contains the C30:29OH VLCFA suggests that another acyltransferase exists that allows the specific transfer of this VLCFA to the lipid A. Sequence genome analysis of ORS278 did not enable identification of another homolog of lpxXL, suggesting that this acyltransferase should be strongly different from BRADO4679. Conversely, a search for genes annotated as containing an acyl transferase domain retrieved more than 40 candidates. This number was too high to envisage the systematic mutagenesis of all of them. On the other hand, a Tn5 mutant in one of them (BRADO0045) was previously described for its nitrogen fixing deficiency in A. indica (Bonaldi et al., 2010) suggesting that this gene plays an important role during symbiosis. This CDS has 42% identity with the ActA protein of S. meliloti, which has been shown to play an essential role in the acid tolerance of the bacteria (Tiwari et al., 1996), and 30% identity with the Apo-lipoprotein acyltransferase Lnt characterized in E. coli, which is involved in the maturation of lipoprotein (Gupta et al., 1993). Together, these different elements prompted us to examine in more detail the role of BRADO0045 in both the physiology and symbiotic properties of ORS278 and to explore the possibility of its direct involvement in the transfer of the C30:29OH VLCFA to the lipid A.

To be sure that the phenotype reported for the Tn5 mutant corresponds to the inactivation of BRADO0045, we constructed a new insertional mutant in this CDS using the non-replicative plasmid pVO155-npt2-GFP. The ability of the two BRADO0045 mutants (Tn5 and pVO155) of ORS278, called 0045T and 0045P, respectively, to cope with various stresses was then analyzed as previously. Analysis of growth kinetics at 28 and 34◦C showed that the two BRADO0045 mutants underwent alteration of their growth at the higher temperature, pointing to destabilization of the membrane, as observed in the lpxXL mutant (**Figures 6A,B**). In addition, the two mutants were found to be more sensitive to salt, oxidative and detergent stresses and to the polymyxin B (**Figures 6C–E**) than the WT strain. These results show that the mutation in BRADO0045 reduces the ability of ORS278 to cope with abiotic stresses in the same way as the lpxXL mutation.

To test if the mutations in BRADO0045 have an impact on the symbiotic properties of ORS278, A. indica plants were inoculated with the WT strain and the two mutants. At 14-dpi, we observed that the plants inoculated with the two mutants were smaller than the plants inoculated by the WT-strain (**Figure 6F**), which is correlated with lower nitrogenase activity (**Figure 6G**). As previously observed in the lpxXL mutant, some nodules elicited by the two BRADO0045 mutants were yellow and hollow (**Figures 6H,L**) and some displayed massive plant defense reactions (**Figures 6I–K**). In addition, cytological analysis showed that the intracellular bacteria were not differentiated or not perfectly spherical, and were of abnormal size, the same as observed in the lpxXL mutant (**Figures 6M,N**). The BRADO0045 protein thus appears to be essential for the establishment of an efficient symbiosis between ORS278 and A. indica.

To determine if this phenotype was due to alteration of the lipid A structure of ORS278, MALDI MS analyses were performed on the lipid A from ORS278 WT and from 0045P mutant. As can be seen in Supplementary Figure S3, the two spectra were very similar, showing that the effects of the BRADO0045 mutation on the phenotypic and symbiotic properties of ORS278 were not due to modification of the lipid A.

### DISCUSSION

The fact that a VLCFA is attached to the lipid A of rhizobia was first described in R. leguminosarum bv. trifolii more than two decades ago (Hollingsworth and Carlson, 1989). Since this discovery, several studies have (i) highlighted the ubiquitous presence of a VLCFA in all the rhizobium lipid A studied, (ii) revealed the importance of this VLCFA in both the free and symbiotic life forms of several rhizobium species and (iii) characterized the genes involved in its biosynthesis (Bhat et al., 1991; Ferguson et al., 2005; Raetz et al., 2007; Haag et al., 2009; Ardissone et al., 2011; Bourassa et al., 2017). More recently, it was shown that the lipid A of several Bradyrhizobium strains differs in the presence of at least two VLCFAs, one of which can be linked to a hopanoid molecule (Choma and Komaniecka, 2011; Komaniecka et al., 2014; Silipo et al., 2014). However, nothing is known on the synthesis and the respective role of these VLCFAs.

In this study, we have shown that Bradyrhizobium genomes contain a gene cluster homologous to the one described in R. leguminosarum that enables the synthesis and the attachment of the VLCFA to the lipid A. In the Bradyrhizobium ORS278 strain, we succeeded in obtaining a mutant in only one gene of this cluster, the BRADO4679 gene homologous to lpxXL and showed that this mutation led to the suppression of only the C26:25OH VLCFA. This indicates that this gene encodes an acyltransferase which catalyzes the transfer of such a VLCFA to the lipid A but not the one of the C30:29OH VLCFA. Despite the fact that the mutant maintains this second VLCFA linked to the lipid A, it displays several phenotypes similar to those reported in lpxXl mutants in other rhizobia species, such as a higher sensitivity to various stresses (NaCl, H202, HCl) and alteration of the efficiency of the symbiosis (Ferguson et al., 2005; Haag et al., 2009; Ardissone et al., 2011). In particular, we observed that the Aeschynomene nodules elicited by the lpxXL mutant of ORS278 displayed several disorders: some nodules were yellowish and the bacteroids were malformed. Furthermore, some A. indica nodules accumulated autofluorescent brown compounds suggesting the induction of plant defense reactions

resistance of ORS278 (white), 0045P (dark gray) and 0045T (gray) mutants, as determined by disk diffusion assays using 5 µl of 5.5 M H2O2, 2N HCl or 10% of SDS (n = 9). (E) Box plots representation of the Polymyxin B resistance of ORS278 and the two 0045P and 0045T mutants, as determined by Etest (Etest <sup>R</sup> bioMérieux) on YM medium (n = 3); (F) Comparison of the growth of A. indica (aerial part), non-inoculated (N.I.) or inoculated with ORS278, 0045P or 0045T mutants. The experiment was carried out in duplicate with 10 plants per condition. (G) Box plots representation of the quantification of ARA and number of nodules per plant inoculated with ORS278 (white), 0045P (dark gray) or 0045T (gray) mutants in A. indica. Whole roots of A. indica inoculed with the 0045P mutant (H) or the 0045T mutant of ORS278 (I); scale bars: 2 mm. (J–N) Nodule thin sections of A. indica elicited by the 0045P mutant and viewed by confocal microscopy; scale bars: 300 µm (J–L) and 20 µm (M,N). (J,K) Whites arrows indicate plant defense reactions. (D,E,G) <sup>∗</sup>P < 0.01, by Tukey's honestly significant difference test.

while the central tissue of some A. evenia nodules was digested. This degradation of the symbiotic tissue is typically observed in senescing nodules such as described in Medicago or other legume species (Puppo et al., 2004). It would be interesting to confirm that the LpxxL mutant can induce plant defense reactions and early nodule degeneracy, respectively, on A. indica and A. evenia, by checking the expression of some senescence and pathogenesis-related genes. Nevertheless, Aeschynomene is just emerging as a new model legume and marker genes of these two cellular programs have to be first characterized for these two species. These severe phenotypes are probably related to the inability of the mutant to cope with the stresses found in the host cell (oxidative stresses, NCR peptides, acid stress) but also to perturbations of the LPS and/or the OM structure that could compromise the recognition of the bacteria by the plant which then reacts by the induction of a defense response or by

triggering senescence. In agreement with this latter possibility, it has been recently identified in Arabidopsis a lectin S-domain receptor kinase that detects the lipid A moiety and modulates immune response to bacterial infection (Ranf et al., 2015). It is also important to note that other mutants which displayed some modifications in the structure of their cell wall, such as the DD-CPase1 mutant of ORS278 strain, which is altered in the level of reticulation of the peptidoglycan, or a hopanoid minus mutant of the BTAi1 strain, exhibited very similar phenotypes to the lpxXL mutant (Silipo et al., 2014; Gully et al., 2016). This suggests that the structural integrity of the bacterial cell envelope of which the attachment of the C26:25OH VLCFA to the lipid A is an important determinant, is essential to maintain chronic intracellular infection during symbiosis.

In this study, we did not succeed in mutating the acpXL, fabZXL, fabF1XL, or fabF2XL genes of the ORS278 strain. Taking into account the fact that an lpxXL mutant was obtained with the previously described features, it can be assumed that these four genes are involved in the synthesis of the two VLCFAs and that in the absence of their synthesis, the ORS278 strain cannot survive. In support of this last hypothesis, a Tn-seqbased study of Rhodopseudomonas palustris CGA009, a strain phylogenetically close to Bradyrhizobium, indicates that these genes are also essential in this bacteria (Pechter et al., 2015). In addition to this common pool of genes, we predict that at least three additional enzymes are necessary to achieve the complete synthesis of the lipid A in its most complex form: (i) an enzyme permitting elongation of the C26:25OH VLCFA to the C30:29OH VLCFA genes, (ii) an acyltransferase specific to the C30:29OH VLCFA allowing its transfer to the lipid A, and finally, (iii) an enzyme catalyzing the attachment of the C35-hopanoid to this VLCFA. The fact that a shc mutant of BTAi1 displayed a lipid A with the two VLCFAs but lacking the hopanoid moiety leads us to think that, first, the C30:29OH VLCFA is attached to the lipid A and, second, the hopanoid is attached to the C30:29OH VLCFA. Some genes of the lipid A biosynthesis pathway in Bradyrhizobium strains therefore remain to be identified. Their discovery will be a real challenge considering that their mutation could be lethal and that no other homolog of lpxXL has been identified in the Bradyrhizobium genomes.

In this context, we examined whether the BRADO0045, that encodes a putative acyltransferase and for which a Tn5 mutant has been reported to be affected in nitrogen fixation, can catalyze the transfer of the C30:29OH VLCFA to the lipid A. MALDI MS analyses of the lipid A clearly showed that BRADO0045 does not play this role since no structural difference was observed with the WT lipid A. On the other hand, the phenotypic properties of the BRADO0045 mutants are very similar to those of the lpxXL mutant indicating that the mutation of BRADO0045 would affect the structure of the cell envelope. The protein BRADO0045 has a low level of identity with the apo-lipoprotein N-acyltransferase Lnt of E. coli. This enzyme plays a role in the maturation of lipoproteins by catalyzing the attachment of a third acyl chain that enables their transfer to the OM by the Lol system in Gram-negative bacteria (Fukuda et al., 2002). It has been shown in E. coli that the mutation of lnt affects the properties of both the inner and OM by reducing the level of incorporation of lipoproteins in the OM and, consequently increasing it in the inner membrane, which is lethal in this bacterium (Robichon et al., 2005; Narita and Tokuda, 2011). It would therefore be interesting to compare the lipoprotein composition of the different membrane compartments in the BRADO0045 mutant and in the WT strain to check whether BRADO0045 corresponds to an apo-lipoprotein N-acyltransferase. It is also to note that the three downstream genes of BRADO0045 (BRADO0046 to BRADO0048) are in the same direction than BRADO0045 (see Supplementary Figure S4), we cannot therefore exclude the possibility that the phenotype observed for the BRADO0045 insertional mutants was due to polar effects.

Numerous studies have revealed an essential role for VLCFAmodified lipid A in bacteria with an intracellular lifestyle, whether they are pathogens or symbionts like rhizobia (Bhat et al., 1991; Zähringer et al., 1995; Becker et al., 2005). One may wonder why, unlike other rhizobia, Bradyrhizobium strains, which have similar lifestyles, have at least two VLCFAs linked to their lipid A instead of one and why a hopanoid molecule is covalently linked to this second VLCFA. It was previously shown by analysis of reconstituted liposomes using electron spin resonance (ESR) spectroscopy, that, thanks to their VLCFAs, HoLA molecules can span the whole OM by placing the hopanoid moiety in the inner leaflet, which results in a higher stabilization of the inner and the outer leaflets of the OM (Silipo et al., 2014). This rigidification of the OM might be a functional advantage by strengthening its barrier role, which could facilitate the survival of the bacteria under stress conditions. This could contribute to the ecological success of the Bradyrhizobium genus which nodulates the widest range of legume species and which is distributed worldwide but predominately in tropical areas and acid soils (Parker, 2015; Sprent et al., 2017).

### AUTHOR CONTRIBUTIONS

NB, EG, AM, and AS conceived the experiments, NB, FDL, AP, LS, FG, JF, DG, and CC conducted the experiments; NB, EG, AM, and AS analyzed the results and wrote the paper.

### FUNDING

This work was supported by the French National Research Agency (ANR-Bugsinacell-13-BSV7-0013).

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb. 2017.01821/full#supplementary-material

### REFERENCES

fmicb-08-01821 September 19, 2017 Time: 16:32 # 12



**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 Busset, Di Lorenzo, Palmigiano, Sturiale, Gressent, Fardoux, Gully, Chaintreuil, Molinaro, Silipo and Giraud. 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.

# Type 3 Secretion System (T3SS) of Bradyrhizobium sp. DOA9 and Its Roles in Legume Symbiosis and Rice Endophytic Association

#### Edited by:

Pascal Ratet, Centre National de la Recherche Scientifique (CNRS), France

#### Reviewed by:

Francisco Javier López-Baena, University of Seville, Spain Brian H. Kvitko, University of Georgia, United States

#### \*Correspondence:

Neung Teaumroong neung@sut.ac.th Eric Giraud eric.giraud@ird.fr

†These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Microbe Interactions, a section of the journal Frontiers in Microbiology

Received: 15 June 2017 Accepted: 05 September 2017 Published: 20 September 2017

#### Citation:

Songwattana P, Noisangiam R, Teamtisong K, Prakamhang J, Teulet A, Tittabutr P, Piromyou P, Boonkerd N, Giraud E and Teaumroong N (2017) Type 3 Secretion System (T3SS) of Bradyrhizobium sp. DOA9 and Its Roles in Legume Symbiosis and Rice Endophytic Association. Front. Microbiol. 8:1810. doi: 10.3389/fmicb.2017.01810 Pongpan Songwattana<sup>1</sup> , Rujirek Noisangiam<sup>2</sup> , Kamonluck Teamtisong<sup>3</sup> , Janpen Prakamhang<sup>4</sup> , Albin Teulet<sup>5</sup> , Panlada Tittabutr<sup>1</sup> , Pongdet Piromyou<sup>1</sup> , Nantakorn Boonkerd<sup>1</sup> , Eric Giraud<sup>5</sup> \* † and Neung Teaumroong<sup>1</sup> \* †

<sup>1</sup> School of Biotechnology, Institute of Agricultural Technology, Suranaree University of Technology, Nakhon Ratchasima, Thailand, <sup>2</sup> National Bureau of Agricultural Commodity and Food Standards, Ministry of Agriculture and Cooperatives, Bangkok, Thailand, <sup>3</sup> The Center for Scientific and Technological Equipment, Suranaree University of Technology, Nakhon Ratchasima, Thailand, <sup>4</sup> Department of Applied Biology, Faculty of Sciences and Liberal Arts, Rajamangala University of Technology Isan, Nakhon Ratchasima, Thailand, <sup>5</sup> Institut de Recherche pour le Développement, LSTM, UMR IRD/SupAgro/INRA/Univ. Montpellier/CIRAD, Montpellier, France

The Bradyrhizobium sp. DOA9 strain isolated from a paddy field has the ability to nodulate a wide spectrum of legumes. Unlike other bradyrhizobia, this strain has a symbiotic plasmid harboring nod, nif, and type 3 secretion system (T3SS) genes. This T3SS cluster contains all the genes necessary for the formation of the secretory apparatus and the transcriptional activator (TtsI), which is preceded by a nod-box motif. An in silico search predicted 14 effectors putatively translocated by this T3SS machinery. In this study, we explored the role of the T3SS in the symbiotic performance of DOA9 by evaluating the ability of a T3SS mutant (rhcN) to nodulate legumes belonging to Dalbergioid, Millettioid, and Genistoid tribes. Among the nine species tested, four (Arachis hypogea, Vigna radiata, Crotalaria juncea, and Macroptilium atropurpureum) responded positively to the rhcN mutation (ranging from suppression of plant defense reactions, an increase in the number of nodules and a dramatic improvement in nodule development and infection), one (Stylosanthes hamata) responded negatively (fewer nodules and less nitrogen fixation) and four species (Aeschynomene americana, Aeschynomene afraspera, Indigofera tinctoria, and Desmodium tortuosum) displayed no phenotype. We also tested the role of the T3SS in the ability of the DOA9 strain to endophytically colonize rice roots, but detected no effect of the T3SS mutation, in contrast to what was previously reported in the Bradyrhizobium SUTN9-2 strain. Taken together, these data indicate that DOA9 contains a functional T3SS that interferes with the ability of the strain to interact symbiotically with legumes but not with rice.

Keywords: type 3 secretion system, effector proteins, Bradyrhizobium, legume, symbiosis

### INTRODUCTION

fmicb-08-01810 September 18, 2017 Time: 15:13 # 2

The establishment of the legume–rhizobium symbiosis relies on the exchange of diffusible signal molecules between the two partners. The first important bacterial signals are lipochitooligosaccharides (LCOs) called Nod factors (NFs) that are synthetized and secreted after activation of bacterial nodulation (nod) genes by the transcriptional regulator protein NodD in specific response to host flavonoids. In return, perception of the NFs signal by the plant triggers nodule organogenesis and bacterial infection processes (Oldroyd et al., 2011). Besides NFs, other bacterial components are important for the successful establishment of the symbiosis, including exopolysaccharides (EPS), lipopolysaccharides (LPS), capsular polysaccharides (KPS), and cyclic β-glucans (Gibson et al., 2008; Janczarek et al., 2015). These bacterial compounds can also act as a symbiotic signal or permit to avoid plant immune responses (Gourion et al., 2015; Kawaharada et al., 2015).

In addition, some rhizobia use a type 3 secretion system (T3SS), which shares homology with the T3SS apparatus of plant and animal pathogenic bacteria, to translocate some putative effectors into the host cell, named nodulation outer proteins (Nops). Those putative effector proteins may have contrasting effects on nodulation depending on the host plant (Staehelin and Krishnan, 2015). Some secreted effector proteins can play a positive role in the establishment and maintenance of symbiosis by suppressing plant defense reactions thereby facilitating the bacterial infection (Bartsev et al., 2004; Jiménez-Guerrero et al., 2015). Conversely, secreted Nop proteins can be recognized by plant receptor proteins [resistant (R) proteins] leading to effector-triggered immunity, which blocks infection and nodulation (Yang et al., 2010; Yasuda et al., 2016).

It is important to indicate that the term Nop has been initially introduced by Marie et al. (2001) to mirror the nomenclature of Yersinia outer proteins (Yops) and that the first identified Nops was done by characterizing the proteins secreted in Sinorhizobium NGR234 culture supernatants in a NodD1 and a flavonoid-dependent manner (Marie et al., 2003). As consequence, this term has been used to name some components of the extracellular appendages or pili of the T3SS, i.e., NopA, NopB, and NopX but also effector proteins secreted through the T3SS that are translocated to the interior of the host cell, i.e., NopL, NopE, and NopC. In this paper, we will consider the term Nop to speak about the effector proteins translocated into the host cell.

The involvment of the T3SS in the establishement of the symbiosis is supported by two facts: (i) the genes that code for T3SS machinery and some Nops are found in the same region as the nod and nif genes (on a symbiotic island or a symbiotic plasmid) and (ii) expression of the T3SS genes is also controlled via a regulatory cascade involving NodD. In the latter case, NodD regulates, in a flavonoid-dependent manner, the expression of the transcriptional activator (TtsI), which, in return, activates the expression of T3SS genes by binding to their promoter region at the level of a specific motif, called tts-box (Krause et al., 2002; Tampakaki, 2014).

The importance of T3SS in the modulation of the host spectrum has been extensively demonstrated in several rhizobial strains (Sinorhizobium sp. NGR234, S. fredii USDA257, S. fredii HH103, Mesorhizobium loti MAFF303099, Bradyrhizobium diazoefficiens USDA110, B. elkanii USDA61) (Viprey et al., 1998; Krause et al., 2002; Krishnan et al., 2003; de Lyra Mdo et al., 2006; López-Baena et al., 2009; Okazaki et al., 2009, 2010). More recently, some non-photosynthetic Bradyrhizobium strains have been shown to be able to nodulate some legume plants (Glycine max cv. Enrei and several Aeschynomene species) in the absence of NF synthesis but in these cases, nodulation requires a functional T3SS. This points to the existence of an alternative Nod-independent symbiotic pathway in which some Nop effectors can directly activate the nodulation signaling pathway (Okazaki et al., 2013, 2016). Moreover, in Bradyrhizobium sp. SUTN9-2, T3SS has been reported to be an important determinant of rice-endophyte colonization (Piromyou et al., 2015b). Taken together, these data show that the role played by the T3SS in rhizobia during the interaction with their host plants is more important and broader than previously believed.

The non-photosynthetic Bradyrhizobium sp. strain DOA9 was originally isolated from a paddy field in Thailand using Aeschynomene americana as trap plant (Noisangiam et al., 2012). In addition to its ability to endophytically colonize rice roots, this strain has been shown to be able to induce the formation of nodules in a large spectrum of legume hosts (Teamtisong et al., 2014). Unlike other bradyrhizobia, the DOA9 is the only strain yet identified to possess a symbiotic plasmid that harbors nitrogen fixation (nif), nodulation (nod), and T3SS genes (Okazaki et al., 2015). In this study, we examined the role of T3SS in the DOA9 strain by testing the symbiotic performance of a T3SS mutant (rhcN) on various legume species as well as on rice.

### MATERIALS AND METHODS

### Bacterial Strains and Culture Conditions

The bacterial strains used in this study are listed in **Table 1**. Bradyrhizobium sp. strain DOA9 and its rhcN derivative were grown in YM medium (Vincent, 1970) at 28◦C. Escherichia coli strains were cultured at 37◦C in LB medium (Sambrook et al., 2001). The media were supplemented with antibiotics at the following concentrations when appropriate: 200 µg/mL streptomycin (Sm), 20 µg/mL nalidixic acid (Nal), and 50 µg/mL kanamycin (Km).

### Construction of the Reporter and Mutant Strains

The single cross-homologous recombination technique was used to construct the insertion mutant in the structural type III secretion system gene (rhcN) of strain DOA9. For this purpose, an internal fragment (263-bp) of rhcN was amplified by polymerase chain reaction (PCR) using the following primers: RhcN.D9p.int.f (5<sup>0</sup> -CATCTCGTCGACTCGCAGCA AAGGATGTCGATAC-3<sup>0</sup> ) and RhcN.D9p.int.r (5<sup>0</sup> -GAGCAG TCTAGACCCGACTGACACTTCCTGCATG-3<sup>0</sup> ). The PCR

#### TABLE 1 | Bacterial strains, plasmids used, and plants tested in this study.


product was then digested by Xbal and SalI and cloned onto the corresponding sites of the plasmid pVO155-Sm-npt2-gfp (Wongdee et al., 2016). This plasmid, which is non-replicative in Bradyrhizobium strains, is a derivative of the plasmid pVO155 (Oke and Long, 1999) and carries the promoterless gusA gene, a constitutively expressed gfp, and a Km- and a Sm/spectinomycinresistance genes (Wongdee et al., 2016). The resulting plasmid was introduced into E. coli S17-1 by electroporation (15 kv/cm, 100 , and 25 µF) and was transferred into the Bradyrhizobium sp. strain DOA9 by biparental mating, using the protocol described by Giraud et al. (2010). The transconjugants were spread on YM medium supplemented with an antibiotic mixture of Sm, Km, and Nal, and the successful insertion of the plasmid in rhcN was checked by PCR.

The GFP-labeled DOA9 strain was constructed by introducing the replicative plasmid pMG103-npt2-Sm-npt2-gfp harboring a constitutive gfp gene by electroporation (15 kv/cm, 100 , and 25 µF). Plasmid-containing strains were selected on YM plates supplemented with Sm.

### Plant Nodulation and Symbiosis Analysis

DOA9 and its derivatives were grown for 5 days as previously described and used as inoculum. Sterilization and germination of the seeds of all plants tested (Supplementary Table S1) were performed as described by Teamtisong et al. (2014). Plants were sown in Leonard's jar filled with sterilized vermiculite (Somasegaran and Hoben, 1994). All the plants were watered with BNM medium (Ehrhardt et al., 1992) and grown under the following controlled environmental conditions: 28 ± 2 ◦C with a 16 h light/8 h dark cycle at light intensities of 300 µE/m<sup>2</sup> S and with 50% humidity. Five days after planting, each seedling was inoculated with 1 mL of a 5 day old inoculum (logphase) after washing and adjusting the optical density to 600 nm to 1 (approximately 10<sup>9</sup> cells). The experiment was conducted with five replicates with the following treatments: control (no inoculation), GPF-labeled DOA9, and rhcN. To check nodulation and nitrogen fixation abilities, five plants were taken at 21 day post-inoculation (dpi) and the number of nodules and nitrogenase activity were assessed using the acetylene reduction assay (ARA). Briefly, all the nodules on each plant were collected and placed in headspace bottles with 10% acetylene, and incubated at 28◦C for 1 h. Gas chromatography was used to measure the peak height of ethylene and acetylene with 1 mL gas samples from the bottles by using a PE-alumina packed column with injector at 150◦C, an oven temperature of 200◦C, and ionization detector (FID) at a temperature of 50◦C (Somasegaran and Hoben, 1994).

### Rice Experiment

Rice (Oryza sativa L. ssp. indica cv. Pathum Thani) seeds were surface sterilized by soaking in 70% (v/v) ethanol for 30 s, then washed twice with 10% (v/v) hydrogen peroxide for 10 min, washed again in 3% (v/v) sodium hypochlorite for 1 h. Finally, the seeds were washed in sterilized water five times and germinated overnight on YM medium containing 0.8% (v/v) agar. The germinated seeds with 1 cm of root were soaked in DOA9 cell suspension (OD<sup>600</sup> = 1) overnight. The rice seedlings were transferred to metal net in 80 mL tubes containing 15 mL Hoagland's plant growth medium (Hoagland and Arnon, 1950) in presence of 0.1 mM NH4NO3. The experiment was conducted with three technical replicates with the following treatments: control (no inoculation), GFP-labeled DOA9, and rhcN. Plants were grown under the following controlled environmental conditions: 28 ± 2 ◦C with a 16 h light/8 h dark cycle at light intensities of 300 µE/m<sup>2</sup> S and with 50% humidity.

### Enumeration of Epiphytic and Endophytic Bacteria in Rice Roots

Two and 3 weeks after inoculation with the DOA9 tagged strains and rhcN containing gfp and Sm<sup>r</sup> , the rice roots were washed in normal saline solution (0.85% of NaCl) to remove debris and any bacteria not firmly attached to the root surface. Next, the bacterial cells remaining on the surface of the roots were harvested by vortex with normal saline containing 0.3% (v/v) tween-80 for 30 s. The washing solution containing the epiphytic bacteria was then spread on YM plates containing Sm. For the enumeration of endophytic bacterial, root tissues were surface sterilized with 70% (v/v) ethanol for 5 min followed by 3% (v/v) sodium hypochlorite for 5 min and rinsed with sterilized water five times. The surface sterilized roots were macerated with a sterilized mortar and pestle and diluted in saline solution prior to being spread on medium as described above (Piromyou et al., 2015a). The epiphyte and endophyte population counted on agar plates are expressed as colony-forming unit per gram of root fresh weight.

### Preparation of Root Exudates and Bacterial Induction

The sterilized legume seeds were germinated and transferred into tubes containing BNM medium (50 mg seeds/mL). Plants were

maintained in controlled environmental conditions as described above for 5 days. The root exudates were obtained from plant medium after filtration using a 0.2 µm filter syringe. Root exudates were stored at −20◦C until use.

The mid-log phase culture of DOA9 was washed and the OD<sup>600</sup> was adjusted to approximately 0.4 with YM supplemented with 1/3 (v/v) of the root exudates or purified flavonoids (20 µM of naringenin or genistein dissolved in DMSO). The sterilized BNM medium and DMSO were used as negative controls. The bacterial cells were cultured at 28◦C for 24 h and then collected by centrifugation (4,000 × g for 10 min, 4◦C) and immediately frozen in liquid nitrogen and stored at −80◦C for further total RNA isolation.

### RNA Isolation and qRT-PCR

Total bacterial RNA was extracted from induced cells using the RNeasy <sup>R</sup> Mini Kit (QIAGEN, United States) according to the manufacturer's protocol. Total RNA was treated with RNase-free DNase I (NEB) for 30 min at 37◦C. cDNA was synthesized from 500 ng total RNA using High Capacity cDNA Reverse Transcription Kits (Applied BiosystemsTM) according to the manufacturers' protocols. Thirty nanograms of cDNA was subjected to PCR amplification using specific primers of rhcN (rhcN.RT.D9.f: 5<sup>0</sup> -CATTGGCGATATGGTAGGCT-3<sup>0</sup> ; rhcN.RT.D9.r: 5<sup>0</sup> -GGACAAGTGTGAACCGTCCT-3<sup>0</sup> ) and nopX (nopX.RT.D9.f: 5<sup>0</sup> -CATCAACCCGAACAACACAG-3<sup>0</sup> ; nopX.RT.D9.r: 5<sup>0</sup> -GGCTCGATAGACAAGGTCCACAA-3<sup>0</sup> ) genes. PCR amplification was performed using QuantStudio 3 Real-Time PCR System Mix (Applied BiosystemsTM) and the following program: an initial denaturation step at 95◦C for 2 min; 35 cycles at 95◦C for 30 s, 55◦C for 30 s, and at 72◦C for 1.5 min; and a final extension step at 72◦C for 10 min. The relative gene expression was analyzed using the comparative Ct method (−11CT) normalized to the endogenous housekeeping gene (16S rRNA) using PBA338F (50 -ACTCCTACGGGAGGCAGCAG-3<sup>0</sup> ) and PRUN518R (5<sup>0</sup> - ATTACCGCGGCTGCTGG-3<sup>0</sup> ). Three biological replicates were pooled and analyzed. At least three replicate PCR amplifications were performed for each sample.

### Microscopy

An Olympus Fluoview FV1000 confocal laser scanning microscope was used to investigate nodule development, bacteroid differentiation, and rice colonization. For this purpose, 40–50 µm thick sections of fresh nodules and rice roots were prepared using a VT1000S vibratome (Leica Nanterre, France). Nodule sections were stained for 20 min with 0.01% calcofluor (to stain the plant cell wall) and with 30 µM propidium iodide (PI) in 50 mM Tris–HCl pH 7.0 buffer to identify dead cells (Haag et al., 2011). Calcofluor was excited at 405 nm and detected with a 460–500 nm emission filter. The GFP-labeled DOA9 and rhcN strains were detected after excitation with the 488 nm laser line and emission signal collection at 490–522 nm, while the PI used to identify dead cells was excited with the 535 nm laser line and emission signals were collected at 617–636 nm. Confocal images were reconstructed with NIS elements software (Nikon), and images were colored and prepared for publication with Adobe Photoshop software. To detect β-glucuronidase (GUS) activity in the nodules elicited by the rhcN mutant the same protocol than Bonaldi et al. (2010) was used.

### Search for Genes Encoding Putative T3 Effectors on the DOA9 Genome

Two complementary approaches were used to identify putative effector genes: (i) a TBlastN search of the genome for known Nops identified in rhizobia (Kimbrel et al., 2013; Staehelin and Krishnan, 2015), the identified homologs were considered as possible candidates when they matched the following parameters (% of identity ≥40% over 80% of the length of the sequence) and (ii) a search for tts-boxes on the chromosome and plasmid of DOA9 genome (GenBank accession numbers DF820425 and DF820426, respectively) using a hidden Markov model initially trained with the sequences of 26 confirmed tts-boxes from Sinorhizobium (Ensifer) sp. strain NGR234, B. diazoefficiens USDA110, and B. elkanii USDA61 (Eddy, 1998; Marie et al., 2004; Zehner et al., 2008; Okazaki et al., 2009). To be considered as a candidate, the gene had to contain a tts-box with a score ≥9 located on the same strand in its upstream region (up to 1 kb).

### Statistical Analysis

The experiment data were subjected to analysis of variance (ANOVA). When confirming a statistically significant value in the F-test (p ≤ 0.05), a post hoc test (Duncan's multiple-range test at p ≤ 0.05) was used as a multiple comparison procedure (Duncan, 1955) using SPSS1 software for WINDOWSTM, Version 14.0.

### RESULTS

### Gene Organization and Comparison of the T3SS Genes of Bradyrhizobium sp. DOA9 and Other Rhizobial Strains

Sequence genome analysis revealed the presence of a T3SS gene cluster organized in several operons on the pDOA9 plasmid (**Figures 1A,B**). The main putative operon preceded by a ttsbox contained most of the genes coding for components of the secretion machinery in the order nopB, rhcJ, nolU, rhcL, rhcN, rhcU, rhcQ, rhcR, rhcS, rhcT, and rhcU. This genetic organization is similar to that found in other Bradyrhizobium or Sinorhizobium strains and a very high level of AA identity was observed with the corresponding homologous proteins. However, unlike these strains, a large intergenic region of 650 bp was observed between rhcS and rhcT also containing a tts-box (**Figure 1A**), raising the possibility that the rhcT and rhcU genes are transcribed independently.

As described in other rhizobia, we could assume that this main cluster is under the dual control of TtsI and NodD since a ttsI homolog was found at the vicinity of the main operon preceded by a nod-box (**Figure 1A**). However, it is to highlight the presence of a gene encoding a putative transposase between this nod-box and ttsI. To confirm that the presence of this putative

transposase does not interfere with the formation of the T3SS secretion apparatus, we analyzed the expression of two structural T3SS genes (rhcN and nopX) in the presence of two flavonoids (genistein or naringenin) previously reported to induce T3SS genes in rhizobial strains (Krause et al., 2002; Okazaki et al., 2010). As shown in Supplementary Figure S1, both flavonoids strongly induced nopX and rhcN expression, showing that despite this transposase, the T3SS genes in DOA9 strain are classically regulated by flavonoids, most probably via the control of NodD and TtsI.

To identify putative effectors that are possibly translocated by this T3SS machinery, we combined two in silico searches: (i) a TBlastN search of the genome for Nops previously identified in other rhizobia and (ii) a search of the tts-box motif using a hidden Markov model. After eliminating the Nops proteins predicted to correspond to components of the secretory apparatus (NopA, NopX, and NopB) (Staehelin and Krishnan, 2015), this analysis retrieved 20 putative translocated effectors. Of these, six were located on the chromosome (**Figure 1C**). However, all six candidates remain doubtful because homologous proteins for all of them can be identified in photosynthetic Bradyrhizobium strains that lack a T3SS gene cluster (Supplementary Table S2). Interestingly, among the other putative effectors found on the plasmid, three correspond to new effectors not previously identified in other rhizobia. They were preceded by a tts-box and all contained the small ubiquitin-like modifier (SUMO) protease domain of the C48 peptidase [ubiquitin-like protease 1 (Ulp1)] family. Several effectors containing this functional domain have already been characterized in pathogenic and symbiotic bacteria (Hotson et al., 2003; Hubber et al., 2004; Rodrigues et al., 2007; Tsurumaru et al., 2015), reinforcing the hypothesis that these three effector candidates are bona fide new effectors.

### Symbiotic Role of the T3SS

To analyze the role of the T3SS in DOA9 strain during symbiosis, we constructed a T3SS mutant by inserting the nonreplicative plasmid pVO155-Sm-npt2-gfp into the 5<sup>0</sup> -region of rhcN. The rhcN gene was selected as target because it encodes an ATPase that is indispensable for the functioning of the T3SS injectisome. Furthermore, the pVO155 plasmid used for inactivation of the gene contains a constitutive expressed GFP that enables monitoring of the bacteria inside the nodules. The symbiotic performance of the WT and rhcN mutant was compared on several legume species belonging to Genistoid (Crotalaria juncea), Dalbergioid (Aeschynomene americana, A. afraspera, Arachis hypogaea, and Stylosanthes hamata), and Millettioid tribes (Macroptillium atropurpureum, Vigna radiata, Indigofera tinctoria, and Desmodium tortuosum) by analyzing several symbiotic parameters (number of nodules per plant, nitrogen fixation estimated using the ARA, plant dry weight, and cytological aspect of the nodules).

Depending on the plant species, the rhcN mutant displayed contrasted responses compared to the WT strain. The responses were classified in three categories. In category 1, non-responsive phenotype (called T3SS-no effect group), the plants inoculated with the mutant displayed the same number of nodules and the same nitrogen fixing capacity as the plants inoculated with the WT strain. Furthermore, no cytological differences were observed between the mutant and WT nodules. These plants

were A. americana, A. afraspera, I. tinctoria, and D. tortuosum (**Figure 2**).

In category 2, the rhcN mutation had a positive effect on one of the symbiotic parameters analyzed (**Figure 3**). This group comprised A. hypogaea, V. radiata, C. juncea, and M. atropurpureum. Inoculation with the rhcN mutant led to an almost 65% increase in the number of nodules on A. hypogea compared with WT (**Figure 3F**). This result suggests that in DOA9, T3SS compromises either the infection or nodule organogenesis process in this species. **Figure 3** shows that the higher number of nodules is not correlated with an increase in N<sup>2</sup> fixation. We assume that the plant compensates for the small number of nodules elicited by the WT strain by stimulating their expansion, as can be seen in **Figures 3B,C**. The same mechanism has been reported in Medicago and is related to plant N demand (Laguerre et al., 2012). The deleterious effect of the T3SS on the establishment of the symbiosis was more marked in the cases of C. juncea and V. radiata. Indeed, whereas the WT strain only elicited bumps or small completely necrotic nodules in C. juncea and V. radiata, respectively, the rhcN mutant formed perfectly developed nodules on these two plants (**Figures 3H–U**). This suggests that some effectors translocated by the T3SS could activate plant defense reactions in these two species, thereby preventing infection and the development of nodules. Similar stimulation of plant immunity probably also occurred in M. atropurpureum but to a lesser extent, since the nodules induced by the WT strain were well formed, but displayed brown necrotic areas. The bacteroids in these areas were dead as revealed by the red PI staining, whereas the rhcN nodules were perfectly normal and the host cells were filled with viable bacteroids, as revealed by the green of GFP tagging cells (**Figures 3V–AB**). Nevertheless, while the T3SS mutation made it possible to restore nodule formation capacity in V. radiata and C. juncea, only very weak nitrogenase activity was detected, and no real benefit for plant growth was observed (**Figures 3N,U** and Supplementary Table S3). This shows that other restrictions exist between DOA9 and these two plant species resulting in inefficient symbiosis.

