Sinorhizobium meliloti succinylated high‐molecular‐weight succinoglycan and the Medicago truncatula LysM receptor‐like kinase MtLYK10 participate independently in symbiotic infection

Summary The formation of nitrogen‐fixing nodules on legume hosts is a finely tuned process involving many components of both symbiotic partners. Production of the exopolysaccharide succinoglycan by the nitrogen‐fixing bacterium Sinorhizobium meliloti 1021 is needed for an effective symbiosis with Medicago spp., and the succinyl modification to this polysaccharide is critical. However, it is not known when succinoglycan intervenes in the symbiotic process, and it is not known whether the plant lysin‐motif receptor‐like kinase MtLYK10 intervenes in recognition of succinoglycan, as might be inferred from work on the Lotus japonicus MtLYK10 ortholog, LjEPR3. We studied the symbiotic infection phenotypes of S. meliloti mutants deficient in succinoglycan production or producing modified succinoglycan, in wild‐type Medicago truncatula plants and in Mtlyk10 mutant plants. On wild‐type plants, S. meliloti strains producing no succinoglycan or only unsuccinylated succinoglycan still induced nodule primordia and epidermal infections, but further progression of the symbiotic process was blocked. These S. meliloti mutants induced a more severe infection phenotype on Mtlyk10 mutant plants. Nodulation by succinoglycan‐defective strains was achieved by in trans rescue with a Nod factor‐deficient S. meliloti mutant. While the Nod factor‐deficient strain was always more abundant inside nodules, the succinoglycan‐deficient strain was more efficient than the strain producing only unsuccinylated succinoglycan. Together, these data show that succinylated succinoglycan is essential for infection thread formation in M. truncatula, and that MtLYK10 plays an important, but different role in this symbiotic process. These data also suggest that succinoglycan is more important than Nod factors for bacterial survival inside nodules.


INTRODUCTION
Sinorhizobium meliloti 1021 is a nitrogen-fixing symbiont of the plants Medicago truncatula cv. Jemalong, M. truncatula ssp. tricycla R108, and Medicago sativa (alfalfa) (Hoffmann et al., 1997;Oldroyd et al., 2011). On roots of compatible hosts, S. meliloti induces nodule organogenesis, and invades and colonizes the developing nodule 1888; Burris, 1974). Invasion of host roots by rhizobia requires that compatible symbiont/host pairs exchange multiple signals that promote bacterial entry. S. meliloti produces a lipo-chitooligosaccharide signal called Nod factor (NF) (Peters et al., 1986;Lerouge et al., 1990) responsible for inducing root hair curling around attached rhizobia. This traps S. meliloti microcolonies within an infection chamber (Fournier et al., 2015). NFs also induce cell division in the root cortex leading to formation of nodule primordia (Timmers et al., 1999;Xiao et al., 2014). Rhizobia access root cortical cell layers through structures called infection threads that initiate from the colonized curled root hairs (CCRHs). An infection thread is a progressive ingrowth of root hair cell membrane that forms a tube populated with S. meliloti and filled with a matrix of bacterial exopolysaccharide (EPS), secreted bacterial proteins, and plant cell wall material (Brewin, 2004;Gage, 2004). S. meliloti propagate in the infection thread during its extension, a process that requires bacterial production of both NFs and the EPS succinoglycan (Jones et al., 2007). Infection threads then extend through each successive cell layer until they reach the inner cortex, where bacteria are endocytosed by proliferating cells of the nodule primordium (Timmers et al., 1999;Xiao et al., 2014).
On the host M. truncatula, S. meliloti succinoglycan is required for symbiosis and cannot be substituted by other polysaccharides produced by this bacterium (Glazebrook and Walker, 1989). It is needed for infection thread progression on M. truncatula A17 (Jones et al., 2008) and on alfalfa (Cheng and Walker, 1998). Both M. truncatula A17 (Jones et al., 2008) and alfalfa (Niehaus et al., 1993) show signs of plant defense when inoculated with the succinoglycan-deficient exoY mutant. Succinoglycan is a polymer of an octasaccharide composed of seven glucose and one galactose sugar that S. meliloti produces in both high molecular weight (HMW) and low-molecular-weight (LMW) forms (Reinhold et al., 1994;Gonz alez et al., 1998). Production of the LMW form is dependent on cleavage by the glycanases encoded by the exoK and exsH genes (York and Walker, 1997;Mendis et al., 2016). Succinyl, pyruvyl, and acetyl groups are added to S. meliloti succinoglycan by, respectively, the products of the exoH, exoV, and exoZ genes (Reuber and Walker, 1993).
Not only is there a requirement for succinoglycan for host plant invasion, but the material that is produced must be properly modified by succinylation. If succinoglycan is not succinylated by ExoH, no functional nodules are formed by S. meliloti 1021 on M. truncatula A17 (Mendis et al., 2016) or on alfalfa (Leigh et al., 1987). Unsuccinylated succinoglycan produced by exoH mutants cannot be cleaved by exoK-or exsH-encoded glycanases, leaving all the unsuccinylated polymer produced by exoH mutants in the HMW form (York and Walker, 1998). We determined that the strict requirement for the succinylated form of succinoglycan for functional nodule formation on M. truncatula A17 is independent of the effect that succinylation has on production of the LMW form by the ExoK and ExsH glycanases (Mendis et al., 2016). S. meliloti strains lacking both glycanases can form a successful symbiosis on M. truncatula A17 despite producing only HMW succinoglycan. However, these 'double-glycanase mutants' have reduced symbiotic efficiency relative to control strains (c. 70% shoot fresh weight of wild-type (WT)-inoculated plants) (Mendis et al., 2016). Strikingly, strains that lack the exoH-encoded succinyltransferase but are otherwise isogenic to the exoKdel/exsH double-glycanase strains, have the same plant growth phenotype as non-inoculated plants or plants inoculated with the succinoglycan-deficient exoY mutant (c. 18% of shoot fresh weight of S. meliloti WT-inoculated plants (Mendis et al., 2016)). Thus, loss of the LMW form of succinoglycan results in reduction of symbiotic efficiency, but the HMW form plays an even more critical role in the symbiosis as long as it is succinylated.
We set out to further characterize the symbiotic phenotypes of exoY succinoglycan-deficient and exoH succinyltransferase-deficient S. meliloti mutants. Specifically, to determine if these strains are equally deficient in invasion, and whether the defect occurs at the same stage in infection. We also aimed to determine how these rhizobial mutants and the LMW succinoglycan-deficient strains interact with M. truncatula plants carrying a mutation in a LysM-receptor-like kinase (LysM-RLK), MtLYK10 (Medtr5g033490). This plant protein is the ortholog of the exopolysaccharide receptor 3 (EPR3) protein from the determinate-nodule-forming legume Lotus japonicus (Buendia et al., 2018). In the symbiotic interaction between the L. japonicus ecotype 'Gifu' and its symbiont Mesorhizobium loti R7A, M. loti EPS facilitates more efficient infection thread formation and nodulation, but EPS is not strictly required for symbiosis (Kelly et al., 2013). Ljepr3 null mutants permit invasion of M. loti exoU mutants that produce a truncated, pentasaccharide form of EPS that blocks M. loti invasion on WT plants ( Kawaharada et al., 2015). This suggests that LjEPR3 plays a role in EPS recognition and restricts infection of bacteria that produce inappropriate EPS (Kawaharada et al., 2015).
We identified a M. truncatula mutant in MtLYK10, and determined how this line interacts with WT S. meliloti and S. meliloti mutants that are completely succinoglycan deficient, those that do not produce LMW succinoglycan and those that produce only unsuccinylated HMW succinoglycan. Our results suggest that S. meliloti succinoglycan has a function in infection of M. truncatula that is independent of LYK10, but is entirely dependent on succinylation of the polysaccharide.

