MtGRF8 interacts with MtSymCRK and prevents early bacteroid death during Medicago – Sinorhizobium symbiosis

Summary

• In Medicago littoralis R‐108, Symbiotic CYSTEINE‐RICH RECEPTOR‐LIKE KINASE (MtSymCRK) is required to prevent defense reactions in symbiotic nodules during chronic infection. MtSymCRK represents a key legume symbiotic gene controlling nodule immunity and early bacteroid survival, but its mode of action in the cell signaling remains poorly understood.

• This study investigated proteins interacting with MtSymCRK by screening an Arabidopsis thaliana open reading frame (ORF) library using yeast‐two‐hybrid followed by translational genetics towards R‐108. Finally, we conducted a reverse genetic approach in R‐108 using Transposon of Nicotiana tabacum 1 (Tnt1) insertional mutants.

• We identified a 14‐3‐3 protein, AtGENERAL REGULATORY FACTOR 8 (AtGRF8) interacting with MtSymCRK kinase domain (MtSymCRK^KIN) and confirmed the interaction between MtSymCRK^KIN and MtGRF8, the closest homolog of AtGRF8 in R‐108. Moreover, other members of the 14‐3‐3 family interact with MtSymCRK^KIN but with a lower affinity than MtGRF8. We also demonstrated that the corresponding Tnt1 mutant line Mtgrf8 displays symbiotic dysfunctions in defense and senescence responses, resulting in a reduction of bacteroid differentiation. These dysfunctions compromise the intracellular survival of the rhizobia, thereby impairing the nitrogen fixation of rhizobia.

• Our findings show that MtGRF8 and MtSymCRK are required for effective functioning of symbiosis and prevention of bacteroid death.

MtSymCRK–MtGRF8‐14‐3‐3a, e, h and i functional model in R‐108 nodules.

Introduction

Under limited nitrogen conditions, legumes such as Medicago excrete flavonoids that are perceived by rhizobia and subsequently activate the expression of bacterial nodulation (nod) genes and the synthesis of bacterial nodulation factors (NFs; Lerouge et al., ¹⁹⁹⁰; Peck et al., ²⁰⁰⁶; Cooper, ²⁰⁰⁷; Oldroyd & Downie, ²⁰⁰⁸; Oldroyd et al., ²⁰¹¹). The perception of NFs by legume LYSIN MOTIF RECEPTOR‐LIKE KINASES and LEUCINE‐RICH REPEAT RECEPTOR‐LIKE KINASES complexes initiates a signaling cascade that ultimately activates the transcription of a panel of symbiotic genes required for nodule formation (Limpens et al., ²⁰⁰³; Madsen et al., ²⁰⁰³; Radutoiu et al., ²⁰⁰³; Arrighi et al., ²⁰⁰⁶; Smit et al., ²⁰⁰⁷; Broghammer et al., ²⁰¹²; Roy et al., ²⁰²⁰; Yang et al., ²⁰²²; Zhang et al., ²⁰²⁵). These signaling events lead to the development of root organs termed nodules, within which rhizobia infect host cells to establish nitrogen‐fixing symbiosis (Oldroyd & Downie, 2008; Roy et al., ²⁰²⁰; Shumilina et al., ²⁰²³; Zhang et al., ²⁰²⁴). In Medicago, nodules are indeterminate and spatially organized in distinct zones with specialized functions, ranging from meristem (ZI) at the nodule apex to the senescence zone (ZIV) near the root. The meristem (ZI) responsible for nodule growth, is followed by the infection zone (ZII), where the rhizobia penetrate into the host cells. Between ZII and the fixation zone (ZIII) where the nitrogen fixation occurs, there is the interzone II‐III infected by rhizobia entering into the terminal differentiation phase. Finally, in the senescence zone (ZIV), symbionts and host cells undergo degradation and recycling. In Medicago, bacteroids are subjected to terminal differentiation, a process, driven by the action of small antimicrobial peptides. These peptides, designated as Nodule Cysteine‐Rich peptides (NCRs) are produced by the host plant and secreted into the symbiosomes (corresponding to the bacteroids surrounded by the host membrane: e.g. peribacteroid membrane). NCRs modulate bacteroid physiology by inducing genome endoreduplication, cell elongation and membrane permeabilization (Mergaert et al., ²⁰⁰⁶; Van de Velde et al., ²⁰¹⁰; Alunni & Gourion, ²⁰¹⁶).

The host plant modulates nodule functioning (N₂ fixation by bacteroids) primarily through two interconnected processes: senescence and defense. Senescence is a programmed decline in cellular activity aiming at recycling cellular components that is often associated with the induction of SENESCENCE‐ASSOCIATED GENES (SAGs), including CYSTEINE PROTEASES (CPs) 2–6 involved in cell degradation (Pérez Guerra et al., ²⁰¹⁰; Gan, ²⁰¹⁸). Defense mechanisms regulate nodule infection by activating defense‐related genes, oxidative bursts and antimicrobial compound accumulation (Gourion et al., ²⁰¹⁵; Yu et al., ²⁰¹⁸). Interestingly, senescence and defense responses are not independent. A complex interplay between these two processes is observed in legume nodules, and the host can trigger the activation of senescence or/and defense to fine‐tuned nodule functioning (Berrabah et al., ²⁰²⁴). Prolonged or excessive defense activation can accelerate senescence, while premature senescence can itself amplify defense signaling through increased ROS production, protease activity and loss of bacteroid integrity (Berrabah et al., ^{2023, 2024}). This bidirectional crosstalk allows the plant to adjust nodule longevity and infection levels in response to symbiont performance.

Several legume genes involved in suppressing nodule immunity have been characterized in Medicago littoralis (previously known as M. truncatula R‐108, Choi et al., ²⁰²²). Among those, MtDEFECTIVE IN NITROGEN FIXATION 2 (MtDNF2) encodes a phosphatidylinositol phospholipase C‐like protein (Bourcy et al., ²⁰¹³; Berrabah et al., ^{2014a, 2015}), MtNODULES WITH ACTIVATED DEFENSE 1 (MtNAD1) encodes a protein of unknown function (Wang et al., ²⁰¹⁶; Domonkos et al., ²⁰¹⁷) and MtREGULATOR OF SYMBIOSOME DIFFERENTIATION (MtRSD) encodes a C₂H₂ transcription factor (Sinharoy et al., ²⁰¹³). Finally, MtSymbiotic CYSTEINE‐RICH RECEPTOR‐LIKE KINASE (MtSymCRK) encodes a cysteine‐rich receptor‐like kinase belonging to the multigenic family of plant cysteine‐rich kinases (Berrabah et al., ^2014b). MtSymCRK is a 654 amino acid protein with a cysteine‐rich extracellular domain exhibiting two Domain of Unknown Function 26 (DUF26) motifs potentially involved in sensing the redox state of the symbiotic cell (Chen, 2001; Berrabah et al., ^2014b). This domain is separated by a single transmembrane segment from the intracellular serine/threonine protein kinase domain, which exhibits a non‐arginine–aspartate (non‐RD) motif that is a feature of innate immunity receptors (Dardick et al., ²⁰¹²; Berrabah et al., ^2014b). MtSymCRK is specifically expressed in nodule‐infected cells after rhizobia internalization and prevents defense and early senescence responses at least through blocking the ethylene pathway (Berrabah et al., ^{2014b, 2015, 2018}). In M. littoralis R‐108 (hereafter R‐108), Mtdnf2, Mtnad1, Mtrsd and Mtsymcrk mutants develop nitrogen fixation‐deficient (fix‐) nodules showing exacerbated defense responses, including necrosis of the nodule tissues, accumulation of defense compounds (e.g. phenols) and expression of defense genes, such as PATHOGENESIS‐RELATED PROTEINS (PR) 5, 8 and 10 (Bourcy et al., ²⁰¹³; Sinharoy et al., ²⁰¹³; Berrabah et al., ^{2014a, 2014b, 2015}; Wang et al., ²⁰¹⁶; Domonkos et al., ²⁰¹⁷). The developed defenses compromise the intracellular survival of symbionts in the mutants and lead to the accumulation of undifferentiated dead bacteroids. Gene expression analysis showed that CP2, 3, 4 and 5 were mainly downregulated in Mtdnf2 and Mtsymcrk, whereas an upregulation of these CPs was observed in other mutants like Mtnad1, suggesting that defenses can be often associated with senescence response (Wang et al., ²⁰¹⁶; Berrabah et al., ²⁰²⁴b). Based on mutant phenotypes, genetic and gene expression analyses, symbiotic genes preventing immune response seem to act in a sequential mode, in which MtSymCRK and MtRSD act after MtDNF2 and likely after MtNAD1 (Berrabah et al., ²⁰¹⁵; Domonkos et al., ²⁰¹⁷). Moreover, MtRSD can influence bacteroids differentiation and immunity by controlling the expression of MtSymCRK through its interaction with MtNODULE INCEPTION (MtNIN) and MtNIN‐LIKE PROTEIN 2 (MtNLP2; Bhardwaj et al., ²⁰²⁶). Despite the crucial role of MtSymCRK in defense prevention, its precise function in nodule immune signaling remains misunderstood.

The 14‐3‐3 protein family is an eukaryotic‐specific group of phosphopeptide‐binding proteins regulating signal transduction and plant immunity (Ferl, 1996; Wu et al., ^1997a; Lozano‐Duran & Robatzek, ²⁰¹⁵; Sheikh et al., ²⁰²⁴). The role of 14‐3‐3 proteins in modulating immune responses arises primarily from their capacity to act as interaction hubs that coordinate protein–protein interactions (PPI) and influence the activity of their target proteins (Sheikh et al., ²⁰²⁴). This interaction hub function of the 14‐3‐3 members is based on their capability to form homodimers and/or heterodimers by establishing bridges with themselves and/or target proteins involved in signaling pathways (Paul et al., ²⁰¹²). However, few examples provide direct evidence on the mechanism coordinating the interaction between kinase receptors and 14‐3‐3 proteins in plants. In Arabidopsis, the λ and κ isoforms of 14‐3‐3 proteins play a key role in both antibacterial and antifungal immunity. These proteins directly interact with MAP KINASE KINASE KINASE 5 (MAPKKK5) to enable their activation by upstream RECEPTOR‐LIKE CYTOPLASMIC KINASES (RLCKs; Dong et al., ²⁰²³). In tomato, the 14‐3‐3 protein Tomato Fourteen‐Three‐Three isoform 1 (TFT1) plays a protective role against bacterial infection (Taylor et al., ²⁰¹²). To our knowledge, no studies directly examined how 14‐3‐3 proteins coordinate defense suppression with senescence through interactions with kinase receptors.

To better understand how MtsymCRK modulates the nodule immune response, we aimed at deciphering the signaling pathway operating downstream MtSymCRK. We first screened an Arabidopsis thaliana ORFeome for proteins interacting with the MtSymCRK kinase domain. This approach allowed the identification of AtGENERAL REGULATORY FACTOR 8 (AtGRF8), a 14‐3‐3 protein, as a direct interactor. We subsequently searched for the closest homolog in R‐108, namely MtGRF8, and confirmed an interaction between the kinase domain of MtSymCRK and MtGRF8. Further yeast two‐hybrid (Y2H) analyses revealed that MtSymCRK interacts with three other Mt14‐3‐3 proteins (namely 14‐3‐3b, e and i). Besides MtGRF8, two of them, Mt14‐3‐3e and i are expressed in nodules but MtGRF8 is the strongest interactant of MtSymCRK in Y2H.