Only one species, S. hamata, was attributed to category 3, corresponding to a negative effect of the rhcN mutation.

The number of nodules and nitrogenase activity of nodules induced by the rhcN strain were reduced by approximately 30% compared with the wild-type strain (**Figures 4A–G**). These data suggest that the T3SS of strain DOA9 also plays a positive role in the establishment of the symbiosis, most probably by translocating effectors that weaken the plant immune system.

We also examined the effect of root exudates on the expression of rhcN and nopX genes to determine whether the absence of

phenotype observed for the rhcN mutant in some legume species could be linked with the absence of formation of the T3SS apparatus. As expected, a significant level of expression of the two T3SS genes was detected in the five species displaying a positive or negative phenotype with the rhcN strain (Supplementary Figure S1). In contrast, of the four species that displayed no phenotype with the rhcN mutant, the root exudates of two of them (I. tinctoria and D. tortuosum) displayed a similar level of expression to the control without inducers raising the possibility that the T3SS machinery is not formed during the early steps of the interaction between the DOA9 strain and these two legumes species. We also took advantage of the promoterless gusA reporter gene present in the insertional pVO155 plasmid used to construct the rhcN mutant to determine whether the T3SS is expressed in mature nodules as already reported in B. diazoefficiens USDA110 nodulating soybean (Zehner et al., 2008). As shown in Supplementary Figure S2, for the three tested species (A. americana, M. atropurpureum, and I. tinctoria), a GUS activity was detected in the infected nodule cells, in contrast to the WT nodules, in which no activity was observed (Supplementary Figure S2). Altogether, these data suggest that T3SS genes of DOA9 strain are active during the early and/or late stages of symbiosis on the different species tested.

### Role of the T3SS in Rice Colonization and Infection

We also investigated the effect of the T3SS mutation on the ability of DOA9 to colonize the surface of rice roots and to infect deep root tissue. These observations were made on a Thai rice (cultivar Pathum Thani 1) at the second and third week postinoculation (wpi) (**Figure 5F**). The total population density of WT and rhcN strains on the surface of the rice roots was almost the same (around 8 log<sup>10</sup> CFU/g of root fresh weight), and this level of population remained constant at 2 and 3 wpi. When the roots were surface sterilized to conserve only the endophytic bacteria cells, the estimated population of both WT and mutant strains was also very close (around 4 log<sup>10</sup> CFU/g root fresh weight). We also took advantage of the GFP tag added on the WT and rhcN strains to monitor invasion of the roots. As can be seen in **Figures 5A–C,E**, both WT and rhcN cells were attached to the root hairs and to the surface of the root epidermis. Analysis of root sections revealed endophytic bacterial cells in the intercellular space of the root cortex and endodermis. Taken together, these data suggest that the T3SS mutation did not affect the ability of the DOA9 strain to colonize and infect the rice root tissue intercellularly. This absence of effect of the T3SS mutation was not related to the absence of T3SS genes expression considering that a GUS activity could be detected at the surface of the rice roots inoculated with the rhcN mutant (**Figure 5D**).

### DISCUSSION

The Bradyrhizobium sp. DOA9 strain isolated from a paddy field has been shown to have the ability to induce the development of symbiotic nodules in various legume species (Teamtisong et al., 2014). Unlike other bradyrhizobia, this strain contains a symbiotic megaplasmid (pDOA9) that harbors the nod and nif genes as well as a T3SS gene cluster (Okazaki et al., 2015). This T3SS cluster contains all the genes necessary for the formation of the secretory apparatus and the transcriptional activator (TtsI), which is preceded by a nod-box promoter motif. This genetic organization is similar to that described in other bradyrhizobial strains. Why DOA9 harbors a symbiotic plasmid while all the other Bradyrhizobium strains sequenced contain a symbiotic island remains an open question. In both cases, the lower GC content and the codon usage of the genes present in the symbiotic region suggest an acquisition by lateral transfer. This indicates that, in addition to the possibility of a difference in the origin of the symbiotic region, the mechanisms of its transfer and its maintenance inside the bacterial genome could also differ from one Bradyrhizobium strain to another.

Despite high conservation of T3SS genes encoding the injectisome, the T3SS effector contents of DOA9 strain differ from those in other bradyrhizobia, notably in their number,

which is comparatively lower. Indeed, in the DOA9 strain, only 14 candidate effector proteins have been predicted versus between 30 and 35 putative effectors in other bradyrhizobia (Zehner et al., 2008; Kimbrel et al., 2013). In particular, the effectors NopM and NopE, classically found in Bradyrhizobium strains, often as multicopies, are clearly absent in the DOA9 genome. Conversely, three putative effectors that were not previously identified in rhizobia are predicted in the DOA9 strain. These three new putative effectors proteins (WP\_042337156, WP\_025038828, and GAJ3805) most probably correspond to SUMO peptidases that contain the catalytic domain of members of the family of C48 cysteine proteases involved in the de-ubiquitinization of eukaryotic proteins (Hotson et al., 2003). Interestingly, a SUMO peptidase protein named NopD in S. fredii HH103 was identified to be secreted through the T3SS in the culture supernatant indicating that this family protein could constitute bona fide effectors translocated into the host cell (Rodrigues et al., 2007). Furthermore, more recently, other SUMO protease putative effectors were reported to be responsible for the incompatibility of some Bradyrhizobium strains (B. japonicum Is-1 and B. elkanii USDA61), which prevents them from interacting symbiotically with Rj4 genotype soybeans (Faruque et al., 2015; Tsurumaru et al., 2015). Taken together, this reinforces the hypothesis that the three new Nop candidats identified in DOA9 strain are effectors that could also affect the ability to interact with some plants.

It has previously been shown that T3SS in bradyrhizobia can have a positive or negative effect on nodule formation by stimulating or repressing the plant immune system depending on the host plant. For example, the deletion of T3SS in B. diazoefficiens delayed nodulation on soybean and reduced the number of nodules on M. atropurpureum (Krause et al., 2002), while the absence of T3SS in B. elkanii USDA61 induced the formation of nodules on V. radiata L. cv. KPS1 and on the soybean cultivar Hill, which contains the Rj4 allele (Okazaki et al., 2009). In the present study, we observed that the T3SS mutation affected the ability of DOA9 strain to interact with five out of the nine species tested. In particular, the T3SS mutation dramatically improved nodule formation and development in C. juncea and V. radiata and had a more moderate but also positive impact

on nodulation in A. hypogaea and M. atropurpureum. This suggests that among the cocktail of effectors that are translocated into the host cell, some are recognized as virulence factors by R-proteins and block infection and nodule organogenesis to a greater or lesser extent. As previously mentioned above, some SUMO protease effectors and also NopT have been observed to act as negative effectors in some hosts (Dai et al., 2008), which raises the possibility that the effector(s) responsible for this incompatibility could be one or some of the new effectors identified in the DOA9 strain or even NopT, which is also present in DOA9.

Conversely, the T3SS mutation had a negative impact on the number of nodules elicited on S. hamata, suggesting, in this case, that some effectors act positively on the symbiosis most probably, either by helping the bacteria to overcome plant defense reactions that limit the interaction or by directly inducing the nodulation signaling pathway as recently described for some nonphotosynthetic bradyrhizobia shown to nodulate some legume species in the absence of NFs synthesis but thanks to the T3SS (Okazaki et al., 2013, 2016).

Identifying the effectors responsible for these contrasted symbiotic effects depending on the host plant, and understanding their role are real challenges. To date, a specific role in the symbiosis has been described for only a few effectors (Staehelin and Krishnan, 2015). One limitation is that the different effectors translocated into the host cell mostly act synergistically by interfering in different plant defense pathways. The DOA9 strain may be an ideal model to study the specific role of effectors individually or in combination. Indeed, all the symbiotic determinants are found on the pDOA9 plasmid and it has been possible to create a plasmid-free DOA9 strain (Tittabutr, personal communication). We can imagine designing plasmids encompassing a minimum set of canonical nod, nif, and T3SS injectisome genes supplemented by different combinations of effectors and to study the symbiotic properties of a DOA9 strain lacking pDOA9 in which these synthetic plasmids are reintroduced. This kind of approach may seem too ambitious and time consuming, but with the development of synthetic biology and a reduction in the cost of synthetizing DNA, it could be feasible in the near future.

Among the four species in which the T3SS mutation had no effect on the symbiosis, we observed that the root exudate of two of them (I. tinctoria and D. tortuosum) did not induce the expression of nopX and rhcN genes, suggesting that the T3SS apparatus is not formed during the early stages of the symbiosis between the DOA9 strain and these two species. These results were surprising considering that these two species were found to be nodulated by the DOA9 strain, indicating that the nod genes are expressed. It would be interesting to explore more deeply the regulation of the T3SS and nod genes in presence of flavonoids and root exudates to determine if decoupling between their expression is possible. In the same vein, we observed that some putative T3SS effector genes are not preceded by a tts-box, which raises the question of whether these genes are expressed when the T3SS is formed or if their regulation involves another activator than TtsI. The observation that T3SS genes were found expressed in mature nodules while they were not found activated by root exudates is also puzzling. This could be related to the fact that we used different technics to analyze genes expression or that the technic of extraction of root exudate was not adapted for I. tinctoria considering that this very small plant do not grow well in liquid medium. We cannot also exclude the possibility that the regulation of the T3SS genes could be more complex involving other input signals than flavonoids and additional regulators than NodD and TtsI leading to different gene expression controls during the early and late stages of symbiosis.

Finally, given that a previous study focusing on the SUTN9-2 strain showed that the T3SS is an important determinant in rice infection (Piromyou et al., 2015b), we also investigated if the T3SS plays a similar role during the interaction between the DOA9 strain and rice. However, despite the use of the same cultivar (Pathum Thani 1), our observations indicate that the T3SS mutation does not affect the ability of DOA9 strain to colonize the root surface and to infect the root tissue intercellularly. This result was surprising given that the two strains were isolated from similar ecological niches (Thai paddy fields using A. americana as trap legumes) and considering that we observed that the T3SS genes are active during the interaction of DOA9 strain with rice. These two strains differ in the origin of the symbiotic region, which harbors a different cortege of nod and nop genes. We can suppose that during evolution, each strain adapted this symbiotic toolbox differently as a function of the variety of plant partners it encountered during its life history. This might explain why, via a different cocktail of effectors, the same T3SS machinery can lead to different responses from one strain to another and from one host to another.

### AUTHOR CONTRIBUTIONS

PS, PT, NB, EG, and NT conceived the experiment(s). PS, RN, KT, JP, AT, PT, and PP conducted the experiment(s). PS, PT, EG, and NT analyzed the result(s) and wrote the paper. All authors reviewed the manuscript.

### FUNDING

This work was financially supported by Suranaree University of Technology (OROG), the Franco-Thai Cooperation Program in Higher Education and Research (PHC SIAM 2017 – No. 38280WF) and by the ANR, grant "SymEffectors" number ANR-16-CE20-0013.

### ACKNOWLEDGMENT

The authors would like to thank Dr. Issra Pramoolsook for advice and comments on the language of the manuscript.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb.2017. 01810/full#supplementary-material

### REFERENCES

fmicb-08-01810 September 18, 2017 Time: 15:13 # 11


identification using LC-MALDI and LC-ESI with a monolithic capillary column. J. Proteome Res. 6, 1029–1037. doi: 10.1021/pr060519f


**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 Songwattana, Noisangiam, Teamtisong, Prakamhang, Teulet, Tittabutr, Piromyou, Boonkerd, Giraud and Teaumroong. 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.

# Distinct Lotus japonicus Transcriptomic Responses to a Spectrum of Bacteria Ranging From Symbiotic to Pathogenic

#### Simon Kelly<sup>1</sup> , Terry Mun<sup>1</sup> , Jens Stougaard<sup>1</sup> , Cécile Ben<sup>2</sup> and Stig U. Andersen<sup>1</sup> \*

<sup>1</sup> Department of Molecular Biology and Genetics, Aarhus University, Aarhus, Denmark, <sup>2</sup> ECOLAB, Université de Toulouse, CNRS, INP, UPS, Toulouse, France

#### Edited by:

Benjamin Gourion, UMR 2594, Laboratoire des Interactions Plantes-Microorganismes (LIPM), France

#### Reviewed by:

Ulrike Mathesius, Australian National University, Australia Yangrong Cao, Huazhong Agricultural University, China Jeremy Dale Murray, Shanghai Institutes for Biological Sciences (CAS), China

> \*Correspondence: Stig U. Andersen sua@mbg.au.dk

#### Specialty section:

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

Received: 22 March 2018 Accepted: 30 July 2018 Published: 20 August 2018

### Citation:

Kelly S, Mun T, Stougaard J, Ben C and Andersen SU (2018) Distinct Lotus japonicus Transcriptomic Responses to a Spectrum of Bacteria Ranging From Symbiotic to Pathogenic. Front. Plant Sci. 9:1218. doi: 10.3389/fpls.2018.01218 Lotus japonicus is a well-studied nodulating legume and a model organism for the investigation of plant-microbe interactions. The majority of legume transcriptome studies have focused on interactions with compatible symbionts, whereas responses to non-adapted rhizobia and pathogenic bacteria have not been well-characterized. In this study, we first characterized the transcriptomic response of L. japonicus to its compatible symbiont, Mesorhizobium loti R7A, through RNA-seq analysis of various plant tissues. Early symbiotic signaling was largely Nod factor-dependent and enhanced within root hairs, and we observed large-scale transcriptional reprogramming in nodule primordia and mature nitrogen-fixing nodules. We then characterized root transcriptional responses to a spectrum of L. japonicus interacting bacteria ranging from semi-compatible symbionts to pathogens. M. loti R7A and the semi-compatible strain Sinorhizobium fredii HH103 showed remarkably similar responses, allowing us to identify a small number of genes potentially involved in differentiating between fully and semi-compatible symbionts. The incompatible symbiont Bradyrhizobium elkanii USDA61 induced a more attenuated response, but the weakest response was observed for the foliar pathogen Pseudomonas syringae pv. tomato DC3000, where the affected genes also responded to other tested bacteria, pointing to a small set of common bacterial response genes. In contrast, the root pathogen Ralstonia solanacearum JS763 induced a pronounced and distinct transcriptomic pathogen response, which we compared to the results of the other treatments. This comparative analysis did not support the concept that an early defense-like response is generally evoked by compatible rhizobia during establishment of symbiosis.

Keywords: symbiosis, nodulation, RNA-seq, pathogen, legume, nitrogen fixation, plant-microbe interaction, rhizobia

## INTRODUCTION

Lotus japonicus develops determinate root nodules in association with its compatible rhizobia Mesorhizobium loti. Establishment of an effective nitrogen-fixing symbiosis requires molecular communication to ensure compatibility and coordinate the developmental processes of rhizobial infection and nodule organogenesis with legume LysM-receptor kinases playing an integral role in

these processes (Madsen et al., 2010; Oldroyd et al., 2011; Kelly et al., 2017a). Nod factor (NF) is the key signal molecule produced by rhizobia (Long, 1996) and is perceived by a NF receptor complex, which in Lotus consists of NFR1, NFR5 and SYMRK (Madsen et al., 2003; Radutoiu et al., 2003; Broghammer et al., 2012; Antolin-Llovera et al., 2014). Perception of compatible NF results in rapid physiological and transcriptional responses in the host (Desbrosses and Stougaard, 2011; Oldroyd, 2013). An additional level of compatibility scrutiny in Lotus occurs through the perception of rhizobial exopolysaccharides (EPSs) by the EPR3 receptor (Kelly et al., 2013; Kawaharada et al., 2015, 2017b).

Significant resources have been established to assist with investigations of L. japonicus. The genome sequence is available (Sato et al., 2008) as is an extensive LORE1 retrotransposon mutant resource consisting of 150,000 lines that provides potential mutants for an estimated 90% of active protein coding genes (Malolepszy et al., 2016). A Lotus japonicus gene expression atlas (LjGEA) was established that incorporated and extended on available Lotus transcriptome data in response to rhizobia and abiotic stresses (Verdier et al., 2013). All of these genomic resources can now be accessed through Lotus Base (Mun et al., 2016). Comparable transcriptomic resources are available for the indeterminate nodulating model legume Medicago truncatula (Benedito et al., 2008) and additional significant transcriptome data for this host was provided by the specific analysis of root hairs, which revealed a role for auxin signaling in infection thread (IT) formation (Breakspear et al., 2014).

Transcriptome responses of L. japonicus to compatible rhizobia and arbuscular mycorrhiza form the bulk of host transcriptional data available, while responses to incompatible rhizobia and pathogens have been less well-defined. Microarray analysis has been performed on two L. japonicus ecotypes that show differing leaf phenotypes when challenged with Pseudomonas syringae pv. tomato DC3000 and the response to pathogenic fungal exudates has also been investigated, identifying potential defense related genes (Bordenave et al., 2013; Giovannetti et al., 2015).

The interplay between symbiotic and pathogenic responses in legumes has largely been investigated through coinoculation experiments, which revealed a negative impact of pathogens and defense eliciting compounds on symbiotic efficiency (Lopez-Gomez et al., 2012; Chen et al., 2017). Analysis of a cDNA array of expressed sequence tags from L. japonicus at various time points after inoculation with M. loti TONO indicated that an initial defense-like transcriptional response in L. japonicus is subsequently dampened through symbiotic signaling. The identified defense-like genes encode hyper-sensitive related proteins, pathogenesis-related (PR) proteins, and proteins associated with phytoalexin biosynthesis and cell wall modification (Kouchi et al., 2004).

In this study, a comprehensive symbiotic transcriptomic response of L. japonicus to its compatible symbiont M. loti R7A and a spectrum of interacting bacteria was obtained through RNA-seq analysis, revealing distinct transcriptional responses and challenging the concept that an early defense-like response is evoked by compatible rhizobia in L. japonicus during the establishment of symbiosis.

### MATERIALS AND METHODS

### Plant Material and Growth Conditions

Lotus japonicus ecotype Gifu (Handberg and Stougaard, 1992) was used as the wild-type plant. Seed sterilization and plant-growth setups were as previously described (Kawaharada et al., 2015). For tissue-specific analysis of transcriptome responses to M. loti R7A, plants were grown at 21◦C with a 16 h day and 8 h night cycle. For the analysis of transcriptome responses to a spectrum of interacting bacteria, plants were grown at 25◦C with a 16 h day and 8 h night cycle. Plant growth plates, each containing 10 seedlings, were inoculated with 750 µL of OD<sup>600</sup> = 0.02–0.05 bacterial suspensions along the length of the root. Purified M. loti R7A NF was obtained as previously described (Rodpothong et al., 2009) and used at 10−<sup>8</sup> M for plant inoculation.

### Bacterial Strains

Bacterial strains used in this study are listed in **Supplementary Table 1**. All strains were cultured at 28◦C. M. loti R7A, Bradyrhizobium elkanii USDA61 and Sinorhizobium fredii HH103 were cultured in yeast mannitol broth (YMB) (Vincent, 1970), Pseudomonas syringae pv. tomato DC3000 in NYG medium (Daniels et al., 1984) and Ralstonia solanacearum JS763 in BGT medium (BG medium plus 0.005% tetrazolium chloride) (Saile et al., 1997). Antibiotics were added to media as required at the following concentrations: polymyxin, 50 µg mL−<sup>1</sup> ; rifampicin, 50 µg mL−<sup>1</sup> ; kanamycin, 50 µg mL−<sup>1</sup> .

### Plant Phenotyping

Phenotypic responses of L. japonicus to the spectrum of interacting bacteria was determined though observation of plants at weekly intervals following inoculation. Symbiotic phenotypes at 21 dpi were observed using a Zeiss Discovery V8 stereomicroscope. Disease responses, root-browning and wilting symptoms, of L. japonicus to R. solanacearum JS763 were analyzed at 5 and 29 dpi.

### Plant Tissue Harvesting

Whole root samples were harvested by removing the root tip (∼3 mm), to minimize hormone and meristematic activity specific to the root tip, and then collecting roots in tubes immersed in liquid nitrogen. Each whole root sample consisted of 10 roots. Root hairs were harvested using a previously described method (Sauviac et al., 2005). Roots were separated from shoots and root tips were removed. Ten roots were transferred to 50 mL Falcon tubes immersed in liquid nitrogen and after 1 min were twice vortexed for 15 s, tubes were returned to liquid nitrogen between vortexing. Falcon tubes were gently tapped to collect root hairs in the bottom and tweezers were used to remove roots stripped of root hairs. The process was then repeated with another 10 roots and a total of 100 roots was processed for each sample. Nodule primordia were collected at 7 dpi using a

scalpel to slice through the root above and below the developing primordia. Excised primordia were collected in tubes immersed in liquid nitrogen and each sample represents the total primordia collected from 30 roots. Nodules collected at 21 dpi were gently separated from the root using tweezers and collected in tubes immersed in liquid nitrogen. Each sample represents the total nodules collected from 30 roots.

### RNA Isolation, Library Preparation, and Sequencing

Total RNA was isolated from whole root, nodule primordia, and nodule tissue samples using a NucleoSpin <sup>R</sup> RNA Plant kit (Macherey-Nagel) according to the manufacturer's instructions. mRNA was directly isolated from root hair samples using a Dynabeads mRNA DIRECT kit (Thermo Fisher Scientific) according to the manufacturer's instructions. RNA quality was assessed on an Agilent 2100 Bioanalyser and samples were sent to GATC-Biotech<sup>1</sup> for library preparation and sequencing. Sequencing data is deposited at the Short Read Archive with the BioProject ID SRP127678. Sequencing data for roots and shoots 3 dpi with R7A was obtained previously (Munch et al., 2018) and is deposited at the Short Read Archive with the BioProject ID PRJNA384655.

### Bioinformatics and Statistical Analysis

Reads were mapped to the L. japonicus v. 3.0 genome and differential gene expression was analyzed using CLC Genomics Workbench 9.5.3 (Qiagen). For each sample a minimum of 30 million reads were obtained with >90% of reads mapped to the reference genome. A criterion of fourfold change with a FDR p-value ≤ 0.05 was applied to determine differential gene expression. Results of the analysis are available on Lotus Base<sup>2</sup> (Mun et al., 2016). Multidimensional scaling (MDS) analysis was performed in R (R Core Team, 2017) from total read count data using the script outlined in **Supplementary Data 1**. Heatmaps were generated in R using Heatmap and Heatmap3 functions (Zhao et al., 2014), an example script is provided in **Supplementary Data 1**. Venn diagrams were constructed using BioVenn (Hulsen et al., 2008). Detection of differentially expressed genes was also performed using the R package DEGseq (Wang et al., 2010) with a criterion of fourfold change with a FDR p-value ≤ 0.05 to examine the effects of combining replicates from different time points.

### RESULTS

### Tissue-Specific Responses of Lotus japonicus to Its Compatible Symbiont Mesorhizobium loti R7A

A comprehensive transcriptome response of L. japonicus ecotype Gifu to its compatible symbiotic partner M. loti R7A was obtained through RNA-seq analysis of specific tissues at various

<sup>1</sup>https://gatc-biotech.com

<sup>2</sup>https://lotus.au.dk

stages of symbiosis. Initial interactions between rhizobia and the legume were investigated through the isolation of root hairs following R7A inoculation at 1 and 3 dpi. The NF-deficient strain R7AnodC was included to reveal the role of NF in these early responses, which was further examined through the isolation of purified NF-treated root hairs 2 days post-treatment. Root hair deformations of L. japonicus are evident as early as 24 h following inoculation with M. loti or treatment with purified NF (Radutoiu et al., 2003; Maekawa-Yoshikawa and Murooka, 2009). Transcriptome responses during nodule colonization were examined through the harvesting of 7 dpi nodule primordia that are in the early stages of colonization by R7A and mature nitrogen-fixing nodules that are fully colonized by R7A. RNAseq of whole roots (minus root tips) and shoots collected at 3 dpi to provide a basis for tissue-specific responses was previously performed (Munch et al., 2018). For each condition examined, biological triplicate samples were harvested and total RNA was isolated. The reads were mapped to the L. japonicus v3.0 genome and all data is available from the Lotus Base resource website (Mun et al., 2016) 2 .

Multidimensional scaling analysis clearly separated the samples based on their tissue origin (**Figure 1A**). A closer examination of whole root samples revealed a clear shift following R7A inoculation (**Figure 1B**). Most root hair samples separated into distinct groups depending on their treatment (**Figure 1C**). H2O and R7AnodC samples clustered together, indicating that the majority of early root hair transcriptome responses to R7A are dependent on NF production. R7A treated root hairs formed a cluster with separation between the 1 and 3 dpi samples, while NF-treated root hairs formed a distinct cluster removed from the R7A inoculated samples.

The expression of a set of known symbiotic genes was examined in the different tissues in response to inoculation with R7A or NF treatment compared to H2O-treated controls (**Figure 1D**). No difference in gene expression was observed in shoots 3 dpi with R7A or root hairs 1 dpi with the R7AnodC mutant compared to their respective controls, 3 dpi H2O-treated shoots and 3 dpi H2O-treated root hairs. Similar expression patterns were observed for R7A and NF treated root hair and root samples compared to control 3 dpi H2O-treated root hairs. Symbiotic transcription factors, Nin (Schauser et al., 1999), NfyaA1 (Soyano et al., 2013), and Ern1 (Cerri et al., 2017; Kawaharada et al., 2017a; Yano et al., 2017) were strongly induced while Nsp1 and Nsp2 (Heckmann et al., 2006; Kawaharada et al., 2017a) showed only slight induction. The EPS receptor Epr3 (Kawaharada et al., 2015, 2017b), pectate lyase Npl (Xie et al., 2012) and Lotus homologs of Vapyrin (Murray et al., 2011) and Rpg (Arrighi et al., 2008) that were identified as required for IT formation in Medicago were all strongly induced. Notably, Rpg was not induced until 3 dpi with R7A. The receptors involved in NF perception, Nfr1, Nfr5, and Symrk (Stracke et al., 2002; Radutoiu et al., 2003) showed little change in expression other than a slight downregulation in root hairs 3 dpi.

Nodule primordia, compared to control 3 dpi H2O-treated roots, shared a similar expression pattern to root hair and root samples with the exception that the NF receptors Nfr1 and Nfr5 were downregulated. This downregulation of the NF receptors

was stronger in the 21 dpi mature nodule samples compared to control 3 dpi H2O-treated roots. The Nsp2 transcription factor also showed downregulation in mature nodules and expression of Epr3, Npl, and Vapyrin was reduced compared to primordia, root hairs, and roots.

### Early Symbiotic Responses Are Enhanced in Root Hairs and Largely Depend on Nod Factor Signaling

Predictably, significant overlap was observed in differentially expressed genes between root hairs harvested at 1 and 3 dpi compared to control 3 dpi H2O-treated root hairs and whole roots at 3 dpi compared to control 3 dpi H2O-treated whole roots (**Figure 2A**). Although expected symbiotic gene induction was observed in both tissues, the root hair samples showed enhanced transcriptional responses compared to whole roots (**Figure 1D** and **Supplementary Figure 1**). Among the most highly induced of the 93 genes identified as specifically differentially expressed within root hairs at 1 and 3 dpi compared to whole roots, we identified an aspartic protease and a calmodulin-binding protein as well as several blue copper proteins, expansins and pectinesterase/pectinesterase inhibitors (**Supplementary Table 2**). The root hair transcriptional changes observed in response to R7A are largely NF-dependent with root hairs inoculated with the NF-deficient R7AnodC strain sharing minimal overlap with R7A or NF treated samples

(**Figure 2B**). Nodule primordia and mature nodules show a large transcriptional response compared to 3 dpi root samples with downregulation of genes more prominent than in other tissues investigated (**Figure 2C**).

Altogether, the tissue transcriptome data of L. japonicus responses to R7A provided here represents an important resource for future research into this determinate nodulating model system.

various known symbiotic genes following inoculation with the indicated strains compared to H2O controls.

### Lotus japonicus Transcriptome Responses to a Spectrum of Interacting Bacteria

Transcriptional responses of L. japonicus to interacting bacteria representing a spectrum from compatible symbiont to pathogen were investigated to determine similarities or differences in the host transcriptome response to these diverse bacteria (**Figure 3A**). M. loti R7A (Ml R7A) represents an adapted symbiont that forms nitrogen-fixing nodules on L. japonicus. Sinorhizobium fredii HH103 (Sf HH103) is a broad host-range rhizobium that forms nitrogen-fixing nodules of some species of Lotus but induces uninfected primordia on L. japonicus (Acosta-Jurado et al., 2016) and (**Figure 3B**). Bradyrhizobium elkanii USDA61 (Be USDA61) is a symbiont of soybean that induces minimal responses on L. japonicus roots, with only an occasional root swelling near lateral root junctions observed (**Figure 3B**). Pseudomonas syringae pv. tomato DC3000 (Pst DC3000) is a well-known foliar phytopathogen (Xin and He, 2013). We observed no responses to root inoculation with Pst DC3000 indicating that the strain is non-pathogenic to L. japonicus

Gifu, as has been previously reported (Bordenave et al., 2013). Ralstonia solanacearum causes bacterial wilt in over 200 plant species and represents one of the most damaging bacterial pathogens in plants (Hayward, 1991). We observed that Rs JS763 causes brown discoloration of the roots from 3 dpi followed by discoloration at the base of the hypocotyl as well as chlorosis and wilting symptoms on leaves in L. japonicus Gifu (**Figure 3C**). These are typical plant disease symptoms due to R. solanacearum infection, confirming it to be a genuine root pathogen of L. japonicus, as previously described (Nagata et al., 2008).

Whole roots (minus root tips) treated with the spectrum of diverse bacterial strains or water were harvested at 1 and 3 dpi in biological duplicates. RNA isolation, Illumina sequencing and read mapping were performed as described above for the tissue samples. All data are available from the Lotus Base resource website (Mun et al., 2016) 2 . We detected outlier replicates for each inoculum and these were removed before proceeding with downstream analysis. The remaining three datasets, consisting of two 3 dpi and one 1 dpi samples, were combined to allow robust detection of differentially expressed genes. The effects of combining replicates from two 3 dpi and one 1 dpi samples were investigated using the R package DEGseq (Wang et al., 2010), which allows for differential gene expression analysis to be performed with fewer than three replicates. Differentially expressed genes were identified using the combined three replicates as well as using just the two 3 dpi replicates. A strong overlap in the genes identified between the three and two replicate analysis was observed. Almost all genes identified using the combined three replicates in our CLC genomics workbench analysis were also identified as differentially expressed in both the two and three replicate analyses using DEGseq, indicating that they likely represent genuine differentially regulated genes (**Supplementary Data 2**). However, genes oppositely regulated at 1 and 3 dpi or differentially expressed at only one time point may not have been detected due to the combined analysis of 1 and 3 dpi time points. We used the conservative lists of differentially expressed genes obtained from CLC genomics workbench in the downstream analyses. MDS analysis clustered Ml R7A and Sf HH103 together, Pst DC3000 clustered with H2O while Be USDA61 was slightly removed from them and Rs JS763 samples formed a less defined group removed from the other samples (**Figure 3D**).

To assess the symbiotic response induced by the diverse strains, expression of the same set of known symbiotic genes analyzed in the various tissue samples following inoculation with R7A or NF was examined through comparing bacteria inoculated root samples to control H2O-treated roots (**Figure 3E**). Ml R7A and Sf HH103 induced similar responses that are comparable to those obtained in analysis of the various tissues examined following R7A or NF treatment (**Figure 1D**) with the exception of Rpg expression, which was not induced in the Ml R7A and Sf HH103 samples. Only the Npl gene showed any expression change following inoculation with Be USDA61 and no response for any of the genes was observed following Pst DC3000 or Rs JS763 inoculation.

### L. japonicus Responses to a Non-adapted Pathogen and Incompatible Rhizobia

Pst DC3000 induced no discernible phenotypic responses on L. japonicus. Only a small set of genes (44) showed significant expression changes following Pst DC3000 treatment, 8 of which showed similar regulation following Ml R7A inoculation (**Figure 4A** and **Supplementary Table 3**). Similar expression of the identified Pst DC3000 differentially regulated genes was observed following inoculation with the spectrum of strains investigated, indicating that these genes may represent a common response to bacteria by L. japonicus (**Figure 4A** and **Supplementary Table 3**). Of interest in the identified genes are two leucine-rich repeat (LRR) receptor like proteins (Lj0g3v0095839 and Lj0g3v0331049), an unknown-conserved GYF-domain containing protein (Lj5g3v0101610) and an exportin-7 like protein (Lj1g3v2392240) that show upregulated expression in response to all of the diverse strains. LRR receptors represent the largest receptor family in plants and are involved in diverse developmental, defense and symbiotic processes (Torii, 2004). Exportin-7 proteins are reportedly involved in nuclear export processes (Mingot et al., 2004).

Be USDA61 and Sf HH103 represent two rhizobial species that are incompatible for the establishment of nitrogen-fixing symbiosis with L. japonicus (**Figure 3B**). Significant gene expression changes were limited following Be USDA61 inoculation, whilst Sf HH103 inoculation induced a greater response and shared a large overlap with Ml R7A inoculation (**Figure 4B**).

Genes showing significant differential expression following Ml R7A inoculation but not Sf HH103 or Be USDA61 represent candidates involved in specifically promoting interaction with the compatible symbiont. The 34 genes identified that fit these criteria are listed in **Supplementary Table 4**. Of particular interest in this list are Lj5g3v2288900 that encodes a legume specific chalcone isomerase that has recently been shown to be involved in the biosynthesis of an Ml R7A NodD1 activating inducer, specifically within ITs (Kelly et al., 2017b) and Lj4g3v2365210 that encodes for the Nfy-B1 subunit, which forms part of the Nuclear Factor Y transcriptional factor known to be involved in nodulation (Soyano et al., 2013). Lj4g3v1983610 encodes an ENOD-like protein with a predicted cupredoxin domain. The protein shares 46% amino acid identity to ENOD16 of M. truncatula, which has been shown to be induced during symbiosis. No clear roles for ENOD16 have been demonstrated but it has been proposed that they may be involved in cell wall reorganization (Greene et al., 1998), which is required for IT development.

### Interplay Between Symbiotic and Defense Responses in L. japonicus

Rs JS763 was shown to be a root pathogen of L. japonicus (**Figure 3B**). Rs JS763 inoculation resulted in a distinct transcription response, with 347 (277 upregulated and 70 downregulated) differentially expressed genes identified

(**Supplementary Data 3**). Among the genes upregulated by Rs JS763 are five genes annotated as WRKY transcription factors (Lj0g3v0074419, Lj0g3v0244439, Lj1g3v1815540, Lj1g3v1787630, and Lj1g3v1220240). WRKY transcription factors belong to a large family of regulatory proteins involved in various plant processes but most notably in plant immunity against pathogen attack, including R. solanacearum infection (Mukhtar et al., 2008; Wang et al., 2013, 2014; Dang et al., 2014; Cai et al., 2015; Liu et al., 2017). Two 1-aminocyclopropane-1-carboxylate oxidase (ACO) 1-like genes (Lj2g3v0911860 and Lj6g3v0086450) were also induced. ACO proteins are involved in biosynthesis of ethylene, a plant hormone that is involved in many developmental and physiological responses including defense (Broekgaarden et al., 2015) and symbiosis (Reid et al., 2018). Ethylene signaling plays a critical role in bacterial wilt disease development in A. thaliana (Hirsch et al., 2002), M. truncatula (Moreau et al., 2014) and in tomato resistance to R. solanacearum (Zhang et al., 2004). In addition to inducing ethylene production by the host plant, R. solanacearum is also capable of producing ethylene itself (Weingart et al., 1999). Other defense related genes induced by Rs JS763 included pathogenesis-related proteins [such as PR1a: Lj6g3v2170740, beta-1,3-glucanase (PR-2): Lj0g3v0278459; PR10: Lj0g3v0286359; peroxidase 1-like: Lj0g3v0336889; two chitinases: Lj5g3v1961260 and Lj6g3v1078650] and several protease inhibitor proteins potentially involved in plant immunity (Lj0g3v0174139, Lj0g3v0174109, Lj2g3v0604160, Lj0g3v0265049, Lj5g3v1174410). Homologs of some of these genes were also reported as differentially expressed in response to R. solanacearum in M. truncatula according to proteomic and transcriptomic analyses (Yamchi et al., 2018). Interestingly, a gene annotated as a WAT1 (Walls Are Thin1)-related gene (Lj1g3v0913340) was downregulated in response to Rs JS763. In A. thaliana, inactivation of WAT1 gene, which is required for secondary cell-wall deposition, conferred broad-spectrum resistance to vascular pathogens, including R. solanacearum (Denance et al., 2013).

A transient defense response in L. japonicus roots during the early stages of symbiosis with compatible rhizobia had previously been reported (Kouchi et al., 2004). We investigated expression of the defense related genes identified by Kouchi et al. (2004) but did not observe comparable expression in response to symbiotic or pathogenic strains in our dataset (**Figure 5A** and **Supplementary Data 4**). Since symbiosis genes were consistently induced across all our experiments, we take this as an indication that the set of genes with defense-related annotations identified by Kouchi et al. (2004), are not generally associated with symbiotic or pathogenic interactions and our data does not support the notion that an early defense-like response is evoked by compatible rhizobia in the establishment of symbiosis with L. japonicus.