RESULTS
Sinorhizobium meliloti succinoglycan-deficient and succinoglycan-succinyltransferase-deficient mutants form arrested infection threads in both M. truncatula cv. Jemalong and M. truncatula ssp. tricycla R108 ecotypes Analyses of infection threads formed by S. meliloti exo mutants have previously been performed in alfalfa cv. Iroquois (Leigh et al., 1987;Yang et al., 1992;Cheng and Walker, 1998), but the symbiosis between S. meliloti 1021 and M. truncatula has important differences from the symbiosis with alfalfa. Increased production of succinoglycan by S. meliloti 1021 enhances infection on M. truncatula Jemalong, but not on alfalfa (Jones, 2012). A second EPS (called EPSII or galactoglucan) produced by some strains of S. meliloti (but not S. meliloti 1021) can function in place of succinoglycan on alfalfa cv. Iroquois (Gonz alez et al., 1996;Pellock et al., 2000), but not on M. truncatula (Glazebrook and Walker, 1989). Indeed in a similar comparison performed in our growth conditions, we confirmed that EPSII cannot substitute for succinoglycan on M. truncatula cv. Jemalong A17 ( Figure S1). Therefore, we first examined infection threads formed on M. truncatula by S. meliloti exo mutants to determine how the different types of exo mutants are impaired in invasion.
In M. truncatula A17, plants inoculated with an exoY mutant or with S. meliloti 1021 WT look similar at 3 days post-inoculation (dpi), with exoY-inoculated plants not yet having produced the large numbers of failed infection threads that will accumulate over time (Jones et al., 2008). (See Table 1 and Figure S2 for strain descriptions.) However, using cerulean Cyan Fluorescent Protein (cCFP)-expressing bacteria at the very low culture density of OD 600 = 0.001, the near-absence of background fluorescence makes it possible to see subtle differences in infection threads between exoY-inoculated and WT-inoculated roots. On M. truncatula Jemalong super numeric nodules-2 mutant (sunn-2) inoculated with S. meliloti 1021 WT expressing cCFP, infection threads can be observed at 3-5 dpi extended all the way to the base of a root hair cell (Figure 1a). (The M. truncatula sunn-2 mutant is routinely used for infection thread analysis because it produces more infection events without any qualitative differences compared with M. truncatula WT plants (Schnabel et al., 2005, Fournier et al., 2015). In contrast with WT S. meliloti 1021, the S. meliloti exoY mutant at the same time point has formed only short, probably aborted infection threads . The microcolony-containing infection chambers formed by exoY within the CCRHs are much larger than those usually formed by WT bacteria (Figure 1b-d, asterisks).
We next investigated the infection thread phenotypes of an S. meliloti strain that forms only HMW unsuccinylated succinoglycan. To this aim, we used the non-polar exoHKdel/exsH-1345 triple mutant whose symbiotic phenotype mainly results from the lack of the ExoH succinyltransferase (see next paragraph). In M. truncatula sunn-2 at 3 dpi, the exoHKdel/exsH-1345 mutant has formed only excessively large CCRHs (Figure 1e) or aborted infection threads ( Figure 1f). Thus, the infection threads formed by exoY and exoHKdel/exsH-1345 on the M. truncatula sunn-2 line are quite similar in that they initiate some tubular growth but then elongate slowly and struggle to progress. No differences were observed in the infection thread phenotypes of these exo mutants on M. truncatula sunn-2 versus M. truncatula Jemalong A17, and infections initiated by exoY or exoHKdel/exsH-1345 are never observed penetrating a nodule primordium in M. truncatula Jemalong A17 ( Figure S4). Note that the isogenic exoHKdel/exsH mutants 1345, 1343, and 1349 are lacking function of not only the exoH succinyltransferase, but also of the glycanases encoded by exoK and exsH ( Figure S2). Complementation with a plasmid carrying exoH returns these strains to the level of symbiotic function of a non-polar exoK mutant (70% shoot fresh weight of WT) ( Figure S5).
exo mutant infection thread development and nodule formation were also analyzed on the host M. truncatula ssp. tricycla R108. This is the genotype for most Tnt1 retrotransposon-induced plant mutants available for M. truncatula (d 'Erfurth et al., 2003;Tadege et al., 2008). Two S. meliloti strains, 961 and exsH-1317, which have the exo-LAMON operon transcribed from a heterologous promoter and retain the WT copy of the exoH and exoK genes (Figures S2 and S3) were first tested. The exoLAMON operon has been placed under heterologous regulation in these strains to isolate it from polar effects from mutations in the upstream exoH and exoK Mendis et al., 2016). The control strains have the same altered genetic regulation as the exoK and exoHK deletion strains ( Figures S2 and S3). Both strains form fully invaded nodules by 17 dpi on M. truncatula R108, similarly to WT S. meliloti 1021 ( Figure 2). In addition, the double-glycanase exoKdel/exsH-1325 mutant, which only makes HMW succinoglycan (Mendis et al., 2016) also forms invaded nodules on M. truncatula R108, although significantly fewer nodules form compared with inoculation with the two WT 1021, 961 or single mutant exsH-1317 strains ( Figure 2). This is consistent with the fact that inoculation with the double-glycanase exoKdel/exsH-1325 mutant ultimately produces plants with c. 70% the shoot fresh weight of those inoculated with control strains (Mendis et al., 2016). In contrast with these strains, neither the S. meliloti exoHKdel/exsH-1345 mutant nor the S. meliloti exoY mutant forms any invaded primordia or nodules on M. truncatula R108 ( Figure 2). Instead, both strains have an arrested infection thread phenotype on M. truncatula R108 that closely resembles the phenotype on M. truncatula Jemalong. Thus, at 17 dpi, both exoHKdel/exsH-1345 and exoY form excessively large microcolonies in CCRHs, and only very short infection threads that are blocked within root hairs and/or in the outer cortical cell layer above uninvaded nodule primordia ( Figure 2). These infection blocks presumably result in the arrest of nodule development, such that only nodule primordia are formed. The complete lack of successful infection of nodule primordia explains the absence of shoot fresh weight gain by either strain (Mendis et al., 2016).

