P. vulgaris cultivar Negro Jamapa was used in this study. Seeds were surface sterilized and germinated as reported in Reyero-Saavedra et al. (2017). Two-day-old seedlings were transferred to 2 L pots containing moist perlite, kept in a growth chamber at 25-27°C, and watered with Summerfield nutrient solution (Summerfield et al., 1977) every three days.

Rhizobium tropici CIAT 899 strain was used to inoculate P. vulgaris seedlings. R. tropici cells were grown for two days at 30°C on PY medium (5 g/L peptone; 3 g/L yeast extract) supplemented with 0.7 M CaCl₂ and 20 μg/mL nalidixic acid. After two days, R. tropici cells were harvested and resuspended in sterile water at O.D_600nm= 0.3. One mL of this bacterial suspension was used to inoculate P. vulgaris seedlings individually.

Empty binary vectors pKGWFS7 and pTDT-DC-RNAi were propagated in Escherichia coli DB3.1, promoterPvAGO5::GUS::GFP, and the RNA interference (RNAi) against PvAGO5 (see below for details of these genetic constructs) were propagated in DH5α.

Agrobacterium rhizogenes K599 strain was used to generate transgenic roots in P. vulgaris plants (see below for details). A. rhizogenes cells were grown on Luria-Bertani (LB) plates for two days at 30°C. 100 μg/mL spectinomycin was added to select for the presence of plasmid vectors.

To analyze PvAGO5 promoter activity, a 1,800 bp DNA fragment upstream of the start codon was PCR-amplified from genomic DNA of P. vulgaris var. Negro Jamapa using specific primers. The amplified fragment was then cloned into the pENTR-D-TOPO (Thermo Fisher Scientific) vector. The resulting pENTR-pPvAGO5 plasmid was recombined into the pKGWFS7 binary vector containing GUS and GFP CDS, yielding the transcriptional pPvAGO5::GUS::GFP fusion.

A previously generated and reported RNAi construct was used to silence the expression of PvAGO5 (Reyero-Saavedra et al., 2017).

All constructs were verified by DNA sequencing. Primer sequences for plasmid constructions are shown in Supplementary Table 1 .

Binary vectors with pPvAGO5:GUS-GFP or PvAGO5-RNAi constructs were mobilized into A. rhizogenes K599 strain by electroporation. The empty vectors pKGWFS7 or pTDT-DC-RNAi were used as controls. A. rhizogenes-mediated transformation was performed according to Estrada-Navarrete et al. (2007). P. vulgaris composite plants (plants with the transformed root system and untransformed shoot system) were grown in 2 L pots containing wet perlite. Tandem Double Tomato (TDT) or GFP fluorescence in transgenic roots was observed with a fluorescence stereomicroscope.

To evaluate the effect of the gene silencing of PvAGO5 in the lateral root development under non-symbiotic conditions, A. rhizogenes-mediated pPvAGO5-RNAi transgenic roots were generated in the P. vulgaris as described above. Transgenic roots expressing an empty vector were used as a control. Composite plants were watered with 5 mM KNO₃-supplemented Summerfield nutrient solution. After three weeks, transgenic roots showing TDT fluorescence were collected to evaluate the number, density and length of lateral roots. Lateral root density was calculated by dividing the number of lateral roots on each transgenic root with the length of the root. For these experiments, ten biological replicates, each one containing ten transgenic roots from independent composite plants, were included.

Two-day-old P. vulgaris wild-type or composite plants expressing the empty vector or the PvAGO5-RNAi construct were transferred to 2 L pots containing wet perlite and inoculated with 1 mL of R. tropici (O.D._600nm= 0.3). Plants were watered with nitrogen-free Summerfield nutrient solution (Summerfield et al., 1977) every three days. Inoculated plants were kept in a growth chamber at 25-27°C. Ten and twenty days after rhizobial inoculation, roots with nodule primordia and nitrogen-fixing nodules (mature nodules) were collected separately and immediately processed for PvAGO5 Immunoprecipitation assays or RNA-seq analyses (see below for details). For these experiments, three biological replicates containing six independent wild-type or composite plants were included.