We further characterized the Mtgrf8 Transposon of Nicotiana tabacum 1 (Tnt1) insertional mutant line (Mtgrf8) in R‐108. Mtgrf8 nodules displayed a necrotic zone and accumulated phenols as observed in Mtsymcrk, supporting the hypothesis that MtGRF8 and MtSymCRK function within the same signaling pathway and that they are both required to maintain nodule chronic infection.

Based on our results, we proposed a theoretical model in which the MtSymCRK–MtGRF8 interaction regulates nodule defense and senescence in order to turn down plant defense responses and enable bacteroid persistence for a successful symbiosis. This work contributes to the understanding of the regulation of nodule immunity and the persistence of the symbiont in R‐108.

Materials and Methods

Yeast two‐hybrid vector construction

Primer design

For forward primers, attB1 sequence was added at the 5′‐end of the specific primer sequence, with two nucleotides being inserted in between (the two nucleotides should not be AA, AG or GA). For reverse primers, attB2 sequence was added at the 5′‐end of the specific primer sequence containing a stop codon, with one nucleotide inserted in between.

CDS amplification

Sequences encoding 5′ signal peptide were identified using Signal P 3.0 Server and not considered for cloning (www.cbs.dtu.dk/services/SignalP‐3.0/). Sequences of interest were amplified using cDNA from A. thaliana leaves or Medicago littoralis R‐108 nodules, using Phusion High Fidelity DNA polymerase (Thermo Scientific). PCR products were purified on gel using NucleoSpin™ Gel and PCR Clean‐up Kit (Macherey‐Nagel).

BP and LR recombination

Purified PCR products were sub‐cloned into pDONR221 (kanamycin resistant) using Gateway BP Clonase II Enzyme Mix (Thermo Scientific, Les Ulis, France) and subsequently cloned in pDEST:BD and pDEST:AD destination vectors (ampicillin resistant; Dreze et al., ²⁰¹⁰) using Gateway LR Clonase II Enzyme Mix (Thermo Scientific). Recombinant plasmids were introduced in Escherichia coli One Shot TOP10 chemically competent cells (Thermo Scientific) and genotyped by colony PCR using DreamTaq DNA polymerase (Promega) and cloning primers. Constructs were sequenced at each step (www.eurofinsgenomics.eu; Supporting Information Tables S1, S2).

Saccharomyces cerevisiae transformation

Yeast transformation was performed using the lithium acetate (LiAc) method (Gietz & Schiestl, 1995) with some modifications. Yeast strains Y8930 (MATα) and Y8800 (MATa) were transformed with pDEST:DB‐X and pDEST:AD‐Y constructs, respectively (Dreze et al., ²⁰¹⁰). The details of the transformation method are described in Methods S1.

Pairwise yeast mating and yeast two hybrid assay

Y8930:pDEST:BD‐X and Y8800:pDEST:AD‐Y strains were grown in 100 μl of YPD for 1 d at 30°C. Mating was performed by incubating 10 μl of each strain in 100 μl of YPD for 1 d at 30°C. Yeast selection was performed by incubating 10 μl of each mating in 100 μl of minimal SD – Leu (‐L) – Trp (‐W) for 2 d at 30°C. Five microliters of each yeast mating were dropped on the following selective media: Minimal SD ‐L‐W, minimal SD ‐L‐W ‐ His (‐H), minimal SD ‐L‐W‐H with 3‐amino‐1,2,4‐triazole (3‐AT; 1, 10, 25, 50, 100 mM; Sigma‐Aldrich) and minimal SD ‐L‐W‐H ‐Ala (‐A) (Clontech (Takara Bio Europe, Saint‐Germain‐en‐Laye, France)).

Interactome‐based Y2H approach

The improved Y2H high‐throughput binary interactome mapping liquid pipeline described here is an adaptation of a previously developed interactome (Dreze et al., ²⁰¹⁰; Castandet et al., ²⁰²³). The same yeast expression vectors pDEST:BD‐X and pDEST:AD‐Y and the yeast strains Y8930 and Y8800 were used in this study. Reporter genes GAL2‐ADE2 and LYS2::GAL1‐HIS3 were integrated into the yeast genome. Expression of the GAL1‐HIS3 reporter gene was tested with 1 mM 3‐AT. Y8800 and Y8930 were transformed with pDEST:AD‐Y and pDEST:DB‐X constructs, respectively. Before Y2H screening, pDEST:DB‐X strains were tested for auto‐activation of the GAL1‐HIS3 reporter gene in the absence of pDEST:AD‐Y plasmid. In case of autoactivation, pDEST:DB‐X were physically removed from the collection of baits and screened by mating against the (DB)‐AtORFeome collection using their pDEST:AD‐X constructs. Briefly, yeasts expressing pDEST:DB‐X baits were individually grown at 30°C for 72 h in 5 ml of fresh SC ‐L selective media, pooled and 50 μl were plated into 384‐wells low profile microplates. Glycerol stocks of the 12K_space_(AD)‐AtORFeome collection (Arabidopsis Interactome Mapping Consortium, 2011; Weßling et al., ²⁰¹⁴) corresponding to 127 96‐well plates were thawed, replicated using the colony picker Qpix2 XT into 32 384‐well plates filled with 50 μl of fresh SC ‐W selective media and incubated at 30°C for 72 h. Culture plates corresponding to the DNA binding domain (DB)‐baits pools and activation domain (AD) collection were replicated into mating plates filled with YPD media and incubated at 30°C for 24 h. Mating plates were replicated into screening plates filled with 50 μl of fresh SC ‐L‐W‐H with 1 mM 3AT media and incubated at 30°C for 5 d. To identify primary positives, the OD₆₀₀ of the 384‐well screening plates was measured using a microplate‐reader Tecan Infinite M200 PRO. Yeast cultures identified as positive interactions were picked from selective media and protein pairs identified by depooling the DB baits in a targeted matricial liquid assay in which all the DB baits were individually tested against all the positive AD proteins. Identified pairs were cherry‐picked and checked by DNA sequencing.

Transient expression of fluorescent proteins in tobacco leaf epidermis cells

Cloning strategies

Constructs for bimolecular fluorescence complementation (BiFC) assays and enhanced GREEN FLUORESCENT PROTEIN (eGFP)‐tagged protein subcellular localization were performed by Gateway cloning (Thermo Scientific) using pBiFc2 and pBiFc3 vectors, and pR81 (pK7WGF2) vectors, respectively (Karimi et al., ²⁰⁰²; Azimzadeh et al., ²⁰⁰⁸). Constructs were introduced into Agrobacterium tumefaciens (A. tumefaciens) GV3101 (Koncz & Schell, 1986). Transient expressions of enhanced YELLOW FLUORESCENT PROTEIN (eYFP) moiety‐ or eGFP‐tagged proteins in tobacco (Nicotiana benthamiana) leaf epidermis cells were performed by agroinfiltration as described in Sparkes et al. (2006) with minor modifications.

Agrobacterium tumefaciens suspension preparation

A total of 1.5 ml of overnight culture of transformed A. tumefaciens were transferred in a microtube. Cells were centrifuged at 1000  g for 10 min, washed with 1 ml of infiltration medium (for 50 ml of infiltration medium: D‐glucose, 250 mg; MES 500 mM, 5 ml; Na₃PO₄.12H₂O 20 mM, 5 ml; acetosyringone 1 M, 5 μl; qsp 50 ml with H₂O), centrifuged at 1000  g , 10 min and finally resuspended in 1 ml of infiltration medium. For BiFC and eGFP subcellular localization, each construct was adjusted to OD_600nm: 0.2 with the infiltration medium to a final volume of 500 μl. Agroinfiltrations were performed in presence of A. tumefaciens GV3101 expressing the viral protein helper component‐proteinase (HC‐Pro) RNA silencing inhibitor (Mallory et al., ²⁰⁰²) at OD_HC‐Pro = ∑ OD_constructs tested.

Agroinfiltration

One hour before infiltration, tobacco plants were placed under white light and large third and fourth leaves from apex were chosen for agroinfiltration. Leaf abaxial surfaces were pierced, A. tumefaciens mixtures were infiltrated using 1‐ml syringe.

Light microscopy

Four days after agroinfiltration, leaf discs of 1 cm of diameter were observed. Fluorescent signals from eYFP and eGFP (excitation wavelength, 488 nm; detection wavelength, 508–562 nm) were detected by confocal laser scanning microscopy (LSM880; Zeiss). Images were acquired and processed using the ZEN2.3 lite software (Zeiss).

3D structure prediction

The 3D models were predicted with the AlphaFold3 software installed locally on a GPU/CPU cluster (Abramson et al., ²⁰²⁴). The default parameters as implemented in the local version of the software were used (Table S3). Formally, the protocol consists of 3 steps of jackhmmer search on (1) uniref90_2022_05, (2) uniprot_all (version Apr.2021) and (3) bfd‐first_non_consensus_sequences followed by hhmmbuild for building the corresponding multiple sequence alignment and finally hhmmsearch on pdb_seqres database (version Aug. 2022) to identify putative structural templates. For each sequence, 10 runs with different model seeds were executed and for each run, five different models were generated corresponding to different seeds. Thus, in total, 50 models were built. The ranking score for each model is reported in Table S4. For each run, the best model was kept for further analyses and figure representation. The studied proteins are as follows:

• The kinase domain of MtSymCRK (residues 341–612) to the C‐terminus (Uniprot ID: G7J6M8): FSEDNKIGEGGFGDVYKGMFPNGYEIAVKRLIRNSSQGAVEFKNEVLLIAKLQHRNLVRLLGFCIQRNEKILIYEYMHNKSLDYYLFSPENHRKLTWHARYKIIRGIARGILYLHEDSHLKIIHCDLKPSNILLDDKMNAKISDFGLARIVAIDQMQGNTSIIAGTYGYMSPEYAMLGQFSVKSDVFSFGVIMLEIVSGKRNVDYNGVNSIDDLVSHAWKKWTENKQMELLDPALTYSFSETEVSRCIQLGLLCVQENPDQRPTMATIALYFNIDSIDLPLPQQPPFYMRGKIESKVASKKTMSGRPRSYSVTRF

• The GRF8 monomer (Uniprot ID: G7J567):

• MGGAIPENLNREQYVYLAKLAEQAERYEEMVSFMQKLVVGSTPSSELSVEERNLISVAYKNVIGSLRAAWRIVSSIEQKEEGRKNEDHVVLVKDYRSKVESELTNVCGSILELLDSNLIPSASSSESKVFYYKMKGDYHRYLAEFKIGDEKKSAAEDTMLSYKAAQDIAAADLPSTHPIRLGLALNFSVFYYEILNQSDKACDMAKQAFEEAIAELDTLGEESYKDSTLIMQLLRDNLTLWTSDVQDQLDEP.

The figures have been made with the ChimeraX software (Meng et al., ²⁰²³). The contacts were calculated with the tools implemented in ChimeraX.