M. truncatula lyk10 mutant plants show an early infection phenotype with WT S. meliloti
Having established these phenotypes on M. truncatula R108 plants, we studied an R108 line with a Tnt1 insertion in the LjEPR3 ortholog MtLYK10 (Medtr5g033490) (Cheng et al., 2014). Phylogenetic and synteny analysis confirmed the orthology between MtLYK10 and LjEPR3, and revealed that the genomes of some plants, including some  Figure S2. Fabaceae species, encode two MtLYK10 homologs, but not M. truncatula ( Figure S6). qRT-PCR analysis confirmed that the Tnt1 insertion results in a null mutation in MtLYK10 ( Figure S7). WT S. meliloti 1021 was able to form infection threads with a normal appearance on Mtlyk10 mutant plants. However, compared with WT M. truncatula R108 plants, most infection threads remained in epidermal cells on Mtlyk10 mutant roots and significantly fewer infection threads progressed into the cortex at 11 dpi with WT S. meliloti 1021. Consequently, significantly fewer successfully invaded nodule primordia formed on mutant roots (Figure 3a-g). Numbers of uninvaded nodule primordia ( Figure 3a) were not significantly different between genotypes (P = 0.0962), and more were seen on R108 compared with previous work on A17. However, a higher proportion of all nodules/ primordia were uninvaded for Mtlyk10 (66.09%) compared with R108 (28.13%) (Figure 3a). Young nodules of Mtlyk10 plants appeared normal and well-invaded (Figure 3d,g). We also found that nodulation was delayed on Mtlyk10, and that more developed nodules are also equivalently well-invaded in mutant and WT plants ( Figure 4).
We also compared the nodulation ability of two independent lines carrying the Mtlyk10 mutation, selected from the backcross to WT R108 plants. Both Mtlyk10 lines similarly formed fewer invaded primordia and nodules, and fewer Figure 2. Infection and nodulation invasion phenotypes of the succinoglycan-deficient exoY mutant and the exoHKdel/exsH-1345 mutant on Medicago truncatula ssp. tricycla R108. Infection and nodule invasion phenotypes of M. truncatula WT R108 plants were scored 17 dpi following inoculation with six different Sinorhizobium meliloti strains; 1021, 961, exsH-1317, exoKdel/exsH-1325, exoHKdel/exsH-1345 and exoY. The first three of these strains correspond to the unmodified WT strain (1021), a modified WT strain (961), which serves as a control for the expression of exoLAMON genes from a heterologous promoter, and exsH-1317, which is identical to 961 except it lacks the exsH-encoded glycanase. All three strains display an identical symbiotic phenotype. The double-glycanase exoKdel/exsH-1325 mutant (deficient for both the exoK and exsH glycanases) makes only HMW succinoglycan. The exoHKdel/exsH-1345 mutant is isogenic to exoKdel/exsH-1325, except that 1345 is missing the exoH succinyltransferase, and thus makes non-succinylated HMW succinoglycan. exoY is completely deficient for production of succinoglycan.
(a) Quantification of uninvaded nodule primordia, infection threads, and invaded nodule primordia and nodules. Data are the averages of two independent experiments, n = 28 for each strain. Error bars = SEM. Statistically significant differences in invaded primordia and nodules between exoKdel/exsH-1325 and other strains are indicated. *P < 0.05; ***P < 0.001. (b-g) Photographs illustrating these infection phenotypes, all inoculated strains carrying a constitutive hemA-lacZ reporter gene fusion (pXLGD4) for visualization of bacteria in blue. Fully invaded nodules were seen with 1021, 961, exsH-1317 and exoKdelexsH-1325 (b-e, respectively), while only uninvaded nodule primordia were observed with exoHKdel/exsH-1345 and exoY (f and g, respectively). In these latter two cases, rhizobia appear restricted to the epidermis, but apparently continue to multiply in these unsuccessful infection compartments, resulting in excessively large root hair microcolonies. Scale bars: 100 µm.
© 2019 The Authors. cortical infection threads than WT plants ( Figure S8). We also showed that the infection and nodulation phenotype of a homozygous WT line selected from the F2 population of the backcross between the Mtlyk10 mutant and R108 was comparable to the R108 WT line ( Figure S8). Taken together, these data show that MtLYK10 plays a role in the symbiotic process of rhizobial infection thread growth, and by consequence the rate of successful nodulation in M. truncatula.