It remains possible, however, that overlaps in gene expression responses could exist between infection-competent symbiotic and pathogenic bacteria. To further examine such potential similarities, a list of commonly differentially expressed genes in early symbiotic samples was compiled from Ml R7A inoculated roots and root hairs. Expression of the 71 genes was analyzed in the spectrum of interacting bacteria, including Rs JS763 (**Figure 5B**). In a complementary analysis, expression of the 347 differentially expressed genes following Rs JS763 inoculation was compared in response to the diverse strains (**Figure 5C**). Rs JS763 inoculated roots did not show similar gene expression patterns with the symbiotic samples in either of these analyses. However, some overlapping expression changes were observed between Rs JS763 responsive genes and a compiled list of early symbiotic genes that were responsive to either Ml R7A or NF in whole roots or root hairs (**Figure 5D** and **Supplementary Table 5**).

This overlap included the cytokinin oxidase Ckx3 (Lj5g3v0692300) that is involved in cytokinin breakdown during nodulation (Reid et al., 2016) and the sulfate transporter Sst1 (Lj2g3v0776860.1) that is required for nitrogen-fixation within nodules. In addition, putative defense-related genes identified were a PR10-like gene (Lj0g3v0286359), the WAT1-related gene (Lj1g3v0913340) and a gene encoding a salicylate O-methyltransferase-like protein (Lj6g3v0509430) that is involved in the biosynthesis of methyl salicylate which was found to participate in root responses against fungal pathogens (Ament et al., 2010; Boba et al., 2016). Further studies will be required to determine the roles of the symbiosis-related genes in the Lotus–Ralstonia interaction and of the putative defense genes in Lotus–rhizobium interactions. However, the identification of genes responding to both Ralstonia and rhizobium treatments does not necessarily indicate that symbiotic rhizobia are initially perceived as pathogenic by Lotus. In contrast, our data indicate that the pathogenic and symbiotic responses are well-separated as early as 1 dpi (**Figure 3D**).

## DISCUSSION

Lotus japonicus has been extensively utilized for the study of symbiotic interactions with rhizobial bacteria and mycorrhizal fungi. Analysis of the host transcriptional response during these interactions has been beneficial in developing our understanding of the molecular processes behind the symbiotic interactions and identifying potential genes important for these.

Transcriptome data collected from numerous microarray based studies of L. japonicus tissues at various stages of development or in association with biotic and abiotic treatments has been collated together in the LjGEA and is now accessible through Lotus Base<sup>2</sup> (Sanchez et al., 2008, 2011; Guether et al., 2009; Hogslund et al., 2009; Diaz et al., 2010; Verdier et al., 2013; Mun et al., 2016). We have expanded on this through the generation of a comprehensive RNA-seq based transcriptome analysis of the response of L. japonicus to its compatible symbiont M. loti R7A. Previous studies in M. truncatula have highlighted the benefits of isolating root hairs to amplify early symbiotic transcriptome responses (Breakspear et al., 2014). In our analysis, early symbiotic signaling was similarly amplified in root hair samples compared to whole roots. Npl (Xie et al., 2012), Vapyrin (Murray et al., 2011), and Rpg (Arrighi et al., 2008) represent genes known to be required for IT development that were highly induced in our root hair samples. Additional genes identified as showing enhanced expression within root hairs represent potential candidates that may be involved in

IT development. An interesting candidate from this analysis is the aspartic peptidase APN1 (Lj3g3v0950730.1) that was recently discerned to be expressed in nodules and demonstrated to play a crucial role in nodule functioning. apn1 mutant nodules fail to fix nitrogen and undergo early senescence in a M. loti strain-depended manner (Yamaya-Ito et al., 2017). Our identification of root hair enhanced expression of Apn1 compared to whole roots suggests that IT phenotypes of apn1 inoculated with R7A would be of interest to examine. As expected, early symbiotic signaling was largely NF-dependent and although considerable overlap was observed for differentially expressed genes following R7A or NF treatment, including most of the known symbiotic genes, there remains clear differences in the responses. A comparable difference in differentially expressed genes following NF treatment or S. meliloti inoculation was observed in M. truncatula (Breakspear et al., 2014). The observed differences in responses to purified NF compared to rhizobial inoculation may be due to the effects of other rhizobial factors, for example surface glycans, or differences in the concentration of NF roots are exposed to following rhizobial inoculation compared to purified NF treatment. Rhizobia produce an array of NF species and it is also possible that this induces differences in transcriptional responses compared to application of one particular purified NF species. Interestingly, Rpg induction was not observed in 2 dpi NF-treated or at 1 dpi with R7A, suggesting that Rpg induction is delayed compared to the other symbiotic genes investigated. This delay was not observed in M. truncatula, where Rpg induction was detected in roots at 1 dpi with S. meliloti or NF-treated root hairs (Arrighi et al., 2008; Breakspear et al., 2014). Nodule primordia and mature nodules exhibited large transcriptional reprogramming as has previously been observed for Lotus nodules (Takanashi et al., 2012). Highly upregulated genes include various transporters, auxin and cytokinin responsive genes as well as genes known to be required for nodule functioning such as the sulfate transporter Sst1 (Krusell et al., 2005) and homocitrate synthase Fen1 (Hakoyama et al., 2009).

Model legume resources, developed largely through the study of symbiotic interactions, have not been exploited to such an extent in interactions with diverse microbes. In this

study, we identified distinct transcriptome responses to a spectrum of interacting bacteria ranging from symbiotic to pathogenic. The broad host-range rhizobium Sf HH103 forms small nodule primordia that remain uninfected on L. japonicus (Sandal et al., 2012; Acosta-Jurado et al., 2016). A remarkably similar early transcriptome response of L. japonicus to Ml R7A and Sf HH103 was observed, indicating that NF-induced signaling is comparable between the two strains. Both strains produce a variety of NFs with one notable difference being the substitution on the reducing terminal residue, which is generally an acetylfucosyl for Ml R7A NF and a methylfucosyl for Sf HH103 NF (Gil-Serrano et al., 1997; Rodpothong et al., 2009). Genes identified as potentially limiting the symbiotic capacity of Sf HH103 based on differential expression compared to Ml R7A include a legume specific chalcone isomerase, Nfy-B1 and an ENOD16 encoding gene. Be USDA61 induced only occasional root swellings on L. japonicus and a limited transcriptome response was observed, suggesting that the NF species produced by this rhizobial strain are not well-perceived by L. japonicus. We also detected minimal transcriptome responses to Pst DC3000 inoculation, which appeared to be non-pathogenic to L. japonicus roots. Differentially expressed genes identified from these non-symbioitic/non-pathogenic strains may represent a common response to bacteria as similar transcription responses were overserved across the spectrum of interacting bacteria.

Rs JS763 was characterized as a genuine pathogen of L. japonicus with typical disease symptoms observed following root inoculation and a clear and distinct root transcriptome response was detected, allowing us to investigate the interplay between pathogenic and symbiotic signaling. Transcriptome-based studies have previously suggested that an initial defense-like response to rhizobia is subsequently dampened through symbiotic signaling in soybean/B. japonicum, M. truncatula/S. meliloti, and L. japonicus/M. loti symbioses (Kouchi et al., 2004; Lohar et al., 2006; Libault et al., 2010). Kouchi et al. (2004) identified a set of potential defense-related genes that were induced at early time points before being suppressed at later stages of L. japonicus/M. loti interactions. We did not observe similar induction of the identified genes in our RNAseq transcriptome analysis following M. loti R7A inoculation, indicating that the gene set is not reproducibly associated with symbiotic infection. Furthermore, only two of these genes (Lj0g3v0229819 and Lj5g3v1203340) encoding carboxylesterase 6-like and basic 7S globulin-like proteins respectively, showed a significant response following Rs JS763 inoculation in our analysis, suggesting that although the genes identified by Kouchi et al. (2004) have potential defense associated annotations they may not represent genes that are generally pathogen responsive in L. japonicus. It is possible that differences in host and M. loti genotypes as well as growth setups used in our analysis and by Kouchi et al. (2004) may be responsible for the differences in transcriptional responses observed.

We did however identify a small overlap between early time point Ml R7A responsive genes and Rs JS763 differentially expressed genes. The only known symbiotic genes identified as commonly transcriptionally regulated are the cytokinin oxidase Ckx3 and the sulfate transporter Sst1. Cytokinin is a key plant hormone and is tightly regulated in root development and symbiotic processes (Reid et al., 2017). Transcriptional changes in the expression of cytokinin related genes have previously been reported following R. solanacearum inoculation of M. truncatula (Moreau et al., 2014). SST1 functions during the later stages of symbiosis to transport sulfate from the plant to rhizobia where it is essential for protein synthesis, including nitrogenase biosynthesis (Krusell et al., 2005). It is unclear if a similar role is performed by SST1 in earlier stages of interactions with rhizobia or other bacteria. The only obvious potentially defense-related genes identified in the list of commonly induced genes were a PR10-like gene (Lj0g3v0286359), WAT1-related gene (Lj1g3v0913340) and a gene encoding a salicylate O-methyltransferase-like protein (Lj6g3v0509430). Lj6g3v0509430 is one of several salicylate O-methyltransferase-like proteins with high similarity encoded by L. japonicus and additional studies are required to determine if it is involved in salicylic acid responses.

Altogether, our analysis of transcriptome responses to a spectrum of interacting bacteria indicates that distinct transcriptome responses are observed in response to symbiotic and pathogenic bacteria and does not support the concept that an early defense-like response is generally evoked by compatible rhizobia in L. japonicus during the establishment of symbiosis.

### AUTHOR CONTRIBUTIONS

SK and TM performed the plant experiments, isolated RNA, and analyzed the sequencing data. SUA conceived and supervised the study. CB and JS devised the experiments and contributed resources. SK, CB, and SUA wrote the manuscript with input from all authors.

## FUNDING

The work was supported by the Danish National Research Foundation grant DNRF79 (JS) and a fellowship from the Institut Français du Danemark in the frame of the Blåtand Program (CB).

## ACKNOWLEDGMENTS

We thank J. Luisetti for providing strain JS763 of R. solanacearum used in this work, Finn Pedersen for greenhouse assistance and Clive Ronson, John Sullivan, and Nicolai Maolanon for purified Nod factor.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01218/ full#supplementary-material

### REFERENCES

fpls-09-01218 August 16, 2018 Time: 19:15 # 12


of root nodule formation in a model legume, Lotus japonicus. DNA Res. 11, 263–274. doi: 10.1093/dnares/11.4.263



**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 Kelly, Mun, Stougaard, Ben and Andersen. 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.

# Mini-Review: Nod Factor Regulation of Phytohormone Signaling and Homeostasis During Rhizobia-Legume Symbiosis

William P. Buhian and Sandra Bensmihen\*

LIPM, Université de Toulouse, INRA, CNRS, Toulouse, France

The rhizobia-legume symbiosis is a mutualistic association in which bacteria provide plants with nitrogen compounds and the plant provides bacteria with carbon sources. A successful symbiotic interaction relies on a molecular dialog between the plant and the bacteria, and generally involves rhizobial lipo-chitooligosaccharide signals called Nod factors (NFs). In most cases, specific NF perception is required for rhizobia to enter root cells through newly formed intracellular structures called infection threads (ITs). Concomitantly to IT formation in root hairs, root cortical cells start to divide to create a new root organ called the nodule, which will provide the bacteria with a specific micro-environment required for symbiotic nitrogen fixation. During all these steps of plant–bacteria interaction, new plant cellular compartments and developmental programs are activated. This interaction is costly for the plant that tightly controls symbiosis establishment and functioning. Phytohormones are key regulators of cellular and developmental plasticity in plants, and they are influential endogenous signals that rapidly control plant responses. Although early symbiotic responses were known for decades to be linked to phytohormone-related responses, new data reveal the molecular mechanisms involved and links between phytohormones and the control of early symbiotic events. Reciprocally, NF signaling also targets phytohormone signaling pathways. In this review, we will focus on the emerging notion of NF and phytohormone signaling crosstalk, and how it could contribute to the tight control of symbiosis establishment in legume host plants.

### Edited by:

Pascal Ratet, Centre National de la Recherche Scientifique (CNRS), France

#### Reviewed by:

Jeanne Marie Harris, University of Vermont, United States Brendan Keith Riely, University of California, Davis, United States

> \*Correspondence: Sandra Bensmihen sandra.bensmihen@inra.fr

#### Specialty section:

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

Received: 06 April 2018 Accepted: 06 August 2018 Published: 03 September 2018

#### Citation:

Buhian WP and Bensmihen S (2018) Mini-Review: Nod Factor Regulation of Phytohormone Signaling and Homeostasis During Rhizobia-Legume Symbiosis. Front. Plant Sci. 9:1247. doi: 10.3389/fpls.2018.01247 Keywords: auxin, cytokinins, gibberellin, ethylene, strigolactone, infection, nodule organogenesis, LCOs

### INTRODUCTION

Legume plants can interact with soil bacteria, named rhizobia, to establish the rhizobium legume (RL) symbiosis. Legumes host Rhizobia in specific root organs called nodules, where they fix atmospheric nitrogen. This symbiosis provides the plant with nitrogen compounds (ammonium) and the plant provides the bacteria with carbon sources. The efficiency of this interaction relies for a great part on the massive intracellular infection of plant cells by rhizobia and the protective structure of the nodule. The plant host must invest a great deal of energy to both house and support nitrogen fixing bacteria, and as such, must tightly regulate this process.

Nod factors (NFs) are lipo-chitooligosaccharide (LCO) molecules produced by rhizobia in response to flavonoids present in root exudates. NFs are generally essential for the onset and the

maintenance of the RL symbiosis (Dénarié, 1996). Intriguingly, NFs can also stimulate lateral root formation via the symbiotic signaling pathway implying that similar host mechanisms are involved in both nodule and lateral root development (Olah et al., 2005). Genetic pathways governing NF perception and signaling are now quite well understood (Gough and Cullimore, 2011). NFs are perceived by the plant at the root epidermis and are likely produced by rhizobia all along the infection process (Sharma and Signer, 1990).

Rhizobia first attach to root hairs (RHs) in a susceptible zone of host roots (Bhuvaneswari et al., 1981) and then enter RHs through a membrane invagination called an infection thread (IT). Even before the IT is formed, pericycle cells are activated and divide to start organogenesis of the nodule (Timmers et al., 1999). NFs are usually essential for the infection process and, in Medicago sativa, purified NFs can trigger pericycle activation and initiate nodule organogenesis (Truchet et al., 1991; Timmers et al., 1999). Defects in the infection process often result in defective nodule organogenesis. Infection and organogenesis events of the RL symbiosis are thus tightly coordinated. This tight coordination and signaling across root layers, and the events of cellular reactivation and cell division associated with symbiotic establishment strongly imply the involvement of hormonal pathways.

Recent transcriptomic data suggest that NF signaling regulates the host hormone biosynthesis and signal transduction pathways. Hormones function in the pico to nano-molar range (Gaspar et al., 2003). Their activity depends on a tight balance of activation (biosynthesis) and inactivation (conjugation, degradation) pathways, as well as transport and signaling. In this mini-review, we will focus on data showing positive and negative feedback regulatory mechanisms between NF signaling and hormonal pathways. We present data from legumes producing both indeterminate (Medicago truncatula, pea) and determinate [Lotus japonicus, Glycine max (soybean)] nodules. These two types of nodule differ in the site of initial cortical divisions, persistency of the nodule meristem and auxin sensitivity (Bensmihen, 2015; Ng and Mathesius, 2018). Effects of NF signaling on both hormone homeostasis and hormone signaling will be discussed.

### NF SIGNALING AND REGULATION OF HORMONE HOMEOSTASIS

Phytohormones are key regulators of plant growth and responses to biotic and abiotic stresses. Several transcriptomic studies show that hormone biosynthesis, activation or degradation genes are differentially expressed upon NF treatment (see **Table 1** and **Figure 1**).

### Cytokinins Biosynthesis and Activation

Numerous studies document the role of CKs as key regulators of nodule organogenesis (Gonzalez-Rizzo et al., 2006; Murray et al., 2007; Tirichine et al., 2007) and IT formation/progression (Held et al., 2014; Jardinaud et al., 2016). Recent transcriptomic data show that expression of CK biosynthesis genes such as isopentenyl transferases (IPT) and CYP735A that contribute to the biosynthesis of the bioactive CK form trans-zeatin, as well as CK activation genes such as LONELY GUYs (LOGs), can be induced by a 3 h NF treatment (van Zeijl et al., 2015b). Moreover, MtIPT1and MtCYP735A1 regulation is independent of the CK receptor MtCRE1, suggesting a direct (and possibly local) production of CK as an early event following NF signaling. Similarly, LjIPT2 and LjLOG4 are regulated by NF application independently of the CK receptor LjLHK1 in L. japonicus (Reid et al., 2017). Accumulation of bioactive CKs is also detected in the root susceptible zone following a 3 h NF treatment, and this is dependent on the NF signaling gene MtDMI3 (van Zeijl et al., 2015b). Several genes from the trans-zeatin metabolic network are regulated in RHs after a 24 h NF treatment, suggesting accumulation of the bioactive CK trans-zeatin upon NF treatment. However, bioactive CK levels have not been measured in NF treated RHs (Jardinaud et al., 2016).

### Inactivation and Degradation

Interestingly, CK inactivation enzymes adenyl phosphoribosides transferases (APT) and cytokinin oxidases (CKX) are also induced by a 3 h NF treatment but in an MtCRE1 dependent manner, suggesting downstream negative regulatory feedback loops on CK accumulation (see **Figure 1**). Indeed, expression of the AtCKX3 gene in M. truncatula plants under an epidermis specific promoter enhanced the number of ITs and nodules formed upon rhizobium inoculation. In contrast, expression of AtCKX3 under a cortical specific promoter reduced nodulation (Jardinaud et al., 2016). In L. japonicus, NFs also induce expression of the LjCKX3 gene but its promoter activity is specific to inner root tissues. Inactivation of LjCKX3 leads to enhanced production of bioactive CK forms and reduces nodule organogenesis and IT formation (Reid et al., 2016). Altogether, this suggests that reducing the levels of active CKs may increase the efficiency of the infection process. This fits with the model suggested by Held et al. (2014) that argues for an inhibitory feedback loop driven by CK accumulation in the epidermis at late stages of nodulation. This is also consistent with antagonistic effects of CK application on the induction of the NF responsive gene MtENOD11. Indeed, a 3 h CK pretreatment of M. truncatula roots significantly reduces the induction of MtENOD11 by NF. Conversely, epidermal expression of the AtCKX3 gene enhanced the NF induction of MtENOD11 (Jardinaud et al., 2016). In summary, growing evidence suggests that NFs induce CK production, which first controls nodule organogenesis, then rapidly activates negative feedbacks on NF signaling and infection processes, notably in the epidermis.

### Gibberellins

Increasing evidence also suggests a role for gibberellins (GAs) in controlling early events of symbiosis. A 24 h NF treatment induces both GA metabolic and biosynthetic genes in M. truncatula RHs, suggesting accumulation of the bio-active GA precursor GA12 in the root epidermis upon NF treatment (Breakspear et al., 2014; Jardinaud et al., 2016). An important NF-dependent and transient activation of GA biosynthesis



Data from Breakspear et al. (2014) and Jardinaud et al. (2016) were obtained from epidermal samples. All other transcriptomic data were obtained from whole root samples of Medicago truncatula (Larrainzar et al., 2015; van Zeijl et al., 2015b; Herrbach et al., 2017), Glycine max (Hayashi et al., 2012), and Lotus japonicus (Miyata et al., 2013). Unless otherwise indicated, samples were treated with Nod factors. <sup>a</sup>Root samples inoculated with rhizobia. CK, cytokinins; GA, Gibberellins; SL, strigolactones; JA, jasmonic acid; BR, brassinosteroids; NF, Nod factors. <sup>1</sup>Jardinaud et al., 2016. Breakspear et al., 2014. Damodaran et al., 2017. Plet et al., 2011. Ng et al., 2015. Larrainzar et al., 2015. 7 van Zeijl et al., 2015b. Franssen et al., 2015. Herrbach et al., 2017. Murray et al., 2007. Gonzalez-Rizzo et al., 2006. Reid et al., 2017. Mortier et al., 2015. Reid et al., 2016. Hayashi et al., 2012. Op den Camp et al., 2011. Penmetsa and Cook, 1997. Penmetsa et al., 2003. Miyata et al., 2013. van Zeijl et al., 2015a.

genes was observed in soybean roots 12 hpi with rhizobia (Hayashi et al., 2012). Interestingly, 10−<sup>7</sup> M GA3 treatment was previously shown to suppress NF-induced expression of the transcription factors NSP2 and NIN in L. japonicus and 10−<sup>6</sup> M GA3 suppressed NF induced RH deformation (Maekawa et al., 2009). Likewise, pretreatment of M. truncatula roots with 10−<sup>6</sup> M GA3 suppressed NF dependent ENOD11 induction. NF treatment also induced expression of the GA catabolic and biosynthetic enzymes GA2 and GA20ox, respectively (Fonouni-Farde et al., 2016). Moreover, this GA2ox catabolic gene was positively regulated by a 10 h NF treatment (Herrbach et al., 2017) and down-regulated in RHs after a 24 h NF treatment

(Zaat et al., 1989; Breakspear et al., 2014). These data suggest that NFs can rapidly induce the biosynthesis of bioactive GA in RHs, which consequently triggers a negative feedback leading both to the downregulation of NF signaling and the activation of GA catabolic enzymes (see **Figure 1**). Increasing evidence also shows that bioactive GAs negatively regulate nodulation and infection in both determinate and indeterminate nodules (Maekawa et al., 2009; Fonouni-Farde et al., 2016). Thus, NF induced GAs could help fine-tune NF signaling and rhizobium infection during symbiosis.

### Ethylene

Previous genetic, physiologic, and pharmacologic studies highlight the key regulatory role ethylene plays in early symbiotic processes (Zaat et al., 1989; Heidstra et al., 1997; Oldroyd et al., 2001; Penmetsa et al., 2008). NFs can induce the expression of several ethylene biosynthetic genes, including two 1-aminocyclopropane-1-carboxylate (ACC) synthase genes (MtACS1 and MtACS2; van Zeijl et al., 2015b). Moreover, three ACS genes are upregulated by NF signaling at 6 hpi with rhizobium (Larrainzar et al., 2015). Similarly, NF dependent ethylene production is detected in L. japonicus roots as early as 6 hpi (Reid et al., 2018). MtACS3 is also synergistically regulated by a 10 h treatment with a combination of NFs and auxin (Herrbach et al., 2017). Promoter:GUS fusions showed RH expression of three ACS genes in M. truncatula (Larrainzar et al., 2015). This is consistent with previous data obtained in L. japonicus where a 1 h treatment with 10−<sup>8</sup> M NF increased the expression of the ACC oxidase LjACO2, which was synergistically regulated by the combined application of NFs and ACC (Miyata et al., 2013). In contrast, NF-dependent downregulation of ethylene biosynthesis was observed at 48 hpi in the susceptible zone of soybean roots (Hayashi et al., 2012). Altogether, these data suggest that NF signaling rapidly and transiently induces the production of a negative regulator of infection, ethylene.

### Auxin

### Auxin Metabolism

Nod factor-dependent induction of genes encoding auxin biosynthetic enzymes such as YUCCA and ANTHRANILATE SYNTHASE (ASA) was observed early after Sinorhizobium meliloti inoculation in whole roots (Larrainzar et al., 2015). Several auxin signaling genes are activated in RHs after a 24 h NF treatment (Breakspear et al., 2014), which argues for an effect of NFs on auxin production and signaling (see section "Auxin Signaling"). Similarly, we recently described a synergistic effect of NFs and auxin on the transcription of a large number of hormone biosynthesis genes, including an auxin biosynthesis TRYPTOPHAN AMINOTRANSFERASE-related (TAR2) homolog (Herrbach et al., 2017). Interestingly, a GH3.1 gene, predicted to encode an auxin conjugation and inactivation enzyme (Yang et al., 2015) is expressed both in NF treated RHs as well as developing nodules (Breakspear et al., 2014; Ng et al., 2015; Jardinaud et al., 2016). We found this same GH3 gene to be synergistically upregulated by NFs and auxin (Herrbach et al., 2017). This suggests fine tuning of bioactive pools of auxin, in RHs and probably inner cortical tissues for indeterminate nodules, upon NF perception and rhizobium infection.

### Auxin Transport

Several studies link NF signaling to auxin transport. Spot inoculation of compatible rhizobia (but not a non-host strain) or micro-targeting of specific LCOs can modify auxin gradients, as measured by the GH3:GUS reporter gene in white clover (Mathesius et al., 1998). Likewise, flavonoid application also inhibits auxin transport. NF application or S. meliloti infection inhibits auxin transport from shoot to root at 24 h, as well as regulating expression of some MtPIN auxin efflux transporter genes, in an MtCRE1 dependent manner (van Noorden et al., 2006; Plet et al., 2011; Ng et al., 2015). Moreover, MtCRE1 dependent pathways also control the accumulation of flavonoids in M. truncatula roots upon infection and flavonoid application can rescue the Mtcre1 nodulation phenotype. These data suggest that NF induced, CK signaling triggers flavonoid induction and the subsequent inhibition of polar auxin transport. The resulting accumulation of auxin initiates cortical cell division and nodule organogenesis (**Figure 1**).

### Strigolactones

Strigolactones (SGLs) show a dose-dependent effect on nodulation in M. truncatula (Gomez-Roldan et al., 2008; De Cuyper et al., 2015). Interestingly, several studies (Larrainzar et al., 2015; van Zeijl et al., 2015a; Herrbach et al., 2017) showed a direct NF regulation of the expression of the SGL biosynthesis gene DWARF27 (D27). The promoter of the D27 gene is active in the root epidermis after a 3 h NF treatment and upon early steps of nodule organogenesis (van Zeijl et al., 2015a). Moreover, we observed that a combined auxin+NF treatment reduced the NF induction of D27 expression, suggesting that auxin can antagonize NF effects (Herrbach et al., 2017).

### Jasmonic Acid (JA) and Brassinosteroids (BR)

Jasmonic acid (JA) and BR regulate plant stress responses and plant growth, but their role in RL symbiosis is not well understood. Conflicting reports indicate that host responses to these hormones vary depending on the legume species, hormone dose, or type of treatment studied (see Ferguson and Mathesius, 2014 for review and Conclusion and Perspectives section below). Some evidence suggests NF regulation of JA biosynthetic enzymes. Larrainzar et al. (2015) observed NFdependent induction of two JA biosynthesis genes at 3–6 hpi and Breakspear et al. (2014) observed downregulation of a few JA biosynthetic genes in RHs after a 24 h NF treatment. In contrast, NF-dependent upregulation of several JA biosynthesis genes was observed in soybean at 48 hpi, while there seemed to be a reduction in expression of BR biosynthetic genes (Hayashi et al., 2012).

FIGURE 1 | Recapitulative spatio-temporal scheme of the interactions between NF signaling and downstream hormonal pathways. Rhizobia produce NFs that are perceived in root hairs (RHs) through LysM-RLK receptors such as NFP and LYK3. NF perception leads to calcium spiking and activation of DMI3 that acts upstream of the CK receptor MtCRE1. The signaling cascade involving DMI3 and CRE1 is likely involved in both epidermal and cortical signaling (black boxes). NF treatment triggers early CK biosynthesis gene (IPT/CYP735A/LOG) expression and CK accumulation in M. truncatula roots downstream of DMI3 but independently of CRE1 signaling (van Zeijl et al., 2015b). Although tissue specificity of this CK production was not determined, evidence from L. japonicus and M. truncatula suggests that epidermal CK accumulation is a negative regulator of RH infection (1) and NF signaling (2) (Held et al., 2014; Jardinaud et al., 2016). In contrast, cortical CK is a positive regulator of nodule organogenesis (3) (Gonzalez-Rizzo et al., 2006; Murray et al., 2007; Reid et al., 2017). Bioactive CKs are perceived by CRE1 and induce expression of NIN and ERN1 (Ariel et al., 2012), which are positive regulators of both infection and nodule organogenesis (Andriankaja et al., 2007; Marsh et al., 2007). This induction might be partly through regulation of DELLA activities. GA is a negative regulator of DELLA protein stability. Bioactive GA pools are likely present in both epidermis and cortex early after NF treatment (Fonouni-Farde et al., 2016; Jardinaud et al., 2016; Herrbach et al., 2017). DELLAs play a positive regulatory role on symbiotic gene expression such as ERN1 and they negatively regulate CK degradation (Fonouni-Farde et al., 2016, 2017; Jin et al., 2016). In contrast, CK positively regulate GA inactivation enzymes and down-regulate expression of the GA20ox activation enzyme (Fonouni-Farde et al., 2018), suggesting a negative feedback of CK on GA active pools. NF signaling induces ethylene production, both independently of the LHK1/CRE1 CK pathway (4) (Reid et al., 2018) and downstream of CK perception (5) (van Zeijl et al., 2015b). Ethylene reduces CK production in M. truncatula roots, possibly through negative feedback on NF signaling (6) (van Zeijl et al., 2015b). Ethylene negatively regulates NF-induced calcium spiking, RH infection, and nodule organogenesis (Heidstra et al., 1997; Penmetsa et al., 2008). Regulation of auxin biosynthetic and conjugation enzyme (GH3) genes occurs in NF treated RHs and upon S. meliloti inoculation in an NF-dependent manner (Breakspear et al., 2014; Larrainzar et al., 2015; Jardinaud et al., 2016; Herrbach et al., 2017). Reciprocal positive and negative feedback regulatory loops between some auxin and NF regulated genes (7) was shown by comparing the combined effect of auxin and NFs to either treatment alone (Herrbach et al., 2017). Downstream of CK perception, control of auxin transport in the cortex seems regulated by differential expression of PIN genes (Plet et al., 2011) or accumulation of flavonoid compounds. Auxin accumulation is also mediated by MtLAX2 that is induced upon S. meliloti infection in vasculature and early nodule primordia, and which is required for nodule organogenesis (Roy et al., 2017). In parallel, epidermal auxin signaling controls infection thread (IT) formation, at least partly through ARF16a (Breakspear et al., 2014). Exogenous application of high concentrations (µM range) of ABA or JA inhibits NF-induced calcium spiking (Sun et al., 2006; Ding et al., 2008) but this inhibition is so far not supported by any transcriptomic data. Plain bars represent negative and plain arrows positive regulations. Dashed lines are hypothetical relationships, and solid lines have evidence from the literature. Different hormones are highlighted by different colors. CK, cytokinins; GA, gibberellins; ABA, abscisic acid; JA, jasmonic acid; NF, Nod factors. Bacteria entrapped in curled root hair and IT are shown in blue.

Hence, these data suggest that NF-signaling regulates several hormonal biosynthesis and activation pathways, with subsequent negative feedback loops that are rapidly activated after a brief NF treatment or upon rhizobium inoculation. Further evidence for a temporal regulation of bioactive hormone pools by NF signaling comes from the downstream regulation of many hormonal signaling genes (**Table 1**).

### NF REGULATION OF HORMONE SIGNALING GENES

## CK Signaling

CK signaling utilizes a two-component signaling system comprised of a histidine-kinase (HK) receptor, phospho-transfer

proteins (AHPs), and response regulators (RRs). Two types of RRs operate in CK signal transduction: A-type RRs are transcriptionally regulated upon CK perception but do not contain a DNA binding domain, and are generally considered as negative regulators of the CK response. In contrast, B-type RRs contain a Myb DNA binding domain, function as transcription factors and are generally considered as positive regulators (Kieber and Schaller, 2014). In M. truncatula and L. japonicus, the HK genes MtCRE1 and LjLHK1 have positive roles in cortical cell divisions and nodule organogenesis (Gonzalez-Rizzo et al., 2006; Murray et al., 2007; Tirichine et al., 2007). Recent transcriptomic and promoter fusion analyses demonstrate the induction of these CK receptor genes upon rhizobium infection or NF application; in Lotus, this epidermal expression inhibits infection (Held et al., 2014; Jardinaud et al., 2016).

Independent data suggest NF induced, RR gene expression occurs downstream of CK receptor activation. For instance, the A-type MtRR4, MtRR8, MtRR9, MtRR11, and the B-type MtRR1 are expressed in M. truncatula roots following inoculation with either NF or wild-type S. meliloti, but not with nod-S. meliloti mutants (Gonzalez-Rizzo et al., 2006; Op den Camp et al., 2011; van Zeijl et al., 2015b; Jardinaud et al., 2016). Likewise, an HK and an RR (GmRR5) gene are also induced in soybean in an NF-dependent manner (Hayashi et al., 2012). Finally, the CK signaling reporter construct, TCSn:GUS, can be induced in M. truncatula RHs upon a 4 h NF treatment (Jardinaud et al., 2016).

Altogether, an early activation of CK signaling probably occurs both in RHs and cortical cells, directly downstream of the NF signaling pathway (**Figure 1**).

### Auxin Signaling

Accumulating evidence suggests that NFs regulate the expression of auxin signaling genes. NF application induces the auxin related transcription factors genes MtARF16a and MtPLETHORA3 in RHs, while several ARF genes are downregulated in RHs after a 24 h NF treatment (Breakspear et al., 2014; Jardinaud et al., 2016). These genes are also induced by auxin in M. truncatula roots (Herrbach et al., 2017). This suggests that NF perception could lead to auxin accumulation that would in turn activate some specific auxin signaling genes (**Figure 1**). Such genes could control cell divisions or IT formation (Breakspear et al., 2014).

### GA Signaling

DELLAs are well-known negative regulators of GA signaling (Sun, 2011). Recent data showed a positive effect of DELLAs on nodulation, and more specifically on IT formation (Fonouni-Farde et al., 2016; Jin et al., 2016). DELLAs reside in the NSP1/NSP2/IPD3 transcription factor complex that mediates NF signaling (Fonouni-Farde et al., 2016; Jin et al., 2016). External application of GAs can repress NF induction of several genes, such as ERN1, ENOD11, NSP2, and NIN (Maekawa et al., 2009; Fonouni-Farde et al., 2016). However, NFs do not induce DELLA gene expression (Fonouni-Farde et al., 2016).

### Ethylene and JA Signaling

Larrainzar et al. (2015) observed a biphasic regulation of expression of 47 AP2/ERF transcription factor genes in M. truncatula upon S. meliloti infection. The first wave of AP2/ERF expression takes place 1 hpi independent of NF perception, whereas the second wave of NF dependent, ethyleneregulated transcription occurs 6 hpi (Larrainzar et al., 2015). Although little is known about JA signaling during symbiosis, a 24 h NF treatment downregulates the expression of the putative JA receptor COI1 in M. truncatula RHs implying that NF perception may attenuate this signaling pathway (Breakspear et al., 2014).

### CONCLUSION AND PERSPECTIVES

Growing evidence shows that early events of NF signaling lead to the production and/or activation of hormones such as CK and GAs and the activation of their downstream signaling pathways. These hormones, in turn, interfere with other hormonal pathways, such as ethylene, ABA, and auxin, which can impinge upon NF signaling itself. For example, external application of CK or GA inhibits NF induction of MtENOD11 (Fonouni-Farde et al., 2016; Jardinaud et al., 2016). Similarly, JA, ABA, and ethylene are well-known negative regulators of NF signaling (Oldroyd et al., 2001; Sun et al., 2006; Ding et al., 2008). We observed that auxin both positively and negatively regulated different NF responsive genes in a combined auxin + NF treatment (Herrbach et al., 2017) providing further evidence for the regulation of symbiosis signaling pathways by hormones. These negative effects may underpin the mechanism by which the host controls nodulation following the initial NF/rhizobium recognition.

It should, however, be noted that the concentrations of hormones applied in these studies are often in the micromolar range. In the absence of quantitative data reporting endogenous hormone levels during symbiotic responses, it is difficult to assess if these studies reflect biologically relevant responses to plant hormones and this field requires such analyses.

Furthermore, recent evidence suggests that hormones function in a tissue-specific manner during symbiotic responses (see Gamas et al., 2017) but the specificity of these hormonal pathways is only partly resolved. Future work should address the respective contribution of different root tissues in producing, sensing, and responding to NF induced hormones, using tissue specific tools or quantitative hormone biosensors (Jones et al., 2014; Liao et al., 2015).

Finally, the NF signaling pathway overlaps with other symbiotic and developmental pathways suggesting that hormones may also impinge upon these processes. Similar to Rhizobia, arbuscular mycorrhizal fungi produce LCOs and utilize the same core signaling pathway to initiate symbiosis (Maillet et al., 2011). Future work should address whether LCO/ hormone crosstalk regulates mycorrhization, where evidence

already implicates a role for auxin, SGLs, GA, and LCO pathways (Lauressergues et al., 2012; Floss et al., 2013; Etemadi et al., 2014). Likewise, LCOs stimulate LR development and some NF induced genes are also involved in LR development (Herrbach et al., 2017). How LCO/hormone crosstalk could be involved in LCO stimulation of lateral root formation and how this could compare to its role in nodulation still requires further investigation.

### AUTHOR CONTRIBUTIONS

SB and WB drafted the review and produced **Figure 1**. SB wrote the review with help of WB. WB produced **Table 1**.

### REFERENCES


### FUNDING

Research conducted by WB and SB at the LIPM was partly supported by the "Laboratoire d'Excellence" (LABEX) entitled TULIP (ANR-10-LABX-41). WB's Ph.D. is funded by a joint French Embassy and Commission on Higher Education from the Philippines Government (CHED-Phil France) program.

### ACKNOWLEDGMENTS

We thank Clare Gough and Frédéric Debellé for critical reading of the manuscript. We apologize for colleagues whose work was not cited due to space limitation or our focus on crosstalk between NF and hormone signaling aspects only.



**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 Buhian and Bensmihen. 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.