Plasmid constructions for hairy root transformation strategies

For hairy root complementation assay, the complete cDNA of MtGRF8 was amplified by PCR and cloned in the pCAMBIA2200 vector under the control of the 35S promoter by Golden Gate strategy and introduced in A. rhizogenes AR1193 for hairy root transformation (named Prom35S::MtGRF8).

For hairy root RNAi constructions, a specific region of 201‐bp cDNA which begins 6‐bp upstream 3′UTR of MtGRF8 was amplified by PCR using the couple primers RNAi–MtGRF8. This sequence was introduced into the pFGC5941 vector (named RNAi::MtGRF8). The pFGC5941 vector containing GUS was used as a control. Primer sequences are provided in Table S5.

Agrobacterium rhizogenes‐mediated root transformation

Hairy root transformations of the R‐108 root system were performed in vitro as described in Boisson‐Dernier et al. (2001) using A. rhizogenes AR1193. Transgenic roots were grown for 2 wk on Fahraeus medium (Fahraeus, 1957) containing kalys agar (1.5%, w/v), 0.5 μM AVG and kanamycin 25 mg l⁻¹. Transgenic roots were transferred on BNM plates for 3 d before inoculation with S. medicae WSM419.

Plant lines, bacterial strains, culture and symbiotic assay

The wild‐type (WT) Medicago truncatula (Gaertn.) cv. Jemalong (line A17 annotated M. truncatula A17) was used in experiments (Young et al., ²⁰¹¹). The WT M. littoralis R‐108 (formerly M. truncatula (Gaertn.) cv. R‐108; Hoffmann et al., ¹⁹⁹⁷; Choi et al., ²⁰²²) and its corresponding mutant Mtgrf8 (nf2687) were also used in this study. Mtgrf8 has been identified from the Medicago truncatula Mutant Database (https://medicago‐mutant.dasnr.okstate.edu/mutant/index.php). The Tnt1 transposon tagged‐line nf737 (Mtsymcrk, Berrabah et al., ^2014b) was used in this study.

R‐108 seeds were scarified using sand paper and surface‐sterilized with NaOCl (1 pellet per 1 l of water) and one droplet of liquid soap for 30 min. Three washes were performed with sterile water. Seeds were stratified for 2 d at 4°C under darkness on water agar plates and then transferred for 24 h at 24°C under darkness for germination. Seedlings were grown in vitro on Buffered Nodulation Media (BNM; Ehrhardt et al., ¹⁹⁹²) in the presence of 0.5 μM 2‐aminoethoxyvinylglycine (AVG) or in perlite : sand mixture (3 : 1, v/v). Plants were grown in controlled environmental chambers under 16 h : 8 h, 24°C : 24°C, light : dark, with 60% relative humidity and 200 μE light intensity. Plants grown in perlite : sand mixture were watered with 1 g l⁻¹ of nitrogen‐free nutritive solution (NPK 0‐15‐40; Plantprod). Sinorhizobium medicae WSM419 (Howieson & Ewing, 1986) was used as a symbiotic partner and grown in YEB medium for 48 h at 30°C (Krall et al., ²⁰⁰²). For in vitro nodulation assays, 1 ml of bacterial suspensions adjusted to OD_{600 nm}: 0.1 was used per plate containing eight seedlings. For perlite:sand mixture, 50 ml of bacterial suspension adjusted to OD_{600 nm}: 0.1 was used per pot containing five seedlings.

R‐108 DNA extraction and Mtgrf8 Tnt1 insertional mutants genotyping

R‐108 DNA was extracted from young leaves using the CTAB procedure. Mtgrf8 (nf2687) Tnt1 lines were genotyped by PCR using DreamTaq DNA Polymerase (Thermo Scientific) according to the manufacturer's instructions. Information regarding the primers used for Mtgrf8 genotyping and sequencing is provided in Table S2.

qRT‐PCR gene expression analysis

Quantitative reverse transcription polymerase chain reaction (qRT‐PCR) gene expression analysis was performed as described in Magne et al. (2018). Full‐length cDNAs were synthesized from 500 ng of total mRNA using SuperScript II Reverse Transcriptase Kit (Invitrogen) and diluted 10 times prior utilization. Information concerning qRT‐PCR primers is provided in Table S5.

Acetylene reduction assay

Acetylene reduction assays were performed as described in Magne et al. (2018). Briefly, a single nodulated plant was placed into a 21‐ml airtight glass vial sealed with a rubber septum containing 200 μl of sterile water. Five hundred microliters of acetylene were injected per vial. Plants were incubated in a growth chamber (at 24°C under a 16 h : 8 h, light : dark photoperiod and 60% of hygrometry) for 2 h. Then, 200 μl of gas was collected from the vial and injected into the gas chromatography system (7820A; Agilent Technology) to determine the ethylene production.

Semi‐thin section

Nodules were embedded in agarose (6%, w/v) and sectioned using vibratome (VT1200S; speed, 0.60 mm s⁻¹; amplitude, 2.55 mm; thickness, 60 μm; www.leica‐microsystems.com). Sections were kept in Tris–HCl, 50 mM pH: 7.2 for subsequent processing.

Phenol coloration and quantification

To quantify the phenolic accumulation area, nodules harvested at 21 d post‐inoculation (dpi) were sectioned and subsequently stained with KMnO4 and methylene blue following the procedure described in Berrabah et al. (2015). This staining method visualizes the presence of phenolic compounds, which appear as distinct dark brown regions within the tissue. Images were acquired and processed using the ZEN2.3 lite software (Zeiss). Quantitative image analysis was then performed using the ImageJ software (NIH, Bethesda, MD, USA; Schneider et al., ²⁰¹²). For each biological replicate, the total cross‐sectional area of the nodule and the specific localized area exhibiting the dark brown phenolic staining were manually delineated and measured (in mm²). All data plotting and visualization were subsequently generated using the R software.

LIVE/DEAD

Nodule sections were stained for 20 min under darkness using LIVE/DEAD BacLight Bacterial Viability Kit (SYTO9), 3.34 μM and PI, 20 μM in Tris–HCl, 50 mM pH 7.2 (Thermo Scientific). Sections were washed in Tris–HCl, 50 mM pH 7.2 before observation. Fluorescent signals from SYTO9‐stained alive bacteroids (excitation wavelength, 488 nm; detection wavelength, 508–571 nm) and from IP‐stained dead bacteroids (excitation wavelength, 561 nm; detection wavelength, 602–674 nm) were detected by confocal laser scanning microscopy (LSM880; Zeiss). Images were acquired and processed using the ZEN2.3 lite software (Zeiss).

DAPI

Bacteroids were stained for 10 min with DAPI (4, 6‐diamidino‐2‐phenylindole) 50 μg ml⁻¹. Fluorescent signals from DAPI‐stained bacteroids (excitation wavelength, 405 nm; detection wavelength, 425–514 nm) were detected by confocal laser scanning microscopy (LSM880; Zeiss). Images were acquired and processed using the ZEN2.3 lite software (Zeiss).

Semi‐thin resin section

Semi‐thin resin sections were performed using Technovit (Kulzer) as described in Magne et al. (2018). Briefly, samples were fixed in 0.05 M sodium cacodylate buffer, pH 7, 1% (v/v) glutaraldehyde and 4% (v/v) formaldehyde by 15 min of infiltration under vacuum (c. 500 mm Hg) and incubated overnight at 4°C. Once dehydrated by successive ethanol baths, samples underwent three successive Technovit stock solution (3 : 1 [v/v], 1 : 1 [v/v] and 1 : 3 [v/v]) baths and three 1 h 100% Technovit stock solution baths at 4°C under agitation. Samples were included in Technovit resin using Teflon Histoform S embedding molds (Heraeus Kulzer). Technovit sectioning was carried out using an RM 2155 microtome (Leica) and tungsten disposable blade (TC‐65; Leica).

In situ hybridization

In situ hybridization (ISH) was based on the procedure described in Laufs et al. (1998). Twenty‐one dpi nodule segments were fixed, dehydrated and embedded. RNA antisense and sense probes were generated using, as substrate, the full‐length coding sequence of MtGRF8 and the MtSymCRK kinase domain, amplified by PCR from 21 dpi nodule cDNA of R‐108. Antisense and sense RNA probes were labelled with digoxigenin (DIG)‐11‐UTP (Sigma) by in vitro transcription and used to hybridize nodule sections (8 μm) cut by Leica RM2165 microtome and embedded in paraffin (Surgipath Paraplast Plus; Leica). Immunodetection of the DIG‐labelled probes was performed using an anti‐DIG antibody with phosphatase as described by the manufacturer (NBT + BCIP; Promega). Slides were analyzed with AzioImager Z (Zeiss).

Results

MtSymCRK interacts with AtGENERAL REGULATORY FACTOR 8

To better characterize the MtSymCRK signaling pathway, we searched for direct MtSymCRK protein interactor partners. Here, we hypothesized that MtSymCRK could interact with conserved protein partners in A. thaliana. We thus screened an A. thaliana ORFeome by Y2H using the Medicago protein as bait.

As transmembrane domains might interfere with protein transport and targeting in Y2H assays, we independently cloned the extracellular domain, the kinase domain and the full‐length cDNA of MtSymCRK (Medtr3g079850, SymCRK^EX, SymCRK^KIN and SymCRK^FULL; Fig. 1a; Table S1). These constructions were used as baits and screened against a library of 12 000 A. thaliana sequence‐checked ORFs (12k_space; InterAtome) via a Y2H mapping liquid pipeline to identify PPIs (Castandet et al., ²⁰²³).

Fig. 1.

Description of the Medicago littoralis R‐108 Symbiotic CYSTEINE‐RICH RECEPTOR‐LIKE KINASE (MtSymCRK)‐derived baits used in the yeast two‐hybrid (Y2H) screen and of the identified interacting proteins. (a) Schematic representation of MtSymCRK (annotated SymCRK) coding sequence with corresponding protein domains. The three bait constructs used in the Y2H screen, SymCRK^EX, SymCRK^KIN and SymCRK^FULL, correspond to the extracellular domain, the kinase domain and the full‐length MtSymCRK, respectively. The lengths of the different sequences are indicated between brackets in base pairs. CD, cysteine/aspartate kinase motif; DUF26, domain of unknown function 26; KINASE, kinase domain; SP, signal peptide; TM, transmembrane domain. (b) List of the MtSymCRK interacting proteins identified via Y2H. GRF8, GENERAL REGULATORY FACTOR 8; CSN5B, AtCOP9‐SIGNALOSOME 5B; bZIP, BASIC LEUCINE ZIPPER‐MUK11.16; DRB4, DOUBLE‐STRANDED RNA BINDING PROTEIN 4; PTAC14, PLKASTID TRANSCRIPTIONALLY ACTIVE 14; APC8, ANAPHASE PROMOTING COMPLEX SUBUNIT 8; KIN7.4, KINESIN MOTOR 7.4.