To evaluate PvAGO5 promoter activity under non-symbiotic conditions, A. rhizogenes-mediated pPvAGO5:GUS-GFP transgenic roots were generated in the P. vulgaris as described above. Composite plants were watered with 5 mM KNO₃-supplemented Summerfield nutrient solution. After three weeks, transgenic roots showing GFP fluorescence were collected for GUS staining as reported in (Isidra-Arellano et al., 2020). For this experiment, ten biological replicates containing ten composite plants and ten roots were included.

To evaluate PvAGO5 promoter activity during the root nodule symbiosis, A. rhizogenes-mediated pPvAGO5:GUS-GFP transgenic roots were generated in the P. vulgaris. Composite plants were inoculated with 1 mL of R. tropici (O.D._600nm= 0.3). Rhizobia-inoculated plants were watered with nitrogen-free Summerfield nutrient solution. Upon one, ten and twenty days after rhizobia inoculation, transgenic roots, roots with nodule primordia and mature nodules were collected for GUS staining assays as reported in (Isidra-Arellano et al., 2020). For this experiment, ten biological replicates containing ten composite plants and ten roots were included.

To analyze the expression of genes listed in Supplementary Table 1 , transgenic roots and mature nodules expressing the PvAGO5-RNAi construct and showing TDT fluorescence were immediately harvested in liquid nitrogen and stored at -80°C until used. Total RNA was isolated from roots and nodules from three different composite plants using the ZR Plant RNA miniprep (Zymo Research, USA) following the manufacturer’s instructions. cDNA was synthesized from 1 μg of genomic DNA-free total RNA and used to analyze gene expression by RT-qPCR as we previously described in Isidra-Arellano et al. (2020). RT-qPCR primer sequences used in this study are provided in Supplementary Table 1 . Three biological replicates, each one with six technical replicates, were included for this experiment.

Total RNA was isolated as described in the previous section from 0.5 g of one-day rhizobia- or mock-inoculated transgenic roots and transgenic mature nodules expressing an empty vector, or the PvAGO5-RNAi construct. Stranded messenger RNA-seq (mRNA-seq) libraries were generated from 1 μg of gDNA-free total RNA from each experimental condition and prepared using the TruSeq RNA Sample Prep kit (Illumina, San Diego, CA, USA) according to the manufacturer’s instructions. For each experimental condition, three biological replicates containing six independent composite plants were included. Eighteen libraries were sequenced on an Illumina NextSeq 500 platform with a 150-cycle sequencing kit and a configuration of pair-end reads with a 75 bp read length. Library construction and sequencing were performed by the Unidad Universitaria de Secuenciación Masiva y Bioinformática (Instituto de Biotecnología, UNAM, México).

Adapter and contamination removal were carried out using in-house Perl scripts. Sequences were filtered based on quality (Q33, FASTQ Quality Filter v0.0.13, http://hannonlab.cshl.edu/fastx_toolkit/index.html). About ten million reads per sample were aligned to the P. vulgaris transcriptome (v2.1 from Phytozome v13) using Bowtie2 (v2.3.5) and the recommended parameters to match RSEM analysis input requirements (Langmead and Salzberg, 2012). Gene expression was calculated using the RNA-seq by Expectation Maximization (RSEM) method (v1.3.3) and the default parameters (Li and Dewey, 2011). Significantly Differentially Expressed Genes (DEGs: adjusted p-value ≤ 0.05) were identified using DESeq2, part of the Integrated Differential Expression Analysis MultiEXperiment (IDEAMEX) platform (Jiménez-Jacinto et al., 2019), with the RSEM expected counts. Gene Ontology (GO) term enrichment analysis was performed using AgriGO (v2.0) and default parameters (FDR cutoff = 0.05) (Tian et al., 2017). Protein domain enrichment analysis was performed using PhytoMine tool from Phytozome (v13), with Holm-Bonferroni correction (FDR cutoff = 0.05). Heatmaps were created with ggplot2 and heatmap.2 libraries using R software (v4.1.2).

