Identification and evolution of nsLTPs in the root nodule nitrogen fixation clade and molecular response of Frankia to AgLTP24

Non-specific lipid transfer proteins (nsLTPs) are antimicrobial peptides, involved in several plant biological processes including root nodule nitrogen fixation (RNF). Nodulating plants belonging to the RNF clade establish symbiosis with the nitrogen-fixing bacteria rhizobia (legumes symbiosis model) and Frankia (actinorhizal symbiosis model) leading to root nodule formation. nsLTPs are involved in processes active in early step of symbiosis and functional nodule in both models. In legumes, nsLTPs have been shown to regulate symbiont entry, promote root cortex infection, membrane biosynthesis, and improve symbiosis efficiency. More recently, a nsLTP, AgLTP24 has been described in the context of actinorhizal symbiosis between Alnus glutinosa and Frankia alni ACN14a. AgLTP24 is secreted at an early step of symbiosis on the deformed root hairs and targets the symbiont in the nitrogen-fixing vesicles in functional nodules. nsLTPs are involved in RNF, but their functions and evolutionary history are still largely unknown. Numerous putative nsLTPs were found up-regulated in functional nodules compared to non-infected roots in different lineages within the RNF clade. Here, results highlight that nodulating plants that are co-evolving with their nitrogen-fixing symbionts appear to have independently specialized nsLTPs for this interaction, suggesting a possible convergence of function, which opens perspectives to investigate nsLTPs functions in RNF.

the two models of RNF symbiosis, three AMP families are described; the Nodule Cysteine Rich peptide (NCRs) and NCRs-like peptides secreted by Fabales plants of the IRLC and Dalbergioid clades 10,11 , the defensins secreted by actinorhizal plants [12][13][14][15] , and a third family investigated in this study, the non-specific lipid transfer protein (nsLTPs) secreted by nodulating plants belonging to the Fabales order and described in only one actinorhizal plant, Alnus glutinosa [16][17][18][19] .The nsLTPs are peptides with a hypervariable amino acid sequence of less than 100 residues and an N-terminal signal sequence that allows them to be addressed to target cell compartments as mature peptides 9 .They are characterized by 4 disulfide bridges formed by a conserved 8 Cysteines Motif (8CM) in the mature peptide: "C-X n -C-X n -CC-X n -CXC-X n -C-X n -C" 20 where "X" represents any amino acid residue and "n" the number of amino acids.These disulfide bridges stabilize 4 alpha helices and give rise to a hydrophobic tunnel-like cavity allowing the binding and transport of hydrophobic molecules 21 .This structure allows them to resist heat, denaturing agents, and proteases 22 .This AMPs family is widely distributed in plant tissues among all land plants suggesting that nsLTPs were originally acquired in their common ancestor 23,24 .They may have been gained even earlier, as a putative nsLTPs in a green alga was predicted 25 .In plants, nsLTPs are involved in plant innate immunity and are classified as pathogenesis-related proteins (PR-14) but are also involved in several biological processes such as stress resistance, reproduction, germination, plant defense against pathogen attacks, cuticle formation, pollen tube formation, and RNF symbiosis [17][18][19]21,26 .
In RNF symbiosis, nsLTPs were first described in legumes (Fabales) at the early step of nodulation and in nodules of Medicago truncatula, Astragalus sinicus (Chinese milk vetch), and Phaseolus vulgaris 24,27,28 .The nsLTPs MtN5 and MtLTP7 are secreted by M. truncatula to regulate symbiont entry into the root epidermis and promote infection in the root cortex 16,17,[27][28][29] .In A. sinicus a nsLTPs named AsE246 is also expressed at early and late steps of nodulation and is localized on the symbiosome membrane, which could be involved in membrane biosynthesis and to promote symbiosis efficiency 19 .Little is known about the involvement of AMPs during actinorhizal symbiosis due to the lack of genetic tools.However, a transcriptomic analysis at early and maturing steps of nodulation permitted to identify a gene encoding an nsLTPs up-regulated in deformed root hairs and functional nodule of A. glutinosa in symbiosis with Frankia alni ACN14a compared to non-infected roots 18 .This peptide named AgLTP24 is addressed to deformed root hairs at an early step of symbiosis and targets the nitrogen-fixing vesicle cells of Frankia at a later step in nodules.As AgLTP24 targets the symbiont in the nodule, the effect of this peptide on the physiology of Frankia was tested in a previous article and showed that high concentration (5 µM) decreased metabolic activity and lower concentration (100 nM) inhibited nitrogen fixation 18 .