No PPIs were identified when using the extracellular domain or the full‐length protein of MtSymCRK. By contrast, the use of the kinase domain SymCRK^KIN leads to the identification of seven PPIs (Fig. 1b), thereby demonstrating the feasibility to identify PPIs using an inter‐species genetic translation from Brassicaceae towards Fabaceae. Among the seven identified A. thaliana PPIs using the MtSymCRK kinase domain as bait, the GENERAL REGULATORY FACTOR 8 (AtGRF8; AT5G65430), which belongs to the 14‐3‐3 protein family, was selected for further investigations. AtGRF8 encodes a 14‐3‐3 protein involved in the regulation of A. thaliana immunity under pathogenic interaction with bacteria (Nomura et al., ²⁰⁰⁶; Gampala et al., ²⁰⁰⁷; Lozano‐Durán et al., ²⁰¹⁴). A similar function was also shown for SlTFT1, the homolog of AtGRF8 in Solanum lycopersicum (Kim et al., ²⁰⁰⁹; Taylor et al., ²⁰¹²). In addition, in A. thaliana, Oryza sativa and Hordeum vulgare, GRF8‐LIKE and other 14‐3‐3 proteins were described as positive regulators of ethylene biosynthesis, a hormone known to control senescence and immunity in plants (Alexander & Morris, 2006; Yao et al., ²⁰⁰⁷; Yoon & Kieber, ²⁰¹³; Lee et al., ²⁰²¹).

Since AtGRF8 and MtSymCRK are both associated with immunity, we further studied the role of MtGRF8 in Medicago in relation to MtSymCRK.

MtSymCRK interacts with M. littoralis GRF8 orthologs (MtGRF8)

To perform translational genetics toward Medicago, we searched for the homolog of AtGRF8 in R‐108. In a previous study, Qin et al. (2016) identified ten 14‐3‐3 in A17 and named them 14‐3‐3a to 14‐3‐3j. We applied reciprocal BLAST analysis between A. thaliana and Medicago and carried out a phylogenetic analysis (Fig. S1). Maximum likelihood‐based phylogenetic analysis showed that Medtr3g100620, also known as Mt14‐3‐3f (Qin et al., ²⁰¹⁶), is the closest homolog of both AtGRF8 and SlTFT1, with 83% and 84% identity, respectively. In this work, we changed the name of Mt14‐3‐3f to MtGRF8. Genomic analysis suggests that MtGRF8/Mt14‐3‐3d, Mt14‐3‐3i/Mt14‐3‐3e and Mt14‐3‐3 g/Mt14‐3‐3 h pairs are paralog proteins (Qin et al., ²⁰¹⁶).

To determine whether MtGRF8 interacts with MtSymCRK, the full‐length MtGRF8 protein sequence (252 aa) was cloned and tested in yeast by Y2H against MtSymCRK^EX, MtSymCRK^KIN and MtSymCRK^FULL. Additionally, we tested for potential MtSymCRK and MtGRF8 homodimerization. We showed that MtSymCRK^KIN fused to the GAL4 DNA binding domain interacts with both full‐length MtGRF8 and AtGRF8 proteins fused to the GAL4 activation domain. The use of 3‐AT competitive inhibitor until 25 mM did not affect the interaction, thereby confirming the strong affinity between MtSymCRK and MtGRF8 in yeast (Fig. 2a). MtSymCRK does not interact with itself in the Y2H test (Fig. 2a), suggesting that this receptor acts as a monomer in the nodule cells. Conversely, we observed the dimerization of MtGRF8 on selective medium supplemented with 25 mM of 3‐AT, as well as on the most stringent selective medium lacking both histidine and adenine, suggesting a strong interaction between MtGRF8 dimers (Fig. 2b).

Fig. 2.

Medicago littoralis R‐108 Symbiotic CYSTEINE‐RICH RECEPTOR‐LIKE KINASE (MtSymCRK)–Medicago littoralis GENERAL REGULATORY FACTOR 8 (MtGRF8) and MtGRF8–MtGRF8 protein interactions. (a, b) Pairwise yeast two‐hybrid (Y2H) assays after 6 d of growth. (a) Protein–protein interactions (PPIs) between MtSymCRK extracellular (SymCRK^EX), MtSymCRK kinase (SymCRK^KIN) and MtSymCRK full‐length (SymCRK^FULL) proteins were tested with SymCRK^EX, SymCRK^KIN, SymCRK^FULL and with MtGRF8 and Arabidopsis thaliana GENERAL REGULATORY FACTOR 8 (AtGRF8) full‐length proteins. Empty vectors were used as negative controls. SymCRK^KIN fused to the GALACTOSE‐RESPONSIVE TRANSCRIPTION FACTOR (GAL4) DNA binding domain (DB) interacts with both MtGRF8 and AtGRF8 full‐length proteins fused to the GAL4 activation domain (AD). These interactions were observed on minimal Synthetic Defined medium (SD) – Leu (‐L) – Trp (‐W) – His (‐H) supplemented with 25 mM of 3‐amino‐1,2,4‐triazole (3‐AT). The panel (a) is a mosaic of pictures cropped from a single picture from which some rows and lines were not considered. (b) PPI between MtGRF8 full‐length proteins. MtGRF8 forms homodimers on minimal SD ‐L‐W‐H supplemented with 25 mM of 3‐AT and on minimal SD ‐L‐W‐H‐ Ala (‐A). Empty vectors were used as negative controls. (c) Split enhanced YELLOW FLUORESCENT PROTEIN (eYFP) assays validating MtSymCRK and MtGRF8 interactions in Nicotiana benthamiana. SymCRK^EX, SymCRK^KIN sub‐domains and MtGRF8 full‐length proteins were fused to the C‐terminal (eYFPc), or the N‐terminal (eYFPn) fragments of eYFP. eYFPc and eYFPn empty constructs were used as negative controls. eYFPn‐SymCRK^KIN and eYFPc‐MtGRF8, eYFPc‐SymCRK^KIN and eYFPn‐MtGRF8, and eYFPc‐MtGRF8 and eYFPn‐MtGRF8 showed positive fluorescent signals indicating PPIs. Interactions between SymCRK^EX and SymCRK^KIN sub‐domains were not tested due to lack of biological relevance. eYFPc and eYFPn empty constructs were used as negative controls. Experiments were performed three times. Bars, 20 μm.

To confirm these interactions in planta, the interaction between the MtGRF8 and the MtSymCRK^KIN was further confirmed by BiFC assays through transient expression in tobacco leaves. We performed BiFC assays to assess PPIs by co‐expressing two sets of fusion constructs: (1) SymCRK^EX, SymCRK^KIN, SymCRK^FUL and MtGRF8, fused to the C‐terminal moiety of the eYFP (eYFPc), and (2) the same proteins, fused to the N‐terminal moiety of the eYFP (eYFPn). Both eYFPn‐SymCRK^KIN/eYFPc‐MtGRF8 and eYFPn‐SymCRK^FULL/eYFPc‐MtGRF8 as well as the reciprocal constructs produced positive fluorescent signals at the plasma membrane, thereby confirming the interaction between MtSymCRK and MtGRF8 (Figs 2c, S2). We thus confirmed that both MtSymCRK^FULL and MtSymCRK^KIN interact with MtGRF8, demonstrating that MtGRF8 is also binding to the kinase domain of a full‐length MtSymCRK.

MtSymCRK did not self‐interact, suggesting again that it does not form homodimers and likely mediates its function as a monomer (Fig. 2c). Unlike MtSymCRK, MtGRF8 formed homodimers in plant cells. Interestingly, MtGRF8 dimers were detected in the cell membrane, cytosol and nucleus (Fig. 2c). To determine more precisely the MtGRF8 cell localization, we transiently expressed MtGRF8 fused to the eGFP reporter gene under the control of the 35S promoter (35S:eGFP:GRF8) in tobacco epidermis cells. We confirmed the ability of MtGRF8 to localize also in the cytoplasm and nucleus in addition to the plasma membrane (Fig. S3).

Together, our data demonstrate that MtSymCRK interacts with MtGRF8 in the cytosol–plasma membrane region.

MtSymCRK^KIN interacts with monomer and dimer of MtGRF8

To better understand the interaction between MtSymCRK and MtGRF8, a 3D modeling was performed using the AlphaFold3 (AF3) software. Several models were generated (Table S4) and the best models are shown in Fig. 3. The kinase domain of MtSymCRK adopts the typical fold of a serine/threonine protein kinase in the Protein Data Bank and is composed of nine alpha‐helices (essentially) and one beta‐sheet (Figs 3a, S4a). The AF3‐generated models differed mainly by the conformation of the C‐terminus region (residues 628–655 in the whole sequence) and its relative positioning with respect to the core of the kinase domain. This region is folded as an alpha‐helix followed by an unstructured region. The location of this C‐terminus last segment can differ greatly, being totally exposed or buried within the core of the kinase domain (Fig. 3a). MtGRF8 monomer is folded as an alpha‐helical bundle composed of nine antiparallel helices (Figs 3b, S4b). The alpha‐helical bundle adopts a concave shape forming a wide groove delimited by helices 3, 5, 7 and 9 (Table S6). In the complex formed between MtGRF8 monomer and MtSymCRK^KIN, the MtGRF8 groove is occupied by the C‐terminus region of MtSymCRK^KIN that extends from residues 273–315 (Fig. 3c). Many contacts are made with H3, H5, H7 helices and mostly with H9 of MtGRF8 (Table S7).

Fig. 3.

MtSymCRK^KIN interaction with Medicago littoralis GENERAL REGULATORY FACTOR 8 (MtGRF8). 3D models of the different partners and interactions. (a) Medicago littoralis R‐108 Symbiotic CYSTEINE‐RICH RECEPTOR‐LIKE KINASE kinase domain (MtSymCRK^KIN, MtSymCRK kinase, in green). (b) MtGRF8 (in purple). (c) The complex of MtGRF8 monomer with MtSymCRK^KIN. (d) MtGRF8 dimer and (e) the complex of MtGRF8 dimer with MtSymCRK^KIN. For each model, the backbone (helices) of the proteins is represented with tubes and the molecular surfaces in transparency.

As MtGRF8 can form a homodimer (Fig. 2), we further investigated the ability of the MtGRF8 dimer to interact with MtSymCRK^KIN. AF3 modeling revealed that in the MtGRF8 dimer, the two monomers assembled in an antiparallel way, with the small N‐terminus helices of chain A making contacts with H3 and H4 of chain B symmetrically (Fig. 3d; Table S8). The structure of the monomer is not modified within the dimer, which results in two large accessible grooves. When MtSymCRK^KIN interacts with the MtGRF8 dimer, the number of contacts is larger (69) compared with the monomer (50), which would contribute to an improved stabilization of the complex (Fig. 3e; Table S9). The same MtGRF8 helices are involved in the interaction as observed with a single chain of the dimer, but in addition, the presence of the second chain has resulted in 24 additional residue contacts. The C‐terminus region of MtSymCRK^KIN seems to participate in the interaction with the MtGRF8 dimer. In addition, 44% of the contacts are made with residues outside of the C‐terminus region. Helical faces of MtGRF8 engaged in the interaction with MtSymCRK^KIN are the same, even if some residues involved in the interaction differed between the two MtGRF8 forms.

We further investigated the importance of the C‐terminal regions of MtSymCRK^KIN and MtGRF8 in the interaction between these two proteins.

Indeed, since among the interfaces identified by the AF3 models, contacts between the MtGRF8 C‐ter (H9) and the MtSymCRK^KIN C‐ter were found frequently, we focused on these two regions and made two constructs by deleting the corresponding regions. These include the C‐terminus (C‐ter) regions of MtSymCRK and the H9 region of MtGRF8, respectively, generating the deleted proteins MtSymCRK^KINΔC‐ter and MtGRF8^ΔC‐ter. For MtSymCRK^KINΔC‐ter, the deleted region extends from residues 610 to 655 (end of full‐length MtSymCRK). For MtGRF8^ΔC‐ter, the deleted region contains the helix H9 and corresponds to the deletion of residues 226 to 252 (end of MtGRF8). The new constructs were tested by BiFC using transient expression in tobacco leaves (Fig. S5).