# Local and Systemic Effect of Cytokinins on Soybean Nodulation and Regulation of Their Isopentenyl Transferase (IPT) Biosynthesis Genes Following Rhizobia Inoculation

Celine Mens†‡, Dongxue Li† , Laura E. Haaima, Peter M. Gresshoff and Brett J. Ferguson\*

Centre for Integrative Legume Research, School of Agriculture and Food Sciences, The University of Queensland, Brisbane, QLD, Australia

#### Edited by:

Stig U. Andersen, Aarhus University, Denmark

#### Reviewed by:

Katharina Markmann, Universität Tübingen, Germany Florian Frugier, Centre National de la Recherche Scientifique (CNRS), France

### \*Correspondence:

Brett J. Ferguson b.ferguson1@uq.edu.au †These authors have contributed equally to this work ‡orcid.org/0000-0001-6104-0442

#### Specialty section:

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

Received: 17 December 2017 Accepted: 18 July 2018 Published: 08 August 2018

#### Citation:

Mens C, Li D, Haaima LE, Gresshoff PM and Ferguson BJ (2018) Local and Systemic Effect of Cytokinins on Soybean Nodulation and Regulation of Their Isopentenyl Transferase (IPT) Biosynthesis Genes Following Rhizobia Inoculation. Front. Plant Sci. 9:1150. doi: 10.3389/fpls.2018.01150 Cytokinins are important regulators of cell proliferation and differentiation in plant development. Here, a role for this phytohormone group in soybean nodulation is shown through the exogenous application of cytokinins (6-benzylaminopurine, N<sup>6</sup> -(1<sup>2</sup> isopentenyl)-adenine and trans-zeatin) via either root drenching or a petiole feeding technique. Overall, nodule numbers were reduced by treatment with high cytokinin concentrations, but increased with lower concentrations. This was especially evident when feeding the solutions directly into the vasculature via petiole feeding. These findings highlight the importance of cytokinin in nodule development. To further investigate the role of cytokinin in controlling nodule numbers, the IPT gene family involved in cytokinin biosynthesis was characterized in soybean. Bioinformatic analyses identified 17 IPT genes in the soybean genome and homeologous duplicate gene partners were subsequently identified including GmIPT5 and GmIPT6, the orthologs of LjIPT3. Expression of GmIPT5 was upregulated in the shoot in response to nodulation, but this was independent of a functional copy of the autoregulation of nodulation (AON) receptor, GmNARK, which suggests it is unlikely to have a role in the negative feedback system called AON. Legumes also control nodule numbers in the presence of soil nitrogen through nitrate-dependent regulation of nodulation, a locally acting pathway in soybean. Upon nitrate treatment to the root, the tandem duplicates GmIPT3 and GmIPT15 were upregulated in expression indicating a role for these genes in the plant's response to soil nitrogen, potentially including the nitrate-dependent regulation of legume nodulation pathway. Additional roles for cytokinin and their IPT biosynthetic genes in nodulation and the control of nodule numbers are discussed.

Keywords: autoregulation of nodulation, cytokinin, legumes, nodulation, plant signaling and development, symbiosis, IPT, isopentenyltransferase

## INTRODUCTION

Legumes are economically important food, feed and fuel crops (Gresshoff et al., 2015). The majority are able to engage in a symbiotic relationship with nitrogen-fixing soil bacteria referred to as rhizobia, allowing legumes to grow under low soil nitrogen conditions. While rhizobia fix atmospheric nitrogen (N2) into ammonia that is readily assimilated by the plant, rhizobia in return receive photosynthetic carbohydrates in the protected nodule environment.

Reciprocal plant-microbe signaling is required to establish this relationship, which results in the formation of new root organs called nodules. Both nodulation and nitrogen fixation come at a high energy and resource cost to the plant. It is therefore important for the host to regulate its nodule numbers. This can be achieved via a nitrate-dependent regulation of nodulation mechanism in the presence of high soil nitrogen and through autoregulation of nodulation (AON) (Ferguson et al., 2010; Reid et al., 2011; Ferguson et al., 2018). The AON pathway is a negative feedback system involving extensive signaling between the root and shoot. Following rhizobia inoculation, plant-derived CLAVATA3/embryo-surrounding region (CLE) peptide signals are produced in the root (Reid et al., 2011; Hastwell et al., 2015a,b, 2018). These peptides are transported via the xylem to the shoot where they are perceived by a receptor complex that centers around a CLAVATA1 (CLV1)-like leucine-rich repeat receptor kinase [such as GmNARK in Glycine max (soybean), LjHAR1 in Lotus japonicus and MtSUNN in Medicago truncatula] (Okamoto et al., 2009, 2013; Mortier et al., 2010; Reid et al., 2011). Perception of the CLE peptides results in the differential regulation of a novel shoot-derived signal that travels back down to the roots to regulate nodule numbers (Ferguson et al., 2010; Ferguson et al., 2018). The shoot-derived signal and many other factors acting in the control of nodule numbers, are currently unknown or poorly understood.

Phytohormones are important signals involved in the formation and regulation of legume nodules (reviewed by Ferguson and Mathesius, 2014). Cytokinin hormones are a class of structurally similar N<sup>6</sup> -substituted adenine derivatives with a central role in plant growth and development. Since their discovery in cell proliferation and differentiation, several other functions have been attributed to this class of phytohormones, including maintenance of the shoot apical meristem, branching, organogenesis, delay of senescence, long-distance communication of nutritional status and the plant's response to biotrophic pathogens (reviewed by Kamada-Nobusada and Sakakibara, 2009). Like auxins, they can act both locally and as long-distance messengers. Natural cytokinins are classified into aromatic and isoprenoid cytokinins, the latter are most prevalent in higher plants and include trans-zeatin (tZ), cis-zeatin (cZ), dihydrozeatin (DZ) and N<sup>6</sup> -(1<sup>2</sup> -isopentenyl) adenine (2-iP). Cytokinins show further variation through the addition of side chains, which seem to confer receptor specificity. Kinetin and 6-benzylaminopurine (BAP) are synthetic cytokinins frequently used for research purposes.

In legumes, the exogenous application of cytokinins induces the formation of nodule-like structures through the induction of early nodulin genes and cortical cell divisions (Mathesius et al., 2000; Heckmann et al., 2011). Mutants defective in the cytokinin receptor histidine kinase, Cytokinin Response 1 (CRE1) in M. truncatula and its ortholog Lotus Histidine Kinase 1 (LHK1) in L. japonicus, lack the ability to effectively form nodule primordia, whereas gain-of-function mutants exhibit spontaneous nodule formation (Gonzalez-Rizzo et al., 2006; Murray et al., 2007; Tirichine et al., 2007; Plet et al., 2011). Moreover, Lonely Guy (LOG) genes encoding cytokinin riboside 5 0 -monophosphate phosphoribohydrolase, which are involved in the biological activation of cytokinins, were found to be expressed in the proliferating cells of the nodule primordium (Mortier et al., 2014). Upregulation of LOG genes resulted in less nodules and was independent of the leucine-rich repeat receptor kinase that acts in the shoot to perceive root-derived CLE peptide signals during AON (Mortier et al., 2014).

Isopentenyl transferases (IPTs) carry out the first and ratelimiting step in cytokinin biosynthesis, where an isopentenyl group is transferred to either AMP, ADP or ATP (Kakimoto, 2001). Recently, IPT3 was shown to be required for nodule development in the model legume L. japonicus (Chen et al., 2014; Sasaki et al., 2014; Reid et al., 2017). It has also been proposed that LjIPT3 could have a role in the synthesis of cytokinin molecules that act as the shoot-derived inhibitory factor in AON (Sasaki et al., 2014). LjIPT2 is thought to be responsible for the initial cytokinin build-up required for nodulation initiation, alongside LjLOG4 and independent of the LHK1 receptor (Reid et al., 2017). Interestingly, IPT encoding genes are also differentially expressed following treatment with certain nitrogen sources (Miyawaki et al., 2004), with nitrogen promoting plant development, but also inhibiting nodule organogenesis. However, it is important to remember that in addition to synthesis, posttranscriptional processing and degradation are also important to maintaining homeostasis of biologically active cytokinins.

Here, we report that exogenous application of low concentrations of cytokinin promotes nodule numbers in soybean, while high concentrations reduce them. This was achieved by treating soybean with either BAP, 2-iP or tZ type cytokinins using root-drench or petiole-feeding methods. These findings support the notion that cytokinin promotes nodule organogenesis, but suggest that the hormone is unlikely the shoot-derived factor of AON and that the inhibition we observe by feeding high concentrations is more likely the result of toxicity to the plant. In addition, 17 IPT genes were identified in the soybean genome and genetically characterized. Of these genes, the tandem duplicates GmIPT3 and GmIPT15 were upregulated in the root upon nitrate treatment, indicating a role for these genes in nitrate response. Upon rhizobia inoculation, GmIPT5 (a soybean ortholog of LjIPT3) was significantly upregulated in the shoot following rhizobia inoculation of the root, consistent with previous reports (Chen et al., 2014; Sasaki et al., 2014). Upregulation of GmIPT5 occurred in both the root and shoot in wild-type and GmNARK mutant plants (nts382), indicating that it functions in the leaf and root in response to nodulation, but likely does not have a role in the AON pathway.

### MATERIALS AND METHODS

### General Plant and Bacterial Growth Conditions

Soybean lines used include wild-type Bragg and the GmNARK mutant lines nts1116 (weak allele) and nts382 (strong allele), a hypernodulating and supernodulating line, respectively (Carroll et al., 1985). Seeds were surface-sterilized overnight with chlorine gas or in 70% ethanol/3.5% H2O<sup>2</sup> for 1 min followed by extensive rinsing with sterilized water.

Plants were grown in controlled glasshouse conditions with a 16 h/8 h photoperiod at 28 and 26◦C, respectively and watered as required with a modified nutrient solution (Broughton and Dilworth, 1971; Herridge, 1982). For the nitrate treatments, the nutrient solution was supplemented with 10 mM of KNO<sup>3</sup> in the week before harvest and applied every second day for 6 days prior to sample collection.

Plants were inoculated with Bradyrhizobium diazoefficiens CB1809/USDA110 (formerly Bradyrhizobium japonicum) or the incompatible Nod-factor mutant nodC−. The rhizobia were grown at 28◦C in a yeast-mannitol broth for 3 days and inoculated at an optical density at 600 nm of 0.1. Nodule numbers were determined 9 days after inoculation and subsequently root dry weights were recorded.

### Cytokinin Treatments

Six-week old plants were treated with water as a control or three different types of cytokinin (BAP, 2-iP and tZ) at concentrations of 0.1 or 100 µM. The hormones were dissolved in MilliQ water and the treatments were administered through either root drenching (Lorteau et al., 2001) or petiole feeding (Lin et al., 2010, 2011). For the root-drench assay, the hormone solutions (150 ml) were poured directly onto the vermiculite every second day. For the petiole-feeding assay, a feeding apparatus consisting of a syringe and silicon tubing was constructed that enables the cytokinin solution to be fed continuously into the shoot via a severed petiole. This method directly supplies the solutions into the plant, including the phloem, which mimics the source of the shoot-derived inhibitor. In both cases, plants were inoculated with B. diazoefficiens CB1809 1 day after the start of feeding. Plants were harvested after 10 days of treatment.

### Bioinformatic Analysis

To identify the IPT gene family in soybean, common bean (Phaseolus vulgaris), M. truncatula and L. japonicus, multiple BLASTP searches were conducted against various online databases including Phytozome, NCBI, DOBLAST and Lotus Base using the LjIPT3 and the nine previously identified AtIPT amino acid sequences as a query (Takei et al., 2001; Miyawaki et al., 2004). Functional domains were predicted via the Pfam (EMBL-EBI) and PANTHERTM v11.0 databases (Punta et al., 2012; Mi et al., 2017). Synteny between genomic environments was analyzed using Phytozome JBrowse. The obtained sequences were further characterized through a multiple sequence alignment via Clustal Omega hosted by EMBL-EBI (Goujon et al., 2010; Sievers et al., 2011; McWilliam et al., 2013; Corcilius et al., 2017). Phylogenetic trees were constructed using the PHYML plugin embedded in Geneious Pro v10.0.9 (Guindon and Gascuel, 2003) based on the protein sequence alignment adjusted by an algorithm in Geneious Pro v10.0.9 to retain only the homologous sequences and remove 75% of the gap regions. This plugin implements the maximum likelihood method to generate phylogenetic trees with 1,000 bootstraps supporting the branches.

### Gene Expression Analysis

For gene expression analyses, the first trifoliate leaf and the entire root from rhizobia-inoculated plants or root from 23 day-old nitrate-treated plants, were harvested, snap-frozen and homogenized in liquid nitrogen. The tissue was harvested 10 days after rhizobia inoculation of 2-week old plants. Total RNA was extracted the automated Maxwell <sup>R</sup> LEV simplyRNA Tissue kit (Promega) according to the manufacturer's protocol. The quality and quantity of the RNA samples were assessed using the NanoDropTM One Spectrophotometer (Thermo Fisher Scientific). SuperScript <sup>R</sup> III Reverse Transcriptase (Invitrogen) was used to generate cDNA from 500 ng of DNase-treated RNA. An initial PCR on cDNA was done to identify IPT genes expressed in the different conditions. An RT-qPCR analysis was performed using the Roche LightCycler <sup>R</sup> 96 with SYBR green fluorescence detection in a 96-well plate. All reactions were conducted in duplicate for at least two biological replicates and a target amplicon size of approximately 100 bp (**Supplementary Table S1**). The GmCons6 housekeeping gene was included to normalize gene expression levels (Libault et al., 2008).

### Statistical Analysis

A statistical analysis was performed on all results using the GraphPad Prism 7 software. A Student's t-test was used to determine the statistical significance of differences in nodule numbers assuming normal distribution of the data, root dry weights and relative expression of the target genes.

### RESULTS

### Effect of Cytokinin Treatment on Soybean Nodulation

To determine the effect of cytokinin on soybean nodulation, 0.1 or 100 µM of BAP, tZ or 2-IP or a water control, were applied to the hypernodulating GmNARK mutant, nts1116. Concentrations lower than 0.1 µM were not applied as these are too low to induce a response in soybean. The hypernodulating mutant was used as a possible effect on nodule numbers caused by the cytokinin treatments would be easier to observe and quantify on a hypernodulating nts1116 root system compared with the wild-type. The cytokinin treatments were administered using one of two methods; root-drenching or a petiole-feeding method to introduce the hormone to the shoot (Lin et al., 2010, 2011), after which the nodule number and root dry weight were determined.

Petiole feeding of 100 µM solutions with any of the three types of cytokinin significantly reduced nodule numbers compared with the water control (P < 0.001) (**Figure 1A**). In contrast, low 0.1 µM concentrations of both BAP and 2-IP significantly increase the number of nodules formed (P = 0.0004 and P = 0.030, respectively). A slight increase was also observed using 0.1 µM tZ, but this was not significant (P = 0.096). No significant differences in root dry weight were observed with any of the treatments used.

The root drenching results were similar, but generally less pronounced. Application of 100 µM BAP almost completely abolished nodulation (P < 0.0001), with 100 µM tZ also resulting

in a significant reduction in nodule numbers (P = 0.012) (**Figure 1B**). Treatment with 100 µM 2-IP resulted in a slight, but not significant reduction (P = 0.390). Drenching with 0.1 µM of 2-IP significantly increased nodule numbers (P = 0.019), similar to what was observed using petiole feeding, whereas 0.1 µM of BAP or tZ led to a slight, but not significant, change in nodulation.

### Identification and Sequence Characterization of the IPT Gene Family in Soybean

To determine the role of cytokinin biosynthesis during the development and control of soybean nodulation, IPT gene expression was investigated. First, the complete IPT gene family was identified in soybean via multiple BLAST searches using the LjIPT3 amino acid sequence as a query. LjIPT3 was previously reported to have a role in nodulation and/or AON (Chen et al., 2014; Sasaki et al., 2014; Reid et al., 2017). The BLAST results lead to the identification of 17 IPT gene members in soybean having high sequence identity to LjIPT3 (**Table 1** and **Supplementary Table S2**). All of the translated IPT gene products were predicted via the Pfam database to possess a tRNA delta(2)-isopentenyl pyrophosphate (IPP) transferase domain, which is critical for the enzyme's function in cytokinin biosynthesis, with the exception of GmIPT10, GmIPT15 and GmIPT17. However, additional searches using PANTHERTM v11.0 confirmed the classification of GmIPT10, GmIPT15 and GmIPT17 within the IPP family of proteins. To further confirm their classification within the soybean IPT gene family and provide information on their potential function, synteny between their genomic environments was analyzed (see below, **Figure 3**).

The BLAST searches were broadened to include the databases for common bean, L. japonicus, M. truncatula and the nonlegume Arabidopsis thaliana. This confirmed the presence of 23 putative IPT genes in M. truncatula, making it the largest gene family compared with the eight, six and nine members identified in common bean, L. japonicus and A. thaliana, respectively (Takei et al., 2001; Azarakhsh et al., 2018) (**Supplementary Tables S3**–**S6**).

To further characterize the sequences obtained here, a multiple sequence alignment and phylogenetic tree were produced using Clustal Omega and PHYML. The multiple sequence alignments revealed a high level of conservation of amino acid residues among the different members of both the legume and non-legume IPT gene families (**Supplementary Figures S1**, **S2**). This is most evident between the amino acid residues situated at the positions 102–213 in the alignment at the N-terminal end of the protein sequences where the functional IPP domain is located.

Seven homeologous (i.e., duplicate) gene pairs are present within the soybean genome grouping together in the phylogenetic tree (**Figure 2**). This is a common occurrence with soybean genes due to previous whole genome duplication events 59 and 13 million years ago (Schmutz et al., 2010).

The orthologs of LjIPT3 are the homeologous duplicates, GmIPT5 and GmIPT6, with 75.1 and 72.2% amino acid sequence identity, respectively, while sharing 83.1% identity. Other soybean IPT family members clustering closely to LjIPT3, are the homeologous duplicates GmIPT7 and GmIPT8 with 50.8 and 49.8% amino acid identity. These four soybean genes are present in duplicated regions in the genome showing their close relationship. Homologous versions of LjIPT3 in common bean, M. truncatula and A. thaliana have the identifiers Phvul.007G170100, Medtr1g072540 (MtIPT3) and AT3G63110 (AtIPT3) with a protein sequence identity of 76, 74 and 53%, respectively.

Six IPT genes group into two groups of three genes (GmIPT3/GmIPT4/GmIPT15 and GmIPT11/GmIPT12/GmIPT17) (**Figure 2**). The first group clusters closely to LjIPT2 that was recently shown to be responsible for the initial cytokinin burst required for nodulation initiation (Reid et al., 2017). The genomic regions of GmIPT3 and GmIPT4 display a high level of synteny with five identical genes surrounding the two IPT genes (**Figure 3A**). This indicates

that GmIPT3 and GmIPT4 are true homeologous duplicates. The third gene, GmIPT15, shares a protein sequence identity of 92.9% with GmIPT3. In addition, the two genes are present on the same chromosome (chromosome 17) and lack duplications of their genomic regions, indicating a recent tandem duplication of GmIPT3 resulting in GmIPT15. Interestingly, the predicted GmIPT15 protein is truncated to only 75 amino acid residues and does not feature the highly conserved IPP amino acid sequence motif at the N-terminal end that is required for function. In addition to having a gap upstream of the predicted coding sequence that puts it out of frame, the currently available genome includes a 100 bp stretch of unconfirmed nucleic acid identity (denoted in the sequence as: -NNN-). Similar to GmIPT15, the GmIPT10 product is truncated with a stretch of nucleic acids of unconfirmed identity. This lack of identified nucleic acid residues in GmIPT15 and GmIPT10 may be due to the high level of sequence identity amongst the IPT genes and/or the presence of repetitive sequences within a gene, making the identity of some nucleic acid residues difficult to confirm with absolute certainty.

The second group (GmIPT11/GmIPT12/GmIPT17) is closely related to LjIPT5 as well as Phvul.011G091500 and MtIPT5 in common bean and M. truncatula. The role of LjIPT5 was previously investigated in nodule development and AON by Sasaki et al. (2014) and unlike LjIPT3, no differences in transcript levels were observed in the shoot after inoculation with compatible rhizobia. These orthologs are all characterized by the presence of a high number of introns (6–9 introns), except for GmIPT17, which has only one predicted intron (**Table 1**). A multiple sequence alignment of their genomic sequences using Clustal Omega indicated little conservation. GmIPT11 and GmIPT12 share 69.9% sequence identity and are located in a highly similar genomic region highlighted by the presence of six conserved genes (**Figure 3B**). The GmIPT12 protein is missing part of the N-terminal conserved region. Wrongly predicted intron locations might put the sequence out of frame leading to a truncated protein sequence. The genomic sequence of GmIPT17 is much less conserved compared with GmIPT11 and GmIPT12 showing no conservation of the genomic region. GmIPT17 likely does not have a homeologous duplicate. In addition, GmIPT17 is predicted to be a truncated protein with only the central region translated. Like GmIPT17, GmIPT16 is another gene that appears to completely lack a duplicate partner, indicating that it was lost or that GmIPT16 arose after the most recent whole genome duplication event. A high level of genetic synteny is present for GmIPT16 and orthologs in other legumes species including MtIPT9 and Phvul.005G067800 in contrast to GmIPT17 (**Figure 3C**). Therefore, its homeologous duplicate was most likely lost over time.

### IPT Gene Expression in the Shoot and Root Following Rhizobia Inoculation

Previously, cytokinin was suggested to be the unidentified shoot-derived inhibitor in AON as demonstrated by the upregulation of LjIPT3 in wild-type shoots, but not in LjHAR1 defective mutants (Sasaki et al., 2014). To identify IPT genes in soybean that are differentially regulated following rhizobia inoculation of the root, the expression of all 17 IPT genes was examined in both the root and shoot (i.e., trifoliate leaves) (**Supplementary Figure S3**). The first trifoliate leaf was tested as the GmNARK receptor is highly expressed in mature trifoliate leaves (Nontachaiyapoom et al., 2007). Tissues were harvested from wild-type Bragg and its supernodulating GmNARK mutant (nts382) inoculated with either compatible B. diazoefficiens USDA110 or its incompatible nodC<sup>−</sup> mutant strain as a negative control. GmNARK function


tRNA delta(2)-isopentenyl pyrophosphate transferase domain (IPP) predicted via Pfam.

is completely blocked in the nts382 background, enabling characterization of possible IPT candidates involved in the AON pathway.

The homeologous duplicates GmIPT1 and GmIPT2 respond to inoculation in the root, but not the shoot (**Figure 4**). This change in expression is significant for GmIPT2 in both WT Bragg and GmNARK mutant (nts382) roots (P = 0.002 and P = 0.021, respectively); and is consistent with LjIPT1 expression, the ortholog of GmIPT1 and GmIPT2, which is upregulated in the root and not in the shoot (Chen et al., 2014; Sasaki et al., 2014).

Expression of the second homeologous pair, GmIPT3 and GmIPT4, was root-specific and GmNARK-independent with transcript levels of GmIPT4 increasing following inoculation (P = 0.161 in WT and P = 0.044 in nts382) (**Figure 4**). These genes cluster with GmIPT15, a duplicate of GmIPT3 and possibly a pseudogene, which does not show a clear trend in expression in response to the presence of rhizobia. The ortholog of these genes is LjIPT2; it is expressed in roots in conjunction with LjLOG4 and is reported to be required for the initial cytokinin burst in early nodulation (Reid et al., 2017).

One of the orthologs of LjIPT3, GmIPT5, was found to be significantly induced in the wild-type shoot upon inoculation with compatible rhizobia (P = 0.014) (**Figure 4**). This trend of elevated GmIPT5 expression was observed in the root as well (P = 0.165) and similarly observed in the GmNARK mutant root and shoot; however, these increases in expression were not significant (P = 0.060 and 0.244, respectively). As a trend of upregulation was noticeable, a second set of GmNARK mutant shoot samples was analyzed to determine whether GmIPT5 expression is GmNARK-dependent or not (**Figure 5**). This established a significant increase in GmIPT5 transcript levels in a GmNARK-independent manner (P = 0.0003). Expression of GmIPT6, the homeologous duplicate of GmIPT5, exhibited a similar pattern in response to inoculation with compatible rhizobia, with the exception of GmNARK mutant shoots, but

the transcripts levels of GmIPT6 were much lower compared to GmIPT5 (P = 0.150 and P = 0.143 in shoot and root, respectively) and hence caution should be taken when assessing its expression here (**Supplementary Figure S3**). This difference in expression levels is often observed for soybean homeologous with one of the duplicates often becoming reduced or even silenced over long periods of time (Lynch and Conery, 2000; Granger et al., 2002). Likewise, a decrease in GmIPT14 mRNA levels in the GmNARK mutant shoot was observed, but again the transcript levels were very low for this gene and may not be biologically relevant (**Figure 4** and **Supplementary Figure S3**). The homeologous partner of GmIPT14, GmIPT13, was omitted as its expression could not be detected.

GmIPT7 and GmIPT8 cluster closely to GmIPT5, GmIPT6 and LjIPT3 (**Figure 2**). They are shoot-specific in expression, but unlike GmIPT5, their expression remained constant after inoculation with compatible rhizobia in both wild-type and GmNARK mutant plants (**Figure 4**). GmIPT9 and GmIPT10 are root-specific in expression, but also were not differentially expressed by inoculation with the different rhizobia strains.

The remaining genes (GmIPT11, GmIPT12 and GmIPT16) do not show a noticeable trend in shoot or root gene expression after inoculation with B. diazoefficiens. GmIPT17 was predicted to be truncated and its gene expression could not be detected. This gene is likely a pseudogene and is therefore not shown in **Figure 4**.

### IPT Gene Expression in the Root Following Nitrate Treatment

Nitrate-dependent regulation of nodulation acts locally in soybean as demonstrated by overexpression experiments using GmNIC1a where nodulation was only suppressed in transgenic, but not in non-transgenic roots on the same plant (Reid et al., 2011). Therefore, the expression of all 17 IPT genes of soybean was assessed in the root following nitrate-treatment to identify members of the gene family involved in nitrogen response and possibly nitrate-dependent regulation of nodulation.

Nitrate treatment induced the upregulation of the duplicates GmIPT3 (P = 0.027) and GmIPT15 (P = 0.021) (**Figure 6**). The transcript level of the third copy clustering with these two genes, GmIPT4, also increased, but not significantly (P = 0.095). GmIPT3 is expressed at levels twice that of both GmIPT15 and GmIPT4, which showed a trend of upregulation in the rhizobia inoculated roots. Other rhizobia-responsive genes, e.g., GmIPT2 and GmIPT5, as well as all remaining IPT genes, were not differentially regulated in the root in response to nitrate (**Supplementary Figure S4**).

### DISCUSSION

Cytokinins are important regulators of plant growth and development, fine-tuning the balance between cell proliferation and differentiation. They are essential for rhizobia infection and nodule development through the initiation of cortical cell divisions and the induction of early nodulation transcription factors, such as NSP2 (Nodulation Signaling Pathway 2), NIN (Nodule Inception) and ERN1 (Ethylene-Responsive binding domain factor required for Nodulation 1), acting in an LHK1 dependent manner (Murray et al., 2007; Tirichine et al., 2007; Plet et al., 2011; van Zeijl et al., 2015). Indeed, a lack of cytokinin perception in the early stages of nodulation leads to infection thread formation without subsequent nodule organogenesis (Murray et al., 2007; Held et al., 2014). Here, we set out to further characterize the role of cytokinin and the IPT gene family in the AON and nitrate-regulation control pathways.

Exogenous application of cytokinin promoted nodule development at low concentrations, while high concentrations reduced nodulation. The former result supports a promoting effect of the hormone on nodulation, while the later finding is likely the result of toxicity induced by an excess level of the hormone (e.g., elevated ethylene production; Lorteau et al., 2001; Ferguson et al., 2005). This effect is particularly evident when BAP is directly applied to the root. BAP is a synthetic cytokinin that is very active and more stable in aqueous solution and once taken up by the plant than other cytokinin compounds (Deleuze et al., 1972). Despite the negative effect on plant growth and development, high doses of phytohormones are commonly used in current research and care should be taken in the selection of appropriate physiologically relevant concentrations in relation to the species and growing conditions used. The promoting effect was most evident when using petiole feeding to deliver the hormone directly and continuously into the plant mimicking the action of the shoot-derived inhibitor in AON, compared with root drenching, which is influenced by plant uptake, metabolism and the physical properties of the growing substrate. Exogenous cytokinin supply has previously been shown to induce nodulin genes, amyloplast depositions and the formation of nodule primordia (Bauer et al., 1996; Heckmann et al., 2011). Furthermore, overexpression of cytokinin biosynthesis genes and subsequent secretion by nodulation-defective rhizobia was reported to initiate organogenesis in alfalfa

(Cooper and Long, 1994). Sasaki et al. (2014) also report an inhibitory effect on nodulation following the application of cytokinin (BAP) to the roots of L. japonicus and did not report a promotion of nodulation when using lower concentrations of cytokinin. However, these findings might not be contradictory and may instead reflect differences in cytokinin type and effective concentration range between species. Differences in age at harvest, growing conditions or another experimental parameter may also influence responses.

Cytokinin biosynthesis is catalyzed by IPT enzymes and an initial cytokinin burst is required to trigger nodule organogenesis (reviewed in Ferguson and Mathesius, 2014). Reid et al. (2017) recently demonstrated that LjIPT2, together with LjLOG4, is responsible for the build-up of cytokinins, independent to and upstream of the LHK1 cytokinin receptor in L. japonicus. In addition, LjIPT3 is induced in the root after inoculation with compatible rhizobia (Chen et al., 2014; Sasaki et al., 2014). Sasaki et al. (2014) reported LjIPT3 expression was also induced in the shoot and speculated that cytokinin might be the elusive shootderived inhibiting signal, acting downstream of the leucine-rich repeat receptor kinase LjHAR1 in AON.

In this study, 17 IPT genes were identified in the soybean genome. All members were found to have a homeologous duplicate or to be a duplicate of one of the homeologous copies, with the exception of GmIPT16 and GmIPT17, which have no discernible partner. The comprehensive phylogenetic analysis reported here provides insight into the potential functions of these genes. The 17 family members of soybean are considerably more than the 8 and 6 members identified in common bean and L. japonicus. Having approximately double the number of gene family members of common bean is typical for soybean due to a whole genome duplication event that occurred roughly 13 million years ago (Schmutz et al., 2010; Hastwell et al., 2015a, 2017).

The orthologs of the root-induced LjIPT2 (Reid et al., 2017) were identified as GmIPT3 and GmIPT4, with GmIPT15 identified as a duplicate of GmIPT3. The orthologs of the rootand shoot-induced LjIPT3 (Chen et al., 2014; Sasaki et al., 2014) were found to be GmIPT5 and its duplicate GmIPT6.

IPT gene expression was analyzed in both soybean shoots and roots of rhizobia-inoculated plants to identify potential candidates acting in nodulation and AON (Ferguson et al., 2018). GmIPT5 was significantly induced in wild-type trifoliate leaves following inoculation of the root with compatible rhizobia. This is consistent with findings of Sasaki et al. (2014). Despite not being significant, the trend of increased GmIPT5 transcript levels is also noticeable in the GmNARK mutant (nts382) in both the root and shoot. GmNARK defective plants that were either inoculated or non-inoculated confirmed a significant difference in GmIPT5 transcript abundance. This seems to suggest that the upregulation of gene expression is independent of the GmNARK receptor and is unlikely to have a role in the AON pathway in soybean as suggested for L. japonicus by Sasaki et al. (2014) who found that regulation of LjIPT3 is LjHAR1 dependent in the shoot. This might be due to any number of reasons including species-specific differences, different plant development stages, differences in growing conditions, etc. Expression of the homeologue of GmIPT5, GmIPT6, was considerably lower, indicating that it might not be biologically relevant in this process with GmIPT5 being the dominant copy.

Upregulation of LjIPT3 in the root has been reported to occur as early as 3 h after rhizobial inoculation (Chen et al., 2014) and as late as 10 days after inoculation (Reid et al., 2017). The latter finding is long after the CLE peptides, LjCLE-RS1 and LjCLE-RS2, are induced to initiate the AON process (Okamoto et al., 2009). Furthermore, the differential regulation of nodulation-suppressing CLE peptides in L. japonicus and M. truncatula is dependent on the cytokinin-induced action of NIN, a transcription factor required for infection thread and nodule primordia formation (Schauser et al., 1999; Mortier et al., 2012; Soyano et al., 2014). Cytokinins and their IPT biosynthesis genes therefore seem to play a major role in the nodule development process upstream of AON rather than being a key factor in AON itself. Increased GmIPT5 expression in the shoot after rhizobial inoculation might be the plant's way to enhance shoot growth in preparation of an influx in nitrogen via nitrogen fixation, rather than cytokinins acting as the shoot-derived inhibitor in AON (Rahayu et al., 2005). In support of this is work by Chen et al. (2014) who show that RNA interference of LjIPT3 results in reduced shoot development and increased chlorophyll breakdown resulting in subsequent leaf senescence, even in the absence of rhizobia.

AtIPT3-synthesised cytokinins induce the expression of shoottype nitrate transporters (AtNRT) to translocate and partition nitrate in the shoot in the non-legume A. thaliana when plenty of nitrogen is available in the soil (Kiba et al., 2011).

Another gene, GmIPT2, was found to respond to rhizobial inoculation in the root in a clear GmNARK-independent manner. Its homeologous duplicate, GmIPT1, is root-specific as well and shows the same induction albeit less strong. This is consistent with the increased expression in the root but not the shoot of the ortholog of these genes in L. japonicus, LjIPT1, following rhizobia inoculation. The precise function of this gene remains unclear, but expression of LjIPT1 is highest in the flower and only moderate in the roots (Chen et al., 2014).

The fourth gene responding to compatible rhizobia inoculation is the root-specific GmIPT4, the homeologue of GmIPT3 that shows no clear change in expression. These genes are closely related to LjIPT2, which is required for the initial cytokinin burst during early nodulation events that are induced by rhizobia inoculation (Reid et al., 2017). Interestingly, expression of GmIPT3 and its tandem duplicate GmIPT15 (which is predicted to be truncated), is promoted in the root following nitrate treatment, but GmIPT4 expression is not induced as strongly. Furthermore, transcript levels of GmIPT4 are much higher than those of GmIPT3 following rhizobia inoculation, while the opposite was observed in response to high soil nitrogen status. This might indicate that these genes have undergone genetic divergence (i.e., neofunctionalization) in which they developed a different function, as is sometimes observed with duplicated genes of soybean (Mirzaei et al., 2017).

In addition to the negative feedback system of AON, nodulation is controlled by environmental factors including the soil nitrogen status. In soybean, nitrate-dependent regulation of nodulation inhibits the nodulation process via a local pathway in contrast to systemic signaling required for AON (Hinson, 1975; Reid et al., 2011; reviewed in Ferguson et al., 2018). Therefore, only roots were tested in this study to investigate the role of cytokinins as the root-derived inhibitor in nitrate-dependent regulation of nodulation. In addition, AtIPT accumulation is highest in the root tissue in response to nitrogen, while expression in the shoot is relatively low (Takei et al., 2004). Systemic components of the nitrate-regulation of nodulation (e.g., nitrogen fixation, nodule size and nodule number) have been shown in other model legumes such as M. truncatula and L. japonicus (Jeudy et al., 2010; Soyano et al., 2014; Nishida et al., 2018; reviewed in Ferguson et al., 2018). Therefore, analysis of IPT gene expression in the shoot in response to soil nitrogen status could be investigated in future. The plant's response to low nitrogen in the soil requires root phenotypic plasticity. This can occur through association with symbiotic soil rhizobia or through the proliferation of the lateral root system via a process called foraging. Similar to the establishment of nodule primordia, a precise cytokinin balance is essential for lateral root initiation and development. This process of foraging requires a reduction in cytokinin levels to increase lateral root proliferation. Overexpression of IPT genes in tobacco reduced root growth, while lateral root numbers are reduced in transgenic tobacco plants overexpressing a cytokinin oxidase (Hewelt et al., 1994; Werner et al., 2001). Here, the transcript levels of the cytokinin biosynthesis genes GmIPT3 and GmIPT15 are upregulated in the root in response to nitrogen. This is consistent with findings in M. truncatula showing the cytokinin receptor, MtCRE1, is required for both nodule and lateral root development, indicating mechanisms for nodule organogenesis have diverged from existing molecular mechanisms (Gonzalez-Rizzo et al., 2006). In addition, cytokinins are frequently found to act in an opposite manner during lateral root development and nodulation, requiring low or high cytokinin concentrations, respectively (Lohar et al., 2004). In A. thaliana, AtIPT3 and AtIPT5, the homologous of GmIPT3, GmIPT4 and GmIPT15, are upregulated in response to nitrate (Miyawaki et al., 2004; Takei et al., 2004). While AtIPT3 is important in the short-term response to nitrate, AtIPT5 is expressed after long-term exposure to different sources of nitrogen (Takei et al., 2004). Once plants encounter a high nitrogen patch, systemic nitrogen signaling is required to induce lateral root proliferation into the patch and inhibit lateral root development elsewhere. Several studies suggest a role in systemic nitrogen signaling for root-produced cytokinins (tZ) that move up the shoot where they regulate systemic nitrogen-pathways in response to a heterogeneous supply of soil nitrogen (Rahayu et al., 2005; Poitout et al., 2018). In addition, cytokinins in the root reduce nitrate uptake by inhibiting expression of high-affinity nitrate transporters (e.g., NRT2.1) when nitrogen availability is high, while shoot-type transporters are upregulated to translocate and redistribute nitrogen (Kiba et al., 2011). The results obtained here, together with the findings of Reid et al. (2017), suggest a general role for GmIPT3/GmIPT15 and their ortholog LjIPT2 in response to nitrogen status, possibly indicating that IPT-synthesized cytokinins act as regulators of root architecture and nitratedependent regulation of nodulation.