This study aimed to retrace the evolutionary history of nsLTPs in RNF symbiosis as they are involved in both symbiotic models.For this purpose, putative nsLTPs were identified in proteomes of nodulating and nonnodulating plants belonging to the RNF clade and the differential expressions of nsLTPs in nodules for five nodulating plants of the four orders were retrieved from available transcriptomics data.We showed that the nsLTPs family was widespread in nodulating plants as nsLTPs genes were expressed in the functional nodules (which are nodules with an active nitrogen fixation activity) of plants belonging to the four orders.Regarding their evolution in relation to the RNF symbiosis, this family would have been independently co-opted in different lineages suggesting a possible convergence of function.It is important to note that nsLTPs must share the same function in RNF symbiosis to show convergence, so more functional studies are required to conclude on this point.To deepen our understanding of their function during symbiosis, AgLTP24 secreted by A. glutinosa was further studied by investigating the molecular response of the symbiont F. alni ACN14a to subinhibitory concentrations of this nsLTP.

Results nsLTPs identification and characterization
The identification of nsLTPs was done using 15 proteomes of nodulating and non-nodulating plants distributed in the RNF clade and Arabidopsis thaliana belonging to the Brassicales order (Fig. 1).
The nsLTPs are characterized by a hypervariable amino acid sequence and an N-terminal signal sequence.The signal sequence that is responsible for facilitating peptide secretion, which is cleaved during the secretion, resulting in the generation of mature peptides.The mature nsLTPs possess a conserved 8-cysteine motif (8CM) "C-X n -C-X n -CC-X n -CXC-X n -C-X n -C", where 'X' represents any amino acid residue and 'n' the number of amino acids.Due to the sequence hypervariability, conventional approaches such as keyword searches and BLAST analyses are not suited for exhaustively retrieving these peptides 9 .To identify nsLTPs a wrapper script, nsLTPFinder, was made to identify proteins containing an N-terminal signal peptide, with a mature sequence containing a conserved 8 Cysteine Motif (8CM), characteristic of this peptide family.
From the 15 proteomes, an overall number of 705 putative nsLTPs was identified ranging from 23 in the Casuarina glauca proteome to 91 in the Medicago truncatula proteome (Fig. 1 and Supplementary Table S1).Plants belonging to the Fagales order had between 23 and 48 putative nsLTPs, those belonging to the Cucurbitales between 33 and 36, the Rosales had between 35 and 45 putative nsLTPs and the Fabales between 34 and 91 (Fig. 1).The number of putative nsLTPs in plant proteomes was compared with non-parametric Mann-Whitney tests as the data do not follow a normal distribution (Shapiro test) and all p-values were above the threshold of 0.05.This indicated that the number of putative nsLTPs in plant proteomes was not significantly different across the 5 different plant orders nor different based on the capability of the plant to establish RNF symbiosis.Plant proteomes used in this study did not have the same annotation level, thus, the number of nsLTPs predicted for these proteomes are subject to change with the increasing number of genome sequencing or proteomic studies.
Putative nsLTPs were then grouped according to the classification proposed by Boutrot et al. 20 with the addition of the XI type proposed by Li et al. 30 (Fig. 1, Supplementary Table S1).Only 54% of the putative nsLTPs in this dataset could be classified; the most represented was type I with 24% and type VII was not retrieved in our data.The absence of the type VII in our data can be attributed to its specificity to monocotyledons 31 .Among the 326 nsLTPs unassigned to a type, some had a large domain rich in proline, aspartic acid and histidine composed of more than 40 amino acids between the second and third cysteine of the 8CM.This lack of assignation underlines the fact that the current classification is not exhaustive.

nsLTPs expression during nodulation and evolution history in nodulating plants
The phylogeny of nsLTPs was assessed using putative nsLTPs identified in nodulating and non-nodulating plants belonging to the four orders of the RNF clade and A. thaliana (Brassicales order), as an outgroup.It is worth noting that the clades are more representative of the different types of nsLTP than of the different plant orders.Furthermore, within each plant order, different types of nsLTP were found.In the phylogenetic tree, nsLTPs of all plants were present in all clades suggesting that they would have undergone several duplication events.The untyped nsLTPs were grouped in phylogenetic clades with typed nsLTPs and shared the same conserved protein motifs (Fig. 2).
A conserved protein motif analysis using the MEME suite was conducted and showed that the majority of nsLTPs grouped in the same clade and mostly shared the same conserved protein motifs (Fig. 2).
We then identified nsLTPs that might be involved in symbiosis and investigated their distribution in the phylogenetic tree.For that, the expression levels of nsLTPs in functional nodules were retrieved from published transcriptomic data for five plants belonging to the four orders of the RNF clade: Medicago truncatula (Fabales), Parasponia andersonni (Rosales), Datisca glomerata (Cucurbitales), Alnus glutinosa and Casuarina glauca (Fagales).For A. glutinosa in association with Frankia alni ACN14a, transcriptomic data based on EST microarrays were complemented in this study using qRT-PCR targeting 23 genes encoding putative nsLTPs (AgLTPs) to characterize differential expression in the nodule compared to non-infected roots.