Uninoculated roots, roots bearing nodule primordia, and mature nodules from wild-type P. vulgaris were manually collected on ice. For each experimental condition, material from 100 plants was ground on ice with immunoprecipitation buffer (50 mM Tris-HCl pH7.5; 1.5 mM NaCl, 0.1% Nonidet P40, 4 mM MgCl₂, 2 mM DTT, and Sigma Protease inhibitor cocktail). Cell debris was removed by centrifugation twice for 15 min at 12,000 g at 4°C. Next, supernatants were precleared by incubation for 1 hour with 10 μl Protein A-Agarose (Roche). Samples were centrifuged at 1,500 g for 5 minutes at 4°C. Supernatants were incubated with Protein A-Agarose (Roche) supplemented with 2 μl anti-AGO5 for exactly 16 hours with rotation at 4°C. Samples were centrifuged at 1,500 g for 5 minutes at 4°C. The beads were washed three times with a washing buffer (50 mM Tris-HCl pH7.5, 150 mM NaCl, 0.1% Nonidet P40, 4 mM MgCl₂, 2 mM DTT, and Sigma Protease inhibitor cocktail). Next, beads were resuspended in 0.4 M NaCl and sRNAs were extracted by phenol:chloroform:iso-amyl alcohol extraction.

Nine small libraries (3 for uninoculated roots, 3 for roots bearing nodule primordia, and 3 for mature nodules) were generated from AGO5-bound sRNA using TruSeq small RNA Library Prep kit (Illumina, San Diego, CA, USA) according to the manufacturer’s instructions. Libraries were sequenced on an Illumina NextSeq 500. Library construction and sequencing were performed by the Unidad Universitaria de Secuenciación Masiva y Bioinformática (Instituto de Biotecnología, UNAM, México).

Adapters and reads with quality mean lower than 33 were removed and the sequence redundancy collapsed using the FASTX-Toolkit suite (v0.0.13, http://hannonlab.cshl.edu/fastx_toolkit/index.html). Small RNAs were compared to Viridiplantae miRNAs from miRbase (v22), 277 P. vulgaris small RNAs previously identified and published in (Formey et al., 2015; Formey et al., 2016), and 10 tRFs from Rhizobium etli (Ren et al., 2019). Normalization was performed using DESeq2. Only the sRNAs lying within the top 1% of the most accumulated sequences in a given library, were selected for further analyses. Transcript targets for the identified sRNAs were predicted using psRNAtarget (V2) and default parameters (Dai et al., 2018). Venn diagrams were designed using DeepVenn (Hulsen et al., 2008).

Statistical analyses and graphic generation were conducted using R software 4.1.2. The specific tests performed are indicated in the legend of the corresponding figure.

Our previous transcriptional analyses by RT-qPCR showed that PvAGO5 is preferentially expressed in P. vulgaris roots, compared to leaves (Reyero-Saavedra et al., 2017). However, these data do not provide spatiotemporal insights into PvAGO5 expression in this essential organ for root nodule symbiosis. To tackle this, we cloned a 1.8 kb fragment of the PvAGO5 promoter (pPvAGO5) and generated a transcriptional fusion to the GUS (β-glucuronidase) and GFP coding sequence. The Empty vector:GUS-GFP (control) or the pPvAGO5:GUS-GFP constructs were transfected separately into P. vulgaris using A. rhizogenes-mediated transformation (Estrada-Navarrete et al., 2007), and composite plants were grown and watered with nitrogen-containing Summerfield nutrient solution for three weeks. We observed no GUS activity in transgenic roots expressing the empty vector ( Figures 1A, D ). In contrast, in the absence of rhizobia, 65 out of 70 roots transformed with pPvAGO5:GUS-GFP displayed a weak GUS signal in the whole root, intensified to a strong signal in lateral root primordia and mature lateral roots ( Figures 1B, E ); whereas the other five roots showed the same GUS signal intensity through the entire root, ( Figures 1C, F ). GUS activity was absent in the root hairs of the 70 transgenic roots analyzed ( Figures 1G, H ).