Collectively, our results established that low levels of cytokinins can promote nodulation whereas higher levels inhibit it. We also identified IPT genes that are regulated in either the root or shoot in response to inoculation with compatible rhizobia and additional IPT genes that are regulated in the root in response to nitrate. Our findings do not suggest that cytokinin is likely to be the shoot-derived inhibitor in AON. This is based on the upregulation of GmIPT5, an ortholog of LjIPT3, in a GmNARK-independent manner in the shoot combined with the nodulation promoting effect of low concentrations of cytokinin applied to the root and shoot. Indeed, using the petiole-feeding technique, the hormone was fed directly into the plant where SDI is produced and therefore would not be expected to promote nodule development. It should be noted that differences in species, treatment type, concentration range and cytokinin type might be important factors for differences in results reported here and in previous studies. While instead, IPT expression in the shoot possibly prepares the plant for new growth while awaiting a burst of nitrogen, which is often a limiting factor for plant growth. The comprehensive bioinformatic

and expression analysis reported here can be used as tools to support future research into the role of cytokinin in nodule development. This could include establishing the exact function of various IPT genes in nodulation, including the role of increased GmIPT5 expression in both shoot and root, in conjunction with the LOG genes that are crucial for cytokinin activation.

### AUTHOR CONTRIBUTIONS

BF and PG devised the project and supervised. DL, CM and LH conducted the experiments. DL and CM performed bioinformatic and gene expression analyses. DL and LH performed petiole feeding and root drenching assays. CM analyzed the data, prepared the figures and wrote the manuscript with input from all authors. All authors read and approved the manuscript.

### REFERENCES


### FUNDING

This work was funded by The Hermon Slade Foundation and the Australian Research Council Discovery Project grants (DP130103084 and DP130102266).

### ACKNOWLEDGMENTS

The authors would like to thank Huanan Su, Mengbai Zhang, April Hastwell, Xitong Chu and Candice Jones for their technical assistance during this study.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01150/ full#supplementary-material



Tirichine, L., Sandal, N., Madsen, L. H., Radutoiu, S., Albrektsen, A. S., Sato, S., et al. (2007). A gain-of-function mutation in a cytokinin receptor triggers spontaneous root nodule organogenesis. Science 315, 104–107. doi: 10.1126/ science.1132397

van Zeijl, A., Op den Camp, R. H. M., Deinum, E. E., Charnikhova, T., Franssen, H., Op den Camp, H. J. M., et al. (2015). Rhizobium lipo-chitooligosaccharide signaling triggers accumulation of cytokinins in Medicago truncatula roots. Mol. Plant 8, 1213–1226. doi: 10.1016/j.molp.2015.03.010

Werner, T., Motyka, V., Strnad, M., and Schmülling, T. (2001). Regulation of plant growth by cytokinin. Proc. Natl. Acad. Sci. U.S.A. 98, 10487–10492. doi: 10.1073/pnas.171304098

**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 Mens, Li, Haaima, Gresshoff and Ferguson. 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.

# Acropetal Auxin Transport Inhibition Is Involved in Indeterminate But Not Determinate Nodule Formation

### Jason L. P. Ng\* and Ulrike Mathesius

Division of Plant Sciences, Research School of Biology, Australian National University, Canberra, ACT, Australia

Legumes enter into a symbiotic relationship with nitrogen-fixing rhizobia, leading to nodule development. Two main types of nodules have been widely studied, indeterminate and determinate, which differ in the location of the first cell division in the root cortex, and persistency of the nodule meristem. Here, we compared the control of auxin transport, content, and response during the early stages of indeterminate and determinate nodule development in the model legumes Medicago truncatula and Lotus japonicus, respectively, to investigate whether differences in auxin transport control could explain the differences in the location of cortical cell divisions. While auxin responses were activated in dividing cortical cells during nodulation of both nodule types, auxin (indole-3-acetic acid) content at the nodule initiation site was transiently increased in M. truncatula, but transiently reduced in L. japonicus. Root acropetal auxin transport was reduced in M. truncatula at the very start of nodule initiation, in contrast to a prolonged increase in acropetal auxin transport in L. japonicus. The auxin transport inhibitors 2,3,5-triiodobenzoic acid and 1-N-naphthylphthalamic acid (NPA) only induced pseudonodules in legume species forming indeterminate nodules, but failed to elicit such structures in a range of species forming determinate nodules. The development of these pseudonodules in M. truncatula exhibited increased auxin responses in a small primordium formed from the pericycle, endodermis, and inner cortex, similar to rhizobia-induced nodule primordia. In contrast, a diffuse cortical auxin response and no associated cortical cell divisions were found in L. japonicus. Collectively, we hypothesize that a step of acropetal auxin transport inhibition is unique to the process of indeterminate nodule development, leading to auxin responses in pericycle, endodermis, and inner cortex cells, while increased auxin responses in outer cortex cells likely require a different mechanism during the formation of determinate nodules.

Keywords: auxin, indeterminate, determinate, auxin transport inhibitor, acropetal, legume, nodule

### INTRODUCTION

Many legume species form a symbiosis with nitrogen-fixing rhizobia, resulting in the development of nodules in the host roots. There are variations in the organogenesis of nodules in different species of legumes. Two main nodule types have been reported in the literature, indeterminate and determinate nodules. The model legume Medicago truncatula forms indeterminate nodules, while another model legume, Lotus japonicus, forms determinate nodules. Whereas indeterminate

#### Edited by:

Pascal Ratet, Centre national de la recherche scientifique (CNRS), France

#### Reviewed by:

Sen Subramanian, South Dakota State University, United States Sandra Bensmihen, UMR2594 Laboratoire des Interactions Plantes-Microorganismes (LIPM), France

> \*Correspondence: Jason L. P. Ng jason.ng@anu.edu.au

#### Specialty section:

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

Received: 02 November 2017 Accepted: 30 January 2018 Published: 15 February 2018

#### Citation:

Ng JLP and Mathesius U (2018) Acropetal Auxin Transport Inhibition Is Involved in Indeterminate But Not Determinate Nodule Formation. Front. Plant Sci. 9:169. doi: 10.3389/fpls.2018.00169

nodules involve re-initiation of cell divisions in the pericycle, endodermis, and inner cortex of the root, determinate nodules are characterized by cell divisions mainly in the middle and outer cortex, with some contributions from the pericycle and endodermis (Hirsch, 1992). Another difference between the two nodule types is that indeterminate nodules are typically elongate and maintain an apical nodule meristem; determinate nodules lose the meristematic activity of the nodule meristem and are typically round (Hirsch, 1992). So far it is not known what determines the difference between these two organogenesis programs.

Previous studies have shown that the plant hormone auxin is essential for the initiation of cell divisions in plants and that auxin gradients accompany the formation of new plant organs (Benková et al., 2003). This also seems to be the case for nodule initiation, where increased auxin responses have been localized to nodule primordia in various legume species forming both indeterminate or determinate nodules (van Noorden et al., 2007; Takanashi et al., 2011; Suzaki et al., 2012; Turner et al., 2013).

Auxin represents a group of structurally related phytohormones. The most frequently studied auxin is indole-3-acetic acid (IAA). Auxins can be divided into active and inactive/storage forms, with the latter being the more abundant form in plants (Korasick et al., 2013). Active auxins, such as IAA, activate auxin-response genes, whose gene products regulate cell division and organ growth. Other active auxins include indole-3-butyric acid (IBA), 4-chloro-indole-3-acetic acid (4-Cl-IAA), and phenylacetic acid (PAA). Each of these active auxins can be inactivated by forming conjugates with amino acids (Korasick et al., 2013). For example, IAA-Alanine and IAA-Leucine have been suggested to be IAA storage forms in Arabidopsis thaliana (Kowalczyk and Sandberg, 2001; Novák et al., 2012), while IAA-Asp might lead to auxin degradation and inactivation (Korasick et al., 2013). IAA-Tryptophan, interestingly, was postulated to be an auxin antagonist, inhibiting IAA action (Staswick, 2009). The interconversion between free and conjugated IAA provides one mechanism to fine tune plant development through the spatio-temporal control of active IAA concentrations.

Auxin is transported within plants via two mechanisms – a passive, long distance auxin transport system through the phloem and an active, local cell-to-cell auxin transport machinery. Active auxin transport control plays an important role during root development. It is controlled by a suite of auxin transport carriers and cellular regulators that control their intracellular abundance and localization (Overvoorde et al., 2010). In particular, members of the auxin exporter family PIN (PINFORMED) and auxin importer family AUX1/LAX (AUXIN RESISTANT1/LIKE AUX1) have been shown to contribute to the formation of auxin gradients in Arabidopsis (Overvoorde et al., 2010). In M. truncatula, expression of the PIN auxin efflux carriers is altered during symbiotic interactions with Sinorhizobium meliloti. The expression of PIN2, 4, and 10 was upregulated within 24 h in response to S. meliloti inoculation or Nod factor treatment (Plet et al., 2011; Ng et al., 2015). Knockdown of PIN2, 3, and 4 reduced nodule numbers on transgenic M. truncatula roots (Huo et al., 2006). In situ hybridization of M. truncatula LAX1 (homolog of the Arabidopsis auxin influx carrier AUX1) mRNAs suggested that these auxin influx carriers are involved in early nodule primordia development and vasculature differentiation (de Billy et al., 2001). Such evidence strongly supports the role of auxin transport during nodule development.

Computer modeling suggested that during the early stages of M. truncatula–S. meliloti symbiosis, a temporary decrease in acropetal auxin efflux is the most plausible mechanism to explain the observed patterns of auxin accumulation in the dividing cells that comprise a nodule primordium, although auxin import is also likely to contribute (Deinum et al., 2012; Roy et al., 2017). When changes in acropetal auxin transport through the stele were combined with a hypothesized diffusible signal from the epidermis – mimicking a signal from infecting rhizobia – auxin accumulation occurred in cells where auxin responses have been localized in M. truncatula, i.e., pericycle, endodermis, and inner cortex (Deinum et al., 2016). Interestingly, a temporary decrease in auxin efflux has so far only been documented for indeterminate nodules, such as those forming on M. truncatula, white clover, and vetch (Mathesius et al., 1998; Boot et al., 1999; van Noorden et al., 2006). This auxin transport inhibition was absent in the nodulation defective cre1 (cytokinin response 1) mutant of M. truncatula that has a mutation in a cytokinin receptor (Ng et al., 2015). In this mutant, auxin transport inhibitors (ATIs) could restore both auxin transport inhibition as well as nodule initiation and auxin responses, suggesting that auxin transport inhibition in M. truncatula is required for correct auxin localization in the pericycle, endodermis, and inner cortex, which then leads to their divisions (Ng et al., 2015). Pacios-Bras et al. (2003) reported a temporary increase in auxin transport in response to Nod factor treatment in L. japonicus, although this was not statistically analyzed. Based on findings that lack of flavonoids in M. truncatula abolished auxin transport control and nodulation by rhizobia (Wasson et al., 2006), but lack of auxin-transport reducing isoflavonoids in soybean could be compensated for by addition of nod-gene inducers to infecting Bradyrhizobia (Subramanian et al., 2006), it was suggested that auxin transport control may be specific to indeterminate nodulation (Subramanian et al., 2007). However, no detailed auxin transport measurements have been published in other determinate nodule-forming species. It therefore remains unclear whether auxin transport inhibition is likely a mechanism leading to auxin accumulation in inner cortical cells in indeterminate nodule-forming species, but not in determinate nodule-forming species where auxin accumulation is localized in the middle/outer cortex.

Experiments using synthetic ATIs, such as N-1 naphthylphthalamic acid (NPA) and 2,3,5-triiodobenzoic acid (TIBA), have been used to uncover the roles of polar auxin transport in plant development, including lateral root organogenesis, inflorescence growth, and meristem maintenance (Casimiro et al., 2001; Scanlon, 2003; Wu and McSteen, 2007). Intriguingly, the application of ATIs on the roots of indeterminate nodule-forming legumes, such as M. truncatula, alfalfa, pea, and white sweet clover, can induce the formation of nodule-like structures, broadly termed pseudonodules (Hirsch et al., 1989; Scheres et al., 1992;

each stage. Black arrows indicate pericycle, endodermal, and cortical cell divisions. Dotted black arrow in H indicates a possible cortical cell division. Scale bars represent 200 µm.

Wu et al., 1996; Rightmyer and Long, 2011). Pseudonodules have a globular external structure resembling nodules, but contain undifferentiated cortical, endodermal, and pericycle cells. In addition, they fail to develop the characteristic peripheral vasculature (Guan et al., 2013) and do not house rhizobia. The detailed structure of early developmental stages of pseudonodules is not well described in the literature. On the contrary, ATIs failed to form pseudonodules on the roots of L. japonicus (Kawaguchi et al., 1996; Takanashi et al., 2011) and have only been reported to induce nodule-like structures in one species forming determinate nodules, Macroptilium atropurpureum (siratro), albeit with no description of the structure (Relic et al., 1993 ´ ). This suggests that auxin transport inhibition is a mechanism inducing nodule structures by localizing auxin in the inner cortical region in indeterminate- but likely not determinate-nodule forming species (Kohlen et al., 2017).

Here, we aimed to compare changes in auxin transport and localization during indeterminate and determinate nodule formation. We compared acropetal (toward the root tip) auxin transport in corresponding root segments of M. truncatula and L. japonicus seedlings. We corroborated these findings with localization of auxin responses during nodulation and with direct quantification of auxin concentrations in these root segments, as we hypothesized that auxin transport affects the available pool of auxin (active and conjugated) at the nodule initiation site. We also tested the ability of ATIs to induce pseudonodules in a number of legume species forming indeterminate and determinate nodules, as well as their ability to inhibit auxin transport and induce auxin responses in the root.

### RESULTS

### Auxin Transport Regulation during Nodulation Differs between Medicago truncatula and Lotus japonicus

We were interested to compare auxin transport at the equivalent nodule developmental stages in M. truncatula and L. japonicus during the early stages of nodulation. To coordinate the time point of measurement with the onset of nodule development, we made root cross sections at different nodule developmental stages (**Figure 1**). In M. truncatula, no visible cell divisions were

detected at 0 and 6 hours post-inoculation (h.p.i.) (**Figures 1A,B**). We observed the first cortical cell divisions as early as 24 h.p.i. (**Figure 1C**) and more extensive cortical, endodermal, and pericycle divisions at 48 h.p.i. (**Figure 1D**). In L. japonicus, no cell divisions were detected at 0, 12, or 24 h.p.i. (**Figures 1E–G**). In one root, we found a single cortical cell division at 48 h.p.i., but not in any other roots. The first visible cell divisions appeared at 72 h.p.i. (**Figure 1I**) and visible nodule primordia at 120 h.p.i. in the majority of roots (**Figure 1K**).

We were interested in auxin transport regulation prior to the formation of visible nodule primordia on the roots. In M. truncatula, we measured acropetal auxin transport in the segment just below the inoculation spot at 6, 24, and 48 h.p.i. Auxin transported into the segment below the site of nodule initiation was chosen because if dividing cells inhibit the acropetal export of auxin this would be detectable as a reduced amount of auxin transported into the segment just below these cells. Consistent with our previous findings (Ng et al., 2015), we measured a significant decrease in acropetal auxin transport into the segment below the inoculation site at 24 h.p.i., but not at 6 or 48 h.p.i. (**Figure 2A**).

In L. japonicus, visible nodule primordia only formed after 120 h.p.i. under our growth conditions (**Figure 1K**). Hence, we measured auxin transport at 12, 24, 48, 72, 96, and 120 h.p.i. Similar to the findings reported in Pacios-Bras et al. (2003), we found a significant increase in acropetal auxin transport at 48 h.p.i. in the segment below the rhizobia inoculation spot (**Figure 3A**). In addition, acropetal auxin transport also increased significantly at 12, 72, 96, and 120 h.p.i. (**Figure 3A**).

### Auxin Transport Inhibitors Induce Pseudonodules on Legumes Forming Indeterminate Nodules

Several studies have reported the ability of certain indeterminate nodule-forming legumes to form nodule-like structures, also termed pseudonodules, in response to ATIs. We wanted to investigate if the predisposition to form pseudonodules in response to ATIs is unequivocally confined to indeterminatenodule forming legumes. To simulate the transient nature of the acropetal auxin transport inhibition in M. truncatula, we performed temporary flooding of selected legume species with the ATIs TIBA and NPA (Rightmyer and Long, 2011). We selected M. truncatula and Trifolium subterraneum to represent

indeterminate nodule-forming legume species. On the other hand, L. japonicus, Glycine max (soybean), Vicia faba (broad bean), Vigna unguiculata (snake bean), and M. atropurpureum (siratro) were selected to represent determinate nodule-forming legume species. Sesbania rostrata can form indeterminate- or determinate-type nodules, depending on aeration status of the roots (Fernández-López et al., 1998). Selecting this species posed an interesting question as to whether ATIs could induce pseudonodules on this "dual nodule-type" legume.

Using a temporary flooding system, we found that both TIBA and NPA could induce pseudonodules on the roots of M. truncatula and T. subterraneum. The optimal concentration of TIBA and NPA to induce pseudonodules on M. truncatula was 50 µM (**Figure 4A**). At 100 µM, TIBA induced pseudonodules on M. truncatula roots at a similar frequency as 50 µM, but a similar concentration of NPA reduced pseudonodules formed on the roots of M. truncatula (**Figure 4A**). TIBA, but not NPA, could induce pseudonodule formation at 10 µM on M. truncatula roots (**Figure 4A**). For T. subterraneum, pseudonodules formed at the highest frequency at 50 and 100 µM TIBA treatment (**Figure 4B**). Pseudonodules were observed on the roots of T. subterraneum treated with 1 and 10 µM TIBA, as well as 1, 10, 50, and 100 µM NPA, although pseudonodule numbers were significantly lower (**Figure 4B**). In both legumes, TIBA overall induced significantly more pseudonodules than NPA at the concentrations tested (**Figures 4A,B**; two-way ANOVA, p < 0.001). We also tested primary root growth at various concentrations of TIBA and NPA on M. truncatula and L. japonicus to investigate if they had any adverse effects on root development. At concentrations up to 100 µM TIBA in M. truncatula (Supplementary Figure 1A) and 50 µM TIBA in L. japonicus (Supplementary Figure 1B), primary root growth was not significantly affected when compared with the control treatment. However, in both legume species, a 10 µM NPA treatment already significantly reduced primary root growth (Supplementary Figures 1A,B), suggesting a stronger pleiotropic effect caused by NPA treatment. We did not observe any pseudonodules forming on the roots of L. japonicus, G. max, V. faba, V. unguiculata, or M. atropurpureum in response to the temporary flooding with either TIBA or NPA (Supplementary Figure 2; 15–25 plants analyzed per species). Interestingly, a few pseudonodules formed on the roots of S. rostrata in response to TIBA or NPA treatment (10 out of 96 plants) at a range of concentrations tested (1, 10, 50, or 100 µM) under non-flooded conditions that lead to indeterminate nodule types (Fernández-López et al., 1998).

The outward appearance of pseudonodules on M. truncatula (**Figures 5A,B**), T. subterraneum (**Figures 5C,D**), and S. rostrata (**Figures 5E,F**) resembled rhizobia-induced nodules. Interestingly, although pseudonodules on M. truncatula primarily formed on younger parts of the roots (where rhizobiainduced nodules form), they could also be found on more mature parts of the roots (**Figure 5G**). Cross sections of these structures revealed a more diffuse and randomized nature of cell divisions, as opposed to a more controlled pattern of cell divisions observed in rhizobia-induced nodules. Mature pseudonodules formed on M. truncatula (**Figures 6A,B**), T. subterraneum (**Figures 6C,D**), and S. rostrata (**Figures 6E,F**) all displayed cell divisions in

the pericycle, endodermis, and cortex (**Figure 6**). However, pseudonodules were characterized by more extensive pericycle and endodermal cell divisions, rather than the predominantly cortical cell divisions observed in rhizobia-induced nodules.

The legume S. rostrata can also form nodules on its stems. We did not observe pseudonodules forming on the stems of S. rostrata in response to TIBA or NPA treatment. However, unlike the adventitious root buds found on stems under control treatment (Supplementary Figures 3A,B), we found a higher occurrence of adventitious structures resembling bumps forming on the stems (Supplementary Figures 3D,E,G,H). Cross sections of these structures revealed a thicker layer of tissue with high cell wall autofluorescence (Supplementary Figures 3F,I), visualized under UV light, in comparison to the control (Supplementary Figure 3C).

Previously, we demonstrated that the formation of pseudonodules involved an inhibition of acropetal auxin transport in M. truncatula roots (Ng et al., 2015). Thus, we also tested auxin transport regulation in response to TIBA treatment (50 µM) at 6, 24, and 48 h.p.i., similar to the measurements performed for rhizobia inoculation. Unlike the transient nature of acropetal auxin transport inhibition

in M. truncatula (**Figures 1C**, **2A**), acropetal auxin transport decreased significantly in response to TIBA treatment at 6, 24, and 48 h.p.i. (**Figure 2B**). Although rhizobia inoculation failed to reduce acropetal auxin transport in L. japonicus, TIBA treatment also significantly reduced acropetal auxin transport in this legume (**Figure 3B**).

### Changes in Auxin Response and Concentration during Nodule Formation in Medicago truncatula and Lotus japonicus

We localized auxin responses in both M. truncatula and L. japonicus fully transformed plants carrying the GH3:GUS reporter (Pacios-Bras et al., 2003; van Noorden et al., 2007). In M. truncatula, auxin responses were enhanced in early nodule primordia at 48 h.p.i. in divided pericycle, endodermis, and inner cortex cells (**Figure 7B** and Supplementary Figure 4). In mature nodules, an auxin response was

found in the nodule vasculature and the nodule meristem (**Figure 7C**).

During pseudonodule formation in M. truncatula, the first small nodule primordia were found at 7 days following TIBA application, which was slower than formation of equivalent nodule primordia at 48 h after rhizobia inoculation. We found no enhanced GH3::GUS expression in the cortex at 48 h after TIBA treatment (Supplementary Figure 5). An increased auxin response was localized in the dividing pericycle, endodermis, and inner cortex cells at the early primordium stage at 7 days after TIBA application, similar to rhizobiainduced nodule primordia (**Figure 7E**). However, in more mature pseudonodules (20 days post-treatment), GH3::GUS expression did not resemble that of rhizobia-induced nodules and was confined to pericycle, endodermal, and inner cortical cell layers (**Figure 7F**).

In L. japonicus, induction of auxin responses occurred in cortical cells, mainly those surrounding early nodule primordia or those that appeared to have just divided, while primordia themselves showed very low GH3::GUS expression (**Figure 7H** and Supplementary Figure 4). We did not detect any changes in auxin response before the onset of cell divisions (**Figures 7A,D,G,J**). In mature L. japonicus nodules, auxin responses were restricted to the vasculature (**Figure 7I**). Nodules in L. japonicus do not retain an apical meristem, and no apical GH3::GUS response was seen. In L. japonicus roots treated with TIBA, we detected increased GH3::GUS responses first in outer cortical and epidermal cells (5 days post-treatment) and later

in the whole root (10 days post-treatment), but despite serial sectioning of multiple roots, we never detected any divided cells in the cortex (**Figures 7K,L**).

In parallel to the auxin responses, we quantified auxin content in inoculated root segments encompassing the inoculation site. We previously reported increased IAA content in inoculated M. truncatula root segments at 24 h.p.i., but not at 6 h.p.i. (Ng et al., 2015). Measurements at 48 h.p.i. in this study showed no significant changes in IAA content (Supplementary Figure 6). However, we found a significant decrease in the concentration of IAA-Ala at 48 h.p.i. that was not found at either 6 or 24 h.p.i. (Ng et al., 2015; Supplementary Figure 6). The auxins PAA, 4-Cl-IAA, and IAA-Val were not detected in any of the samples measured.

In contrast to M. truncatula, we measured a temporary decrease in IAA concentration in L. japonicus root segments encompassing the nodulation zone at 24 h.p.i. (**Figure 8A**). No IAA changes were detected between control and inoculated L. japonicus roots at 48 h.p.i. or 5 days post-inoculation (**Figure 8A**). The concentrations of IBA, IAA-Alanine, IAA-Aspartate, and IAA-Leucine/Isoleucine were not altered by rhizobia inoculation (**Figures 8B–E**). The auxins 4-chloro-IAA, IAA-Phenylalanine, and IAA-Tryptophan were not consistently detected in the root segments (**Figures 8F–H**).

## DISCUSSION

In our study, we posed two main questions:


### Is Acropetal Auxin Transport Inhibition during Nodulation Restricted to Indeterminate Nodulation, and What Are Possible Reasons for This?

Consistent with previous results in M. truncatula, we observed a transient reduction in acropetal auxin transport following

rhizobia treatment at 24 h.p.i., coinciding with the very earliest cell divisions (Wasson et al., 2006; van Noorden et al., 2006; Plet et al., 2011; Ng et al., 2015). For L. japonicus, Pacios-Bras et al. (2003) previously demonstrated an increase in acropetal auxin transport capacity in response to Nod factor treatment at 48 h.p.i., although those experiments were not statistically analyzed. Here, we included two additional features to our measurements: (1) we included additional time points to assess if this increase in auxin transport capacity was of a transient or prolonged nature and correlated with the onset of cell divisions and (2) we measured auxin transport in equivalent segments below the inoculation spot. Crucially, this study provides a direct comparison of two corresponding segments in M. truncatula and L. japonicus. Unlike the shortterm nature of acropetal auxin transport inhibition during indeterminate nodule initiation, we found that an increased auxin transport capacity in L. japonicus started at 12 h.p.i., i.e., before the very earliest cortical cell divisions, and continued at 48, 72, 96, and 120 h.p.i., where cell divisions were more extensive.

In M. truncatula, knockdown of MtPIN4 (ortholog of AtPIN1 mediating acropetal auxin transport; Schnabel and Frugoli, 2004; Kohlen et al., 2017) reduced nodulation (Huo et al., 2006), and transcription of this gene was one of the most highly upregulated by Nod factor treatment (Plet et al., 2011), suggesting that acropetal auxin transport regulation might occur through MtPIN4, although it may involve other auxin transporters. So far it is unknown how the increased acropetal auxin transport in L. japonicus is regulated. It would be essential to quantify PIN gene expression in L. japonicus in the future, and analyze mutants of auxin transporters.

The lack of auxin transport inhibition during L. japonicus nodule development was supported by the inability of TIBA or NPA to induce pseudonodules in this species. We observed pseudonodule formation on M. truncatula, subclover, and S. rostrata, and pseudonodules were also reported in alfalfa and white sweet clover (Hirsch et al., 1989; Wu et al., 1996). Although S. rostrata has the capacity to form both nodule types, it typically forms indeterminate-type nodules under nonwaterlogging and well-aerated growth conditions (Fernández-López et al., 1998), and our growth system and ATI assays simulated these conditions. Lack of pseudonodule formation on L. japonicus in response to ATIs had previously been observed (Takanashi et al., 2011), even though TIBA treatment reduced acropetal auxin transport capacity in L. japonicus roots similar to M. truncatula. Curiously, gibberellins were able to induce pseudonodules on L. japonicus, but not alfalfa roots (Kawaguchi et al., 1996). The mode of action or physiological effects of gibberellins were, however, not examined. Nevertheless, gibberellins are not likely to inhibit acropetal auxin transport because of their failure to induce pseudonodules on alfalfa. In addition to L. japonicus, pseudonodules were not found on any of the other examined legume species forming determinate nodules, including G. max, V. faba, V. unguiculata, or M. atropurpureum. Pseudonodules had been reported as "few" in response to TIBA or NPA in M. atropurpureum (Relic et al., 1993 ´ ), but no photos were shown in that study and we could not confirm this finding.

Pseudonodules formed in M. truncatula in response to TIBA showed GH3::GUS expression in the base, but no expression at the pseudonodule "tip," where rhizobia-induced nodules showed strong GH3::GUS expression in the nodule meristem. This suggests the lack of a pseudonodule meristem.

Why would ATIs not induce pseudonodules in determinate nodule-forming species? M. truncatula roots treated with TIBA formed pseudonodules, and at the earliest stages of their formation, these pseudonodules showed high auxin (GH3::GUS)

responses in the pericycle, endodermis, and inner cortex, similar to indeterminate nodule primordia. Modeling has shown that acropetal auxin transport inhibition would produce this pattern of auxin accumulation (Deinum et al., 2012). When we treated L. japonicus roots with TIBA, we found increased GH3::GUS expression in a different pattern – in the outer cortex, and later on in all cortex cells – but no small primordia were formed. Thus, it is likely that the difference in the tissue types involved in the early cell divisions (pericycle, endodermis, and inner cortex in indeterminate, and outer cortex for determinate nodules) dictates whether auxin transport changes contribute to nodule initiation. It is also possible that cortical responses to auxin differ in indeterminate and determinate nodule-forming species, for example in sensitivity, and this will have to be further explored in the future. For instance, it might be possible that auxin concentrations generated in the cortex in response to ATIs become super-optimal for determinate nodules (Turner et al., 2013).

### Which Other Mechanisms Could Contribute to Increased Auxin Responses in the Cortex during Nodule Initiation?

During the earliest stage of nodule initiation, both M. truncatula and L. japonicus showed increased auxin (GH3::GUS) response in the inner or outer/middle cortex, respectively. It is possible that the increased auxin response seen in dividing outer cortical cells in L. japonicus is in response to altered sensitivity to auxin. Studies in soybean have demonstrated the role of a number of microRNAs controlling the expression of auxin receptors and response genes that could change sensitivity toward auxin (Mao et al., 2013; Turner et al., 2013). Another mechanism to explain the initial increased auxin response could include auxin synthesis in the cortex, and evidence for the induction of auxin synthesis genes in L. japonicus was found previously (Suzaki et al., 2012). However, we did not detect any significant increase in auxin concentration in that area in L. japonicus roots. Intriguingly, a reduction in IAA content was measured at 24 h.p.i., although we did not detect any changes in auxin transport or auxin response at that stage. Currently, we cannot explain the significance of this finding; it could be due to increased auxin degradation. In M. truncatula, the increased auxin response at the early primordium stage at 48 h.p.i. was not accompanied by increased auxin concentrations, either. This suggests that it is either difficult to detect increased auxin concentrations in small numbers of cells, as 4 mm root segments were harvested for this assay, or that there is no linear relationship between GH3::GUS expression and auxin concentration.

Genetic studies on the role of auxin transport carriers during nodulation in L. japonicus are lacking. One report highlighted the role of an ABCB-type multidrug resistance protein, LjABCB1, in mature L. japonicus nodules, but did not examine its role during early nodule development. The authors proposed that LjABCB1 exports auxin toward infected cells in the nodule, based on the selective localization of the protein at adjacent uninfected cells (Takanashi et al., 2012). Future studies will clearly have to unravel the regulation of various auxin transporters in L. japonicus (Kohlen et al., 2017). In addition, it will be necessary to model the changes in acropetal auxin transport in L. japonicus computationally, to find out if these observed increases in auxin transport would account for the increased auxin responses in the cortex. Previous modeling has shown that auxin accumulation in the middle/outer cortex could be explained by radial repositioning of PIN proteins (Deinum et al., 2012, 2016).

Apart from the differences between M. truncatula and L. japonicus in auxin transport at the time of the first cortical cell divisions, we also found differences in auxin responses at the primordia stage. While auxin responses remained high in young nodule primordia of M. truncatula, auxin response diminished quickly in L. japonicus nodule primordia as reported by other studies (Mathesius et al., 1998; Takanashi et al., 2011; Suzaki et al., 2012; Turner et al., 2013). It is possible that changes in auxin sensitivity in the cortex differentiate the conditions suitable for nodule development in species forming indeterminate and determinate nodules: While reduced auxin sensitivity increased nodule numbers in soybean (Nizampatnam et al., 2015; Wang et al., 2015), a window of higher auxin sensitivity may be required for indeterminate nodule initiation (Mao et al., 2013). However, it was also demonstrated that silencing of the auxin receptor TIR1 in soybean reduced nodulation, suggesting that auxin responses are necessary for determinate nodule formation (Cai et al., 2017). Thus, correct auxin responses in space and time are crucial for nodule development.

In summary, we found no evidence that inhibition of acropetal auxin transport occurs during nodule initiation in L. japonicus, and it is unlikely to be a mechanism that creates the correct auxin gradients for other legume species forming determinate nodules because ATIs did not induce any cortical cell divisions in a number of tested species, in contrast to species forming indeterminate nodules. While increased auxin responses occurred in the earliest dividing cortex cells of L. japonicus, no significant increase in auxin concentration was measured. In contrast, M. truncatula showed more sustained auxin responses in nodule primordia, although increased auxin concentration was only detected at 24 h, preceding nodule primordia. Thus, it is possible that changes in auxin sensitivity in cortical cells, without measurable changes in auxin content, contribute to increased auxin responses.

### MATERIALS AND METHODS

### Plant Material and Growth Conditions

Plant species used in this study included M. truncatula wildtype cultivar Jemalong A17 (South Australian Research and Development Institute, Adelaide, Australia), L. japonicus ecotype Gifu B-129 (Biological Resource Center in Lotus and Glycine, University of Miyazaki, Japan), T. subterraneum cv. Karridale (Clean Seeds, Bungendore, Australia), G. max cv. Bragg (Prof. Peter Gresshoff, University of Queensland), S. rostrata (Prof. Sofie

Goormachtig, Ghent University, Belgium), M. atropurpureum (Selected Seeds, Pittsworth, Australia), V. unguiculate and V. faba cv. Coles Early Dwarf (both Mr. Fothergill's Seeds Pty Ltd., South Windsor, Australia).

Seeds of M. truncatula, L. japonicus, T. subterraneum, and M. atropurpureum were scarified with sand paper, surface-sterilized in 6% (w/v) sodium hypochlorite for 10 min, then washed with sterilized milliQ water five times. Next, M. truncatula and L. japonicus seeds were imbibed in sterilized milliQ water containing 0.25 mg ml−<sup>1</sup> of the antibiotic augmentin for 6 h with gentle mixing by rotation to reduce bacterial contamination. Sterilized seeds were washed once with sterilized milliQ water and then plated on Fahråeus (F) medium (Fahraeus, 1957) for M. truncatula or 1/4; strength Broughton and Dilworth medium (Broughton and Dilworth, 1971) for L. japonicus. For T. subterraneum and M. atropurpureum, seeds were sown straight onto F plates without the augmentin incubation step. Seeds were incubated at 4◦C in the dark for 48 h. Germination of seeds was synchronized by incubating the plates at 25◦C for 24–48 h with plates inverted. Seedlings with radicle length of approximately 5–10 mm were transferred onto F plates (M. truncatula, T. subterraneum, and M. atropurpureum) or 1/4; B&D plates (L. japonicus). Plates were semi-sealed with Parafilm, placed vertically in a container with a black cardboard interspersed between each plate to shield roots from direct light. Plates were incubated at 25 (M. truncatula, T. subterraneum, and M. atropurpureum) or 20◦C (L. japonicus), with a 16 h light and 8 h dark period at 150 µmol m−<sup>2</sup> s −1 light intensity.

For soybean, the germination protocol was based on Brechenmacher et al. (2009). Seeds were soaked in 0.1 N HCl for 10 min and then washed with tap water five times. Seeds were then surfaced-sterilized with 6% (w/v) sodium hypochlorite for 15 min and then washed with sterilized milliQ water five times, air dried for 20 min, and subsequently sown on 1/4; B&D plates. Seeds were left at 25◦C in the dark for germination over several days. Germinated soybeans were transferred into pots containing autoclaved vermiculite (Grade 3), watered with 1/4; B&D medium, and grown at 16 h light and 8 h dark period at 150 µmol m−<sup>2</sup> s −1 light intensity in a glasshouse.

For S. rostrata, the germination protocol was modified from Goethals et al. (1989). Seeds were immersed in 4 M sulfuric acid for 1 h and washed five times with tap water. This was followed by surface sterilization with 6% (w/v) sodium hypochlorite for 15 min and then five washes with sterilized milliQ water. Sterilized seeds were imbibed overnight in sterilized milliQ water. Imbibed seeds were plated onto F plates and left at 25◦C in the dark for germination over several days. Germinated seedlings were transferred into glasshouse pots containing autoclaved vermiculite (Grade 3) and watered with F medium twice a week.

Seeds of V. unguiculata and V. faba were surface-sterilized with 6% (w/v) sodium hypochlorite for 15 min and then washed with sterilized milliQ water five times. Seeds were directly planted into pots containing autoclaved vermiculite (Grade 3) and thinned to one plant per pot after germination. Plants were watered with F medium twice a week and grown in a glasshouse.

### Bacterial Strains and Inoculation Conditions

The S. meliloti strain A2102, a triple nod mutant for nodD1, nodD2, and nodD3 derived from the WT strain Sm1021, containing the pE65 plasmid encoding a constitutively overexpressed copy of nodD3 (Barnett et al., 2004) was used for all inoculations on M. truncatula (kindly provided by Dr. Melanie Barnett and Prof. Sharon Long, Stanford University, Stanford, CA, United States), hereafter referred to as "E65." This strain was used to compare results from this study to that of Ng et al. (2015). We previously established that the E65 strain (Ng et al., 2015) inhibited auxin transport similarly to the wild-type strain 1021 (Wasson et al., 2006). The E65 strain was maintained on Bergensen's Modified Medium (BMM) (Rolfe and Gresshoff, 1988) supplemented with 10 µg ml−<sup>1</sup> tetracycline and 100 µg ml−<sup>1</sup> streptomycin (Sigma Chemicals). For L. japonicus, inoculation (OD<sup>600</sup> nm = 0.05) was performed with Mesorhizobium loti strain MAFF303099, maintained on Tryptone-Yeast (TY) medium.

For inoculation of M. truncatula, an overnight culture of S. meliloti strain E65 in BMM at 28◦C was used. The optical density (OD<sup>600</sup> nm) of the culture was adjusted to 0.1 for spotinoculation. Spot-inoculation was performed by placing ∼1 µl of S. meliloti culture or BMM, onto the root surface 2 mm above the root tip, corresponding to the nodulation-susceptible zone (Bhuvaneswari et al., 1981). For analyzing GH3::GUS expression, spot-inoculation was performed with a glass capillary pulled into a fine tip over a flame and glued to a hypodermic needle.