Up-and down-regulated genes encoding nsLTPs in functional nodules were identified in plants belonging to the 4 orders of the RNF clade (Log2FoldChange ≥ 1 or ≤ − 1) (Fig. 3).
The databases used (see "Materials and methods"), enabled the identification of the differential expression of 26% to 62% of genes encoding nsLTPs within the functional nodules compared to non-infected roots.In nodules of the Fabales M. truncatula and the Rosales P. andersonni capable of establishing symbioses with rhizobia, 7 and 10 up-regulated and 1 and 2 not-regulated nsLTPs were retrieved, respectively.They were no down-regulated nsLTPs found for P. andersonni while 21 nsLTPs were down-regulated in M. truncatula nodules.Concerning actinorhizal plants, D. glomerata had at the nodule step, 9 up-regulated, 4 not-regulated and 3 downregulated DgLTPs.For the Fagales, C. glauca, no CgLTPs was up-regulated, 5 were not-regulated and 5 CgLTPs were down-regulated in nodule.For A. glutinosa, previous transcriptomic data based on EST microarrays led to the identification of four genes (corresponding to 4 ESTs) up or down regulated in the nodule compared to non-infected roots (Log2FoldChange ≥ 1 or ≤ − 1) (see Supplementary Table S2).One of these EST matches with www.nature.com/scientificreports/two putative AgLTPs (AgLTP1 (Alngl907S06353) and AgLTP3 (Alngl66059S34270)) with a high percentage of identity.Thus, for A. glutinosa in association with F. alni ACN14a EST data were refined using qRT-PCR targeting genes encoding putative AgLTPs.The expressions of 23 AgLTPs, including 8 up-regulated, 8 not-regulated and 7 down-regulated (Log2FoldChange ≥ 1 or ≤ − 1), in the functional nodule compared to uninfected roots were assessed and the differential expression of AgLTP1 and AgLTP3 was refined (Fig. 3 and Supplementary Table S2).We also confirmed that AgLTP24 (Alngl424615S03856) is the most up-regulated gene encoding an nsLTP at the functional nodule step.The nsLTPs described as involved in symbiosis in the literature and those up-regulated in the functional nodule were retrieved in different phylogenetic clades, had different conserved protein motifs and different isoelectric points and molecular weights (Fig. 2 and Supplementary Table S1).MtN5 (MtrunA17_Chr5g0445131), AgLTP24 and MtnsLTP54 (MtrunA17_Chr7g0234401), (The annotation of MtnsLTPs was done according to the one proposed in the LEGOO database : MtnsLTP54 corresponds to MtLTP7 described by Santi et al. 28,32,33 ) already described as involved in RNF symbiosis were grouped in the same phylogenetic clade with the same conserved protein motif predicted by MEME.MtN5 and AgLTP24 mature peptides had close isoelectric points (IP) and molecular weights but shared only 21% of sequence identity 18 .MtN5 and MtnsLTP54 mature peptides had a higher sequence identity (38%) but a different molecular property with an IP of 8.8 and 4. 4, respectively (see Supplementary Table S1).Overall, these results indicate that nsLTPs up-regulated in functional nodules compared to uninfected roots have diverse protein motifs in their protein sequence and that nsLTPs described in the literature as functionally involved in symbiosis share the same conserved protein motifs.Typed nsLTPs are represented by colored circles.The unannotated ones are those that do not correspond to any type.The triangles represent the nsLTPs that are up-regulated in functional nodules, of M. truncatula (MtnsLTPs), P. andersonii (PaLTPs), A. glutinosa (AgLTPs) and D. glauca (DgLTPs).For A. glutinosa, AgLTPs up-regulated in nodules compared to non-infected roots are represented by qRT-PCR data (this study).For the tree other plant, up-regulated MtnsLTPs, PaLTPs and DgLTPs in functional nodules were retrieved from databases (see "Materials and methods").Names of nsLTPs already described in the literature as involved in RNF symbiosis are specified next to the gene name.The scale bar represents the number of substitutions per site (under the selected evolutionary model).The schematic representation of conserved protein motifs predicted using the MEME suite is shown outside the tree.