Our results showed that MtSymCRK^KINΔC‐ter is able to interact with MtGRF8, suggesting that the interaction is maintained despite the C‐ter deletion of MtSymCRK. We can hypothesize that the protein extension corresponding to the deleted region is not absolutely required for the interaction between both partners, suggesting that MtGRF8 interacts before the C‐terminus region of MtSymCRK. In addition, no interaction between MtSymCRK^KINΔC‐ter and MtGRF8^ΔC‐ter was observed, indicating the requirement of the C‐terminus region, particularly the H9 helix of MtGRF8, for this interaction.

Our previous results showed that MtGRF8 can form a homodimer. The new results showed that the interaction between MtGRF8 and MtGRF8^ΔCter is not affected by the C‐ter deletion of one of the two MtGRF8, suggesting that the interaction occurs between the N‐terminus regions of the two MtGRF8 in the homodimer as suggested by the contacts deduced from the models (Table S8).

Altogether, these new data allow us to better understand how MtSymCRK and MtGRF8 interact. Both the experimental results and the models tend to support a role for the MtGRF8 C‐terminal helix in its interaction with the kinase domain of MtSymCRK.

MtSymCRK and MtGRF8 are co‐expressed in nodules

To characterize the interaction between MtSymCRK and MtGRF8 in nodules, we investigated the expression patterns of MtSymCRK and MtGRF8 during symbiosis. MtGRF8 expression was downregulated in the primary roots at 2 and 5 dpi. Then, from 8 dpi the expression of MtGRF8 in nodules stopped decreasing and tended to remain stable whereas the expression of MtSymCRK increased (Fig. 4a). These data indicate that MtGRF8 is expressed both in roots and nodules, whereas MtSymCRK is specifically expressed in nodules (Fig. 4a).

Fig. 4.

Expression kinetics and transcript localization of Medicago littoralis GENERAL REGULATORY FACTOR 8 (MtGRF8) and Medicago littoralis R‐108 Symbiotic CYSTEINE‐RICH RECEPTOR‐LIKE KINASE (MtSymCRK). (a) Eff^−ΔΔCT quantitative reverse transcription polymerase chain reaction (qRT‐PCR) gene expression kinetics of MtGRF8 (Medtr3g100620, red circles) and MtSymCRK (Medtr3g079850, black squares) during nodulation. Left and right y‐axis correspond to MtGRF8 and to the MtSymCRK fold changes, respectively. Gene expression levels (Eff^−ΔΔCT) were normalized against the constitutively expressed genes MtRNA RECOGNITION MOTIF and MtACTIN as well as against the uninoculated primary root samples 0 dpi. Results represent means ± SE of four biological replicates. (b) Spatial expression profiles of MtGRF8 and MtSymCRK in the nodule. MtGRF8 and MtSymCRK are co‐expressed, especially in ZIIp, IZ and ZIII, with total read counts of 11 601 and 22 076, respectively, for the nodule. Expression is shown as the percentage of total reads per gene. Data are derived from available public RNA‐seq transcriptomic data (Roux et al., ²⁰¹⁴). ZI, zone I; ZIId, distal zone II; ZIIp, proximal zone II; IZ, interzone; ZIII, zone III. (c, d) In situ hybridization (ISH) on 21 d post‐inoculation (dpi) nodule sections for MtSymCRK (i and iii) and MtGRF8 (ii and iv). (c) entire nodule, (d) zoom in the ZIII are indicated by red dotted squares. Digoxigenin‐labelled antisense (i, ii) and sense RNA probes (iii, iv) were used for ISH. Purple precipitate as a 5‐bromo‐3‐chloro‐3‐indolylphosphate/nitroblue tetrazolium (BCIP/NBT) reaction product indicates transcript localization. Bars correspond to (c) 200 μm and (d) 100 μm. ZI, meristematic zone; ZII, infection zone; ZIII, nitrogen fixation zone. IC, infected cell; UC, uninfected cell.

We further investigated the expression profiles of MtSymCRK and MtGRF8 across the different nodule developmental zones. Based on available public RNA‐seq transcriptomic data, MtSymCRK and MtGRF8 were found to be highly expressed in A17 nodules, with 22 076 reads and 11 601 reads, respectively. MtSymCRK and MtGRF8 were indeed co‐expressed in different nodule zones, especially in the proximal infection/differentiation zone (ZIIp), in the interzone (IZ) and in the nitrogen fixation zone (ZIII, Fig. 4b; Roux et al., ²⁰¹⁴). The localization of MtSymCRK and MtGRF8 transcripts was investigated in 21 dpi mature nodules by ISH (Fig. 4c). ISH revealed that MtSymCRK and MtGRF8 were both expressed in nodule infected cells (Fig. 4d). These results show that MtGRF8 and MtSymCRK are co‐expressed in nodules and support the physiological importance of this interaction in the symbiotic organ.

MtSymCRK interacts with other Mt14‐3‐3 proteins

To determine whether MtSymCRK interacts with other members of the 14‐3‐3 family, we first analyzed the expression pattern of the nine other Mt14‐3‐3 genes during R‐108 ‐ S. medicae WSM419 (hereafter WSM419) symbiosis using qRT‐PCR analysis.

Mt14‐3‐3a, c, e, h and i genes were highly expressed in root tip, in uninoculated and inoculated primary roots, in nodule primordia and nodules as observed for MtGRF8, while Mt14‐3‐3b, d and j were weakly expressed, and Mt14‐3‐3 g expression was not detected (Fig. S6a). In inoculated primary roots, at 2 and 5 dpi, all expressed Mt14‐3‐3 were downregulated compared with the inoculated primary roots at 0 dpi (Fig. S6a). From 8 dpi and later, in nodule primordia and nodules, Mt14‐3‐3a, e, h and i tended to be upregulated compared with non‐inoculated plants (Fig. S6a). Together, these data indicate that Mt14‐3‐3 genes (except for Mt14‐3‐3b, d, g and j) are expressed in nodules, and Mt14‐3‐3i is the most highly expressed member in the symbiotic organ.

We further investigated the Mt14‐3‐3 expression patterns in the different zones of A17 nodules using available public RNA‐seq transcriptomic data (Fig. S6b; Roux et al., ²⁰¹⁴). Mt14‐3‐3a, b, c, e, f, h and i were expressed in nodules with read counts ranging from 4830 to 181 824, with the highest read count observed for Mt14‐3‐3i. Mt14‐3‐3d, g and j were relatively low expressed in nodules with less than 500 reads. Similar expression patterns were observed in Mt14‐3‐3a, c, e, f and i as in MtSymCRK, with low expression in the meristematic zone (ZI) and high expression in the fixation zone (ZIII), while Mt14‐3‐3b and h show opposite expression patterns (Fig. S6b; Roux et al., ²⁰¹⁴).

At last, we tested the PPIs between SymCRK^KIN and Mt14‐3‐3a, b, e, h and i proteins in yeast by Y2H (Fig. S6c). SymCRK^KIN, fused to the GAL4 DNA binding domain, interacts with Mt14‐3‐3b, e, i and MtGRF8 full‐length proteins fused to the GAL4 activation domain on minimal SD‐L‐W‐H with 1 mM of 3‐AT. At 50 mM 3‐AT, only MtGRF8 interacts with SymCRK^KIN, revealing it to be the strongest interacting 14‐3‐3 partner of MtSymCRK.

Our results indicate that, in addition to MtGRF8, MtSymCRK can interact with other Mt14‐3‐3 members. Among them, Mt14‐3‐3 e and i are expressed in nodules and showed the highest binding affinity. We thus tested the interaction of SymCRK^FULL with Mt14‐3‐3e and Mt14‐3‐3i using a BiFC assay (Fig. S2). eYFPn‐SymCRK^FULL/eYFPc‐Mt14‐3‐3i produced a fluorescent signal at the plasma membrane, confirming the interaction between MtSymCRK and Mt14‐3‐3i. A very weak fluorescent signal was also detected for eYFPn‐SymCRK^FULL/eYFPc‐Mt14‐3‐3e. Altogether these results confirm that in addition to MtGRF8, Mt14‐3‐3e and Mt14‐3‐3i are also able to interact with MtSymCRK.

MtGRF8 contributes to the regulation of nodule chronic infection

To characterize the roles of MtGRF8 during root nodule symbiosis, we isolated a Transposon of Nicotiana tabacum1 (Tnt1) insertional mutant line for R‐108 MtGRF8, namely Mtgrf8 (nf2687). Sequencing analysis of the Mtgrf8 locus revealed a Tnt1 insertion in the first exon of MtGRF8, 406 bp after the ATG start codon (Fig. 5a). This insertion results in a predicted truncated protein of 135 amino acids disrupted in its 14‐3‐3 super family domain (www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). The impact of the Tnt1 insertion on the MtGRF8 transcript accumulation was determined by qRT‐PCR in R‐108 and Mtgrf8 backgrounds. Under both in vitro and perlite:sand conditions, the accumulation of MtGRF8 transcripts was reduced by 69 and 78%, respectively (Fig. 5b,c). These results confirmed the knockout nature of the Mtgrf8 Tnt1 insertional mutant line.

Fig. 5.

Molecular and phenotypic characterization of the Medicago littoralis general regulatory  factor 8 (Mtgrf8) mutant. (a) Scheme of the MtGRF8 (Medtr3g100620) genomic sequence with the position of the Tnt1 insertion. Exons and introns are represented by grey boxes and by black lines, respectively. Numbers below exons indicate nucleotide positions. The position of the Tnt1 insertion in the transgenic line Mtgrf8 NF2687 is indicated by a black triangle. The Tnt1 insertion is located 406 bp downstream of the ATG in the first exon. (b, c) Relative expression of MtGRF8 in 21 dpi nodules from R‐108 and Mtgrf8 inoculated with S. medicae WSM419, assessed by quantitative reverse transcription polymerase chain reaction (qRT‐PCR) under in vitro (b) and perlite:sand (c) conditions. Asterisks indicate significant differences (**, P ≤ 0.01; ***, P ≤ 0.001; Mann–Whitney test). Results in (b) and (c) represent means ± SE of two and three biological replicates, respectively. (d) Proportion of different nodule types in R‐108 and Mtgrf8 plants at 14, 21, 28 and 35 d post‐inoculation (dpi) grown in perlite:sand. Nodules were classified as nodule primordia (white), pink fixative nodules (pink), nodules showing a greenish coloration associated with normal senescence (green) and nodules showing brownish coloration (brown). Results represent percentage means ± SE of three biological replicates (n = 60 plants per genotype per time point). Asterisks indicate significant differences (**, P ≤ 0.01; ***, P ≤ 0.001; Student's t‐test). (e) Nitrogenase activity measured by acetylene reduction assay at 14, 21, 28 and 35 dpi in nodulated root systems from R‐108 and Mtgrf8 plants grown under perlite:sand condition. Results represent percentage means ± SE of three biological replicates (n = 60 plants per genotype per time point). Asterisks indicate significant differences (**, P ≤ 0.01; ***, P ≤ 0.001; Student's t‐test). (f) Representative 28 dpi nodules from R‐108 and Mtgrf8. R‐108 nitrogen‐fixing nodule showing a typical pink coloration. Mtgrf8 nodule showing a characteristic brownish coloration at the base of the nodule (black dashed‐line zone). (g) Representative nodule sections of 21 dpi nodules from R‐108 and Mtgrf8 stained with KMnO₄ and methylene blue revealing the presence of phenolic compounds in Mtgrf8 nodules (yellow dashed‐line zone). Bars, (f, g) 500 μm. (h, i) Measurement of phenolic areas in R‐108 and Mtgrf8 nodules at 21 dpi (h) and 28 dpi (i). Each dot corresponds to a nodule section with y‐axis and x‐axis representing the corresponding whole nodule area and phenolic accumulation area, respectively. For R‐108 21 dpi, Mtgrf8 21 dpi, R‐108 28 dpi and Mtgrf8 28 dpi, n = 10, 9, 10, 13 nodule sections, respectively. (j, k) 28 dpi nodules from Mtgrf8 mutant complemented with Prom35S::MtGRF8 (full‐length CDS). The empty vector Prom35S::empty serves as control. (j) global view of nodules from Mtgrf8 mutant expressing Prom35S::MtGRF8 and Prom35S::empty, showing absence and presence (yellow dashed‐line area) of brownish coloration at the base of the nodules, respectively. (k) methylene blue staining of the phenolics in nodules from Mtgrf8 mutant expressing Prom35S::MtGRF8 and Prom35S::empty, showing absence and accumulation (yellow dashed‐line area) of phenolic compounds, respectively. Bars, (j, k) 200 μm.