For spot-inoculation of L. japonicus, a liquid culture of M. loti in TY medium was incubated for 3 days at 28◦C. The OD<sup>600</sup> nm of the cultures was adjusted to 0.05. Spot-inoculation was performed by placing ∼1 µl of rhizobia culture or TY as a negative control, 2 mm above the root tip.

### Flood Treatment with Auxin Transport Inhibitors

Seeds were germinated as described above. Seedlings of M. truncatula, L. japonicus, T. subterraneum, and M. atropurpureum were treated while growing on agar plates. At 1 week posttransfer onto agar medium, seedlings were treated by flooding, as described in Rightmyer and Long (2011). Diluted solutions of NPA and TIBA (Sigma Chemicals) were made in sterile 50 ml Falcon tubes. Control treatments contained equivalent dilution of methanol (which was used as a solvent for stock solutions). The concentrations of ATIs were chosen because they were previously demonstrated to induce pseudonodules on M. truncatula (Rightmyer and Long, 2011). Seedlings were flooded with 20–30 ml of diluted ATIs for 10 s, and then the solution was decanted. Pseudonodules were analyzed 3 weeks after treatment with ATIs.

For plants grown in pots (G. max, S. rostrata, V. faba, V. unguiculata, and M. truncatula as a positive control for pot experiments), 50 ml of ATI diluted in growth medium was applied to each pot. This was repeated after 1 week and plants were then left to grow for another 3 weeks before uprooting the plants and analyzing for pseudonodule formation.

### Auxin Quantification

fpls-09-00169 February 13, 2018 Time: 15:55 # 11

Auxin quantification was based on Ng et al. (2015). Commercial auxin standards and a deuterated internal standard were used to determine elution times, collision energies, detection limits, and for absolute quantification. Auxin standards were obtained from OlChemim (IAA-Phenylalanine, IAA-Leucine, IAA-Valine, IAA-Tryptophan, 4-Cl-IAA), Sigma (IAA-Aspartate, IAA-Alanine, IAA-Isoleucine, IAA, IBA, PAA) and Cambridge Isotope Laboratories (indole-2,4,5,6,7-d5-3-acetic acid).

Plant roots were collected at 6, 24, and 48 h.p.i. with rhizobia. Root segments of 4 mm spanning the spot-inoculation site were collected and snap-frozen immediately in liquid nitrogen. A total of 30–40 root segments were collected for each treatment, for each biological replicate. Between 5 and 10 biological replicates were analyzed for each time point and species. The frozen tissue samples were mechanically lysed with stainless steel beads in a Qiagen TissueLyser LT with a pre-cooled tube holder. To each tube 20 µl of the internal standard (1 µg ml−<sup>1</sup> of 3-[2H5]indolylacetic acid) followed by 500 µl extraction solvent (methanol:propanol:glacial acetic acid, 20:79:1, v/v/v) were added and auxin extraction was performed in a sonicator bath for 30 min at 4◦C. Samples were then centrifuged at 16,000 × g for 15 min. The supernatant was transferred to a fresh tube and subsequently dried in a Speedvac centrifuge. Extraction was repeated once, and the supernatant combined with the (dried) supernatant from the first extraction, and subsequently evaporated in a Speedvac centrifuge. Vacuumdried samples were resuspended with 100% methanol, vortexed for 5 s, and filtered through a Nanosep MF GHP 0.45 µm filter (Pall Life Sciences) by centrifugation at 16,000 × g for 1 min. The resuspension step was repeated once. The eluent containing the auxin extracts was transferred to an amber vial and vacuum-dried. Samples were stored at −80◦C until analysis. Prior to analysis, samples were taken out from the freezer to equilibrate to room temperature. Each sample was resuspended in 50 µl methanol (Acros Organics):water (60:40, v/v).

Tandem mass spectrometry was performed using an Agilent 6530 Accurate Mass LC–MS Q-TOF (Santa Clara, CA, United States). Samples were subjected to ESI in the Jet Stream interface in both ion positive and negative polarities. Based on optimized LC-ESI-Q-TOF parameters using auxin standards, the auxins IAA, IBA, and IAA-Ala had better sensitivity in the positive mode. The other auxin species were better detected in the negative mode. Optimized conditions in the positive mode were as follows: gas temperature 250◦C, drying gas 5 l min−<sup>1</sup> , nebulizer 30 psig, sheath gas temperature 350◦C and flow rate of 11 l min−<sup>1</sup> , capillary voltage 2500 V, nozzle voltage 500 V, and fragmentor voltage 138 V. Conditions in the negative mode were as follows: gas temperature 300◦C, drying gas 9 l min−<sup>1</sup> , nebulizer 25 psig, sheath gas temperature 350◦C and flow rate of 11 l min−<sup>1</sup> , capillary voltage 3000 V, nozzle voltage 500 V, and fragmentor voltage 140 V. Samples were injected (7 µl) onto an Agilent Zorbax Eclipse 1.8 µm XDB-C18 2.1 × 50 mm column. Solvent A consisted of 0.1% aqueous formic acid and solvent B, 90% methanol/water with 0.1% formic acid. Free auxins and conjugates were eluted with a linear gradient from 10 to 50% solvent B over 8 min, 50–70% solvent B from 8 to 12 min (then held at 70% from 12 to 20 min) at a flow rate of 200 µl min−<sup>1</sup> . The quadrupole time-of-flight (Q-TOF) was run in targeted MS/MS mode using collision-induced dissociation (CID; N<sup>2</sup> collision gas supplied at 18 psi with m/z 1.3 isolation window) where the MS extended dynamic range (2 Hz) was m/z 100–1000 with an acquisition rate of 3 spectra s−<sup>1</sup> and MS/MS at m/z 50–1000 at 3 spectra s−<sup>1</sup> . Data were analyzed using Agilent Technologies MassHunter software (ver. B.5.0).

### Histochemistry and Microscopy

Beta-glucuronidase (GUS) staining was performed as described in van Noorden et al. (2007). Sections of 100 mm thickness were made using a Vibratome 1000 (Vibratome Company, St. Louis, MO, United States) and viewed under bright-field optics using a DMLB microscope (Leica Microsystems, Wetzlar, Germany), and images were collected with a mounted CCD camera (RT Slider; Diagnostic Instruments, Sterling Heights, MI, United States).

### Auxin Transport Studies

Tritium-labeled indole-3-acetic acid (3H-IAA) solution (7.5 µl of 1 mCi ml−<sup>1</sup> ) (American Radiolabeled Chemicals, St. Louis, MO, United States) was diluted in 20 µl ethanol and mixed with 1.5 ml of melted and cooled 1% agarose at pH 4.8, in a 3 cm diameter Petri dish. The pH was chosen as it is close to the isoelectric point of IAA. Small blocks with dimensions of 2 mm × 2 mm × 2 mm were cut with a scalpel. This standardized the amount of <sup>3</sup>H-IAA supplied to plants. Seedlings were pretreated with rhizobia/ATIs prior to auxin transport study, as described below.

Acropetal auxin transport was performed in relation to the spot-inoculation site. Roots were cut 8 mm from the inoculation spot in the shootward direction (∼12 mm from the root tip). The shoot-containing segment was discarded, and a small <sup>3</sup>H-IAA block (2 mm × 2 mm × 2 mm) placed on the cut end of the root-tip containing segment. A Parafilm strip was placed underneath the root segments to prevent diffusion of <sup>3</sup>H-IAA from the growth media directly into parts of the root. Samples were incubated vertically for 6 h (M. truncatula) or 12 h (L. japonicus) in the dark to allow <sup>3</sup>H-IAA to diffuse from the agar block through the cut end. The first 4 mm segment touching the <sup>3</sup>H-IAA agar block was discarded. The root segments located just below the inoculation site were transferred into individual scintillation vials containing 2 ml scintillation fluid (Perkin-Elmer).

Samples were incubated on an INOVA 2100 Platform Shaker (New Brunswick Scientific) overnight at room temperature. Radioactivity was measured in a scintillation counter (Tri-Carb <sup>R</sup> Liquid Scintillation Analyzer B2810TR, Perkin-Elmer) over 1 min each. The default settings for tritium decay measurement were used. A vial containing just the scintillation fluid was used as a blank for background subtraction during analysis.

### Statistical Analyses

fpls-09-00169 February 13, 2018 Time: 15:55 # 12

Statistical analyses were carried out with Genstat 15th Edition (VSN International, Hemel Hempstead, United Kingdom) and Prism version 5.02.

### AUTHOR CONTRIBUTIONS

JN and UM: conceived, acquired data, analyzed data, and drafted the work.

### REFERENCES


### FUNDING

This project was supported by a grant from the Australian Research Council (DP150102002 to UM).

### ACKNOWLEDGMENTS

We thank Prof. Herman Spaink and Prof. Jens Stougaard for kindly providing seeds of the L. japonicus GH3::GUStransformed plants. We also thank Dr. Eva Deinum and Dr. Wouter Kohlen for stimulating discussions during the preparation of this manuscript.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00169/ full#supplementary-material


Kohlen, W., Ng, J. L. P., Deinum, E. E., and Mathesius, U. (2017). Auxin transport, metabolism, and signalling during nodule initiation: indeterminate and determinate nodules. J. Exp. Bot. 69, 229–244. doi: 10.1093/jxb/erx308

Korasick, D. A., Enders, T. A., and Strader, L. C. (2013). Auxin biosynthesis and storage forms. J. Exp. Bot. 64, 2541–2555. doi: 10.1093/jxb/ert080



**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 Ng and Mathesius. 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.

# Dual Color Sensors for Simultaneous Analysis of Calcium Signal Dynamics in the Nuclear and Cytoplasmic Compartments of Plant Cells

Audrey Kelner 1†, Nuno Leitão2†, Mireille Chabaud<sup>1</sup> , Myriam Charpentier <sup>2</sup> \* and Fernanda de Carvalho-Niebel <sup>1</sup> \*

<sup>1</sup> Laboratory of Plant Microbe Interactions, Université de Toulouse, Institut National de la Recherche Agronomique, Centre National de la Recherche Scientifique, Castanet-Tolosan, France, <sup>2</sup> Department of Cell and Developmental Biology, John Innes Centre, Norwich, United Kingdom

#### *Edited by:*

Pascal Ratet, Centre National de la Recherche Scientifique (CNRS), France

#### *Reviewed by:*

Jiahui Wu, Weill Cornell Medical College, Cornell University, United States Frantisek Baluska, University of Bonn, Germany

#### *\*Correspondence:*

Myriam Charpentier myriam.charpentier@jic.ac.uk Fernanda de Carvalho-Niebel fernanda.de-carvalho-niebel@inra.fr

† These authors have contributed equally to this work.

#### *Specialty section:*

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

*Received:* 17 December 2017 *Accepted:* 12 February 2018 *Published:* 27 February 2018

#### *Citation:*

Kelner A, Leitão N, Chabaud M, Charpentier M and de Carvalho-Niebel F (2018) Dual Color Sensors for Simultaneous Analysis of Calcium Signal Dynamics in the Nuclear and Cytoplasmic Compartments of Plant Cells. Front. Plant Sci. 9:245. doi: 10.3389/fpls.2018.00245 Spatiotemporal changes in cellular calcium (Ca2+) concentrations are essential for signal transduction in a wide range of plant cellular processes. In legumes, nuclear and perinuclear-localized Ca2<sup>+</sup> oscillations have emerged as key signatures preceding downstream symbiotic signaling responses. Förster resonance energy transfer (FRET) yellow-based Ca2<sup>+</sup> cameleon probes have been successfully exploited to measure the spatiotemporal dynamics of symbiotic Ca2<sup>+</sup> signaling in legumes. Although providing cellular resolution, these sensors were restricted to measuring Ca2<sup>+</sup> changes in single subcellular compartments. In this study, we have explored the potential of single fluorescent protein-based Ca2<sup>+</sup> sensors, the GECOs, for multicolor and simultaneous imaging of the spatiotemporal dynamics of cytoplasmic and nuclear Ca2<sup>+</sup> signaling in root cells. Single and dual fluorescence nuclear and cytoplasmic-localized GECOs expressed in transgenic Medicago truncatula roots and Arabidopsis thaliana were used to successfully monitor Ca2<sup>+</sup> responses to microbial biotic and abiotic elicitors. In M. truncatula, we demonstrate that GECOs detect symbiosis-related Ca2<sup>+</sup> spiking variations with higher sensitivity than the yellow FRET-based sensors previously used. Additionally, in both M. truncatula and A. thaliana, the dual sensor is now able to resolve in a single root cell the coordinated spatiotemporal dynamics of nuclear and cytoplasmic Ca2<sup>+</sup> signaling in vivo. The GECO-based sensors presented here therefore represent powerful tools to monitor Ca2<sup>+</sup> signaling dynamics in vivo in response to different stimuli in multi-subcellular compartments of plant cells.

Keywords: *Medicago truncatula*, *Arabidopsis thaliana*, root symbiosis, root hairs, root elongation zone, calcium, GECO sensors, biotic and abiotic stimuli

### INTRODUCTION

Divalent calcium (Ca2+) ions are versatile second messengers regulating many biological processes in eukaryotic and prokaryotic cells (Dodd et al., 2010; Domínguez et al., 2015). In plants, Ca2<sup>+</sup> mediates developmental and physiological responses to various biotic and abiotic environmental cues (McAinsh and Pittman, 2009; Dodd et al., 2010; Kudla et al., 2010; Stael et al., 2012). Stimuli-induced Ca2<sup>+</sup> releases result in transient or repetitive oscillations, distinct in their frequency, amplitude, duration, and spatial distribution. The sensing and decoding of these so-called Ca2<sup>+</sup> signatures (McAinsh and Pittman, 2009) by Ca2+-binding proteins assure the transduction of the signal to downstream transcriptional and metabolic responses (Galon et al., 2010; Reddy et al., 2011; Whalley and Knight, 2013).

The analysis of Ca <sup>2</sup><sup>+</sup> signatures and their contribution to signaling in different cellular compartments requires tools adapted for monitoring Ca2<sup>+</sup> dynamics with high spatial and temporal resolution. Bioluminescent aequorin-based probes can successfully quantify variations in Ca2<sup>+</sup> signaling in plants, but their low light emission limits the spatiotemporal resolution of these responses during imaging (Knight et al., 1991; McCombs and Palmer, 2008). Cellular resolution can be achieved by the use of fluorescent dyes and geneticallyencoded Ca2<sup>+</sup> indicators (GECIs), consisting of fluorescent proteins (FPs) fused to sensor domains. Compared to fluorescent dyes, GECIs have the advantage of being accessible to any cell type and offer the possibility of measuring Ca2<sup>+</sup> variations in subcellular compartments, by specific targeting of the sensor. The most popular GECI sensors are the ratiometric Förster Resonance Energy Transfer (FRET)-based Yellow Cameleons (YCs) composed of a fluorescent donoracceptor pair, respectively, CFP and YFP, linked by a calmodulin (CaM) and the CaM-binding peptide of myosin light-chain kinase M13 (Miyawaki et al., 1997; Allen et al., 1999). Upon Ca2<sup>+</sup> binding to CaM, CaM and M13 interact which triggers protein conformational changes that result in increased FRET between the donor-acceptor pair. More recently, improved FRET-based pairs or hybrid sensors such as BRET (bioluminescent aequorin coupled with a FP) have been engineered and successfully used in plants (Rodriguez-Garcia et al., 2014; Xiong et al., 2014; Bajar et al., 2016).

Targeting FRET-based probes to different subcellular compartments revealed precise Ca2<sup>+</sup> variations in different organelles, including the nucleus, the endoplasmic reticulum, the chloroplast, peroxisomes, and mitochondria (Iwano et al., 2009; Sieberer et al., 2009; Costa et al., 2010; Krebs et al., 2012; Loro et al., 2012, 2016; Stael et al., 2012; Bonza et al., 2013). The Ca2<sup>+</sup> dynamics in specific compartments are likely to reflect the differential activation of Ca2<sup>+</sup> channels and the contribution of diverse intra and extracellular Ca2<sup>+</sup> stores in the control of specific functions, such as gene transcription (Ranty et al., 2012). A number of stimuli activates Ca2<sup>+</sup> responses simultaneously in different subcellular compartments (van Der Luit et al., 1999; Wood et al., 2001). As such, in legumes, the perception of diffusible nodulation (Nod) and mycorrhizal (Myc) factors released by nitrogen-fixing bacteria and arbuscular mycorrhiza (AM), fungi respectively, induces calcium oscillations in the nucleus and adjacent cytoplasm (Ehrhardt et al., 1996; Miwa et al., 2006a; Sieberer et al., 2009). Symbiotic calcium spiking was monitored using fluorescent dyes, cytoplasmic FRET-based YC probes (Ehrhardt et al., 1996; Catoira et al., 2000; Wais et al., 2000; Shaw and Long, 2003; Miwa et al., 2006a,b; Kosuta et al., 2008) or nuclear-tagged YFP-FRET probes (Sieberer et al., 2009, 2012; Chabaud et al., 2011; Genre et al., 2013; Sun et al., 2015). Although monitored independently, cytoplasmic and nuclear Ca <sup>2</sup><sup>+</sup> responses were hypothesized to be synchronized based on their relative time period. Nod factors (NF), consisting of lipo-chitooligosaccharides, and AM fungi chitin oligomers (COs), both induce Ca2<sup>+</sup> spiking, essential for the transduction of downstream symbiotic signaling in the nucleus (Zipfel and Oldroyd, 2017). More recently, major components involved in the generation of Ca <sup>2</sup><sup>+</sup> spiking were identified in legumes. In the model legume Medicago truncatula, the potassium channel DMI1 (Ané et al., 2004), the Ca2+-ATPase MCA8 (Capoen et al., 2011), and CNGC15 Ca2<sup>+</sup> channels (Charpentier et al., 2016), are required for the generation of symbiotic Ca2<sup>+</sup> oscillations (Granqvist et al., 2012; Charpentier and Oldroyd, 2013; Charpentier et al., 2016). Interestingly, they all locate to both the outer nuclear membrane (ONM) and the inner nuclear membrane (INM) (Capoen et al., 2011; Charpentier et al., 2016). This observation raises the issue of the direction of the observed Ca2<sup>+</sup> release. Ca2<sup>+</sup> could be simultaneously released across the INM and the ONM or, sequentially, through the INM or the ONM to the nucleoplasm and cytoplasm, respectively. Thus, simultaneous investigation of nuclear and cytoplasmic Ca2<sup>+</sup> dynamics is required to gain information on the origin of the specific symbiotic factor-induced nuclear Ca2<sup>+</sup> signals, and clarify the spatial activation of the ion channels.

Recent studies in animal cells revealed that highly sensitive Ca2<sup>+</sup> GECI sensors can discriminate stimulus-dependent directional propagation of Ca2<sup>+</sup> signals (Nakao et al., 2015). These improved GECIs, called GECOs (for GECIs for optical imaging) are circularly permuted single fluorescent intensiometric Ca2+-sensing proteins with a CaM and a M13 peptide domains. Upon Ca2<sup>+</sup> binding, changes in protein conformation result in a visible increase in its fluorescence intensity (Zhao et al., 2011). GECOs have to date only been targeted to the cytoplasm to study Ca2<sup>+</sup> signaling in plant cells (Ngo et al., 2014; Keinath et al., 2015; Wang et al., 2015; Tunc-Ozdemir and Jones, 2017; Waadt et al., 2017). However, the available palette of GECO fluorescent protein colors now opens the possibility of concomitantly imaging Ca2+-signaling in different compartments of plant cells (Zhao et al., 2011). Moreover, improved GECO versions were generated through directed evolution and screened for higher intensity signal change upon Ca2<sup>+</sup> binding, but not yet used in plants (Zhao et al., 2011; Wu et al., 2013).

In this study, we generated dual Ca2<sup>+</sup> sensors that simultaneously monitor the Ca2<sup>+</sup> dynamics in the nucleus and cytoplasm, by targeting highly sensitive GECO Ca2<sup>+</sup> sensors to these subcellular compartments. We demonstrated that the new dual GECO sensors are powerful tools to concomitantly monitor in vivo Ca2<sup>+</sup> dynamics in adjacent nuclear and cytoplasmic compartments of both M. truncatula and Arabidopsis roots in response to both biotic and abiotic signals. We unraveled that the symbiotic Nod factor-induced nuclear calcium oscillations preferentially start in the nucleus. Additionally the dual GECO sensor revealed that biotic and abiotic stimuli trigger distinct cytoplasmic and nuclear Ca2<sup>+</sup> dynamics in A. thaliana root cells.

### RESULTS

### The Highly Sensitive GECO Sensor Enables Real-Time Visualization of Nuclear Ca2<sup>+</sup> Spiking in *M. truncatula*

In order to evaluate the performance of GECO as a Ca2<sup>+</sup> sensor for the symbiotic factor-induced nuclear Ca2<sup>+</sup> oscillations in M. truncatula, we designed a compatible binary vector to constitutively express the red fluorescence R-GECO1 (Zhao et al., 2011) in the nucleus of plant cells (Figure S1). This construct, referred to as NR-GECO1, comprises a viral SV40 nuclear localization signal (NLS) fused to the R-GECO1 and driven by the double CAMV 35S promoter (2xp35S) (Figure S1), and was used together with the previously characterized yellow fluorescence cameleon sensors NUP-YC2.1 (Sieberer et al., 2009) and NLS-YC3.6 (Krebs et al., 2012) to generate M. truncatula composite plants via Agrobacterium rhizogenes-mediated transformation. Confocal laser scanning microscopy (CLSM) revealed that both the NR-GECO1 and the two cameleon sensors were targeted to the nuclear compartment of M. truncatula root cells (Figure S2). NR-GECO1 fluorescence was undetectable in the cytoplasm, and appeared homogeneously distributed within the nucleus (Figure S2A).

Symbiosis-related nuclear Ca2<sup>+</sup> responses were assessed in the nucleus of growing root hair cells using CLSM (Figure S3). The basal fluorescence signal intensity was recorded for 5–10 min for each sensor before treatment of roots with purified Sinorhizobium meliloti NFs (10−<sup>9</sup> M) (**Figures 1A,C,E**). Following NF application, sustained Ca2<sup>+</sup> spikes were observed within the nucleus of cells transformed with the three sensors (**Figures 1B,D,F**). They all generated comparable patterns of NFinduced Ca2<sup>+</sup> spiking, with spike periodicity of about 100–120 s (**Figure 1G**).

To evaluate whether NR-GECO1 was suitable for detecting other Ca2<sup>+</sup> responses in M. truncatula, we challenged M. truncatula composite plants and root organ cultures (ROCs) with tetra-chitooligosaccharides (CO4) previously shown to elicit nuclear Ca2<sup>+</sup> oscillations (Genre et al., 2013; Sun et al., 2015). Similarly to NFs, NR-GECO1 allowed the detection of CO4-induced nuclear Ca2<sup>+</sup> spiking in both composite plants and ROCs (Figure S4). As previously observed with a cameleon probe (Genre et al., 2013), the NF receptor nfp mutant no longer exhibited NF-elicited Ca2<sup>+</sup> spiking but still displayed CO4-induced Ca2<sup>+</sup> spiking responses (Figures S4E–G). Taken together, NR-GECO1 can efficiently measure symbiosisrelated nuclear Ca2<sup>+</sup> spiking responses in M. truncatula roots.

To investigate sensor performance, we compared the maximum signal change to the basal fluorescence signal (before the spike), by calculating the signal-to-noise ratio (SNR) values for each individual probe. The fractional fluorescence changes (1F/F) for NR-GECO1 and the fractional ratio changes (1R/R) for the cameleon sensors were measured as described (Figure S3B and Material and Methods), from a high number of individual root hairs. FRET-based probes exhibited rather similar dynamic changes, although NLS-YC3.6 appeared to be

with the FRET probes NUP-YC2.1 (A,B), NLS-YC3.6 (C,D), and the single NR-GECO1 sensor (E,F) before (Ctrl) and after Nod Factor (+NF) addition. n represents the number of individual root hair cells analyzed from 2 to 3 independent experiments (AU, Arbitrary Units). Note that the y-axis scales are different for each sensor. (G,H) Calcium spiking periodicity (G) and (Continued)

slightly superior to NUP-YC2.1 with mean SNR values of 0.41 and 0.34, respectively (**Figure 1H**). However, there was a striking difference in the dynamic responses observed between the nuclear NR-GECO1 and the cameleon sensors with a mean SNR value of 1.27 (**Figure 1H**). Thus, NR-GECO1 allows monitoring of NF-induced nuclear Ca2<sup>+</sup> spiking with a greater sensitivity than the cameleon sensors. Our results confirm the previously described higher sensitivity of the cytoplasmic R-GECO1 in comparison to the YC3.6, in response to the ATP-induced Ca2<sup>+</sup> signals in A. thaliana roots (Keinath et al., 2015). Additionally, using CLSM, NR-GECO1 allowed real time visualization of the Ca2<sup>+</sup> changes in contrast to the cameleon probe for which the YFP/CFP fluorescent ratio have to be calculated from a pre-selected region of interest (ROI) (Figure S5 and Movies S1, S2). Thus, in comparison to cameleon sensors, NR-GECO1 has a higher dynamic range, which greatly facilitates the rapid monitoring of cell-autonomous dynamics of Ca2<sup>+</sup> spiking in various M. truncatula root hair and non-hair cells.

### A Dual GECO Sensor for Simultaneously Monitoring Ca2<sup>+</sup> Signals in Both Nuclear and Cytoplasmic Compartments of *M. truncatula* Root Cells

Taking advantage of the high sensitivity of GECOs, we generated a dual fluorescence color sensor for imaging Ca2<sup>+</sup> signal dynamics simultaneously in the cytoplasm and the nuclear compartments of the same root hair cell. We therefore combined the nuclear-tagged red GECO1 and the cytoplasmic green GECO1 in a single construct (Figure S1). In this dual sensor, both the nuclear red and cytoplasmic green GECOs (NRCG-GECO1) are driven by 2xp35S promoters. CLSM imaging revealed the dual labeling of nuclear and cytoplasmic compartments of M. truncatula epidermal cells (**Figure 2A**). In a non-spiking state (**Figure 2A**), the red and green fluorescence proteins are found, respectively, in the nuclear and cytoplasmic regions demonstrating the correct targeting of the GECO to the expected subcellular compartments. However, during a Ca2<sup>+</sup> spike a merged orange-to-yellow fluorescence signal is observed in the nuclear region, due to the acquisition of increased nuclear R-GECO1 and perinuclear green G-GECO1 fluorescence signals (**Figure 2B**).

Imaging conditions were set up to avoid crossover between fluorescence emissions of the two probes. The excitation of G-GECO1 and R-GECO1 is achieved using an argon 488 nm

FIGURE 2 | A dual GECO Ca2<sup>+</sup> sensor targeted to both nuclear and cytoplasmic subcellular compartments reveals coordinated oscillatory Ca2<sup>+</sup> responses. (A,B) Subcellular localization of the dual NRCG-GECO1 sensor in M. truncatula root hair cells. Confocal fluorescent images of NRCG-GECO1 sensor before (A) and during a spike (B) are represented as separated red, green, and merged images with respective bright fields. (C) Relative fluorescence intensity traces of NR-GECO1 (red) and CG-GECO1 (green) at 0.25 s intervals from ROIs selected around the nuclear (ROI n) and peri-nuclear cytoplasmic (ROI c) regions of a M. truncatula root hair treated with a 10−<sup>9</sup> M Nod Factor (+NF) solution. Dotted lines mark the initiation of a spike, n corresponds to the number of root hairs responding positively/total number of analyzed root hairs. Data were collected from two biological experiments. Scale bars represent 20µm (A,B) and 10µm (C).

laser and a 561 nm diode, respectively, and emission is detected in the respective fluorescence window ranges of 500–550 and 580–643 nm (Zhao et al., 2011). By using the optimum excitation parameters for G-GECO1 (488 nm excitation, Figure S6A), the green cytoplasmic fluorescence was clearly observed in the expected 500–550 nm emission window in NF-treated M. truncatula root hairs. However, a slight bleed-through of the G-GECO1 was observed in the R-GECO1 emission window of 580–643 nm protein. To avoid this, we therefore reduced the fluorescence window emission range for R-GECO1 from 580– 643 to 600–643 nm. Importantly, no bleed-through of the R-GECO1 was observed in the G-GECO1 500–550 nm emission window. Thus, the dual NRCG-GECO1 sensor expressed in transformed roots of M. truncatula composite plants was able to monitor concomitant nuclear and cytoplasmic Ca2<sup>+</sup> spiking in an individual NF-treated root hair (**Figure 2C**). Similar coordinated Ca2<sup>+</sup> responses in the two subcellular compartments were monitored in M. truncatula transgenic roots treated with CO4 (Figure S7A). To ensure the specificity of these responses, we used the dual NRCG-GECO1 sensor in the dmi1 and dmi3 mutants (Catoira et al., 2000; Arrighi et al., 2006). As expected, coordinated NF-elicited Ca <sup>2</sup><sup>+</sup> spiking was only observed in the dmi3 mutant (Figure S7). Taken together, the dual NRCG-GECO1 Ca2<sup>+</sup> sensor targets GECO fluorescent proteins to two subcellular compartments and is suitable for co-imaging of specific Ca2<sup>+</sup> responses in real time in two distinct subcellular compartments without fluorescence bleed-through.

### Ca2<sup>+</sup> Spiking Preferentially Initiates in the Nucleus of Medicago Root Hairs Responding to Symbiotic Nod Factors

The dual sensor made it possible to simultaneously visualize highly coordinated Ca2<sup>+</sup> signals in both nuclear and cytoplasmic compartments and therefore resolve the initiation dynamics of the symbiotic factor-induced nuclear Ca2<sup>+</sup> spiking. To increase temporal resolution, we reduced the image acquisition intervals to below 0.5 s. A qualitative frame-by-frame analysis illustrated the rise in both red and green fluorescence of respective GECOs during a single spike (**Figure 3A**). The selection of regions of interest (ROI) in the nucleoplasm and in different cytoplasmic-associated areas allowed us to precisely follow the progressive rise in fluorescence intensity of the sensors over time (**Figures 3B,C**). In both compartments, the start of a Ca2<sup>+</sup> spike was recognized by a steep increase in fluorescence intensity to baseline fluorescence levels (**Figures 3B–D**). These analyzes revealed a strong Ca2<sup>+</sup> rise in the nucleus and in the nuclear periphery, followed by sequential signal propagation toward the root hair tip (**Figures 3B,C**). To further evaluate the dynamics of Ca2<sup>+</sup> spiking initiation in the perinuclear region, we simultaneously recorded fluorescence data from nuclear NR-GECO1 and cytoplasmic CG-GECO1 from delimited ROI zones (**Figure 3B**) of individual NF-treated root hairs. The precise time for peak initiation (t1) for both GECO probes was compared by calculating the 1t1 value (CG-GECO1t1-NR-GECO1t1) for a large number of individual root hairs (**Figure 4A**). These analyses revealed that the majority of Ca2<sup>+</sup> spiking (65%) initiated primarily and significantly in the nuclear compartment, while only 2% initiated first in the cytosol. In 30% of the cases, it was not possible to clearly distinguish differences between the Ca2<sup>+</sup> spiking initiation sites.

To exclude intrinsic differences between the sensors in the targeted compartments we generated a reverse dual sensor expressing R-GECO1 in the cytoplasm and the G-GECO1 in the nucleus (Figures S1, S8). Using the dual NGCR-GECO1 sensor we confirmed the preferential initiation of Ca2<sup>+</sup> spiking in the nuclear compartment (**Figure 4A**). The time interval (1t1) for the initiation of a spike between the nuclear and cytoplasmic compartments was similar for both dual sensors (**Figure 4B**), with median values of 250 ms. Taken together, the dual GECO1 Ca2<sup>+</sup> sensors generated here allowed us to measure coordinated Ca2<sup>+</sup> responses in different subcellular compartments of individual cells (Movies S3, S4) and to demonstrate that Ca2<sup>+</sup> spiking primarily initiates in the nuclear compartment of root hair cells responding to rhizobial Nod factors.

### Ca2<sup>+</sup> Signal Dynamics in Both Nuclear and Cytoplasmic Compartments in *A. thaliana* Root Elongation Zone Cells

To investigate the use of the dual sensor in plant species other than M. truncatula, we stably transformed A. thaliana with the NRCG-GECO1.2 sensor (Figure S1) comprising improved versions of R-GECO1 and G-GECO1 (Zhao et al., 2011; Wu et al., 2013). Stable transgenic lines were propagated until the T3 generation under hygromycin selection and selected for expression of the dual sensor. CLSM imaging of A. thaliana roots demonstrated that CG-GECO1.2 and NR-GECO1.2 localized to the cytoplasmic and nucleoplasmic compartments, respectively (Figure S9). The stable transgenic line is phenotypically like wild type plants, indicating that the constitutive expression of the sensor does not interfere with cellular Ca2<sup>+</sup> signaling and homeostasis (Figure S9). In A. thaliana roots transformed with aequorin or cameleon YC3.6, several studies have reported that biotic and abiotic stimuli induce Ca2<sup>+</sup> release. As such, chitin, NaCl, ATP, and cold stress have been shown to induce Ca2<sup>+</sup> release in the root elongation zone (Kiegle et al., 2000; Tanaka et al., 2010; Keinath et al., 2015). However, the subcellular dynamics of these stimuli-induced Ca2<sup>+</sup> releases is lacking.

To monitor the dynamics of cytoplasmic vs. nucleoplasmic Ca2<sup>+</sup> release in response to chitooctaose (CO8), NaCl, ATP and cold stress, we focused our study on the root epidermal cells of the elongation zone. Treating 5 day-old roots with CO8 to a final concentration of 10−<sup>7</sup> M elicited an increase in cytoplasmic Ca2<sup>+</sup> within 2 min of application in 73% of the plants tested (8/11 plants) (**Figures 5A–C**). Ca2<sup>+</sup> release was detected only in the cytoplasmic region without any propagation to the nucleoplasm. The Ca2<sup>+</sup> signature was characterized by a sharp increase followed by progressive smaller oscillations returning to baseline levels over a period of 15 min (**Figure 5B**). The Ca2<sup>+</sup> signal was not detected after treatment with the buffer control (**Figures 5D–F**). Similarly, ATP to a final concentration of 100µM induced a specific cytoplasmic Ca2<sup>+</sup> release 7 min after application (Figure S10).

FIGURE 3 | Spatiotemporal dynamics of Ca2<sup>+</sup> spiking in nuclear and cytoplasmic compartments of a M. truncatula root hair responding to NFs. (A) A sequential frame-by-frame co-imaging (0.25 s intervals) of nuclear-targeted NR-GECO1 and cytosolic CG-GECO1 revealed coordinated responses during a single Ca2<sup>+</sup> spike. (B) Relative fluorescence intensity traces of NR-GECO1 (red) and CG-GECO1 (green) from ROIs selected around the nuclear (ROI n) and perinuclear cytoplasmic (ROI c) regions. Images on the right represent the maximum projection of respective fluorescence images merged with bright field. (C) Relative fluorescence intensity (0.3 s intervals) of CG-GECO1 in selected cytoplasmic regions of interest in the perinuclear (pn), middle (md), or root hair tip (tp) regions during a NF-triggered Ca2<sup>+</sup> spike. (D) Close view of selected fluorescence frames shown in (A). Time points representing baseline fluorescence [b in (A,B,D)], the start of the spike [t1 and arrow (A,B,D)], and subsequent spike rise time points [t2 and t3 in (A,B,D)] are indicated for NR-GECO1 and CG-GECO1. Scale bars represent 10µm.

The ATP-induced Ca2<sup>+</sup> signature was similar to the CO8 induced Ca2<sup>+</sup> release with a sharp peak followed by smaller oscillations, although their frequencies differed (0.6 min−<sup>1</sup> for CO8 and 1.94 min−<sup>1</sup> in the case of ATP) (Figure S10). In contrast to CO8 and ATP, which induced a specific cytoplasmic Ca2<sup>+</sup> response with an oscillatory pattern, NaCl and cold induced Ca2<sup>+</sup> release in both the cytoplasmic and nucleoplasmic compartments (**Figures 6**, **7**, Movies S5, S6). Each of these stimuli induced a transient Ca2<sup>+</sup> release first in the cytoplasm followed by a transient Ca2<sup>+</sup> increase in the nucleoplasm (**Figures 6**, **7**). However, the dynamics of each nucleoplasmic Ca2<sup>+</sup> increase differ in their amplitudes and timing of activation. Cold shock induced a transient nucleoplasmic Ca2<sup>+</sup> release 20 to 30 s after the cytoplasmic Ca2<sup>+</sup> response (**Figure 7**). In contrast, NaCl treatment induced a Ca2<sup>+</sup> response in the nucleus 2–3 s after cytoplasmic Ca2<sup>+</sup> release (**Figure 6**).

FIGURE 4 | Monitoring NF-elicited Ca <sup>2</sup><sup>+</sup> signaling with dual GECO sensors reveals a preferential Ca2<sup>+</sup> spike start in the nuclear compartment. Time values corresponding to the start of individual spikes (t1, illustrated in Figure 3B) were collected for each sensor and used to calculate the relative delay in the start of a spike. The time interval (1t1 in s), between peak start measured by sensors, was calculated by subtracting nuclear t1 from cytoplasmic t1-values (t1C-t1N). A <sup>1</sup>t1 value below 0 indicates a cytoplasmic-to-nucleus spike start (C→N), a 1t1-value close to 0 indicates coordinated cytoplasmic and nuclear spike start (C=N), and a positive 1t1-value indicates a nucleus-to-cytoplasmic spike start (C←N). These analyses were done with n = 68 (NRCG-GECO1) and n = 127 (NGCR-GECO1) individual spikes from 26 or 24 individual nuclei, respectively. (A) Frequency of the individual spikes showing C→N, C=N, and C←N patterns in the dual NRCG-GECO1 (dark gray) and NGCR-GECO1 (light gray) sensors. No significant differences were found between the sensors (p > 0.05, chi-square test of goodness-of-fit). Frequency distribution within the different categories (C→N, C=N, and C←N), for each dual sensor, is significantly different and not due to a random distribution (1/3 of observations in each category, p < 0.001, chi-square test of goodness-of-fit). (B) Time interval between spike start (1t1 in ms) measured for each dual sensor. Average/median values for NRCG-GECO1 and NGCR-GECO1 of 197/250 and 174/250 ms, respectively, are not significantly different. Box plots represent first and third quartile (horizontal box sides), minimum and maximum (outside whiskers), median (central lines), and mean (solid black circle).