Molecular response of Frankia alni ACN14a to AgLTP24
To further investigate the role of nsLTPs in symbiosis, we investigated the function of AgLTP24 secreted by A. glutinosa in symbiosis with F. alni ACN14a.Our previous work demonstrated that AgLTP24 was highly expressed in A. glutinosa both at an early step of infection with F. alni ACN14a and at the functional nodule step 18 .In planta, AgLTP24 is secreted at deformed root hairs during the early step of symbiosis when the host recognizes Frankia and later when it targets the nitrogen-fixing vesicles of the symbiont inside the nodule cells.This previous work has showed that AgLTP24 at 5 µM impacted Frankia physiology by inhibiting cellular activity and nitrogen fixation at 100 nM and above.As Frankia in the nodule is viable and has an active nitrogen fixation to provide nitrogen to the plant, we investigated the molecular response of F. alni ACN14a under N-free conditions in contact or not with a sub-inhibitory concentration of AgLTP24 (1 nM) using RNAseq method (Table 1).Physiological measurements such as nitrogen fixation (ARA), respiration (IRA), and growth (OD 600nm ) were conducted on these assays and confirmed that AgLTP24 at this concentration did not affect Frankia physiology as shown earlier (see Supplementary Fig. S1) 18 .
Transcriptomic analysis identified 107 up-regulated genes and 35 down-regulated genes (Table 1) when F. alni ACN14a was in contact with a sub-inhibitory concentration of AgLTP24 compared to the control condition.Some encoded chaperones involved in refolding proteins and proteins for repairing DNA damages were up-regulated, such as groL, groS, lon, and a gene cluster (FRAAL6639-FRAAL6643) with dnaK, grpE, dnaJ, and clpB.Genes encoding putative proteins involved in cell wall/membrane/envelope biogenesis were up-regulated such as FRAAL6118 and FRAAL6119 encoding a glycosyltransferase and a succinoglycan biosynthesis protein, respectively.Other upregulated genes encoding membrane transporters such as ABC transporters, manganese transport, cation transporting P-type ATPase A, and a citrate transporter (fecD) were retrieved.Several genes involved in energy conversion and metabolism, aerobic respiration (succinate dehydrogenase sdhC, sdhA, sdhB), cytochromes, and nitrogen fixation (nif genes, nifB, nifK, nifX, nifH, nifV, nifZ, nifD) were up-regulated while narK allowing nitrate and nitrite import was repressed.Among the down-regulated genes, only one gene was annotated (narK), the others were not described enough to provide further information.

Discussion
Plant AMPs are involved in many plant functions, such as innate immunity or RNF symbiosis.Some legumes belonging to the IRLC and Dalbergioids clades secrete NCRs and NCR-like respectively to coordinate the terminal differentiation of rhizobia into polyploid bacteroids in the nodule.These NCRs and NCR-like are characterized by a conserved cysteine motif in their protein sequence that is close to the cysteine motif of defensins and neurotoxins.In actinorhizal symbioses, less information is available due to the lack of genetic engineering tools developed.Based on transcriptomic analysis of nodules, AMPs of the defensins family have been identified in the three actinorhizal plants Ceanothus thrysiflorus (Rosales), D. glomerata (Cucurbitales), and A.   www.nature.com/scientificreports/glutinosa (Fabales).These in silico analyses were complemented with functional analyses of AgDef5, a defensin secreted by A. glutinosa at the early step of symbiosis and in the nodule.In vitro, AgDef5 permeabilizes F. alni ACN14a's nitrogen-fixing vesicles, leading to the leakage of nitrogen-rich metabolites, which could improve trophic exchanges between the two partners in planta 34 .The nsLTPs family is involved in RNF symbiosis in both rhizobia/legumes and Frankia/actinorhizal symbioses.These peptides are secreted early in the symbiosis and in the nodule.In P. vulgaris, nsLTPs have a putative role and possible interaction with respiratory burst of oxidase homologs (RBOH)-dependent reactive oxygen species (ROS) production 24 .In M. truncatula and A. sinicus, MtN5 and AsE246 respectively, could regulate symbiont invasion, promote root cortex entry, membrane biosynthesis, and symbiosis efficiency 16,17,19,[27][28][29] .In actinorhizal symbiosis, only one nsLTP has been studied in A. glutinosa, AgLTP24, which targets Frankia's nitrogen-fixing vesicle in nodules 18 .Purified AgLTP24 peptide inhibited F. alni ACN14a nitrogen fixation activity above 100 nM and reduced metabolic activity above 5 µM in vitro.