M. truncatula lyk10 mutant plants show more severe infection phenotypes with S. meliloti succinoglycan mutants
To address the possible implication of MtLYK10 in EPS recognition, we next characterized the symbiotic phenotypes of Mtlyk10 mutant plants with S. meliloti exo mutants. First, we compared the double-glycanase exoKdel/exsH-1325 mutant (only makes HMW succinoglycan) with the appropriate modified WT S. meliloti 961 strain (makes both HMW and LMW succinoglycan), on Mtlyk10 and R108 plants at 14 dpi. As we had observed with the unmodified 1021 WT S. meliloti (Figure 3  Data are the averages of two independent experiments, n = 38 for each genotype. Error bars = SEM. On Mtlyk10 plants compared with R108, significantly fewer cortical infection threads, and significantly fewer invaded primordia and nodules formed. **P < 0.01; ***P < 0.001. The number of uninvaded primordia formed was not significantly different (P = 0.0962) between the two genotypes, but 28.13% (for R108) and 66.09% (for Mtlyk10) of all nodules and primordia were uninfected, as indicated.
(b-g) Photographs illustrating these infection phenotypes, the inoculated strain carrying a constitutive hemA-lacZ reporter gene fusion (pXLGD4)  succinoglycan together with an MtLYK10 mutation creates a stronger infection defect than the absence of LMW succinoglycan alone. WT R108 and Mtlyk10 mutant plants were next inoculated with succinoglycan-deficient S. meliloti exoY, and the S. meliloti exoHKdel/exsH-1345 mutant that only produces non-succinylated HMW succinoglycan. In order to assess both infection and nodulation, we observed plants at 21 dpi. As previously shown ( Figure 2) neither strain successfully infects R108, and in addition, we observed no successful infections with either strain on Mtlyk10 plants (Figure 6a-i). Indeed, on Mtlyk10 mutant plants both strains induced mainly uninfected bumps corresponding to uninvaded nodule primordia with large, abnormal colonies of rhizobia that remain superficial to these structures (Figure 6g,i). These infection events were scored as short epidermal infection threads (Figure 6a), illustrated in Figure 6(f) (exoY/Mtlyk10) and Figure 6(h) (exoHKdel/exsH-1345/Mtlyk10). Some more developed, but nonetheless also aborted, infection threads ('long epidermal infection threads') were observed with both strains on WT R108 plants (for example, Figure 6d shows a long epidermal infection thread for exoHKdel/exsH-1345/R108), but significantly fewer of such long, aborted infection threads were observed on Mtlyk10 mutant roots compared with WT R108, with both exo mutants ( Figure 6a). Thus, the severe infection phenotype of exoY and exoHKdel/exsH-1345 on WT plants was even stronger on Mtlyk10 mutant plants. Taken together, none of the rhizobial mutant phenotypes were suppressed by the mutation in MtLYK10, and in fact, the phenotypes were worse with each of the three S. meliloti exo mutants (exoKdel/exsH 1325, exoY and exoHKdel/exsH-1345).
Rescue in trans by a strain producing normal succinoglycan indicates that succinoglycan-defective mutants are at a disadvantage in mixed inoculation nodules, with non-succinylated succinoglycan being even more of a disadvantage than the absence of succinoglycan Rescue of failed infection by succinoglycan-deficient exo mutants has previously been accomplished by providing WT succinoglycan by co-inoculation with NF-deficient mutants (Klein et al., 1988). Each strain individually is impaired in invasion, but when NF and succinoglycan are provided in trans within an infection thread, both strains succeed in invading. We used this in trans rescue method to investigate whether the non-succinylated succinoglycan produced by exoHKdel/exsH-1345 has a restrictive effect on the invasion of a strain that produces normal succinoglycan. To determine if a NF-deficient S. meliloti mutant rescues the exoHKdel/exsH-1345 mutant with the same efficiency with which it rescues the succinoglycan-deficient exoY mutant, we performed co-inoculations with a mutant in the NF N-acetylglucosaminyltransferase gene, nodC (Jacobs et al., 1985;Atkinson and Long, 1992). If the nodC mutant rescues exoHKdel/exsH-1345 less efficiently than it rescues exoY, it suggests that any restrictive interaction between non-succinylated succinoglycan and the plant is dominant over the reaction with WT succinoglycan. If nodC rescues exoHKdel/exsH-1345 equal to or more efficiently than it rescues exoY, it suggests that any restrictive interaction is not dominant. M. truncatula A17 WT plants were inoculated with 1:1 mixtures of either nodC + exoY or nodC + exoHKdel/exsH-1345, and nodule formation tracked over 28 days. As shown in Figure 7 formation, in contrast with the negative controls of individual strain inoculations. Compared with nodC + exoY, the nodC + exoHKdel/exsH-1345 mix was less efficient, with fewer plants forming nodules. In both cases relatively few nodules were observed (<1 nodule/plant) (Figures 7 and  S9). To determine strain occupancy, mature nodules were surface sterilized, then crushed and bacterial growth was quantified on antibiotic selection plates (for nodC) or on Calcofluor (for exo strains). This showed that both types of nodules contained a clear majority of nodC bacteria (90-95%), and that exoY was significantly more present in mixed inoculation nodules compared with exoHKdel/exsH-1345 (Figure 7b). To visualize nodule occupancy by the exo mutant strains, plants were co-inoculated with the same mixes of S. meliloti strains, but using exo strain versions expressing beta-galactosidase. Unlike for recoverable rhizobia, the in situ staining via lacZ of rhizobia inside nodules also enables visualization of terminally differentiated bacteria that are internalized within host cells. Coloration of well developed nodules of both types confirmed the presence of the exo strains in nodules, and again indicated that exoHKdel/exsH-1345 is less present compared with exoY (Figure 7c-j). Thus, nodC + exoY nodules clearly showed the presence of exoY, usually throughout inner nodule tissue (Figure 7c-f). The dense staining of the exoY rhizobia within nodules suggests that they are internalized into cortical cells. In contrast, nodC + exoHKdel/exsH-1345 nodules showed only quite limited, patchy areas of the exoHKdel/ exsH-1345 mutant within inner nodule tissue (Figure 7g-j). This suggests that the exoHKdel/exsH-1345 mutant has a very limited, if any, ability to become terminally differentiated into bacteroids. In both cases, smaller nodule-like structures that were visualized along with the mature nodules at 18-21 dpi, were not invaded by the exo strains. Co-inoculation studies using R108 were only feasible in the presence of the ethylene biosynthesis inhibitor aminoethoxyvinylglycine (AVG) (extremely few nodules formed without AVG). Ethylene is a well known negative regulator of rhizobial infection, and symbiotic responses and nodulation are often better when AVG is present (Suzaki et al., 2015). In these conditions, co-inoculated R108 plants did not show clear differences in nodulation ability by the two mixes of rhizobial strains. Nonetheless, coloration of nodules clearly showed exoHKdel/exsH-1345 was less persistent in mixed inoculation nodules compared with exoY ( Figure S10), as observed on M. truncatula A17. Taken together, these data indicate that early events of infection by trans-complementation of NF-deficient and succinoglycan-deficient/aberrant rhizobial mutants lead to fully formed nodules in which the main occupant is the strain with normal succinoglycan (nodC). Furthermore, the exoHKdel/exsH-1345 mutant producing non-succinylated succinoglycan appears at even more of a disadvantage than the succinoglycan-deficient exoY strain in mixed inoculation nodules.