Additionally, the amplitude of NaCl-induced nucleoplasmic Ca2<sup>+</sup> release is higher than the cytoplasmic Ca2<sup>+</sup> release, which contrasts with the cold-induced nucleoplasmic Ca2<sup>+</sup> signal, which is lower than the cytoplasmic Ca2<sup>+</sup> signal. The differences in timing of activation and amplitude of the Ca2<sup>+</sup> signals between both compartments strongly suggest that cold and NaCl induce distinct cytoplasmic and nucleoplasmic Ca2<sup>+</sup> signals. The nucleoplasmic response might be a consequence of the cytoplasmic Ca2<sup>+</sup> release, or induced independently. All together our results demonstrate that the dual sensor can effectively monitor Ca2<sup>+</sup> dynamics simultaneously in the cytoplasm and nuclear regions in A. thaliana roots cells, and identify various Ca2<sup>+</sup> signatures generated in response to biotic and abiotic stimuli.

### DISCUSSION

Recent years have witnessed powerful advances in geneticallyencoded fluorescent Ca2<sup>+</sup> sensors, notably with the development of single fluorescent protein GECIs, the GECOs (Zhao et al., 2011). These sensors were engineered to emit fluorescence at different wavelengths, thus enabling multicolor and multiparameter imaging in eukaryotic cells (Zhao et al., 2011; Ngo et al., 2014; Odaka et al., 2014; Waadt et al., 2017). Moreover, the GECOs have a higher sensitivity compared to the FRET-based cameleon YC3.6 (Keinath et al., 2015), which we confirmed in response to symbiotic factors in M. truncatula roots. The availability of multicolor Ca2<sup>+</sup> sensors with greater dynamic ranges increases the possibility of exploring Ca2<sup>+</sup> dynamics at the subcellular level and, notably, the interconnection between Ca2<sup>+</sup> releases from different compartments. As such, the relation between cytoplasmic and nuclear Ca2<sup>+</sup> signals, which remains unclear at the cellular level in planta, can now be addressed with increased resolution. Thus, we took advantage of the separate fluorescence emission windows of G-GECO (500–550 nm) and R-GECO (600–643 nm) to simultaneously assess Ca2<sup>+</sup> dynamics in those contiguous cellular compartments. By targeting each of them to either the cytoplasm or the nucleoplasm, we developed the new dual sensors, NRCG-GECO1, NGCR-GECO1, and NRCG-GECO1.2 that allowed the simultaneous monitoring of nuclear and cytoplasmic Ca2<sup>+</sup> dynamics in M. truncatula and A. thaliana root cells in response to biotic and abiotic stimuli. Despite the different association rates (Kon) of R-GECO and G-GECO (Zhao et al., 2011), these values are in the order of 10<sup>9</sup> -10<sup>15</sup> M−<sup>1</sup> ·s −1 , which means that binding of either sensor to Ca2<sup>+</sup> occurs faster than image capture. Therefore, any differences observed in the response times between the nucleus and the cytosol are intrinsically biological.

In legume symbioses, nitrogen-fixing rhizobial NF and mycorrhizal factors induce Ca2<sup>+</sup> oscillations in perinuclear and nuclear compartments. Although the importance of cytoplasmic Ca2+-mediated signal transduction is unknown, nuclear-localized Ca2<sup>+</sup> is essential to activate a nuclearlocalized Ca2<sup>+</sup> and calmodulin-dependent kinase (CCaMK), that assures downstream transcriptional regulation of endosymbiotic programs (Zipfel and Oldroyd, 2017). Due to the lack of

resolution, nuclear Ca2<sup>+</sup> oscillations appeared synchronized between the nucleoplasm and the adjacent cytoplasmic region (Capoen et al., 2011). Furthermore, the ion channels and pump required to generate the symbiotic factor-induced Ca2<sup>+</sup> oscillations were reported to localize to both the INM and ONM, although preferential localization of DMI1 to the INM was demonstrated (Capoen et al., 2011; Charpentier et al., 2016). Thus, the nuclear Ca2<sup>+</sup> oscillations could start simultaneously or specifically from either nuclear membrane (Capoen et al., 2011). Using the NRCG-GECO1 and NGCR-GECO1 sensors, we demonstrated that Ca2<sup>+</sup> release initiates inside the nucleoplasm in the majority of M. truncatula root hair cells elicited by NFs. This result unravels that the ion channels are activated first at the INM. This further suggests that the secondary messenger activating CNGC15 or DMI1 must either diffuse via the nucleopores or be produced inside the nucleus. Several nucleoporin mutants have been shown to be impaired in the generation of the nuclear Ca2<sup>+</sup> oscillations in a temperature sensitive manner (Kanamori et al., 2006; Saito et al., 2007; Groth et al., 2010). However, their precise roles in this regulation remains unclear. Although the nucleoporins have been proposed to regulate the trafficking of the ion channels to the INM, our results suggest that they might be involved in the trafficking of the secondary messenger or proteins required for the activation of nuclear calcium oscillations.

Our studies further revealed the advantage of the dual GECO sensor to monitor subcellular Ca2<sup>+</sup> dynamics in adjacent cytoplasm and nucleoplasm compartments of A. thaliana cells. By using a stably transformed dual sensor line we compared the dynamics of Ca2<sup>+</sup> simultaneously in the cytoplasm and nucleoplasm of epidermal cells of the root elongation zone in response to diverse stimuli. We demonstrated that different stimuli (CO8, NaCl, ATP and cold shock) can generate, within the same cell type, diverse Ca2<sup>+</sup> responses in the cytoplasmic and nucleoplasmic compartments. Notably, CO8 and ATP trigger cytoplasmic-associated oscillatory Ca2<sup>+</sup> releases that do not diffuse to the nucleoplasm, suggesting that Ca2<sup>+</sup> buffering proteins or Ca2<sup>+</sup> uptake mechanisms are rapidly recruited or activated, respectively, to avoid Ca2<sup>+</sup> diffusion into the nucleus. Interestingly, the ATP-induced Ca2<sup>+</sup> signals in root cells of the elongation zone differ from a previous study showing dual cytoplasmic and nuclear Ca2<sup>+</sup> releases (Krebs et al., 2012). This discrepancy might be due to the analysis of distinct root cells displaying specific ATP-induced Ca2<sup>+</sup> responses, as the same stimulus can elicit different cell type-specific Ca2<sup>+</sup> signatures (Kiegle et al., 2000). These findings further highlight the importance of monitoring Ca2<sup>+</sup> dynamics in the same cell type to aptly compare stimuli-induced Ca2<sup>+</sup> signatures and correlate the Ca2<sup>+</sup> signature to a biological response. In contrast to CO8 and ATP, cold- and NaCl-induced transient Ca2<sup>+</sup> signals first in the cytoplasm and then in the nucleoplasm. However, the NaCl- and cold-induced Ca2<sup>+</sup> signatures in both compartments differ in amplitude and timing of activation, suggesting the involvement of different channels and/or pumps. Notably, the

NaCl-induced nucleoplasmic Ca2<sup>+</sup> signal was observed earlier (within 2 s) than the cold-induced Ca2<sup>+</sup> signal (within 25 s), following the cytoplasmic Ca2<sup>+</sup> release. Moreover, The NaClinduced nucleoplasmic Ca2<sup>+</sup> signal had a higher amplitude than the cytoplasmic Ca2<sup>+</sup> signal, which contrasts with the cold-induced Ca2<sup>+</sup> signal, smaller in amplitude. These results support the induction of nuclear Ca2<sup>+</sup> signals independently of the cytoplasmic Ca2<sup>+</sup> release, as implied by a recent study regarding the NaCl-induced cytoplasmic/nuclear Ca2<sup>+</sup> dynamics (Huang et al., 2017). Alternatively, the cytoplasmic Ca2<sup>+</sup> release might be contributing to the generation of the nuclear Ca2<sup>+</sup> signal either by diffusion or by triggering a Ca2+-induced Ca2<sup>+</sup> release mechanism at the nuclear envelope.

In summary, the cytoplasmic and nucleoplasmic dual sensors developed in this work are powerful tools to analyze the subcellular Ca2<sup>+</sup> dynamics between the cytoplasm and nucleus in planta. These sensors will unequivocally be of interest in exploring Ca2+-mediated signaling in response to diverse stimuli in different plant species.

## MATERIALS AND METHODS DNA Constructs

The single and dual GECO Ca2<sup>+</sup> sensor constructs (Figure S1) were assembled using Golden Gate cloning. All GECO1.0 sensors were cloned under the control of the double cauliflower mosaic virus 35S promoter (2xp35S) and the 35S terminator in the Golden Gate compatible vector pCambiaCR1 1DsRed (Fliegmann et al., 2016) kindly provided by C. Rosenberg (LIPM). PCR amplification of the individual sequence modules was done using Phusion Taq high fidelity DNA polymerase (New England Biolabs) and respective DNA templates and primer pairs are listed in Table S1. Amplified DNA fragments flanked by BsaI and specific cohesive protruding ends were cloned into pBlueScript II (Agilent) and validated by sequencing before being used in Golden Gate assembly reactions (Engler and Marillonnet, 2014). All level 0 modules used in the dual NRCG-GECO1.2 sensor were synthesized by Life TechnologiesTM (ThermoFisher Scientific). All assembled Golden Gate binary vectors were verified by PCR, DNA sequencing and restriction digestion before transformation

into Agrobacterium. The cameleon Ca2<sup>+</sup> sensors NUP-YC2.1 and NLS-YC3.6 used here were previously generated in Sieberer et al. (2009) and Krebs et al. (2012).

### Plant Material and Bacterial Strains

Four M. truncatula cv Jemalong A17 lines were used in this study: the wild type and the symbiotic mutants nfp-2 (Arrighi et al., 2006) dmi1-1 and dmi3-1 (Catoira et al., 2000; Wais et al., 2000). M. truncatula seeds were scarified with sulfuric acid, then surface-sterilized prior to germination and grown on inverted soft Campbel agar plates (http://www.noble.org/ medicagohandbook). Arabidopsis seeds were surface-sterilized in 1.5% bleach for 15 min, followed by five washes in sterile water, and then plated in Murashige and Skoog (MS) solid medium (1% sucrose). After 3–5 days at 4◦C, plates were moved to a growth cabinet (23◦C, 16-h photoperiod, and 300 µmol·m−<sup>2</sup> ·s −1 light intensity) for germination and growth. The Escherichia coli DH5α strain was used for plasmid propagation, the A. rhizogenes Arqua1 strain (Quandt et al., 1993) strain for the generation M. truncatula composite plants and the A. tumefaciens GV3101 (Koncz and Schell, 1986) strain for stable transformation of A. thaliana.

### Generation and *in Vitro* Culture of *Agrobacterium*-Transformed *M. truncatula* and *A. thaliana*

A. rhizogenes-mediated transformation hairy root transformation of M. truncatula was performed as described in Boisson-Dernier et al. (2001) with minor modifications. Germinated seedlings were sectioned ∼3 mm from the root tip and inoculated with a 3–4 µL drop of A. rhizogenes bacterial suspension adjusted to an OD600nm of 1 in water. Inoculated seedlings were placed on Fahraeus/agar plates supplemented with 0.5 mM ammonium nitrate and kanamycin (25 mg.L−<sup>1</sup> ). Plates were partially sealed with parafilm to allow crucial gas exchanges and placed in a 20◦C growth chamber (16-h photoperiod and a light intensity of 70 µE·m−<sup>2</sup> ·s −1 ) for 1 week before transferring to 25◦C. About 2 weeks-post-inoculation, kanamycin-resistant composite plants were transferred to a beaker containing sterile water supplemented with 200 mg·L −1 augmentin (amoxicillin:clavulanic acid [5:1], GlaxoSmithKline), and incubated for at least 30 min, in order to reduce the growth of A. rhizogenes. Plants were rinsed twice before being transferred to square plates containing a nitrogen-free modified Fahraeus medium with 0.5% [w/v] phytagel (Sigma-Aldrich) and supplemented with 50 nM 2-amino-ethoxyvinyl-Gly (AVG), adapted for in vivo microscopic observations as described in Fournier et al. (2008). Transgenic roots were covered with a sterile, gas-permeable and transparent plastic film (Lumox film, Starsted) that allows the use of water-immersion objectives during imaging. Square plates with composite plants were slightly tilted to encourage the growth of the roots along the plastic film. The lower part of the square plate was wrapped in black plastic to protect the roots from light. M. truncatula ROCs expressing the NR-GECO1 sensor were obtained from transgenic composite plants as described by Boisson-Dernier et al. (2001).

ROCs were cultured on M medium (Boisson-Dernier et al., 2001) and propagated vertically until used for microscopy observations.

A. thaliana stable transgenic plants were generated using the floral dip method (Clough and Bent, 1998) and selected in 50 µg·mL−<sup>1</sup> hygromycin B.

### Selection and Symbiotic Treatments of *M. truncatula* Transformed Roots

Kanamycin-resistant A. rhizogenes-transformed composite plants or ROCs transformed with cameleon or GECO1 constructs were screened for expression of the respective fluorescent sensors using fluorescence stereomicroscopes (Leica MZFLIII and Zeiss axiozoom V16) and a Zeiss Axiophot epifluorescent microscope. Selection of transgenic roots was done using filters adapted to detect yellow/green YFP/GFP (excitation: 470–510 nm; emission: 525–575 nm) or red mApple (excitation: 577–587 nm, emission: 620–680 nm) fluorescent proteins from the different cameleon or GECO1 sensors. Only composite plants with fluorescent transgenic roots were transferred to plates containing nitrogen-free modified 0.5% [w/v] phytagel Fahraeus medium for imaging 3–5 days later. Transgenic roots lying on the medium and covered by a gaspermeable film were treated with 1 to 2 mL of a freshly diluted aqueous solution of purified S. meliloti NFs (10−<sup>9</sup> M) or CO4 (10−<sup>5</sup> M or 10−<sup>7</sup> M) that were added to plates between the plastic film and the medium with the help of a micropipette. In the case of ROCs, imaging was performed with excised root sections that were mounted between a slide and a cover slip. Confocal imaging was performed before treatments to assess the background fluorescence levels, and then initiated about 5 min after treatment for a total period of up to 1 h.

### Confocal Imaging of Calcium Responses in Transgenic *M. truncatula* and Arabidopsis Roots

The comparison between the nuclear NR-GECO1, the NUP-YC2.1 and the NLS-YC3.6 Yellow Cameleon sensors was performed using a Leica TCS SP2 AOBS confocal laser-scanning microscope equipped with a long-distance 40x water-immersion objective (HCX Apo L 0.80). The 458 nm argon laser was used to excite the CFP of the cameleon sensors (pinhole diameter at 4 Airy units). Fluorescence emission was collected in the 470– 500 nm range for CFP and 530–570 nm range for YFP. Images were acquired at 5 s intervals from 5 to 60 min either before or after treatment with a 512 × 512 pixels resolution. The 561 nm diode was used to excite the mApple red fluorescence protein of the NR-GECO1 sensor (pinhole diameter at 3 Airy units) and fluorescence was visualized in the specific 580 to 643 nm emission window. The laser intensity used for the cameleon probes (80% power setting, Sieberer et al., 2009) was reduced to 20% for monitoring Ca2<sup>+</sup> responses with the NR-GECO1 sensor in order to avoid photo bleaching of the red fluorescence during long acquisitions. Confocal imaging of transgenic roots expressing the dual NRCG-GECO1 and NGCR-GECO1 sensors was done using a beam-splitter Leica TCS SP8 AOBS confocal laser-scanning microscope equipped with a 25x water-immersion objective. The 488 nm argon laser was used to excite the G-GECO1 and the 561 nm diode to excite the R-GECO1. Fluorescent images were recorded simultaneously for NRCG-GECO1 and NGCR-GECO1, using emission windows of 500– 550 nm and 600–643 nm for G-GECO1 and R-GECO1 signals, respectively. Fluorescence emission ranges were specifically set for each fluorescent protein in order to avoid overlap between the G-GECO1 and R-GECO1 signals, as detailed in Figure S6. Images were acquired in 200 to 500 ms intervals with a scanning resolution of 512 × 128 pixel. Images were acquired using the Leica confocal software.

The subcellular localization of the GECO1.2 Ca2<sup>+</sup> reporters was assessed in 6-day-old roots by CLSM using a Zeiss LSM 780 equipped with a 25x/0.8 water objective. G-GECO1.2 was excited at 488 nm and the emitted light was captured between 463 and 500 nm. R-GECO1.2 was excited at 561 nm and emitted fluorescence captured in the 571–640 nm range. For Ca2<sup>+</sup> imaging, plants were analyzed 5–6 days after germination and carefully mounted in a perfusing (cold treatment) or nonperfusing chamber. Perfusion of cold media was performed using a RC-21BR chamber (Warner Instruments). For ATP, CO8, and NaCl treatments, the elicitors were applied at a 10x concentration to an open chamber made with cover glass and sealed with vacuum grease. After mounting in liquid MS (1% sucrose), samples were incubated at room temperature for at least 15 min before imaging. Ca2<sup>+</sup> imaging was performed using a Nikon ECLIPSE FN1. The fluorophores were excited at a wavelength of 470 ± 24 nm. Emitted fluorescence was separated by an image splitter and passed through an emission filter of 520 ± 40 nm for G-GECO1.2, and 632 ± 60 nm for R-GECO1.2 (Optosplit, Cairn Research, UK). Images were collected every 2 or 3 s.

### Image Analysis

Processing, final cropping, and mounting of the Medicago images were performed using Fiji. The fluorescence, merged images and Supplementary time-lapse movies were obtained after applying a median filter of 2 pixel radius to fluorescence image series. Intensity data was calculated from selected ROI, delimited in individual root hairs within the nuclear region (for nuclearlocalized sensors) or tailored in the perinuclear, middle or root hair tip regions (for cytoplasmic sensors). To quantify relative Ca2<sup>+</sup> spiking amplitudes, the average fluorescence for each ROI was calculated and used to determine the SNR of individual spikes collected from independent roots and biological experiments. The SNR of a given spike measures the difference between the fluorescence baseline and the maximum fluorescence peak value. The SNR values of individual spikes were obtained by calculating the 1F/F (Fmax-Fmin/Fmin) or 1R/R (Rmax-Rmin/Rmin), in which Fmax and Rmax correspond to the maximum fluorescence value or ratio change of a spike and Fmin and Rmin correspond to the data point before the onset of a spike.

For image processing of the Arabidopsis data the following steps were conducted using ImageJ 1.48v: background subtraction, registration using MultiStackReg v1.45 (http:// bradbusse.net/sciencedownloads.html), and application of a lookup table. Image data were obtained from processed images

using Time Series Analyzer V3\_2 (https://imagej.nih.gov/ij/ plugins/time-series.html). Normalized datasets (1F/F) were calculated as (F–F0)/F0, where F<sup>0</sup> represent the average of at least 2 min of baseline values (before treatment).

### Statistical Analyses

Statistical analyses were performed using R (http://r-project. org). Data were evaluated for normality with the Shapiro-Wilk test and homogeneity of variance with Levenne or Bartlett tests. Parametric (ANOVA) or non-parametric (Kruskal-Wallis) statistical tests were used, respectively, for normally-distributed data and non normally-distributed data. Signal periodicity values of calcium sensors were first transformed into normality using Box-Cox (λ = 0.0606), before statistical analysis using oneway ANOVA followed by Tukey honest difference (HSD) tests. Statistical analysis of signal-to-baseline ratio was done using the non-parametrical Kruskal-Wallis test. For the dual GECO sensors, statistical analyses of the start of a calcium spike was done using a chi-square test (p < 0.05) to (i) compare the frequency of observations in each category to an expected frequency of 1/3 per category if the distribution was random and (ii) to compare the number of observations per category for each sensor grouped in a contingency table.

### AUTHOR CONTRIBUTIONS

MyC and FdC-N: conceived the project; AK, NL, MyC, and FdC-N: designed the experiments; AK: performed the M. truncatula and NL: the A. thaliana experiments; MiC: contributed to the

### REFERENCES


NLS-R-GECO1 in ROCs experiments; AK and NL: prepared the figures; AK, NL, MyC, and FdC-N: wrote the manuscript; MyC and FdC-N: are co-responsible for the work.

### ACKNOWLEDGMENTS

We thank David Barker and Joelle Fournier for critical reading of the manuscript, Christian Mazars and Martin Parniske for valuable discussions, Lisa Frances for contributing to the cloning of the R-GECO1 version, Fabienne Maillet for kindly providing purified S. meliloti NFs, and Charles Rosemberg for the pCambia Golgen Gate vectors. The CMV-NLS-R-GECO (Addgene plasmid #32447) and CMV-G-GECO1.0 (Addgene plasmid #32462) were generated and provided by Robert Campbell via Addgene, while the YC3.6 construct was kindly provided by Melanie Krebs and Karin Schumacher. AK was supported by a Ph.D. grant of the French Ministry of Education and Research, NL was supported by a Ph.D. studentship from the John Innes Foundation. MyC was supported by BBSRC (BB/P007112/1) and John Innes Centre, and FdC-N lab work was supported by the ANR COME-IN (grant no. ANR-14-CE35-0007-01) and TULIP (grant no. ANR-10-LABX-41) grants.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018. 00245/full#supplementary-material

components of a nod factor transduction pathway. Plant Cell 12, 1647–1666. doi: 10.1105/tpc.12.9.1647


considerations for optogenetic applications. ACS Chem. Neurosci. 4, 963–972. doi: 10.1021/cn400012b


**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 Kelner, Leitão, Chabaud, Charpentier and de Carvalho-Niebel. 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.

# W342F Mutation in CCaMK Enhances Its Affinity to Calmodulin But Compromises Its Role in Supporting Root Nodule Symbiosis in Medicago truncatula

#### Edgard Jauregui<sup>1</sup> , Liqun Du1,2, Cynthia Gleason<sup>3</sup> and B. W. Poovaiah<sup>1</sup> \*

<sup>1</sup> Laboratory of Molecular Plant Science, Department of Horticulture, Washington State University, Pullman, WA, United States, <sup>2</sup> College of Life and Environmental Sciences, Hangzhou Normal University, Hangzhou, China, <sup>3</sup> Department of Plant Pathology, Washington State University, Pullman, WA, United States

### Edited by:

Jianping Wang, University of Florida, United States

#### Reviewed by:

Muthu Venkateshwaran, University of Wisconsin–Platteville, United States Dong Wang, University of Massachusetts Amherst, United States Keiko Yoshioka, University of Toronto, Canada

> \*Correspondence: B. W. Poovaiah poovaiah@wsu.edu

#### Specialty section:

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

Received: 15 June 2017 Accepted: 24 October 2017 Published: 16 November 2017

#### Citation:

Jauregui E, Du L, Gleason C and Poovaiah BW (2017) W342F Mutation in CCaMK Enhances Its Affinity to Calmodulin But Compromises Its Role in Supporting Root Nodule Symbiosis in Medicago truncatula. Front. Plant Sci. 8:1921. doi: 10.3389/fpls.2017.01921 The calcium/calmodulin-dependent protein kinase (CCaMK) is regulated by free Ca2<sup>+</sup> and Ca2+-loaded calmodulin. This dual binding is believed to be involved in its regulation and associated physiological functions, although direct experimental evidence for this is lacking. Here we document that site-directed mutations in the calmodulinbinding domain of CCaMK alters its binding capacity to calmodulin, providing an effective approach to study how calmodulin regulates CCaMK in terms of kinase activity and regulation of rhizobial symbiosis in Medicago truncatula. We observed that mutating the tryptophan at position 342 to phenylalanine (W342F) markedly increased the calmodulin-binding capability of the mutant. The mutant CCaMK underwent autophosphorylation and catalyzed substrate phosphorylation in the absence of calcium and calmodulin. When the mutant W342F was expressed in ccamk-1 roots, the transgenic roots exhibited an altered nodulation phenotype. These results indicate that altering the calmodulin-binding domain of CCaMK could generate a constitutively activated kinase with a negative role in the physiological function of CCaMK.

Keywords: Ca2+/calmodulin-dependent protein kinase, calcium, calmodulin, symbiosis, Medicago truncatula

### INTRODUCTION

Calcium/calmodulin (Ca2+/CaM)-mediated signaling plays important roles in sensing and transducing environmental stimuli and developmental cues (Du and Poovaiah, 2005; Poovaiah et al., 2013), plant immune responses (Du et al., 2009) and mutualistic plant–microbe interactions (Gleason et al., 2006; Yuan et al., 2017). One important effector protein of this Ca2+/CaMmediated signaling in plants is the Ca2+/CaM-dependent protein kinase, CCaMK (Patil et al., 1995). CCaMK is nuclear localized and lies downstream of the cellular calcium spiking response that occurs in the symbiosis pathway. After activation by Ca2+/CaM, CCaMK phosphorylates CYCLOPS or IPD3 in Medicago truncatula, which are required for symbiosis (Yano et al., 2008).

CCaMK has been well studied not only because of its unique protein structure, but also due to its critical role in the symbiotic interaction between leguminous plants and bacteria and/or fungi (Patil et al., 1995; Ramachandiran et al., 1997; Harper et al., 2004). The protein structure of CCaMK

**96**

contains a catalytic domain followed by a CaMbinding/autoinhibitory domain and three EF hand motifs (Sathyanarayanan et al., 2001; Sathyanarayanan and Poovaiah, 2002; Singh and Parniske, 2012). CCaMK is regulated by free Ca2<sup>+</sup> which binds to the EF hand motifs and Ca2+-loaded CaM, which binds to the autoinhibitory domain (Lévy et al., 2004; Tirichine et al., 2006; Pumplin et al., 2010). This two-step regulatory mechanism allows CCaMK to become active and promote phosphorylation of target proteins that regulate root nodule symbiosis and mycorrhizal arbuscule development (Wang and Poovaiah, 1999; Sathyanarayanan and Poovaiah, 2002; Parniske, 2004). Both regulatory domains are important during symbiotic response (Tirichine et al., 2006; Pumplin et al., 2010). Previous reports demonstrated that expressing only the kinase domain (KD) of this protein or expressing a mutant of the threonine at position 271 (Thr271), located in the KD, resulted in the formation of spontaneous nodules in the absence of bacteria in Medicago roots (Sathyanarayanan et al., 2001; Gleason et al., 2006; Tirichine et al., 2006). This indicates that CCaMK's autoinhibitory domain and the EF-hand motifs negatively regulate the kinase activity. Another report has also shown the importance of the EF hand motifs in negative regulation of CCaMK. This study indicated that the EF hand motifs bind free Ca2+and keep the CCaMK inactive at basal Ca2<sup>+</sup> concentrations (Shimoda et al., 2012). Conversely, if free Ca2<sup>+</sup> is blocked from binding to the EF hand motifs, CCaMK becomes auto-activated (Hayashi et al., 2010; Miller et al., 2013). Recent reports have demonstrated that CCaMK might be regulated by DELLA proteins (Jin et al., 2016) and/or TOR (the target of rapamycin) protein kinase (Nanjareddy et al., 2016) which are essential for symbiotic rhizobial pathway. Furthermore, CCaMK gene is very well conserved among phytozome species (Wang et al., 2015). CCaMKs have been found in peanuts (Arachis hypogaea) and tomato (Solanum lycopersicum), where it may play roles in the symbiotic pathway (Peng et al., 2017) and/or disease resistance (Wang et al., 2015).

Generally, the key to the activation of the kinase requires the release of the protein's own autoinhibition, which can be accomplished by the binding of CaM to the autoinhibitory/CaMbinding domain (Hayashi et al., 2010; Antolín-Llovera et al., 2012). In 2011, researchers using an auto activated form of CCaMK concluded that CaM-binding is essential for nodulation development during rhizobial symbiosis, but it is not required for fungal symbiosis (Horváth et al., 2011). However, since this auto-activated form could no longer interact with either CaM or free Ca2+, it is critical to study the relevance of the CaMbinding alone during the activation of CCaMK (Takeda et al., 2012). Furthermore, CCaMK protein structure is unique to plants and understanding of its activation and regulation is far from clear. There are reports which have focused in its KD and visinin-like domain and have determined their importance of specific amino acids that regulate its kinase activity in addition to calmodulin and Ca2<sup>+</sup> binding. Studies on the relationship between calmodulin and Ca2<sup>+</sup> binding and kinase activity could reveal how this kinase interprets the Ca2<sup>+</sup> signature during the establishment of plant–microbe symbioses.

In this study, we used a site-directed mutagenesis approach to alter the Calmodulin-binding/autoinhibitory domain of CCaMK in order to observe any variation of the CaM-binding capacity of this protein kinase in the presence of Ca2+. We generated a series of mutations at a critical residue in the autoinhibitory domain (Trp-342) and measured the effects of the mutations on CaM-binding capacity. We found that the mutant W342F has increased CaM-binding capacity, even in the absence of Ca2+. Interestingly, roots expressing W342F developed nodules after rhizobial inoculation, but half of these nodules were smaller than normal and were poorly colonized by bacteria.

### MATERIALS AND METHODS

### Site-Directed Mutagenesis

Site-directed mutagenesis of CCaMK was performed using a high fidelity KOD DNA polymerase enzyme to construct the desired plasmid with the mutations, following protocol as described (Liu and Naismith, 2008). The mutated CCaMK cds were confirmed by sequencing.

### Expression and Purification of CCaMK and Site-Directed Mutants

The full-length of CCaMK and its site-directed mutants were cloned into the bacterial expression system pET28b and transformed into Escherichia coli strain BL21 (DE3)/pLysS. The bacteria carrying the above plasmids were grown in LB liquid media containing kanamycin at 37◦C until OD<sup>600</sup> of the culture reached 0.5 units. Once the liquid culture reached this optimal density, 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to induce the recombinant protein. After 3 h induction, cells were harvested and broken using lysozyme treatment (1 mg/ml) followed by sonication. The recombinant protein was purified with Ni-NTA agarose affinity beads (Qiagen) as described in the manufacturer's manual. The purified proteins were dialyzed against buffer containing 40 mM Tris pH 7.6, 1 mM dithiothreitol (DTT), 1 mM EDTA, and 10% ethylene glycol. Dialyzed proteins were quantified by Bradford assay and stored at −80◦C with 15% glycerol.

### CaM-Binding Assays

The CaM-2 from Arabidopsis conjugated with horseradish peroxidase (CaM-HRP) was used to study the CaM-binding property of CCaMK mutants. The induced proteins of CCaMK and its mutants in pET28b were separated by SDS–PAGE (15%) and transferred onto PVDF membrane. The membrane was blocked in binding buffer (10 mM Tris pH 7.5, 150 mM NaCl, 1 mM CaCl2) containing 5% non-fat dry milk for 1 h at room temperature, then incubated with milk containing binding buffer supplemented with AtCaM2-HRP (1:1000 dilution) for 1 h at room temperature. The membrane was then washed three times in binding buffer for 10 min each. To detect the CaM signal, the BM chemiluminescence Western blotting kit (Roche Applied Science) was used according to instructions from manufacturer.

### Autophosphorylation Assays

fpls-08-01921 November 14, 2017 Time: 15:48 # 3

The autophosphorylation assay was performed in 10 µl reaction mixture using 0.4 µg of CCaMK protein and its mutated versions. The reaction buffer contained 50 mM HEPES pH 7.5, 10 mM magnesium acetate, 1 mM DTT, 10 µM ATP and 0.5 µCi/µl [γ-<sup>32</sup>P] ATP, in the presence of 5 mM EGTA with or without 1 µM of bovine brain CaM (Sigma); and 0.5 mM of CaCl<sup>2</sup> with or without bovine CaM. Samples were incubated at 30◦C for 30 min. To stop the reaction, SDS–PAGE sample buffer was added, followed by boiling the samples for 2 min. Samples were separated by a 12.5% SDS–PAGE. Protein gel was then dried and exposed to autoradiography film (Kodak).

### Substrate Phosphorylation Assays

All of the in vitro kinase assays used 0.4 µg of purified protein in a 10 µl total volume. The buffer contained 50 mM HEPES pH 7.5, 10 mM magnesium acetate, 1 mM DDT, 100 µM ATP, and 0.5 µCi/µl [γ-<sup>32</sup>P] ATP in the presence of 5 mM EGTA or 0.5 mM CaCl<sup>2</sup> or 0.5 mM CaCl<sup>2</sup> with 1 µM of bovine brain CaM (Sigma). In order to determine substrate phosphorylation, two micrograms of a bovine myelin basic protein (MBP) was used as substrate. Reactions were stopped by adding SDS–PAGE sample buffer and then boiled for 1 min in water bath. Subsequently, sample reactions were analyzed in SDS–PAGE and gel as dried. The difference of substrate phosphorylation patterns at different conditions was observed by exposing gel to the Kodak autoradiography film.

### Hairy Root Transformation

The generation of transgenic roots was performed using a binary vector (pDL28-DPR) containing the native promoter of CCaMK and a red fluorescent protein encoding region from Discosoma sp. (DsRed) driven by Arabidopsis ubiquitin-10 promoter. The full-length CCaMK and mutant gene (W342F) were digested with SpeI and SalI restriction enzymes and ligated into the binary vector. The verified constructs were transformed into Agrobacterium rhizogenes strain K599 to transform the rootless plantlets of wild-type and ccamk-1 mutant of M. truncatula A17. The control groups used in this experiment were wildtype M. truncatula explants were transformed with the binary vector without any gene (empty vector). In the other control group, the ccamk-1 mutant plants were introduced with the empty vector. Our experimental groups were as follows: The third and fourth group used ccamk-1 plants which were transformed with the binary vector carrying the full-length CCaMK gene and W342F mutant gene, respectively. The seedlings were grown on buffered nodulation medium (BNM) agar plates with Kanamycin selection. After 3 weeks, the plants were screened for transgenic roots by using a Kodak imaging system (Model 4000MM) with corresponding filters for the expression of DsRed fluorescent protein. For the nodulation experiments, the plants with transgenic roots were inoculated with Sinorhizobium meliloti 2011 carrying a green fluorescent protein (GFP) reporter gene and transferred to sterilized growth pouches (Mega International). The inoculated seedlings were watered with 1/10 dilution of nitrogen-deprived nutrient media (BNM) and grown under 16/8-h light/dark cycle at 21◦C for 28 days. At 28-days post-inoculation, nodules were counted and observed under a Leitz fluorescent stereomicroscope using optical filters for GFP. Green fluorescence verified the presence of living rhizobia inside the developed nodule.

### Confocal Microscopy

Selected nodules were analyzed by Zeta 510 Meta confocal microscope in order to determine the location of the bacteria in the nodules formed on the roots of ccamk-1 complemented with wild-type and mutated version of CCaMK (W342F). Nodules were excised in half and were stained with Calcofluor white, a special fluorescent stain that allows observation of cell walls of tissues due to its binding to cellulose and chitin. For this fluorescent stain, we used excitation filter of 365 nm and emission filter 420 nm in 10× microscope lenses. To determine GFP, we used excitation filter 380 nm and emission filter 480 nm. Merged pictures were obtained and subsequently analyzed.

## RESULTS

### A W342F Mutation in the CCaMK Autoinhibitory Binding Domain Can Positively Increase Its CaM-Binding Capacity

The CaM-binding domain in CCaMK interacts with CaM, therefore, site-directed mutations were generated in the CaMbinding/autoinhibitory domain to investigate whether the CaM-binding capacity and kinase activity are correlated. Four residues in the calmodulin-binding/autoinhibitory domain were selected: F327, L333, I338, and W342. Mutations in the latter one produced some interesting results. W342 was mutated to five different residues with less mass but similar hydrophobic property (**Figure 1A**). These mutants were tested for CaM-binding capability. The CaM-binding assay showed that one mutant, W342F, exhibited increased CaM-binding capacity as compared to wild-type CCaMK. W342L and W342I showed less binding to CaM in comparison to CCaMK and W342F. The mutant W343V no longer interacted with CaM, however, W342A mutant was able to bind to CaM with a similar capacity as wild-type CCaMK (**Figure 1B**).

### Significance of W342F Mutant on Its Kinase Activity

In CCaMK, the binding of CaM to the autoinhibitory domain releases its autoinhibition and activates its kinase activity. Since the W342F mutant showed an increase in CaM-binding ability, we hypothesized that the kinase activity of this mutant could be increased. Therefore, we specifically focused on this potential "gain of function" mutant, and an in vitro kinase assay was carried out.

Normally, the autophosphorylation of CCaMK is very low in the absence of Ca2+, and it increases drastically once Ca2<sup>+</sup> is added to the reaction mixture. Similar to previously published reports, we found that the autophosphorylation of

the wild-type CCaMK was decreased after the addition of CaM (Takezawa et al., 1996; Miller et al., 2013). Interestingly we found the W342F mutant was autophosphorylated even in the absence of Ca2+. The W342F mutant was also autophosphorylated in the presence of CaM and 5 mM EGTA. In the presence of Ca2+, both wildtype CCaMK and the W342F mutant underwent autophosphorylation. However, the autophosphorylation intensity of W342F mutant was reduced in comparison to wild-type CCaMK (**Figure 2A** and Supplementary Figures 1, 2). It seems that the activation of autophosphorylation in W342F mutant is no longer dependent on Ca2+, on the contrary, it responds to the Ca2<sup>+</sup> signal in an opposing manner as compared to wild-type CCaMK.