We then performed symbiotic assays using Mtgrf8 and R‐108 with WSM419 as a symbiotic partner. Nodule phenotypes were analyzed at four different time points: 14, 21, 28 and 35 dpi. Mtgrf8 presented significantly higher percentages of nodules exhibiting brownish coloration at the base of nodules, with 12%, 30%, 21% and 14% more brownish nodules at 14, 21, 28 and 35 dpi, respectively, when compared to R‐108 (Fig. 5d). The presence of a brownish coloration in the nodule is commonly considered as a defense‐like reaction. In addition to necrotic nodules, Mtgrf8 also forms pink nodules, in contrast to Mtsymcrk, which produces only white or brown nodules (Berrabah et al., ^2014b). Acetylene reduction assay revealed that at 21, 28 and 35 dpi, the nitrogenase activity of Mtgrf8 nodules was significantly reduced relative to R‐108 (Fig. 5e). However, at 14 dpi, there was no difference between Mtgrf8 and R‐108 nitrogenase activity despite the presence of more brownish nodules in Mtgrf8 (Fig. 5e).

To characterize defense responses in Mtgrf8, nodule phenotyping was performed. Brownish nodules consist of a range of phenotypes, in which the nodule itself and its brownish area associated with necrosis can vary in size (Fig. 5f). Semi‐thin sections of Mtgrf8 brownish nodules, stained with KMnO₄ and methylene blue, revealed the accumulation of phenolic compounds (Fig. 5g). At 21 and 28 dpi, nodule areas accumulating stained phenolic were increased in Mtgrf8 compared with R‐108 (Fig. 5h,i). This indicates that defense reactions take place in Mtgrf8 nodules, a phenotype reminiscent of Mtsymcrk (Berrabah et al., ^2014b.

To ascertain that the Mtgrf8 phenotype was due to the loss of MtGRF8 function, genetic complementation was performed through hairy root transformation. A construct expressing MtGRF8 (Prom35S::MtGRF8) was introduced into Mtgrf8 roots. Roots transformed with the corresponding empty vector (Prom35S::empty) served as a negative control. Complemented nodules showed a WT phenotype with absence of brownish coloration (Fig. 5j) and phenolic compounds accumulation (Fig. 5k).

MtGRF8 is required for bacteroid differentiation during symbiosis

To evaluate the impact of the suppression of MtGRF8 on bacteroid survival, we assessed the viability of the bacteroids in 28 dpi nodules of Mtgrf8 and R‐108 by LIVE/DEAD staining. R‐108 nodules mainly showed alive differentiated bacteroids, while a mixture of alive and dead differentiated bacteroids was observed in Mtgrf8 nodules (Fig. 6a,b). In addition, histological analysis revealed a drastic reduction in the fixation zone accompanied with a severe extension of the necrotic zone in Mtgrf8 nodules compared with R‐108 (Fig. S7). Within the necrotic zone, infected cells exhibited collapsed vacuoles and patchy colorations, indicating a defense activation and/or premature senescence.

Fig. 6.

Medicago littoralis GENERAL REGULATORY FACTOR 8 (MtGRF8) prevents the death of differentiated bacteroids. (a, b) Live/dead staining of 28 d post‐inoculation (dpi) nodule sections to assess bacteroid viability. (a) R‐108 nodule apex sections showing alive bacteroids (SYTO9, green signal). (b) Mtgrf8 mutant nodule apex sections showing a mixture of alive and dead differentiated bacteroids (propidium iodide (PI), red signal). Bars, upper panels, 200 μm, and lower panels, 10 μm. (c) Measurement of bacteroid length (in μm) in 14, 21 and 28 dpi nodules of wild‐type (WT, R‐108), Mtgrf8 and Medicago littoralis R‐108 symbiotic cysteine‐rich receptor‐like kinase (Mtsymcrk) plants grown in sand/perlite condition by using flow cytometry. Asterisks indicate significant differences (*, P ≤ 0.05; **, P ≤ 0.01; Student's t‐test). (d) Expression analysis of early (NCR121) and late (NCR001, NCR094, NCR099 and NCR109) NODULE‐SPECIFIC CYSTEINE‐RICH (NCR) peptides in 14, 21 and 28 dpi nodules of WT, Mtgrf8 and Mtsymcrk plants grown in sand/perlite condition. For each time point and genotype, data represent the mean expression of three biological replicates. Asterisks indicate significant differences (*, P ≤ 0.05; **, P ≤ 0.01; Student's t‐test). (e, f) Live/dead staining of 28 dpi nodule sections to determine bacteroid viability in complemented Mtgrf8 nodules. (e) Mostly dead bacteroids (PI (Propidium Iodide), red signal, yellow dashed‐line area) are present in nodules of the Mtgrf8 mutant expressing Prom35S::empty. (f) Mostly alive bacteroids (SYTO9, green signal) are detected in nodules of the Mtgrf8 mutant expressing Prom35S::MtGRF8. Bars, 200 μm (g) Nitrogenase activity of 28 dpi nodules from Mtgrf8 mutant expressing Prom35S::empty or Prom35S::MtGRF8, cultivated in sand/perlite condition. Data represent means +/− SE of three biological replicates with 25 plants per genotype and per replicate. Asterisks indicate significant differences (**, P ≤ 0.01; Student's t‐test).

To determine whether bacteroid differentiation is impaired in Mtgrf8 nodules, we measured the bacteroid size at 28 dpi. Based on the length of bacteroid cells, we classified bacteroids as non‐differentiated (cell length ≤ 2 μm; average size of free‐living rhizobia) or as differentiated (cell length > 2 μm). In R‐108 and Mtgrf8, 84 and 39% of the bacteroids were differentiated, respectively (Fig. S8a). To confirm this observation, we performed flow cytometry on R‐108, Mtgrf8 and Mtsymcrk at 14, 21 and 28 dpi. At 14 and 21 dpi, Mtgrf8 and R‐108 exhibited similar differentiated bacteroids, while Mtsymcrk showed significantly less differentiated bacteroids compared with these two genotypes (Fig. 6c). At 28 dpi, in both Mtgrf8 and Mtsymcrk, the size of bacteroids is significantly reduced compared with R‐108 (Fig. 6c). These data suggest that in Mtgrf8 the differentiation process is altered at later stages than in Mtsymcrk.

To further investigate whether the reduction in the differentiated bacteroid population was due to impaired differentiation and/or to an increased death of differentiated bacteroids, we examined the expression of early NODULE‐SPECIFIC CYSTEIN‐RICH PEPTIDES (NCR121) and late NCRs (NCR001, NCR094, NCR099 and NCR109) in 14, 21 and 28 dpi nodules of R‐108, Mtgrf8 and Mtsymcrk (Fig. 6d). In accordance with prior reports (Berrabah et al., ^2014b), Mtsymcrk exhibited persistent repression of late NCR genes at all time points. Only the early NCR121 is transiently upregulated at 14 and 21 dpi in Mtsymcrk but repressed at 28 dpi. By contrast, in Mtgrf8, the expression of all NCR is increased at 14 and 21 dpi while NCR121, NCR099 and NCR109 are expressed at lower levels at 28 dpi. These results suggest that the reduction in the bacteroid size in Mtgrf8 nodules at 28 dpi is associated with a down‐ or mis‐regulation of key late NCR genes, suggesting impaired late‐stage differentiation.

Finally, to confirm the involvement of Mtgrf8 in the alteration of bacteroid survival and differentiation, the Prom35S::MtGRF8 construct was introduced into Mtgrf8. Complemented nodules exhibited restored bacteroid viability, as indicated by LIVE/DEAD staining (Figs 6e,f, S8b), along with recovery of nitrogen‐fixation activity (Fig. 6g). These findings confirm that MtGRF8 is essential to ensure bacteroid differentiation and to prevent bacteroid death.

MtGRF8 is required for CPs expression

To further characterize the immune and senescence responses in Mtgrf8 nodules, we analyzed the expression of CP and PR genes in Mtgrf8 compared with R‐108 at 14, 21, 28 and 35 dpi by qRT‐PCR. The overall expression analysis revealed no significant expression variation of PR genes between Mtgrf8 and R‐108 (Fig. S9a–d). By contrast, a significant induction of CP2, 3 and 4 is observed in Mtgrf8 at 28 dpi compared with R‐108 (Fig. S9c). These results indicate that the defense phenotype of Mtgrf8 is not associated with the upregulation of PR genes, and MtGRF8 is required for the repression of CP genes.

Other Mt14‐3‐3 members potentially compensate partially for MtGRF8 loss of function

Mtgrf8 displayed impaired nodule function and consistently showed weaker induction of PRs compared with Mtsymcrk, suggesting that compensatory mechanisms may modulate defense activation in the absence of MtGRF8. As demonstrated in previous studies, 14‐3‐3 proteins exhibit functional redundancy in plants (van Heusden et al., ¹⁹⁹⁶; Kuromori & Yamamoto, ²⁰⁰⁰; Vasara et al., ²⁰⁰²). Our Y2H assays revealed that MtSymCRK can interact with other Mt14‐3‐3 members (Mt14‐3‐3a, b, e, h and i; Fig. S6c).

In addition, MtGRF8 can interact with Mt14‐3‐3e and Mt14‐3‐3i using a BiFC assay, indicating that MtGRF8 can form heterodimers with these two 14‐3‐3 members in nodules (Fig. S2).

We further analyzed Mt14‐3‐3 expression patterns in Mtgrf8 nodules at 21 dpi. In our experiments, Mt14‐3‐3b transcripts were undetectable, whereas Mt14‐3‐3a, e, h and i genes were expressed at comparable levels to those in R‐108 (Fig. S10). These data show that other Mt14‐3‐3 genes remain transcriptionally active in Mtgrf8 and may partially compensate for loss of MtGRF8.