DISCUSSION
Our results show that the succinoglycan-deficient exoY mutant of S. meliloti is completely blocked in invasion of M. truncatula, and only forms very short infection threads that do not elongate and do not reach beyond the epidermis. The exoHKdel/exsH-1345 mutant that produces succinoglycan missing the succinyl group is as impaired in invasion as the succinoglycan-deficient exoY mutant, with infections being similarly restricted to the epidermis. These infection phenotypes explain the previous finding that inoculation of M. truncatula A17 with either mutant results in the same plant growth phenotype as non-inoculated plants (Jones et al., 2008;Mendis et al., 2016). Thus, no successful infections are possible without succinylated succinoglycan. Unlike these two non-invading mutant strains, we had previously observed that the S. meliloti exoKdel/ exsH-1325 mutant, which only produces succinylated HMW succinoglycan, invades, but has a reduced symbiotic efficiency (Mendis et al., 2016). In this study, we now show that the S. meliloti exoKdel/exsH-1325 mutant forms fewer invaded nodules than WT strains, so the enhancement that production of the LMW form of succinoglycan provides to the symbiotic productivity of S. meliloti on the host M. truncatula is due to increased infection thread progression, which results in more nitrogen-fixing nodules. Thus, succinylated succinoglycan plays a critical and early role in infection thread progression in M. truncatula root hairs. Rhizobial NFs are also essential for the infection process in M. truncatula, as in the majority of rhizobial host plants (D enari e et al., 1996). Mutant M. truncatula plants expressing reduced levels of the M. truncatula NF receptor proteins MtNFP or MtLYK3 exhibit arrested infections with unusually large microcolonies of rhizobia and sac-like infection events (Arrighi et al., 2006;Smit et al., 2007), which resemble to some extent, those we observed with S. meliloti exoY or exoHKdel/exsH-1345. Thus, NF perception and the presence of succinylated succinoglycan are both needed for the correct initiation, maintenance, and direction of polar growth of infection threads, essential processes for successful infection of M. truncatula. In contrast with NFs, succinoglycan is not essential for root hair curling and rhizobial entrapment in root hair curls, neither for infection thread initiation nor the induction of nodule primordia formation, such that S. meliloti exoY or exoHKdel/exsH-1345 mutants induce uninvaded nodule-like structures, unlike NF-deficient S. meliloti mutants (D enari e et al., 1996). This suggests that an early symbiotic role of succinoglycan is probably restricted to a local function for infection thread elongation, unlike the pleiotropic roles of NFs.
The rescue of both S. meliloti exoY and exoHKdel/ exsH-1345 by the NF-deficient nodC mutant shows that the prevention of epidermal penetration by both the succinoglycan-deficient exoY mutant and the exoHKdel/exsH-1345 mutant is relieved when normal succinoglycan is supplied. However, complementation efficiency was relatively low, and low percentages of exo rhizobia were recovered from within mature nodules. This suggests that these mutants were both at a disadvantage compared with the nodC mutant in nodules, and that once nodules are formed, the rhizobial production of NFs is less important than the extracellular rhizobial succinoglycan. Alternatively, differences in the quantities of NFs and succinoglycan could be required during nodulation. Laser-capture microdissection RNA-Seq experiments show that exo and nod genes are expressed in apices and nitrogen-fixing zones of M. truncatula nodules (Roux et al., 2014). A disadvantage for internalization into and/or survival within symbiosomes relative to the nodC mutant might be due to greater sensitivity on the part of exo mutants to conditions within the symbiosome. As the dominance of nodC in exoHKdel/exsH-1345 + nodC nodules (approximately 98%) was even more marked compared with exoY + nodC nodules (approximately 88%), we can hypothesize that the production of only non-succinylated succinoglycan renders rhizobia even more sensitive than the complete absence of succinoglycan production. However, if the plant does actively resist invasion by strains producing structurally aberrant succinoglycan, this is only partially effective and the complete absence of succinoglycan is also resisted.
We have also shown that M. truncatula plants carrying a Tnt1 insertion in MtLYK10 have clear, early infection phenotypes, the severity of which depends on the S. meliloti strain tested. In response to WT rhizobia, Mtlyk10 plants form fewer cortical, successful infection threads and consequently fewer invaded nodules, compared with WT plants. Although Mtlyk10 mutant plants were clearly defective in infection, nodule primordia and some invaded nodules can be formed, indicating that the NF-activated process of nodule organogenesis is independent of MtLYK10. Mtlyk10 mutants are thus symbiotic M. truncatula mutants that can uncouple infection and nodule organogenesis (Murray, 2011), revealing another plant component that is specifically involved in the infection process in M. truncatula. The expression profile of MtLYK10 is compatible with its role in infection, with strong induction in root hairs and nodules, and in response to either NFs or rhizobial inoculation, like many infection-related genes (Breakspear et al., 2014;Roux et al., 2014;Larrainzar et al., 2015;van Zeijl et al., 2015;Damiani et al., 2016;Jardinaud et al., 2016;Liu et al., 2019). Within M. truncatula nodules, MtLYK10 is one of the most highly expressed LysM receptor genes, with strong expression in the distal infection zones of nodules, similarly to MtNFP and MtLYK3 (Roux et al., 2014;Bono et al., 2018).
In response to the S. meliloti exo mutants studied here, the infection phenotype of Mtlyk10 mutant plants was consistently more severe compared with S. meliloti WT strains on these plants. The S. meliloti exoKdel/exsH-1325 mutant forms very few invaded nodules on Mtlyk10 mutant plants, while no invaded nodules and very few extended infection threads were observed with either the exoY or the exoHKdel/exsH-1345 strains. Thus, the infection deficiencies conferred by defects in succinoglycan or mutation in MtLYK10 were apparently additive, indicating different mechanisms by which succinoglycan and MtLYK10 play a role in the infection process. Furthermore, if the infection deficiency of the S. meliloti exoY mutant is due to perception of bacterial molecular patterns exposed by the absence of succinoglycan, the restriction is not accomplished through MtLYK10.