As this family is widespread in RNF symbiosis, the evolutionary history of nsLTPs in symbiosis was analyzed in this study.First, nsLTPs were predicted using 15 plant proteomes of nodulating and non-nodulating plants within the RNF clade and one Brassicales A. thaliana.These data permit to perform a phylogenetic analysis, which showed that nsLTPs from diverse plant species were distributed across all clades.nsLTP are grouped by type, independently of plant order, suggesting their ancient acquisition prior to the emergence of the RNF clade.Furthermore, we observed several copies of the same nsLTP type in each plant species, suggesting that several duplications took place within each node.This also showed that nsLTPs had an evolutionary history marked by both ancient but also recent duplications in plants and groups of plants.This observation suggests that the nsLTPs gain predates the emergence of the RNF clade.This conclusion is consistent with Edstam who argued that nsLTPs would have emerged in the first land plants since no nsLTPs in their dataset were identified in algae at that time 23 .A recent study predicted a novel nsLTP lineage in green alga thus nsLTPs could have emerged in the common ancestor of green plants 24,25 .Our results also showed a high percentage of nsLTPs not corresponding to any of the types proposed by Boutrot   1. Frankia alni ACN14a genes up and down regulated in N-free condition supplemented with AgLTP24 versus N-free condition.p-value adjusted (p adj ) ≤ 0.05, up-regulated genes: Log2FoldChange ≥ 1, Downregulated genes Log2FoldChange ≤ − 1. Gene label, name and annotation come from the Genoscope database.
motifs.As there is no official classification for these peptides, it would be relevant to complete or establish a new classification of nsLTPs from plants covering the whole plant kingdom and improve the identification of these peptides in proteomes.Studying nsLTPs from plants representing the entire plant kingdom, as they have recently been identified in algae, would also improve analysis of the evolutionary history of nsLTPs 25 .
To deepen our understanding of the evolutionary history of nsLTPs in RNF symbiosis, the expression of genes encoding putative nsLTPs in nodules of 5 plants was retrieved with available transcriptomics data.This permitted the identification of genes encoding putative nsLTPs up-regulated in the functional nodule of plants belonging to the four orders of the RNF clade except for C. glauca.The only nsLTP studied in actinorhizal plants targets the nitrogen-fixing vesicle of F. alni ACN14a, however, in the symbiosis model between C. glauca and Frankia casuarinae CcI3, the nsLTPs might not be present in the nodules because F. casuarinae Cci3 does not differentiate cells into vesicles in nodule as the oxygen flow is controlled by the host plant 35 .The differential expression data of CgLTPs in the functional nodule were not available for all putative nsLTPs, further analysis could improve these data as we have done here for the putative nsLTPs found in the proteome of A. glutinosa.The nsLTPs already described in RNF symbiosis in the literature, MtN5, MtnsLTP54 and, AgLTP24 grouped in the same phylogenetic clade and share conserved protein motifs but other putative nsLTPs up-regulated in functional nodules had different type and conserved motif and were distributed in all phylogenetic clades.It should also be noted that among the nsLTPs already studied in RNF symbiosis in the literature, AsE246 is so far the only one described as belonging to the type I 19,20 .This raises the question of whether the motifs conserved between MTN5 and AgLTP24 are crucial for their involvement in nodulation, and whether they have the same functions.Furthermore, functional analyses of nsLTPs with different protein motifs belonging to other clades would be required to determine their involvement in nodulation and whether their function is similar or different.Regarding the evolutionary history of nsLTPs in the RNF symbiosis, within the RNF clade, nsLTPs differentially expressed during nodulation belong to different subclades, suggesting that symbiotic functions may have been independently co-opted in different lineages of RNF symbioses.This independent co-option could suggest a convergence of function however, more functional data are required to conclude on this point.It is important to keep in mind that nsLTPs could exhibit the same or diverse functions during the symbiosis.In order to ascertain whether the peptides derived from various nodulating plants, whose encoding genes are upregulated during symbiosis, possess a single function indicative of functional convergence, or exhibit diverse functions within this association, it is imperative to conduct comprehensive functional studies.
More broadly, concerning the evolutionary history of RNF symbiosis, two hypotheses have been proposed, one with an evolutionary model based on several independent acquisitions 36 of the ability to form nodulation and another based on a single gain of this trait in a common ancestor followed by multiple losses.Recently, strong arguments have been published supporting the second hypothesis and indicated also that some additional functions have been acquired in a convergent manner such as plant hemoglobin [2][3][4] .This convergence of function was also described for AMPs involved in RNF symbiosis.NCRs and NCRs-like secreted respectively by legumes belonging to IRLC and Dalbergioids have different structures, but both induce the differentiation of the symbiont into bacteroids with different shapes in the nodule 10,11 .This convergence of function was recently challenged due to their possible origin from within defensins.A recent phylogenetic study between defensins involved in actinorhizal symbioses and NCRs of legumes shows that these peptides would have a common origin 37 .Concerning nsLTPs, their symbiotic functions may have been independently co-opted in different lineages of RNF symbioses to take part of in the symbiosis process in each nodulating plant.That may be a sign of convergent evolution, but it needs to be established first that all nsLTPs up-regulated in functional nodule share a common function.