To test this hypothesis, a study of conservation of amino acids involved in the interaction with MtSymCRK^KIN was performed between MtGRF8 and the nine members of the 14‐3‐3 family. The multiple sequence alignment revealed that the percent of identity for pairs of proteins ranges between 47 and 91% (Fig. S11). The projection of the conserved amino acids onto the 3D structure of the best model showed that the most conserved residues of the helices are facing the groove (Face A, Fig. 3b) unlike those located in the helices outside of the groove (Face B, Fig. 3b). Interestingly, the residues of Face A are those that are predicted to interact with MtSymCRK^KIN, suggesting that the interactions with MtSymCRK might be conserved within the Mt‐14‐3‐3 family.

Suppression of the nitrogen fixation increases defense responses in Mtgrf8 nodules

We showed that Mtgrf8 and Mtsymcrk diverge in the nitrogen‐fixation process as Mtsymcrk nodules are unable to fix nitrogen, unlike those of Mtgrf8. In this work, we also reported that, based on PR markers, defense responses could be reduced in Mtgrf8 nodules, unlike those of Mtsymcrk (Fig. S9; Berrabah et al., ^2014b). A difference in the nitrogen fixation status of nodules might impact defense/senescence responses. To test this hypothesis, R‐108, Mtgrf8 and Mtsymcrk plants were inoculated with the fix‐ strain S. medicae WSM419ΔnifD (hereafter WSM419ΔnifD), which is unable to produce a functional nitrogenase (Magne et al., ²⁰²⁴). The number of nodules showing necrotic vs non‐necrotic nodules was determined at 14, 21 and 28 dpi in plants inoculated with WSM419ΔnifD or WSM419 (Fig. 7). The inoculation of R‐108 with WSM419ΔnifD resulted in a slight increase in necrotic nodules in comparison with WSM419 at all time points. However, at all time points, the number of necrotic nodules significantly increased in Mtgrf8 when inoculated with WSM419ΔnifD compared with WSM419. These results indicate that the suppression of nitrogen fixation by the use of WSM419ΔnifD leads to an increase in necrosis and defense responses in Mtgrf8.

Fig. 7.

Absence of nitrogen fixation increases nodule necrosis in Medicago littoralis general regulatory factor 8 (Mtgrf8). Nodule phenotypes of R‐108, Mtgrf8 and Medicago littoralis R‐108 symbiotic cysteine‐rich receptor‐like kinase (Mtsymcrk) plants inoculated with Sinorhizobium medicae WSM419 or its fix‐ mutant (WSM419 ΔnifD) were assessed at 14, 21 and 28 d post‐inoculation (dpi) under sand/perlite growth conditions. The total number of necrotic or non‐necrotic nodules was determined for the different backgrounds. The inoculation of R‐108 as well as Mtsymcrk with WSM419ΔnifD does not significantly impact the necrotic nodule numbers compared with WSM419. By contrast, the inoculation of Mtgrf8 with WSM419ΔnifD increased necrotic nodules compared with WSM419 at all time points. Data represent percentage means ± SE of three biological replicates. Asterisks indicate significant differences of necrotic nodules (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; Student's t‐test).

Discussion

In this study, we identified interacting partners of MtSymCRK by coupling a high‐throughput Y2H screen of an A. thaliana ORF library and translational genetics towards R‐108. This strategy allows the identification of AtGRF8, a member of the 14‐3‐3 protein family, interacting with the kinase domain of MtSymCRK. Since GRF8‐LIKE proteins and MtSymCRK are both known to regulate plant immunity, we explored the role of the AtGRF8 ortholog in Medicago‐Sinorhizobium symbiosis. MtGENERAL REGULATORY FACTOR 8 (MtGRF8) is the closest ortholog of AtGRF8 in R‐108 and we confirmed its interaction with the MtSymCRK kinase domain in yeast and in planta. MtSymCRK harbors a (pS/pT)X_1‐2‐COOH signature in C‐terminal, which is a preferred binding motif for 14‐3‐3 proteins (Mode III; Yaffe et al., ¹⁹⁹⁷; Coblitz et al., ²⁰⁰⁵). Interestingly, this motif is systematically found in the 3D models of the interaction between MtSymCRK^KIN and MtGRF8. In the literature, 14‐3‐3s have been reported to interact with RD receptor kinases, including AtSOMATIC EMBRYOGENESIS RECEPTOR‐LIKE KINASE 1 (AtSERK1), AtCOLD‐RESPONSIVE PROTEIN KINASE 1 (AtCRPK1), AtCLAVATA 1 (AtCLV1), AtBRASSINOSTEROID INSENSITIVE 1 (AtBRI1), AtBRI1‐LIKE 2 (AtBRL2) and AtBRI1‐ASSOCIATED RECEPTOR KINASE (AtBAK1; Rienties et al., ²⁰⁰⁵; Karlova et al., ²⁰⁰⁶; Chang et al., ²⁰⁰⁹; Shin et al., ²⁰¹¹; Liu et al., ²⁰¹⁷). Here, we reported an interaction between 14‐3‐3 proteins and a non‐RD kinase receptor in the context of root nodule symbiosis.

We demonstrated that MtGRF8 forms homodimers in yeast and in planta. The 3D model analysis brings some support to the ability of MtGRF8 homodimer to interact with the MtSymCRK kinase domain. We also confirmed by BiFC assays that both MtSymCRK kinase domain and MtSymCRK full length are able to form heterodimers with MtGRF8. In A. thaliana and Zea mays, studies attested the homo‐ and hetero‐dimerization of 14‐3‐3 lead to the formation of diverse molecular modulators or adaptors in multiple signaling pathways (Liu et al., ¹⁹⁹⁵; Wu et al., ^1997b; Chang et al., ²⁰⁰⁹; Shin et al., ²⁰¹¹). In A. thaliana, the phosphorylation status of 14‐3‐3ω not only affects homodimer formation, but also modifies the range of 14‐3‐3 family members involved in the heterodimers (Gökirmak et al., ²⁰¹⁵). Interactions between 14‐3‐3s and kinases are complex since phosphorylated 14‐3‐3 proteins can become targets of other 14‐3‐3s (Muslin et al., ¹⁹⁹⁶; de Boer et al., ²⁰¹³). It is therefore possible that the MtSymCRK–MtGRF8 module may produce hetero‐multimers involving other 14‐3‐3 family members (Mt14‐3‐3a, e, h and i; Qin et al., ²⁰¹⁶).

Public transcriptomic data showed that MtSymCRK and Mt14‐3‐3 were concomitantly and highly expressed in M. truncatula A17 nodules (Roux et al., ²⁰¹⁴). Our qRT‐PCR gene expression analysis supported these results and further showed that Mt14‐3‐3 s are expressed in the root tip, in uninoculated and inoculated primary roots, in nodule primordia and in nodules.

Following the inoculation of R‐108 primary root by WSM419, MtGRF8 gene expression was downregulated in roots. Later, during nodule formation, the expression of MtGRF8 was maintained at a reduced level compared with uninoculated roots while MtSymCRK has an opposite expression profile. ISH analyses revealed co‐localization of MtSymCRK and MtGRF8 transcripts in nodule infected cells as also shown by the publicly available RNA‐seq data. However, a discrepancy on expression levels is observed between ISH and RNA‐seq data, which could be explained by the analysis of nodules at different ages. This result is consistent with their role in the control of the host cell infection by rhizobia.

We also confirmed that MtGRF8 localized both in the cytoplasm and nucleus, allowing interaction with the plasma membrane‐anchored MtSymCRK. The MtGRF8 localization is supported by previous proteomic analyses of R‐108, showing an accumulation of MtGRF8 in the cytosol and the microsome fractions (Berrabah et al., ²⁰¹⁸). Together, these data indicate that in R‐108 nodules, MtGRF8 and MtSymCRK interact together during nodule root symbiosis.

To determine the role of MtGRF8 during symbiosis, we characterized the Mtgrf8 loss‐of‐function mutant. When inoculated with WSM419, at 14 dpi, Mtgrf8 produced nodules showing higher NCR expressions than R‐108 despite a similar nitrogen fixation capacity. At 21 dpi, the number of necrotic nodules increased in Mtgrf8 compared with R‐108 and is accompanied by a decline in nitrogen fixation capacity. Interestingly, at this time point, the expression of NCRs remains higher in Mtgrf8 compared with R‐108, suggesting that the reduction in the nitrogen fixation in Mtgrf8 nodules is potentially associated with bacteroid death rather than the alteration of the bacteroids' differentiation. At 28 dpi, the nitrogen fixation was further decreased and associated with both lower expression of some NCRs (NCR121, NCR109 and NCR99) and a reduction in the bacteroid length, suggesting that bacteroid death co‐occurs with the alteration of the differentiation process. Furthermore, histological analyses of Mtgrf8 nodules revealed a premature degradation of infected cells and a reduction in the size of the fixation zone, along with an accumulation of phenolic compounds.

Taken together, our results show that Mtgrf8 presents a symbiotic dysfunction with a progressive worsening compromising bacteroid persistence, differentiation and nodule viability. Gene expression analyses of defense‐ (PRs) and senescence‐ (CPs) marker genes revealed that Mtgrf8 accumulates CPs. By contrast, no significant variation of the expression of PRs genes is observed between R‐108 and Mtgrf8. Our observations indicate that defenses activated in Mtgrf8 nodules are associated with an upregulation of CPs genes rather than PRs genes. This is also in agreement with Pál et al. (2025) study showing an enhanced expression of CPs genes and no change in defense gene expression in fix‐ mutants (dnf4, dnf7‐2, TR183, TRV36 and TR36) associated with the bacteroid death.

When compared to the fix‐ mutant Mtsymcrk, which produces mainly necrotic nodules overexpressing PRs, Mtgrf8 displays a moderate phenotype: it initially forms functional nodules (14 dpi) but finally exhibits a progressive decrease in nitrogen fixation associated with the apparition of defense (necrosis and phenols) and senescence responses co‐occurring with an altered bacteroids differentiation process (28 dpi). The nitrogen fixing ability of Mtgrf8 could potentially reduce defense responses and explain the reduced defense status of Mtgrf8 nodules. Indeed, Mtgrf8 inoculated with WSM419ΔnifD showed an increased number of necrotic nodules compared with WSM419 while slight‐to‐no variation was observed in R‐108 or Mtsymcrk backgrounds, suggesting that MtGRF8 may be involved in the connection between defense and nitrogen fixation. Indeed, 14‐3‐3 proteins also participate in the regulation of the nitrogen metabolism (Zhao et al., ²⁰²¹). It is also well established that 14‐3‐3 proteins show functional redundancy (Darling et al., ²⁰⁰⁵; Le et al., ²⁰¹⁶; Wilson et al., ²⁰¹⁶). To explain the differences between Mtgrf8 and Mtsymcrk, we assessed the role of other members of the 14‐3‐3 family during Medicago‐Sinorhizobium symbiosis. We showed that Mt14‐3‐3a, e, h and i were co‐expressed with MtGRF8 and that Mt14‐3‐3i displayed a higher expression than MtGRF8 in the nodule. To verify the ability of Mt14‐3‐3 proteins to bind the kinase domain of MtSymCRK, Y2H assays were performed and revealed that Mt14‐3‐3a, e, h and i interact moderately with MtSymCRK. Gene expression analyses revealed that these Mt14‐3‐3 genes are expressed in Mtgrf8 nodules. Based on the ability of MtGRF8 to produce homodimers, we cannot exclude the existence of heterodimer between MtGRF8, 14‐3‐3a, e, h and i which could impact the MtSymCRK signaling pathway. This hypothesis is supported by our BiFC results showing that MtGRF8 can form heterodimers with Mt14‐3‐3e and Mt14‐3‐3i in nodules. Interestingly, the comparison of 14‐3‐3 protein sequences revealed a conservation of key amino acids in MtGRF8 and the other 14‐3‐3 members involved in the interaction with MtSymCRK^KIN.