In L. japonicus, the octasaccharide EPS produced by M. loti is also implicated in symbiosis. However, there are important differences between the two model nodulation systems in the role of EPS in infection thread formation. Notably, a M. loti R7A exoB mutant that does not make EPS can invade L. japonicus, albeit with reduced efficiency (Kelly et al., 2013;Kawaharada et al., 2015). Remarkably, in M. loti, structurally aberrant polysaccharide produced by the exoU mutant creates a much greater L. japonicus invasion defect than the complete absence of the EPS in an exoB mutant (Kelly et al., 2013;Kawaharada et al., 2015). It is not possible to perform a completely identical comparison between the two symbiont/host systems. There is no known mutant of S. meliloti that is comparable with the exoU mutant of M. loti in forming a truncated pentasaccharide EPS monomer. A glycosyltransferase that has been designated exoU in S. meliloti is known to function in the succinoglycan biosynthesis pathway, but this mutant does not secrete any succinoglycan (Reuber and Walker, 1993;Becker et al., 1993c). We have confirmed that this mutant has a succinoglycan-deficient phenotype similar to an exoY mutant, in terms of succinoglycan production ( Figure S11). Nonetheless, we can conclude that the interaction between the exoHKdel/exsH-1345 mutant and M. truncatula is different from the interaction between the exoU mutant of M. loti that produces truncated pentasaccharide EPS and L. japonicus. Another major difference between the two systems is that L. japonicus mutants in the MtLYK10 ortholog, LjEPR3, partially suppress the phenotype of a M. loti exoU mutant (Kawaharada et al., 2015), while none of the S. meliloti exo phenotypes we have characterized here are suppressed by the Mtlyk10 mutation. In fact, infection deficiencies are accumulated by the combination of plant (Mtlyk10) and rhizobial (exoKdel/exsH-1325, exoHKdel/exsH-1345 and exoY) mutations, indicating that each succinoglycan-related infection phenotype is independent of MtLYK10. In other words, unlike the proposed role for LjEPR3, our data suggest that S. meliloti succinoglycan and MtLYK10 perform important, but different, roles in infection thread formation in M. truncatula. Thus, although we cannot exclude a role for MtLYK10 in recognizing a form of EPS not tested here, we favor the hypothesis that MtLYK10 does not intervene in EPS recognition. Despite the apparent difference between plant species for EPS recognition, Ljepr3 and Mtlyk10 mutants have similar phenotypes with WT rhizobia, both forming fewer invaded nodules than their corresponding WT plants, within 14 dpi. Also, these Ljepr3 and Mtlyk10 phenotypes are both weaker than those of the appropriate EPS-minus rhizobial strains, M. loti exoB and S. meliloti exoY, respectively, on their WT host plants. This leaves open the possibility that an unknown EPS perception component exists in L. japonicus. So far, the few LysM-RLKs that have been characterized in different plant species have mostly been linked to roles in recognizing chitin oligomers or lipo-chitooligosaccharides, but as pointed out by Kawaharada et al. (2015), such proteins have LysM domains that are quite divergent from those of MtLYK10/LjEPR3 homologs. With the notable exception of Arabidopsis thaliana (Buendia et al., 2018), these MtLYK10/ LjEPR3 homologs are widespread in Angiosperms. Interestingly, some dicot species have two copies of this gene as the result of ancient duplications ( Figure S6), followed potentially by neofunctionalization or subfunctionalization. This could point toward different functions of MtLYK10type genes in different plant species and this gene is not always involved in nodulation, as shown recently in Parasponia andersonii van Velzen et al., 2018). This may be linked to differences in rhizobial infection mechanisms among plant species.
It is not yet clear what role(s) succinylated succinoglycan plays in infection thread development. exoY, exoH, and exoK transposon-insertion mutants are sensitive to acidic pH (Hawkins et al., 2017). However, acid sensitivity could not be the only cause of the symbiotic defect in these strains. exoK mutants are just as sensitive as exoH mutants to acid, yet the symbiotic defect of exoK mutants is much less severe than that of exoH mutants. There is some evidence for succinoglycan interaction with reactive oxygen species (ROS) produced in the infection thread. One possible role is protection of the bacteria from ROS. When tested in the free-living state, succinoglycan-overproducing strains are c. 1.5-2-fold more resistant than WT S. meliloti to H 2 O 2 , whereas the exoY mutant has c. 0.4-0.5 the resistance to H 2 O 2 of the WT (Lehman and Long, 2013). But the symbiotic phenotypes of S. meliloti mutants in genes involved in coping with reactive oxygen species do not closely resemble those of exo mutants. The extremely H 2 O 2 -sensitive S. meliloti oxyR mutant has no symbiotic defect (Jamet et al., 2005). Also, an S. meliloti double-mutant in two catalase-encoding genes katB and katC has a symbiotic defect, but in contrast with exo mutants, it forms numerous infection threads that do not appear to fail until the time of S. meliloti release into symbiosomes (Jamet et al., 2003). Thus, it seems unlikely that sensitivity to ROS could be the only cause of the symbiotic defect of exo mutants.
Another recent article has shown that strains that produce an excess of succinoglycan are less sensitive to the NCR247 cationic peptide than is S. meliloti WT (Arnold et al., 2018). However, the exoY mutant was found to not be more sensitive than the WT to NCR247, at least in the freeliving state, and addition of non-succinylated succinoglycan was as successful as wild-type succinoglycan in enhancing survival of free-living S. meliloti treated with NCR247 (Arnold et al., 2018). NCR peptides are secreted into the symbiosome by most legumes that form indeterminate nodules, including M. truncatula and alfalfa (Alunni and Gourion, 2016). These peptides play multiple roles in the process of bacteroid differentiation and in cell fate (Alunni and Gourion, 2016). A role has not yet been demonstrated for NCR peptides in root hairs and infection threads, and bacteria within M. truncatula infection threads do not show signs of bacteroid differentiation, but expression of NCR247 is strongly increased in root hairs by 3 dpi with S. meliloti (Breakspear et al., 2014). A difference in sensitivity to NCR peptides might explain the disadvantage in nodule occupancy that the exoY and exoHKdel/exsH-1345 mutants appear to have relative to the nodC mutant in our co-inoculation experiments. More than 600 NCR genes are induced in M. truncatula by S. meliloti, and in vitro tests with NCR peptides show increased sensitivity and different responses of a S. meliloti succinoglycan-deficient mutant compared with WT bacteria (Montiel et al., 2017). Also, EPS-minus mutants of Sinorhizobium fredii HH103 can form functional nodules on soybean and pigeonpea (Parada et al., 2006), species which lack NCR genes (Montiel et al., 2016), and also on Glycyrrhiza uralensis (Margaret-Oliver et al., 2012), in which the expression of only a few NCR genes is detected (Montiel et al., 2017). Taken together, these data suggest a link between rhizobial EPS and the mediation of NCR peptide effects on differentiation and survival of endophytic rhizobia at different stages of nodulation.
Thus, the essential function of succinylated succinoglycan in Medicago infection might involve interaction with an as-yet-unidentified plant receptor, or with another component of the infection thread matrix. Alternatively, succinylated succinoglycan could have a role in protecting the bacteria from NCR peptides or an as-yet-unidentified stress within the infection thread.