This study focuses only on RNF symbiosis, but it would be worthwhile to determine the involvement of nsLTPs in other mutualistic symbioses such as mycorrhizal symbioses that has not been documented to our knowledge.Only one publication reports the overexpression of a gene encoding nsLTPs in Oryza sativa roots during appressoria formation and penetration of the mycorrhizal fungus Glomus mosseae.This gene is subsequently down-regulated upon mycorrhization, during the intracellular development of fungal hyphae in the root and is also induced upon treatment with salicylic acid or with the pathogen Pseudomonas syringae indicating that this nsLTP is not involved in mycorrhization but probably part of the plant's defense system 38 .In M. truncatula, two nsLTPs (MtnsLTP104 and MtnsLTP103 corresponding to Medtr4g077180 and Medtr4g076150; respectively) are up-regulated during mycorrhizal symbiosis with the arbuscular mycorrhizal fungi Rhizophagus irregularis but no functional studies have been performed.The nsLTPs are described as part of the plant's immune response against many pathogenic organisms such as bacteria, fungi, viruses, nematodes, and insects [39][40][41][42][43][44] .More broadly this raises the question of how these diverse interactions have shaped the evolution of nsLTPs in plants.
Focusing on RNF interaction, to further investigate the evolution of nsLTPs in nodulating plants, their functions should be explored in nodulating plants of different lineages.This should permit to identify if nsLTPs of a given phylogenetic clade have similar functions or multiple functions during symbiosis.For this purpose, we studied the function of AgLTP24 which is the most expressed nsLTPs gene in the A. glutinosa nodule 18 .This peptide inhibits the metabolic activity of F. alni ACN14a at 5 µM and inhibits the nitrogen fixation activity at 100 nM, however, the symbiont in the nodule is viable and metabolically active to fix nitrogen to supply the host with nitrogen compound.Thus, in this study, we were interested in the molecular response of the symbiont to subinhibitory concentrations of AgLTP24.
Transcriptomic analysis of F. alni ACN14a under N-free conditions supplemented with subinhibitory concentrations of AgLTP24 compared to N-free medium without nsLTP addition indicated that the bacterium copes with stress to ensure its survival by maintaining nitrogen fixation, growth, and respiration and that it was preparing for symbiosis.Several stress-related genes coding for chaperones were up-regulated, as well as genes coding for transporters and transcriptional regulators involved in the management of oxidative stress 45,46 .These genes involved in stress response were also up-regulated at an early step of symbiosis when the bacterium is in indirect contact with the plant 47 .The response of F. alni ACN14a to AgLTP24 is comparable to that of the Sinorhizobium meliloti symbiont when exposed to NCR247 and NCR335.In both cases, there is an up-regulation of genes encoding heat shock proteins, proteins with potential involvement in transcriptional regulation, and ABC-type membrane transporters 48 .Genes encoding nitrogenase and proteins involved in respiration and the TCA cycle were up-regulated and the gene encoding the nitrite transporter NarK was down-regulated when Frankia was in contact with AgLTP24.Under N-free medium in vitro, F. alni ACN14a nif genes are up-regulated as well as narK encoding a nitrite transporter.The nif genes are up-regulated and narK is down-regulated when the symbiont is in nodule condition compared to an N-free medium 49 .The same expression profile of nif and narK genes was seen when Frankia was in contact with AgLTP24 suggesting that the bacteria had a similar nitrogen management to nodule conditions in planta.
F. alni ACN14a in contact with AgLTP24 at subinhibitory concentrations could undergo stress.To overcome this, Frankia could establish resistance systems to adapt to the effects of AMPs.Some up-regulated genes encoding ABC transporters that could be a mechanism of resistance to nsLTPs.An ABC transporter, BacA, essential for the survival of the symbiont in the nodule, allows Sinorhizobium meliloti to cope with the toxicity of NCR peptides secreted by M. truncatula. 50,51.Two genes encoding peptides possibly involved in succinoglycan synthesis were up-regulated by Frankia in contact with AgLTP24.The succinoglycan produced by rhizobia allows them to resist to NCRs 52 .It was described that sub-inhibitory concentration of AMP can act at the membrane or intracellular level, it would be relevant to identify whether this response is induced following the interaction of AgLTP24 with Frankia membranes or intracellular target molecules 53 .

Conclusion
Nodulating plants that are co-evolving with their nitrogen-fixing symbionts appear to have independently specialized nsLTPs for this interaction, suggesting a possible convergence of function.To better understand the various functions of these nsLTPs in RNF symbiosis we identified genes encoding putative nsLTPs in plants distributed in the four orders of the RNF clade which opens new perspectives.Concerning actinorhizal symbioses, we confirmed that AgLTP24 was the most up-regulated gene in the functional nodule of A. glutinosa in symbiosis with F. alni ACN14a.Thus, the function of this nsLTPs was further investigated with the study of the molecular response of the symbiont to sub-inhibitory concentrations of AgLTP24, which permitted to show a similar response to that found in symbiotic conditions and highlighting possible adaptation mechanisms of Frankia to AgLTP24.