The different behaviors of Mtsymcrk and Mtgrf8 point to an earlier effect of MtSymCRK in the suppression of defenses (before bacteroid differentiation) than MtGRF8 (after bacteroid differentiation). These differences can be explained by the expression of other 14‐3‐3 members in Mtgrf8 nodules, which can compensate for the absence of MtGRF8. However, this compensation remains partial since MtGRF8 is the strongest MtSymCRK interactant. By contrast, MtSymCRK is the sole member of the cysteine‐rich receptor kinase family specifically expressed in the nodules and its absence cannot be compensated resulting in a complete arrest of the signaling pathway. We can conclude that MtSymCRK is responsible for the maximum defense suppression effect, whereas MtGRF8 function can be partially compensated by other 14‐3‐3 proteins.

The partial compensation of MtGRF8 by the other 14‐3‐3 members in Mtgrf8 nodules ends by suppression of differentiated bacteroids and defense responses without PR induction. By contrast, Mtsymcrk shows undifferentiated bacteroids and PR stimulation. It was shown that after bacteroid differentiation, senescence can be activated and resulted in CP induction, which is often associated with a low PR stimulation in fix‐ mutants. We can speculate that the absence of PR induction in Mtgrf8 is correlated to a senescence response rather than a defense response.

Taken together, our results show that MtSymCRK interacts with MtGRF8 to modulate nodule immunity and senescence, and to allow the maintenance of bacteroids in symbiotic cells (Fig. 8a). In Mtsymcrk, MtGRF8 remains functional but it is unable to prevent defense and senescence responses (Fig. 8b), whereas in Mtgrf8, MtSymCRK is functional and can establish interactions with Mt14‐3‐3a, e, h and i, allowing the formation of functional nodules at an early symbiotic stage (until 14 dpi, Fig. 8c). However, these secondary interactions would not be sufficient to maintain the signaling process during nodule lifetime (at least after 28 dpi), ending with the accumulation of fix+/− nodules showing exacerbated defense and senescence responses. This hypothesis is supported by the fact that the downregulation of MtGRF8 in Mtsymcrk has no impact on defense responses of this mutant, suggesting that the function of MtSymCRK remains dominant over that of MtGRF8 in the nodules.

Fig. 8.

MtSymCRK–MtGRF8‐14‐3‐3a, e, h and i functional model in R‐108 nodules. (a) In the R‐108, Medicago littoralis R‐108 Symbiotic CYSTEINE‐RICH RECEPTOR‐LIKE KINASE (MtSymCRK) interacts preferentially with Medicago littoralis GENERAL REGULATORY FACTOR 8 (MtGRF8) to allow the correct control of nodule defense and senescence. Nodules from early (14 dpi) to late (35 dpi) stages are able to fix nitrogen. In our laboratory conditions, nodule senescence occurs at 35 dpi and is accompanied by the reduction in nitrogen fixation. (b) In Mtsymcrk, MtGRF8 is unable to mediate cell signaling associated with MtSymCRK, which leads to early symbiotic dysfunctioning ending by the formation of non‐fixing (fix‐) nodules with exacerbated defense responses. (c) In Mtgrf8, MtSymCRK remains functional and performs interaction with moderate affinity with 14‐3‐3a, e, h and i, enabling the activation of MtSymCRK signaling. The MtSymCRK‐14‐3‐3a, e, h and i interactions are sufficient at the early stage of nodule formation (14 dpi) but failed to maintain correct bacteroid differentiation and persistence in older nodules (21 and 35 d post‐inoculation (dpi)), resulting in progressive increase in senescence/defense responses associated with loss of bacteroid differentiation and increase in the symbiont death.

Our data together with previous studies indicated that defense repression in the nodule involves transcriptional factors, such as MtRSD, which activate the expression of MtSymCRK responsible for nodule defense repression (Bhardwaj et al., ²⁰²⁶). Interaction between MtSymCRK and 14‐3‐3 protein, especially MtGRF8, is therefore essential to prevent cell death and phenolics accumulation.

MtGRF8, 14‐3‐3a, e, h and i potentially participate in multiple signaling pathways to regulate nodule functioning. In A. thaliana, AtGRF8 modulates ethylene signaling pathways through its interaction with At1‐AMINO‐CYCLOPROPANE‐1‐CARBOXYLATE SYNTHASEs (ACSs; Yoon & Kieber, 2013; Lozano‐Durán et al., ²⁰¹⁴; Lee et al., ²⁰²¹). Some 14‐3‐3 proteins can also interact with ETHYLENE‐OVER‐PRODUCER1 (ETO1) and ETO1‐LIKE (EOL) proteins involved in ACS degradation. GRF8/ETO1 and GRF8/EOL interactions prevent ACS degradation (Yoon & Kieber, 2013). Interestingly, the closest homolog of AtETO1 in Medicago, MtETO1‐like 2 (Medtr4g014680), is expressed in nodules (Roux et al., ²⁰¹⁴) indicating that 14‐3‐3‐dependent mechanisms controlling ethylene could be conserved in nodules. MtSymCRK controls symbiotic cell infection at least through the repression of the ethylene pathway and Mtsymcrk shows an upregulation of ETHYLENE RESPONSE FACTOR (ERFs) and Mt1‐AMINO‐CYCLOPROPANE‐1‐CARBOXYLATE OXYDASESs (ACOs; Berrabah et al., ²⁰¹⁸). This suggests that MtSymCRK could control the ethylene response through MtGRF8.

Previous studies on 14‐3‐3 proteins functioning suggest that MtGRF8 could interact with other proteins involved in nodule immunity. 14‐3‐3 proteins can bind and regulate Ca²⁺‐DEPENDENT PROTEIN KINASES (CDPK) and stimulate their Ca²⁺‐dependent kinase activities (Ormancey et al., ²⁰¹⁷). Yu et al. (2018) have shown that the CDPK5 activity is required for the defense response in Mtnad1 nodules having a similar phenotype than Mtsymcrk. 14‐3‐3s also interact with MITOGEN‐ACTIVATED PROTEIN KINASE (MAPK) and influence MAPK signaling pathways (Chang et al., ²⁰⁰⁹; Oh et al., ²⁰¹⁰; Oh & Martin, ²⁰¹¹; Shin et al., ²⁰¹¹). In Lotus japonicus, the inhibition of MAPK signaling is required for an efficient symbiotic interaction. For example, Mesorhizobium loti, the symbiont of L. japonicus, can control immune signaling during early stages of symbiosis by injecting symbiotic effectors, namely Nodule Outer Protein (NOP) L and M, targeting MAPK (Ge et al., ²⁰¹⁶). Together, these observations suggest that MtGRF8 and related 14‐3‐3 proteins may act as regulatory nodes that fine‐tune multiple defense signaling pathways, including calcium, MAPK and phytohormone signaling to maintain nodule homeostasis.

In this study, we present evidence that Mt14‐3‐3 proteins play multiple roles in legume‐rhizobia symbiosis. Our findings indicate that Mt14‐3‐3 proteins contribute to the regulation of nodule defense and senescence. This study contributes to deciphering the signaling pathway of MtSymCRK during the nodule chronic infection.

Competing interests

None declared.

Author contributions

VG, FB, CE, PR and KM conceived the project and designed the experiments. DM performed the Y2H screen. KM and CY made the Y2H constructs, the pairwise Y2H approaches. KM made the BiFC approaches, the sub‐cellular protein localization and the phylogenetic reconstruction. KM and CY performed the qRT‐PCR gene expression analysis. CY performed the Mtgrf8 screening, genotyping, characterization and complementation assay with the help of GB. GB did the in situ hybridization. CY performed the RNAi:MtGRF8 transformation and characterization. CE performed the 3D models. KSM, JW and PR generated the M. truncatula Tnt1 mutant collection. KM, CY, DM, FB, VG, CE and PR analyzed the data. KM, CY and FB made the figures. KM wrote the original draft. KM, FB, CY, PR and VG wrote and edited the final manuscript. CY and KM contributed equally to this work.

Disclaimer

The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.

Supporting information

Fig. S1 Phylogenetic analysis of M. truncatula GRF8 orthologs.

Fig. S2 Interactions between MtSymCRK, MtGRF8, 14‐3‐3e and 14‐3‐3i tested by BiFC in Nicotiana benthamiana.

Fig. S3 Subcellular localization of MtGRF8 in planta.

Fig. S4 Secondary structures delimitations for the kinase domain of MtSymCRK and for MtGRF8.

Fig. S5 Importance of the C‐terminal regions of MtSymCRK^KIN and MtGRF8 in the interaction between these two proteins.

Fig. S6 Expression profiles of 14‐3‐3 family members and their interaction with MtSymCRK.

Fig. S7 Histological analysis of Mtgrf8 nodule structure.

Fig. S8 Bacteroid differentiation in Mtgrf8 nodules.

Fig. S9 Temporal profiling of defense (PR) and senescence (CP) gene expression in Mtgrf8 nodules.

Fig. S10 Expression analysis of Mt14‐3‐3 members in R‐108 or Mtgrf8.

Fig. S11 Sequence alignment of Mt‐14‐3‐3 sequences.

Methods S1 Saccharomyces cerevisiae transformation.

Table S1 PCR primers for gateway cloning strategy.

Table S2 Primers for sequencing and plant genotyping.

Table S3 AlphaFold3 parameters.

Table S4 Confidence scores of 3D models.

Table S5 Primer used for qRT‐PCR.

Table S6 Delimitations of secondary structure of MtGRF8 3D model.

Table S7 Residue contacts between MtGRF8 monomer and MtSymCRK^KIN.

Table S8 Major contacts between chain A and chain B in MtGRF8 dimer.

Table S9 Residue contacts between MtGRF8 dimer and MtSymCRK^KIN.

Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Data availability

The Medicago littoralis Tnt1 mutants described in this manuscript are freely available on request by contacting Prof. Véronique Gruber (veronique.gruber@u-paris.fr) or Dr. Pascal Ratet (pascal.ratet@universite-paris-saclay.fr) from the Institute of Plant Sciences of Paris‐Saclay (IPS2, France). M. truncatula Tnt1 mutants, nf2687 (Mtgrf8) and nf737 (Mtsymcrk), used in this study were generated using funds from the Samuel Roberts Noble Foundation and National Science Foundation (award no. DBI 0703285). They were identified from the Medicago truncatula Mutant Database (https://medicago‐mutant.dasnr.okstate.edu/mutant/index.php). All the relevant data can be found in the manuscript and in its Tables S1 and S5.