Plant genotypes
Two WT and two mutant genotypes of Medicago truncatula were used. Medicago truncatula cv. Jemalong A17 and M. truncatula ssp. tricycla R108 are WT lines. The M. truncatula Jemalong super numeric nodules-2 mutant (sunn-2) carries a mutation in the MtSUNN gene, leading to loss of the autoregulation of nodulation and consequently to increased infection events (Schnabel et al., 2005;Fournier et al., 2008). The M. truncatula ssp. tricycla Mtlyk10 (NF4929) mutant carries a Tnt1 retrotransposon in LysM1 of the MtLYK10 gene (Medtr5g033490). Two independent backcrossed lines of this mutant were used, most of the results presented were obtained using the Mtlyk10#46 line (see Figure S8). A line that is homozygous for the WT MtLYK10 gene was also selected after backcross and checked for its symbiotic phenotype (see Figure S8).

Real time quantitative RT PCR analysis on M. truncatula plants
qRT-PCR expression analysis of MtLYK10 was performed on M. truncatula WT (R108) and the Mtlyk10 mutant. Expression was analyzed in the non-inoculated and nodulated root segments for each genotype, using material pooled from 20 plants for each genotype inoculated or not with S. meliloti 1021. Nodulated material was collected from 14 to 25 dpi. Relative expression was calculated using two housekeeping genes, Medtr3g062450 and Medtr3g065110, and then fold changes were calculated for inoculated compared with non-inoculated as the 2 ÀDCT values. The primers for MtLYK10 were: F: GACCCAGTAGCTGCTTTGGAA; and R: TGACACTGCCACAACGATCTC. The RNA extraction protocol and qRT-PCR conditions and primers for the reference genes have been previously described (Gibelin-Viala et al., 2019).