We used bovine MBP as substrate to test the substrate phosphorylation activities of CCaMK. These reactions were performed under three different conditions: (1) in the absence of Ca2+, (2) in the presence of Ca2+, and (3) in the presence of Ca2<sup>+</sup> and CaM. The W342F mutant phosphorylated the substrate in all three conditions, with a slightly increased intensity in the presence of both Ca2<sup>+</sup> and CaM. Nevertheless, in comparison to CCaMK, this mutant showed a clear reduction in substrate phosphorylation activity even in the presence of Ca2<sup>+</sup> and CaM (**Figure 2B** and Supplementary Figures 1, 2). Thus, in the W342F mutant substrate phosphorylation is Ca2+-independent and also CaM-independent, and barely responds to Ca2<sup>+</sup> and CaM inputs.

### W342F Negatively Regulates Root Nodule Symbiosis

With the clear indication that the phosphorylation activity of W342F mutant differs from wild-type CCaMK in its biochemical property, we wanted to test the effect of the W342F on nodule formation. The loss-of-function ccamk-1 Medicago plants were transformed with W342F or CCaMK by Agrobacterium rhizogenes transformation. The transformed roots were inoculated with S. meliloti 2011 tagged with a GFP. The ccamk-1 plants inoculated with S. meliloti 2011-GFP did not produce nodules. The ccamk-1 mutant complemented with wild-type CCaMK produced pink mature nodules at 28-days post-inoculation with S. meliloti 2011-GFP. Interestingly, at 28-days post-inoculation, the ccamk-1 plants transformed with W342F were able to produce nodules, indicating that the W342F mutant can complement the ccamk-1 plants (**Figures 3**, **4**). The total number of nodules produced on the roots of ccamk-1 mutant complemented with the W342F mutant (ccamk-1 + W342F) was comparable to the number of nodules on roots of ccamk-1 mutant

FIGURE 2 | Kinase activity assays of W342F and CCaMK. Autophosphorylation was performed at 30◦C for 30 min. (A) CCaMK did not significantly phosphorylate itself in the absence of Ca2+. In the presence of Ca2+, CCaMK showed drastically increased autophosphorylation, which remarkably decreased once calmodulin was added. The W342F mutant showed very high levels of autophosphorylation in the absence of Ca2<sup>+</sup> as compared to CCaMK, indicating that the mutant behaves in a Ca2+-independent manner. Upper: Autoradiograph film demonstrating the autophosphorylation activity of CCaMK and W432F. Lower: Coomassie staining of the purified proteins from E. coli (BL21) expressing the pET28-CCaMK and W342F. (B) Substrate phosphorylation of CCaMK and W342F. The substrate phosphorylation of CCaMK and W342F was determined in the absence (5 mM EGTA), or presence of Ca2<sup>+</sup> (0.5 mM CaCl2), or Ca2<sup>+</sup> and calmodulin together (0.5 mM CaCl<sup>2</sup> + 1 µM calmodulin). 2 µg of bovine myelin basic protein (MBP) was used as the substrate. Substrate phosphorylation of W342F in the absence and presence of Ca2<sup>+</sup> did not follow a similar pattern compared to CCaMK. This mutant displayed a Ca2+- and/or calmodulin-independent substrate phosphorylation capacity, but the maximum level of MBP phosphorylation intensity catalyzed by W342F was remarkably lower than substrate phosphorylation intensity catalyzed by CCaMK in the presence of Ca2<sup>+</sup> and calmodulin. The kinase activity assays were repeated three times. Upper: Autoradiograph film showing the substrate phosphorylation activity of CCaMK and W432F; Lower: Coomassie staining of MBP substrate (2 µg).

complemented with wild-type CCaMK (ccamk-1 + CCaMK) (**Figure 3A**).

Although the total number of nodules was equivalent between ccamk-1 + CCaMK and ccamk-1 + W342F plants, the number of mature, pink nodules on the ccamk-1 + W342F plants roots was significantly less than that on the ccamk-1 + CCaMK roots (p-value = 0.01). The ccamk-1 + CCaMK roots produced 100% mature, pink nodules after inoculation with rhizobia. Approximately 50% of the nodules on the ccamk-1 + W342F plants were small abnormal nodules. These small nodules were not observed in the wild-type CCaMK inoculated with rhizobia (p-value = 0.00069) (**Figure 3B**). The abnormal nodules were white, indicating poor colonization by rhizobia (**Figure 5A**). Confocal microscopic analysis revealed that GFP-labeled S. meliloti 2011 was inside these small nodules but the distribution patterns of the bacteria in the nodule differed from that of the normal mature nodules (**Figure 5B**).

### DISCUSSION

Extensive studies have been conducted by using site-directed mutagenesis as a powerful tool to manipulate the coding region of various genes. We took advantage of this wellknown strategy to modify a residue on the autoinhibitory/CaMbinding domain in order to alter its well-defined structure and modify the CaM-binding capacity of CCaMK. Altering the CaM-binding capacity of CCaMK should provide a better understanding of the regulation of CCaMK's autoinhibitory domain.

The CaM-binding domain contains particular structural features: it forms an α helix with a net positive charge and two hydrophobic anchor residues on one side. These characteristics allow CaM to bind easily in the presence of Ca2<sup>+</sup> (Yang et al., 2007). Typically, CaM contains two lobes at its N- and C-terminals that are connected by a helix linker, and in the presence of Ca2+, these two lobes will wrap the CaM-binding domain (Tidow and Nissen, 2013). However, there have been reports that some CaM-binding domains showed increased affinity to either of the terminals (Snedden and Fromm, 2001; Liu et al., 2012). This suggests that manipulating the α helix motif can modify its capacity to bind to CaM (Jang et al., 2011; Gifford et al., 2013).

One amino acid in the autoinhibitory/CaM-binding domain of CCaMK (W342) was replaced by amino acids with similar hydrophobicity but smaller masses in order to alter the α helix structure of the CaM-binding domain. Our data showed that reducing the mass of the selected amino acid could alter the capacity of CCaMK to interact with CaM. The mutant W342F showed an increase of CaM-binding in comparison to CCaMK, in addition, mutations W342L and W342I were able to interact with CaM but at lower levels. The mutant W342V did not interact with CaM. Interestingly, W342A mutant was able to bind to CaM similar to wild-type CCaMK even though its mass is relatively lower than that of the valine residue. Therefore, there is no direct correlation between mass of the residue at amino acid 342 and the CCaMK's CaM-binding activity.

W342F mutant had an increase of CaM-binding in comparison to CCaMK. Hence, we postulated that the kinase activity of the mutant could be significantly affected. In vitro analysis of CCaMK revealed that autophosphorylation of wild-type CCaMK is Ca2+-dependent, and it is able to phosphorylate its substrate only in the presence of Ca2<sup>+</sup> and CaM. In this study, we observed that W342F mutant was autophosphorylated and could phosphorylate MBP substrate in the absence of Ca2<sup>+</sup> and CaM. This strongly suggests that the activity of W342F is Ca2+- and CaM-independent. Even though W342F was able to phosphorylate MBP in vitro in the presence of Ca2<sup>+</sup> and CaM, but the mutant barely responded to Ca2<sup>+</sup> and CaM stimulation, the intensity of maximum level of MBP phosphorylation was reduced in comparison to that catalyzed by wild-type CCaMK. These results clearly indicate that the CaM-binding domain is critical for the activation of CCaMK. Also, they suggest that CCaMK protein structure may be critical for its regulation because of the possible structural changes when Ca2<sup>+</sup> and CaM bind. The proximity of W342 to two well-known autophosphorylation sites S343 and S344 might suggest that of Trp-342 could possibly play a role during the autophosphorylation of these two sites (Routray et al., 2013; Jauregui et al., 2017).

Our physiological results demonstrated that the W342F mutation has a negative effect on the nodule morphology, since a significant portion of the nodules observed in the ccamk-1 + W342F plants were abnormal. These abnormal nodules were smaller and appeared to be infected with fewer rhizobia. This may correlate with the compromised substrate phosphorylation. The W342F mutant exhibited substrate phosphorylation in both the presence and absence of Ca2<sup>+</sup> and CaM, but the intensity of the phosphorylation was less than that of the wild-type in the presence of Ca2<sup>+</sup> and CaM. This suggests that even though this mutant could be activated in the absence and presence of Ca2<sup>+</sup> or CaM, it may not be able to sufficiently phosphorylate the target protein in order to obtain a well-developed, mature nodule in the root. In addition, our mutants did not produce any spontaneous nodules in the absence of bacteria (data not shown). This suggests that enhancing autophosphorylation and substrate phosphorylation in the absence of Ca2<sup>+</sup> and CaM does not result in an "autoactive" kinase that can stimulate a pathway toward spontaneous nodulation. Therefore, we hypothesize that a threshold of kinase activity must be reached and this could

trigger the production of normal, mature nodules. Since our mutant showed some reduction in the intensity of substrate phosphorylation in presence of Ca2<sup>+</sup> and CaM, this suggests that phosphorylation may not reach the putative threshold. Hence, the development of many nodules cannot reach the mature stage. Another possibility is that the stability of W342F mutation could be compromised when expressed in the roots.

This study documents that altering the autoinhibitory domain of CCaMK can significant increase the CaM-binding capacity of CCaMK which leads to a negative impact on root nodule development. Understanding of some target proteins of CCaMK involved in the regulation of nodule development might help to understand how CCaMK is involved not only for decoding Ca2<sup>+</sup> oscillation but also in regulating some down-stream targets of root nodule symbiosis.

### AUTHOR CONTRIBUTIONS

EJ conceived the idea for this project in consultation with BP and LD, performed most of the experiments, analyzed results and wrote most of the paper. LD and CG advised, helped in data analysis and supported with further ideas and experiments for the project. BP provided financial support, helped in planning,

### REFERENCES


executing and leading the project. EJ, LD, CG, and BP were involved in manuscript preparation.

### FUNDING

This research was supported by the National Science Foundation grant (1021344), and the Washington State University Agricultural Research Center.

### ACKNOWLEDGMENTS

The authors thank Prof. R. Geurts for sharing the Sinorhizobium meliloti strain. The authors also acknowledge the WSU Franceschi Microscopy and Imaging Center for assistance with the Confocal microscopy. The authors also thank Lorie Mochel for her help in editing.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2017.01921/ full#supplementary-material

or S344 generates an intramolecular interaction blocking the CaMbinding. Plant Signal. Behav. 12, e1343779. doi: 10.1080/15592324.2017. 1343779



calcium/calmodulin. J. Biol. Chem. 271, 8126–8132. doi: 10.1074/jbc.271.14. 8126


**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 Jauregui, Du, Gleason and Poovaiah. 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.

fpls-08-01548 September 5, 2017 Time: 16:57 # 1

# The Soybean Rfg1 Gene Restricts Nodulation by Sinorhizobium fredii USDA193

Yinglun Fan1,2† , Jinge Liu<sup>2</sup>† , Shanhua Lyu1,2, Qi Wang<sup>2</sup> , Shengming Yang<sup>2</sup> and Hongyan Zhu<sup>2</sup> \*

<sup>1</sup> College of Agriculture, Liaocheng University, Liaocheng, China, <sup>2</sup> Department of Plant and Soil Sciences, University of Kentucky, Lexington, KY, United States

Sinorhizobium fredii is a fast-growing rhizobial species that can establish a nitrogenfixing symbiosis with a wide range of legume species including soybeans (Glycine max). In soybeans, this interaction shows a high level of specificity such that particular S. fredii strains nodulate only a limited set of plant genotypes. Here we report the identification of a dominant gene in soybeans that restricts nodulation with S. fredii USDA193. Genetic mapping in an F2 population revealed co-segregation of the underlying locus with the previously cloned Rfg1 gene. The Rfg1 allele encodes a member of the Toll-interleukin receptor/nucleotide-binding site/leucine-rich repeat class of plant resistance proteins that restricts nodulation by S. fredii strains USDA257 and USDA205, and an allelic variant of this gene also restricts nodulation by Bradyrhizobium japonicum USDA122. By means of complementation tests and CRISPR/Cas9-mediated gene knockouts, we demonstrate that the Rfg1 allele also is responsible for resistance to nodulation by S. fredii USDA193. Therefore, the Rfg1 allele likely provides broad-spectrum resistance to nodulation by many S. fredii and B. japonicum strains in soybeans.

Edited by:

Stig U. Andersen, Aarhus University, Denmark

#### Reviewed by:

Dugald Reid, Aarhus University, Denmark Erik Limpens, Wageningen University and Research, Netherlands

#### \*Correspondence:

Hongyan Zhu hzhu4@uky.edu †These authors have contributed equally to this work.

#### Specialty section:

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

Received: 01 July 2017 Accepted: 24 August 2017 Published: 07 September 2017

#### Citation:

Fan Y, Liu J, Lyu S, Wang Q, Yang S and Zhu H (2017) The Soybean Rfg1 Gene Restricts Nodulation by Sinorhizobium fredii USDA193. Front. Plant Sci. 8:1548. doi: 10.3389/fpls.2017.01548 Keywords: soybean, nodulation, nitrogen fixation, rhizobial symbiosis, symbiosis specificity

## INTRODUCTION

The leguminous plants are able to establish a symbiotic relationship with nitrogen-fixing soil bacteria called rhizobia. The symbiosis is featured by the formation of root nodules where the bacteria in nodule cells can convert atmospheric nitrogen into ammonia and make it available to the plant. This symbiotic partnership has important implications in sustainable agriculture because it reduces the need for nitrogen-based fertilizers.

The legume-rhizobial symbiosis starts with host perception of bacterially derived lipochitooligosaccharides known as nodulation (Nod) factors (Oldroyd et al., 2011). Recognition of Nod factors secreted by compatible bacteria induces cell divisions in the root cortex leading to the formation of nodule primordia, and at the same time initiates the infection process that delivers the bacteria into these primordia. Infection of most legumes such as soybeans (Glycine max) and alfalfa (Medicago sativa) is through root hairs. Bacteria are first entrapped by curled root hairs and multiply to form micro-colonies referred to as infection foci. From these foci, plant-made tubular-like structures, called infection threads, start to develop through local cell wall hydrolysis and invagination of the plant plasma membrane. The infection threads that are colonized by dividing bacteria proceed through the epidermal cell layer into the inner cortex where the nodule primordium has formed. Bacteria are then internalized in an endocytosis-like process fpls-08-01548 September 5, 2017 Time: 16:57 # 2

and surrounded by a host membrane, where they differentiate into nitrogen-fixing bacteroides and are confined in organellelike structures called symbiosomes (Jones et al., 2007; Oldroyd et al., 2011).

While some bacteria can nodulate a wide range of hosts, most bacteria have strict host selectivity. As such, particular rhizobial species or strains nodulate only a narrow group of legume species or genotypes (Broughton et al., 2000; Perret et al., 2000; Wang et al., 2012). Understanding the genetic and molecular basis of this specificity is important for developing strategies to improve the agronomic potential of the root nodule symbiosis in agriculture. It has been reported that plant domestication and breeding processes have led to the reduced ability of the modern cultivars to interact with indigenous soil strains as compared to their wild progenitors (Mutch and Young, 2004; Kiers et al., 2007; Kim et al., 2014). In this case, development of cultivars that are promiscuous with indigenous strains would be beneficial. On the other hand, certain indigenous soil bacteria are highly competitive for nodulation with host legumes but with low nitrogen fixation efficiency. Under this latter scenario, it is desirable to grow plants that can restrict nodulation with low-efficient indigenous strains but nodulate preferentially with the effective inoculant strains (Keyser and Li, 1992; Devine and Kuykendall, 1996).

Establishment of a root nodule symbiosis requires mutual recognition of multiple molecular signals between the symbiotic partners (Jones et al., 2007; Deakin and Broughton, 2009; Oldroyd et al., 2011). Therefore, symbiotic specificity can be regulated by multiple mechanisms at different stages of the nodule development (Perret et al., 2000; Wang et al., 2012, 2017; Liu et al., 2014; Yang et al., 2017). In most legumes, bacterial infection and nodule formation is initiated by host recognition of rhizobial Nod factors (Lerouge et al., 1990; Geurts et al., 1997; Limpens et al., 2003; Radutoiu et al., 2003). Nod factors produced by different bacteria carry specific chemical decorations on the chitin backbone, and this structural diversity has been thought to be a major determinant of nodulation specificity in the legume–rhizobal interaction, particularly at the species level (Bras et al., 2000; Radutoiu et al., 2007). Similar to pathogenic bacteria, symbiotic rhizobia also use conserved microbe-associated molecular patterns (MAMPs) or secreted effectors to facilitate their interaction with the host (D'Haeze and Holsters, 2004; Fauvart and Michiels, 2008; Deakin and Broughton, 2009; Soto et al., 2009; Downie, 2010; Wang et al., 2012; Kawaharada et al., 2015). Accordingly, effector- or MAMPtriggered plant immunity mediated by host receptors also plays an important role in regulating host range of rhizobia (Yang et al., 2010; Wang et al., 2012; Faruque et al., 2015; Kawaharada et al., 2015; Tang et al., 2016).

We have cloned several dominant genes in soybeans (e.g., Rj2, Rfg1, and Rj4) that restrict nodulation with specific rhizobial strains (Yang et al., 2010; Tang et al., 2016). In these cases, symbiosis incompatibility is controlled in a similar manner as 'gene-for-gene' resistance against plant pathogens (Sadowsky et al., 1990; Devine and Kuykendall, 1996). Rj2 and Rfg1 are allelic genes, each encoding a typical Tollinterleukin receptor/nucleotide-binding site/leucine-rich repeat (TIR-NBS-LRR) resistance protein that confers resistance to nodulation by specific strains of Bradyrhizobium japonicum and Sinorhizobium fredii, respectively (Yang et al., 2010), while Rj4 encodes a thaumatin-like pathogenesis-related protein that restricts nodulation by specific strains of B. elkanii (Tang et al., 2016). Moreover, the function of these nodulation-restrictive genes is dependent on the bacterial type III secretion system (Krishnan et al., 2003; Okazaki et al., 2009; Yang et al., 2010; Tsukui et al., 2013; Tsurumaru et al., 2015; Tang et al., 2016; Yasuda et al., 2016). These studies revealed an important role of effector-triggered plant immunity in the regulation of nodulation specificity in soybeans (Wang et al., 2012).

Here we describe the study of symbiotic incompatibility of soybean with S. fredii USDA193. A previous report suggested that restriction of nodulation by this strain is associated with the Rj2 and/or Rj3 loci (Nakano et al., 1997). Our study revealed a single dominant gene responsible for this incompatibility. Genetic mapping in an F2 population showed co-segregation of the underlying locus with the previously cloned Rfg1 gene that confers resistance to nodulation by S. fredii strains USDA257 and USDA205 (Yang et al., 2010). Through complementation tests and CRISPR/Cas9-mediated gene disruption, we demonstrate that the Rfg1 allele also is responsible for restricting nodulation by S. fredii USDA193. Our study suggests that the Rfg1 locus is involved in the determination of nodulation specificity with multiple S. fredii and B. japonicum strains in soybeans.

### MATERIALS AND METHODS

### Plant Material, Nodulation Assay and Genotyping

The F2 mapping population was derived by crossing the North American cultivar 'Williams 82' with the cultivar 'Peking,' an introduction from China. Peking formed nitrogen-fixing nodules when inoculated with S. fredii USDA193 (Nod+) while Williams 82 restricted nodulation by this strain (Nod−) (Keyser et al., 1982). S. fredii USDA193, originally isolated from China (Keyser et al., 1982), was obtained from the National Rhizobium Germplasm Collection (USDA-ARS, Beltsville, MD, United States). The strain was cultured on YEM agar plates (yeast extract, 1.0 g/L; mannitol, 10.0 g/L; dipotassium phosphate, 0.5 g/L; magnesium sulfate, 0.2 g/L; sodium chloride, 0.1 g/L; calcium carbonate, 1.0 g/L; agar, 15.0 g/L) in the dark at 28◦C for 4–5 days, and the bacterial paste was then collected and diluted in sterile water to OD<sup>600</sup> of 0.1. For nodulation assay, each 1-week-old seedling was flood-inoculated with 10 mL of the bacterial suspension. Plants were grown in sterilized 50/50 Perlite-Turface mix in a growth chamber programmed for 16 h light at 26◦C and 8 h dark at 23◦C. Nodulation phenotype was assayed 4 weeks post inoculation. Genotyping was conducted by using a CAPS (cleaved amplified polymorphic sequences) marker developed based on a SNP (single nucleotide polymorphism) between the Rfg1 (Williams 82) and rfg1 (Peking) alleles. The primer pair used was 50TGAGAGTACTGGAATGGTGGAG3<sup>0</sup> and 50TTGCTGATCGAACCACTCTG3<sup>0</sup> , and the restriction enzyme used was HpaI.

### Complementation Tests

fpls-08-01548 September 5, 2017 Time: 16:57 # 3

The Williams 82 allele of Rfg1 was used for complementation tests. Genomic DNA of the Rfg1 allele was derived from the BAC clone Gm\_WBa0019D20 of Williams 82 (Marek and Shoemaker, 1996) by digestion with PstI and BmgBI. The released 10.9 kb fragment included the 4.9-kb coding region, the 4.0 kb upstream of the start codon containing the promoter and 5<sup>0</sup> untranslated region, and the 2.0 kb downstream of the stop codon encompassing the 3<sup>0</sup> untranslated region. The DNA fragment was cloned into the binary vector pCAMBIA1305.1 through blunt end cloning. The binary vector was transferred into the Peking genetic background through hairy root transformation as described below. The transgenic roots were identified by GUSstaining because of the presence of a GUS expression cassette in the pCAMBIA1305.1 vector.

### CRISPR/Cas9-Mediated Gene Knockout

The CRISPR/Cas9 gene knockout constructs were developed based on the pHSE401 vector described by Xing et al. (2014). Two pairs of oligos were designed to specifically target two different sites within the fourth exon of Rfg1. For the first targeted site, we used the oligo pair 50ATTGATGAGGACTTAAAAAGCTC3 0 and 50AAACGAGCTTTTTAAGTCCTCAT3 0 . For the second targeted site, we used the oligo pair 50ATTGACAGTAAGC CTTACTACCT3 0 and 50AAACAGGTAGTAAGGCTTACTGT3 0 . The underlined sequences represent the targeted positions. The oligo pairs were first annealed to produce a double-stranded fragment with 4-nt 5<sup>0</sup> overhangs at both ends, and then ligated into the BsaI-digested pHSE401 vector. The constructs were individually transformed to the Williams 82 (Nod−) background by means of hairy root transformation. To validate the CRISPR/Cas9-mediated gene disruption, the roots that formed nodules were subjected to DNA isolation, PCR amplification, and DNA sequencing. If the initial sequencing suggested the presence of multiple heterogeneous mutant alleles, the PCR product was ligated into pGEM T-Easy Vector System (Promega) and at least 10 colonies were selected for sequencing.

### Construction of a Chimeric Gene of Rfg1 and Rj2

The Rfg1 allele that restricts nodulation by S. fredii USDA257 is allelic to Rj2, an allele that restricts nodulation by B. japonicum USDA122 (Yang et al., 2010). At this locus, there exist three type of alleles in natural populations of soybeans, including the Rj2 (rfg1) allele that restricts nodulation with USDA122 but allows nodulation with USDA257, the rj2 (Rfg1) allele that permits nodulation with USDA122 but restricts nodulation with USDA257, and the rj2 (rfg1) allele that nodulates with both strains. These allelic specificities are defined by seven amino-acid substitutions (Yang et al., 2010). However, we did not identify an Rj2 (Rfg1) allele type that prohibits nodulation with both USDA122 and USDA257 in the surveyed soybean lines (Yang et al., 2010). We thus modified the Rfg1 allele by replacing part of polymorphic sequence of Rfg1 with the corresponding sequence of the Rj2 allele, with the expectation to generate a chimeric gene that could lead to restriction of nodulation by both strains. For this purpose, we amplified two overlapping DNA fragments, one that contained the substitutions responsible for the Rj2 allelic function and another that possessed the substitutions required for the Rfg1 allelic function. We then used an overlapping PCR strategy to assemble the two DNA templates into a single one. The amplified fragment was then cloned into the EcoRI digested genomic construct of Rfg1 mentioned above using the In-Fusion Advantage PCR Cloning Kits (Clontech). We tested the function of this chimeric allele by transferring it into Peking, a genotype that formed nodules with both USDA122 and USDA257. To avoid redundancy, we will provide a further description of this experiment in the "Results" section.

### Hairy Root Transformation

Agrobacterium rhizogenes-mediated hairy root transformation was carried out based on the protocol described by Kereszt et al. (2007). Briefly, bacterial paste of the A. rhizogenes strain K599 that contains individual binary vectors was injected into the cotyledonary node of 1-week-old seedlings using a latex free syringe with a thin needle (0.4 mm × 13 mm) (1 ml 27G1/2, Becton, Dickinson & Co.). The infected seedlings were grown in sterile vermiculite and covered with plastic bags in a growth chamber to maintain high humidity. Two to three weeks after inoculation, when hairy roots were well developed at the infection sites, the main roots were removed, and the composite plants were inoculated with rhizobia. Nodulation assays were performed 4 weeks post inoculation.

### Microscopic Analysis

Assay for root hair curling followed the method described in Yang et al. (2010). For anatomical analysis of nodules, nitrogenfixing nodules of Peking and rudimentary nodules of Williams 82 were harvested 4 weeks post inoculation and immediately fixed in 4% paraformaldehyde (w/v) overnight at 4◦C. The tissues were then dehydrated in a graded ethanol series followed by a graded series of xylene. After infiltrated in 50/50 Epon-Araldite resin and propylene oxide overnight and then in 75/25 Epon-Araldite resin and propylene oxide for 8 h, the samples were embedded in resin. Embedded tissues were sectioned (10 µm thick) with a microtome, stained with Toluidine Blue, and examined with bright-field optics.

## RESULTS

### Characterization of the Soybean–S. fredii USDA193 Interactions

Sinorhizobium fredii USDA193 formed mature nitrogen-fixing nodules on the roots of Peking (**Figure 1A**) but not on the roots of Williams 82 (**Figure 1B**). Despite being unable to form functional nodules in the Williams 82 background, the bacterial strain could frequently induce the formation of rudimentary nodules, bump-like small cortical proliferations on the roots (**Figure 1B**). In contrast to the infected nodules formed in the compatible interaction (**Figure 1C**), the rudimentary nodules on the roots of Williams 82 were completely devoid of infected fpls-08-01548 September 5, 2017 Time: 16:57 # 4

bacteria; as such, the cortical cell division ceased at very early stage of the nodule development (**Figure 1D**). The ability to produce rudimentary nodules with USDA193 on the Williams 82 roots suggested that the early responses of Nod factor perception are not affected. Consistent with this inference, root hair curling, a hallmark of Nod factor responses, occurred in both compatible and incompatible interactions (**Figures 1E,F**). Thus, we concluded that the restriction of nodulation by S. fredii USDA193 in Williams 82 was not due to a failure in Nod factor

Photographs were taken 5 days post inoculation. Bars = 100 µm.

## Restriction of Nodulation with USDA193 Is Controlled by a Single Dominant Gene Mapped to the Rfg1 Locus

perception but caused by the block of bacterial infection.

We carried out genetic analysis of the symbiosis incompatibility involving S. fredii USDA193 in an F2 population derived from the cross between Williams 82 (Nod-) and Peking (Nod+). From

FIGURE 2 | Introduction of the Rfg1 allele of Williams 82 into Peking led to restriction of nodulation on the transgenic roots by USDA193 (blue), but nodulation was normal on the wild-type roots (white).

a small population of 122 plants inoculated by USDA193, 95 were Nod− and 27 were Nod+. The segregation statistically fits the 3:1 (Nod− to Nod+) ratio (χ <sup>2</sup> = 0.54, df = 1, P = 0.46), suggesting that the restriction of nodulation by USDA193 in Williams 82 is controlled by a single dominant gene. The dominant nature of the nodulation-restrictive allele supports our hypothesis that the incompatible interaction between Williams 82 and USDA193 was not due to a failure in Nod factor signaling but resembles 'genefor-gene' resistance in the plant–pathogen interactions.

Williams 82 and Peking showed the same phenotype when inoculated with S. fredii strains USDA257 and USDA193, and we previously reported that the resistance to nodulation with USDA257 in Williams 82 is controlled by the dominant Rfg1 allele (Glyma16g33780) that encodes a TIR-NBS-LRR protein (Yang et al., 2010). We therefore suspected that Rfg1 possibly also confers resistance to S. fredii USDA193. To test this possibility, we started the mapping experiment by using a polymorphic DNA marker developed from Rfg1. Consistent with our hypothesis, linkage analysis in the aforementioned F2 population revealed cosegregation between the nodulation phenotypes and the marker genotypes. Thus, we considered Rfg1 as a candidate gene that restricts nodulation by S. fredii USDA193.

### Rfg1 Is Responsible for Resistance to Nodulation by S. fredii USDA193

We first tested the Rfg1 gene by complementation tests using A. rhizogenes-mediated hairy root transformation. Because the transformation experiments were performed without selection, the hairy roots induced by A. rhizogenes included both transgenic and wild type, which can be readily distinguished by the GUSstaining assay. As shown in **Figure 2**, introduction of the Rfg1 allele of Williams 82 into the Peking background resulted in complete block of nodule formation on the transgenic roots. From >20 composite transgenic plants that possessed both transgenic and wild-type roots, nodules were formed on the wild-type roots but not on the transgenic roots.

We further used the CRISPR/Cas9-based reverse genetics tool (Doudna and Charpentier, 2014) to knock out the Rfg1 gene in the Williams 82 background (Nod−). For this purpose, we designed two gRNA vectors that individually target two different sites of the fourth exon (**Figures 3A,B**). The vectors were introduced to A. rhizogenes K599 for hairy root transformation, followed by assaying the nodulation capacity of the hairy roots by inoculation with S. fredii USDA193. For both vectors, we

boxes and lines, respectively. Arrow indicates the transcription direction. (B) Two targeted sites on the fourth exon. The protospacer adjacent motif (PAM) is 'TGG' for the site 1 and 'AGG' for the site 2. (C) An example showing that knockout of Rfg1 at the site 1 led to the formation of root nodules on a transgenic root (boxed). Sequence analysis revealed that this root contained two mutant alleles, one with a 3-bp deletion and another with a 10-bp deletion (indicated by arrows). (D) An example showing that knockout of Rfg1 at the site 2 also resulted in the formation of root nodules on a transgenic root (boxed). Sequence analysis of the DNA from this root identified two mutant alleles, one with a 4-bp deletion and another with a 26-bp deletion (indicated by arrows).

fpls-08-01548 September 5, 2017 Time: 16:57 # 5

fpls-08-01548 September 5, 2017 Time: 16:57 # 6

obtained >30 putative transgenic roots from more than 50 independent plants that formed mature nitrogen-fixing nodules. DNA sequencing validated the targeted gene disruption in these roots and these roots did not contain the wild-type allele. Two examples are illustrated in **Figures 3C,D**. **Figure 3C** shows that a transgenic root forming nodules resulted from the knockout of Rfg1 at the site 1, and sequence analysis revealed two mutant alleles, one with a 3-bp deletion and another with a 10-bp deletion. **Figure 3D** represents an example showing that knockout of Rfg1 at the site 2 also led to the formation of root nodules on a transgenic root; sequence analysis identified two mutant alleles, one with a 4-bp deletion and another with a 26-bp deletion when compared with the wild-type allele. Taken together, we conclude that Rfg1 is responsible for nodulation restriction by S. fredii USDA193 in Williams 82.

### Functional Analysis of a Chimeric Gene of Rfg1 and Rj2

Rfg1 and Rj2 are allelic genes, each encoding a TIR-NBS-LRR protein of 1052 amino acids (Yang et al., 2010). A survey of a group of soybean lines identified three types of naturally occurring alleles, namely Rj2 (rfg1), rj2 (Rfg1), and rj2 (rfg1) (Yang et al., 2010). The Rj2 (rfg1) allele restricts nodulation with USDA122 but not with USDA257; the rj2 (Rfg1) allele restricts nodulation with USDA257 but not with USDA122; and the rj2 (rfg1) allele allows nodulation with both strains. These allelic specificities are determined by seven amino acid substitutions occurring around the C-terminus of the NBS domain and the sixth LRR repeat (**Figures 4A,B**). Comparing between Rj2 (rfg1) and rj2 (rfg1) and between rj2 (Rfg1), and rj2 (rfg1) suggests that E452 and I490 are required for the Rj2-mediated nodulation restriction against USDA122, while E731, N736, S743, D756, and S758 are essential for Rfg1-mediated resistance against USDA257 and USDA193. If this inference is true, then the chimeric gene of Rj2 and Rfg1, called Rj2 (Rfg1), encoding a protein with E452, I490, E731, N736, S743, D756, and S758, would prevent nodulation with both strains. We generated multiple transgenic roots (from >100 independent plants) expressing the chimeric gene in Peking that carries an rj2 (rfg1) allele and forms nodules with USDA257, USDA193, and USDA122. In consistence with our hypothesis, the transgenic roots restrict nodulation with both USDA257 (**Figure 4C**), USDA193 (**Figure 4D**), and USDA122 (**Figure 4E**). The transgenic roots retained their ability to nodulate with B. japonicum USDA110, a strain that nodulates both Rj2 and Rfg1 genotypes (**Figure 4F**).

**110**

### DISCUSSION

fpls-08-01548 September 5, 2017 Time: 16:57 # 7

Previous studies showed that the soybean Rfg1 gene restricts nodulation with the fast-growing S. fredii strains USDA257 and USDA205 (Devine and Kuykendall, 1996; Yang et al., 2010). The Rfg1-mediated resistance to USDA257 was dependent on the bacterial type III secretion system and presumably resulted from the effector-triggered plant immunity, even though the cognate effector(s) has not yet been identified (Krishnan et al., 2003; Yang et al., 2010). In this paper, we demonstrate that the soybean Rfg1 allele also conditions nodulation restriction with S. fredii USDA193, suggesting that Rfg1 likely provide broad-spectrum resistance against a proportion of S. fredii strains (Keyser et al., 1982).

The soybean Rj2/Rfg1 locus confers nodulation specificity toward many naturally occurring B. japonicum and B. fredii strains (Caldwell, 1966; Devine and Kuykendall, 1994; Trese, 1995; Yang et al., 2010). A survey of 847 soybean genotypes from Asian countries revealed an Rj2 allele frequency of 0.02, mainly originated from southeast China (Devine and Breithaupt, 1981). In contrast, the Rfg1 allele occurred with a much higher frequency of 0.44 in a similar population consisting of 285 plant introductions from Asian countries (Devine, 1985). The distribution of the Rfg1 (0.44) and rfg1 (0.56) allele frequencies suggests that these alleles has not been subjected to strong natural selection. However, a screen of 197 soybean lines from the Midwestern United States revealed an Rfg1 allele frequency of 0.83, suggesting that the compatibility with S. fredii strains may be eroded during the breeding process, likely due to the narrow genetic basis of germplasms used in the soybean breeding programs (Balatti and Pueppke, 1992). The diversification of the Rfg1 and rfg1 alleles appears to predate that of the Rj2 and rj2 alleles. The Rj2 (rfg1) allele likely have derived from the rj2 (rfg1) allele through gain-of-function mutations. The rj2 (rfg1) allelic form is most promiscuous, conferring an unrestricted nodulation phenotype with many strains that are restricted by the Rj2 or Rfg1 alleles. Therefore, the rj2 (rfg1) genotype should be prioritized in soybean breeding programs if the goal is to ensure the cultivar to be able to nodulate with indigenous bacterial strains.

The polymorphic Rj2/Rfg1 locus encodes a TIR-NB-LRR gene with three allelic variants that show differing specificities to yet unknown bacterial effectors. E452 and I490 in or near the C-terminal NBS domain of Rj2 are required for the recognition

### REFERENCES


of the cognate effector from B. japonicum USDA122, while E731, N736, S743, D756, and S758 in the LRR domain of Rfg1 are essential for the recognition of the corresponding effectors secreted by USDA257 and USDA193. This situation is similar to the previous studies showing that both TIR, NBS, and LRR domains could be involved in the determination of resistance specificity (Ellis et al., 1999; Dodds et al., 2001; Mondragón-Palomino et al., 2002; Ellis et al., 2007). The chimeric Rj2(Rfg1) allele composed of E452, I490, E731, N736, S743, D756, and S758 could recognize corresponding effectors of all these strains. Such an allele could exist in natural population if intragenic recombination events occur between the Rj2(rfg1) and rj2(Rfg1) alleles. It is possible that these polymorphic sites play a role in direct effector binding or intra- or intermolecular interactions in protein complexes during recognition and signaling. Further analysis of the mechanisms underlying the recognitions mediated by the Rj2/Rfg1 locus awaits the identification of the cognate bacterial effectors.

### AUTHOR CONTRIBUTIONS

YF, JL, SL, and HZ: conception and design of the work. YF, JL, QW, SY, and SL: performed the work and analyzed data. HZ wrote the paper.

### FUNDING

This work was supported by the Kentucky Soybean Promotion Board to HZ and by the National Natural Science Foundation of China (grant number 31271751) to SL and YF. YF and SL received support from the Shandong Provincial Education Association of China to study at University of Kentucky.

### ACKNOWLEDGMENTS

We thank Patrick Elia (US Department of Agriculture-Agricultural Research Service, Beltsville, MD, United States) for providing the rhizobial strains and Dr. Qijun Chen (China Agricultural University, Beijing, China) for providing the CRISPR/Cas9 binary vectors.


a sulphated and acylated glucosamine oligosaccharide signal. Nature 344, 781–784. doi: 10.1038/344781a0


fpls-08-01548 September 5, 2017 Time: 16:57 # 8

fpls-08-01548 September 5, 2017 Time: 16:57 # 9


**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 DR and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.

Copyright © 2017 Fan, Liu, Lyu, Wang, Yang and Zhu. 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.