A wrapper script, nsLTPFinder, was used to identify putative nsLTPs peptides in plant proteomes (https:// github.com/ jeank eller/ nsLtp Finder.git).As input, a directory containing the proteomes was used to be analyzed in FASTA format.First, a HMMSEARCH from the HMMER v3.3 package was performed using the Hidden Markov Model (HMM) profile of Probable lipid transfer (PF14368.6),Hydrophobic seed protein (PF14547.6)and Protease inhibitor/seed storage/LTP family (PF00234.22)(PFAM34 database).Searches were performed using an e-value threshold of 10 for full and domain hits.The nsLTPs were also searched with the regular expression "C.(6,15)C.(6,80)CC.(8,29)C.C. (8,37)C.(4,25)C" in proteomes.Results from HMMSEARCH and the regular expression search were merged and protein sequences were then extracted from proteomes.Signal sequences were searched using SignalP 5.0 60 .The isoelectric point, molecular weight, and grand average hydropathy (GRAVY) were retrieved for peptides and mature peptides, which correspond to the peptides without signal sequence using Expasy ProtParam tool 61 .Proteins identified by the regular expression search and HMMSEARCH with an identified signal peptide and containing 8 cysteines in the mature sequence were extracted as "top candidates" and proteins identified only with the regular expression search with a signal peptide, 8 cysteines in the mature sequence were extracted as low confidence candidates.Conserved motifs were predicted using the MEME 62 suite on the top and low confidence mature peptides for each plant proteome and on all top and low confidence for all proteomes.
Once nsLTPFinder ran to completion, the 8CMs were manually checked in the mature peptide's amino acid sequences for top and low-confidence candidates.The identified nsLTPs were grouped according to the classification proposed by Boutrot et al. 20 and completed with the type XI proposed by Li et al. 30 .The graphical representation of the number of nsLTPs and their classification in each plant was performed using RStudio 2021.09.2.The comparison of the number of putative nsLTP in plants belonging to four orders of the RNF clade and the comparison of the number of nsLTPs present in nodulating and non-nodulating plants belonging to the RNF clade was performed using Shapiro normality test and Mann-Whitney test to analyze the distribution of data using GraphPad Prism 9.5.0.

Phylogenetic analysis and sequence alignment
Multiple sequence alignments of nsLTPs CDS sequences were performed using Mafft v7 with local pairwise.The alignment was cleaned using TrimAl 1.4.1 63 to remove positions with more than 50% of gaps.A Maximumlikelihood phylogenetic tree was reconstructed using IQ-TREE2 2.1.4_beta 64(SH-aLRT test and ultrafast bootstrap with 10,000 replicates) and the model GTR + F + R9 determined with ModelFinder (https:// doi.org/ 10. 1038/ nmeth.4285) according to the Bayesian Information Criteria.Branch supports were tested using 10,000 replicates of Ultrafast Bootstrap 65 .The tree was visualized with iTOL 6.3.2 platform 66 .The maximum-likelihood phylogenetic tree reconstructed with IQ-TREE2 with bootstrap values in Newick format can be found as Supplementary File S1.

nsLTPs differential expression during nodulation
To analyze differential expressions of genes encoding nsLTPs in the nodule, previously calculated transcriptomic data from five nodulating plants were recovered from available transcriptomics data.For M. truncatula, expression data were obtained after 14 days post-inoculation (dpi) with Sinorhizobium meliloti 1021 67 via the MtSSBPdb platform 32 .Gene annotation correspondence was done using the LeGOO database 33 .The P. andersonii expression data were obtained at stage 3 (corresponding to functional nodule) after inoculation with Mesorhizobium plurifarium BOR2 3 .Differential expressions data of D. glomerata were obtained after 24 dpi with nodule crush 68 .For C. glauca the expression data had been obtained on 21 dpi nodules with Frankia casuarinae Cci3 69 using SESAM database 70 .nsLTPs of A. glutinosa (AgLTPs) genes expression in nodule (21 dpi) infected with F. alni ACN14a were analyzed using EST (Expressed Sequence Tag) database and microarray analysis, which are publicly available on the Gene Expression Omnibus database (www.ncbi.nlm.nih.gov/ geo; accession number GSE24153).Correspondence between EST and A. glutinosa gene name 2 was determined with a Blast search using percentage identity > 90% and EST-gene coverage > 85% parameters.Differential expression of genes encoding AgLTPs was determined using the microarray dataset with a p-value threshold of 0.05.Briefly, Student's t-test was applied to compare nodules versus non-inoculated roots and average Fold Changes (FC) were calculated and false discovery rate (FDR) adjusted p-value (FC are considered as significative if p-value adj < 0.05).To complement and confirm these microarray data, reverse transcription (RT) and quantitative real-time PCR (qRT-PCR) using nodules from 3 plant biological replicates obtained after infection with F. alni ACN14a (21 dpi) were performed.The results obtained were compared to uninfected roots as reference.RT was performed using 5 µg of total mRNA using Transcriptor Reverse Transcriptase and oligo (dT) 15 primer (Roche, Mannheim, Germany).qRT-PCR was run on BioRad QX 100 using iTaq Universal SYBR Green Supermix (Bio-rad) under the following conditions: 95 °C for 5 min; 44 cycles of 95 °C for 20 s, 60 °C for 20 s 72 °C for 15 s.Primer sets were designed using Primer3Plus software and can be found in Supplementary Table S3.Expression values were normalized using the expression level of the Ag-ubi gene that encodes ubiquitin 71 .