Whole-plant assay of nodulation and symbiotic efficiency
Medicago truncatula cv. Jemalong A17 was prepared for inoculation with S. meliloti as previously described .
Seedlings were moved to individual Jensen's medium microcosms and inoculated with S. meliloti of the appropriate strain. Plants were grown in a Percival AR-36L incubator (Perry, IA) at 21°C, with 60-70% relative humidity and 100-175 µmol m À2 s À1 light. Shoot fresh weight was measured after 7 weeks.
Infection thread assays of cCFP-expressing S. meliloti strains on M. truncatula Jemalong Rhizobial infection sites were analyzed in M. truncatula Jemalong sunn-2 or A17 root hairs by in vivo microscopy as previously described (Fournier et al., 2008;Fournier et al., 2015). In brief, plants were placed in 12 cm x 12 cm plates containing modified F ahraeus medium (0.5% (w/v) Phytagel; Sigma-Aldrich) supplemented with 50 nM aminoethoxyvinylglycine (AVG). Roots were covered with gas-permeable plastic film (Lumox Film, Sarstedt, France), and plants were grown with the dishes slightly tilted to encourage the growth of the roots along the film. An aqueous suspension (OD 600 = 0.001) of exponentially growing bacteria constitutively expressing cCFP was used for inoculation of nitrogen-starved plants. Selected infection sites were imaged with a Leica TCS SP2 AOBS confocal laser scanning microscope ( Figure 1) or a Zeiss epifluorescence microscope ( Figure S4) equipped with long-distance 40x HCX Apo L (numerical aperture, 0.80) water immersion objectives. For confocal images, the argon laser band of 458 nm was used to excite cCFP and a 561-nm diode to observe cell wall autofluorescence, respectively. Specific emission windows used for cCFP and autofluorescence signals were 465-485 nm and 620-720 nm, respectively, and emitted fluorescence was false colored in green (cCFP) and magenta (wall autofluorescence). The composite images shown are maximal projections of selected planes of a z-stack. For epifluorescence images, cCFP and cell wall autofluorescence were recorded using combinations of BP436/10 and BP450-480 (for cCFP) or BP573/20 and BP625/30 (for autofluorescence) excitation and emission filters. Images were acquired and projected using Leica confocal software or MetaVue (Molecular Devices) and processed using Leica confocal software, METAVUE and FIJI (http://imagej.nih. gov/ij/; Schindelin et al., 2012;Schindelin et al., 2015). Data are from two experiments in sunn-2 and 2 experiments in A17.
Infection thread assays of b-galactosidase-expressing S. meliloti strains on M. truncatula tricycla R108 Seeds were surface sterilized, germinated, and grown in 12 cm x 12 cm plates containing slanting 0.2 mM NH 4 NO 3 -F ahraeus agar medium covered by growth paper for 6 days at 22°C with day/night cycles of 16/8 h. Each plant was inoculated with 2 9 10 3 bacteria.

Co-inoculation assays
Plants grown as described above were inoculated with 100 µl per plant of a 1:1 mix of nodC + exo mutant at an OD 600 = 0.5. Mature nodules were surface sterilized 5 min in bleach 2.4% active chloride, rinsed 59 in sterile water, crushed with a mortar in a 1.5 ml microtube, and spread on plates containing streptomycin and neomycin. Ten to 12 nodules of each combination were crushed and 52 randomly picked colonies/ nodule were characterized.

Statistical analysis
Data are presented as means AE SEM. Data analyses were performed with R software (v.3.4.1). Statistical comparisons of phenotypic data were performed using a Wilcoxon test or a Student's t-test (when the normality assumption was followed).

Phylogeny analysis
MtLYK10 homologs were found from Buendia et al. (2018) and from NCBI blasts. Synteny was verified using PHYTOZOME (https://phytozome.jgi.doe.gov), SOYBASE (https://www.soybase. org) and the Legume Information System, LIS (http://legumeinfo. org). Protein or gene names are shown in the tree, together with the plant species name. The tree was generated using http:// www.phylogeny.fr using the GBLOCKS program to eliminate poorly aligned positions and divergent regions. 67013-21579. Thanks to TRI-FR 3450 Imaging Platform Facilities, CNRS, France.

CONFLICTS OF INTEREST
The authors have no conflicts of interest to declare.

DATA AVAILABILITY STATEMENT
Supporting Information is listed below and these figures are provided for this article. Image acquisition and modification information has been described in Experimental Procedures or in Figure legends.

SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article. Figure S1. Plant performance and nodulation phenotypes on alfalfa cv. Iroquois and M. truncatula A17 of S. meliloti strains that make EPSII (galactoglucan). Figure S2. Information for all S. meliloti strains and plasmids, both those are discussed in the main text and in Supporting Information. Figure S3. Information on the genetic modifications in strains 961, 1317, exoKdel/exsH-1325 and exoHKdel/exsH-1345 with the exoLA-MON operon under the control of a heterologous promoter. Figure S4. Infections formed by the exoY and exoHKdel/exsH-1345 mutants on Medicago truncatula Jemalong A17. Figure S5. Complementation of strains with an exoHK deletion by a plasmid carrying exoH. Figure S6. Phylogenetic tree of MtLYK10 homologs from a variety of plant species, mostly Rosid dicots, but also two monocots, Oryza sativa and Brachypodium distachyon, and with MtLYK11 as an outgroup. Figure S7. qRT-PCR expression analysis of MtLYK10 in Medicago truncatula WT (R108) and the Mtlyk10 mutant. Figure S8. Comparison of the infection and nodulation phenotype of two independent Mtlyk10 mutant lines (Mtlyk10#2 and Mtlyk10#46) and a homozygous MtLYK10-WT line (MtLYK10-WT), all selected from the backcross between the Mtlyk10 mutant and R108. Figure S9. Kinetics of nodule formation when the nodulation defects of exo mutants are rescued in trans by co-inoculation with a nodC mutant. Figure S10. Nodules formed after mixed inoculation on Medicago truncatula WT R108 plants. Figure S11. The exoU mutant of S. meliloti does not produce detectable succinoglycan.