Strain and plant growth condition
Frankia alni strain ACN14a 72 was grown at 28 °C with 200 rpm stirring in FBM medium with 5 mM ammonium as described earlier 73 to the exponential phase.The cells were then harvested, sedimented by centrifugation (5000×g, 10 min), and washed twice with corresponding NH 4 + -free FBM medium (FBM-).Plant growth, inoculation and nodule harvesting were done as described earlier 13 .

Molecular response of F. alni ACN14a to contact with AgLTP24
AgLTP24 was produced and purified as described earlier 18 .Three independent cultures of F. alni ACN14a were made in 240 ml of FBM-liquid medium (N-free condition) supplemented or not with 1 nM of AgLTP24 and grown for 7 days at 28 °C.The pellets were collected by centrifugation at 5100×g.Then, mRNAs were extracted and converted into cDNA as described previously 49 .Ribosomal RNAs were depleted using Truseq stranded total RNA (Illumina) and the cDNA were sequenced using Novaseq6000 (Illumina at the MGX, Montpellier, France).Bioinformatic and statistical treatments were made by the MGX (Montpellier GenomiX Platform) platform.The sequences were aligned on the F. alni strain ACN14a genome using the BWA 0.7.17-r1188 software 74 .Statistical analyses were made using DESeq2 1.26.0 with R 3.6.1 75 .
Bioassays were conducted by growing F. alni strain ACN14a in FBM-and incubating it for 7 days at 28 °C, 3 replicates per condition were performed as described previously 18 .Frankia's nitrogen fixation activity (or ARA), respiration (IRA), and growth (OD 600nm ) were tested as described in previous work 18 .Statistical analyses were computed using RStudio 4.1.2.The normality of the distribution was tested with a Shapiro-Wilk normality test, variances homogeneity was tested with a Fisher's test.Means comparisons were performed with a Student's t-test.Graphics were made using GraphPad Prism 9.2.0 (GraphPad Software Inc; San Diego, CA, USA).

Figure 1 .
Figure 1.Putative nsLTPs characteristics.Graphical representation of the number of putative nsLTPs retrieved in plants proteomes belonging to the RNF clade and A. thaliana.Plant orders are represented with colored boxes on the y-axis.Nodulating and non-nodulating plants are identified by a schematic representation of a root with or without a nodule.The number of nsLTPs for each plant is represented by the number of typed and untyped (NA) nsLTPs.The percentages of the different types of nsLTPs in this dataset are shown in the legend.

Figure 2 .
Figure 2. Phylogenetic tree representing nsLTPs evolution in the RNF clade.A graphical representation of the unrooted maximum-likelihood phylogenetic tree of nsLTPs from RNF plants was constructed with IQ-TREE.Typed nsLTPs are represented by colored circles.The unannotated ones are those that do not correspond to any type.The triangles represent the nsLTPs that are up-regulated in functional nodules, of M. truncatula (MtnsLTPs), P. andersonii (PaLTPs), A. glutinosa (AgLTPs) and D. glauca (DgLTPs).For A. glutinosa, AgLTPs up-regulated in nodules compared to non-infected roots are represented by qRT-PCR data (this study).For the tree other plant, up-regulated MtnsLTPs, PaLTPs and DgLTPs in functional nodules were retrieved from databases (see "Materials and methods").Names of nsLTPs already described in the literature as involved in RNF symbiosis are specified next to the gene name.The scale bar represents the number of substitutions per site (under the selected evolutionary model).The schematic representation of conserved protein motifs predicted using the MEME suite is shown outside the tree. https://doi.org/10.1038/s41598-023-41117-1www.nature.com/scientificreports/ https://doi.org/10.1038/s41598-023-41117-1 et al. but grouped in the same phylogenetic clade with conserved protein