The presence of PvAGO5 activity in the lateral root primordia prompted us to compare the number, density, and length of lateral roots in P. vulgaris composite plants expressing an empty vector (control) or the PvAGO5-RNAi construct. This analysis indicated that the number of lateral roots and density was increased 15% in PvAGO5-RNAi transgenic roots compared to the empty vector controls, with no significant differences in root length ( Figures 1I–K ). Moreover, no growth defects in the root hairs were observed in PvAGO5-RNAi transgenic roots ( Supplementary Figure 1 ). In summary, these data inform that PvAGO5 is expressed mostly in lateral root primordia in the absence of rhizobia. Furthermore, they also suggest that AGO5 plays a role in lateral root development in P. vulgaris.

We previously demonstrated by RT-qPCR that the expression of PvAGO5 increases at one and three hours after inoculation with rhizobia as compared to non-inoculated roots, as well as in mature nodules (Reyero-Saavedra et al., 2017). To further understand the cellular expression pattern of PvAGO5 across different stages of root nodule symbiosis, we evaluated its spatio-temporal promoter activity in common bean composite plants. These plants were inoculated with R. tropici and transgenic roots were harvested one, ten, and twenty days after inoculation to evaluate the pPvAGO5:GUS activity during the infection process, nodules primordia at the nodule emerging stage (nodule primordia), and in mature nodules, respectively. We observed no GUS activity in mock- and rhizobia-inoculated transgenic roots, or nodules expressing the empty vector ( Figures 2A, B ). Mock-inoculated pPvAGO5:GUS-GFP transgenic roots showed a weak GUS signal in the whole root ( Figure 2A ). After one day of rhizobial inoculation, PvAGO5 is expressed in the entire root ( Figure 2A ). At this time-point, we also observed a faint but consistent GUS signal in the tips of root hairs from 60 out of 80 transgenic roots expressing the pPvAGO5:GUS-GFP construct ( Figure 2A ), whereas the other 20 roots showed no GUS activity in these cells ( Supplementary Figure 2 ). Furthermore, at ten- and twenty-days post-inoculation with rhizobia, we observed a strong GUS activity in nodule primordia, mature nodules, and throughout the vasculature system ( Figure 2B and Supplementary Figure 2 ). Altogether, our results show that PvAGO5 expression is increased at early stages of the root nodule symbiosis. In addition, they demonstrate that the transcriptional activity of PvAGO5 is further enhanced during the nodule development process.

To analyze global transcriptomic effects of PvAGO5 down-regulation, empty vector (control) and PvAGO5-RNAi were analyzed from mock or R. tropici-inoculated roots at one day after inoculation and twenty-days-old nodules. For each time point, three biological replicates were analyzed using RNA-seq.

Overall, 10.7 to 12.2 million short sequence reads (2 X 75 bp) were generated from each of 18 RNA-seq libraries, with alignment rates to the reference genome ranging between 84-88%. Principal component analysis (PCA) showed a clustering of biological replicates and silencing-dependent variation, with the first component explaining an average of ~54% of data variation ( Supplementary Figure 3 ). Comparison of expression levels in each biological condition and at each time point (relative to the empty vector) revealed 2,295 differentially expressed genes (DEGs) with significant change in expression (adjusted P < 0.05), including 1,497 and 798 genes that were up- or down-regulated relative to empty vector, respectively ( Supplementary Tables S3 - S7 ). This transcriptional analysis confirmed the gene silencing of PvAGO5 (two-fold reduction) in PvAGO5-RNAi transgenic roots and nodules. Furthermore, we observed that the expression of other AGO protein-encoding genes was not affected by the PvAGO5-RNAi construct ( Supplementary Table 2 ), which indicates that the gene silencing of PvAGO5 was specific. To validate these transcriptional data sets, we randomly selected 11 genes and their expression levels were assessed by RT-qPCR. The results show similar trends between RNA-seq and RT-qPCR data ( Supplementary Figure 4 ).

To gain insight into the role of AGO5 under non-symbiotic conditions, we compared the transcriptomes of transgenic roots carrying the PvAGO5-RNAi against the empty vector in the mock samples. This analysis led to the identification of 962DEGs, 598 up-regulated and 364 down-regulated genesinPvAGO5-RNAi roots, respectively ( Figure 3 and Supplementary Table 3 ). Gene Ontology term enrichment analysis of the identified DEGs revealed that the most significant GO terms were those involved in cell wall biogenesis ( Supplementary Table 8 ). Their upregulation in PvAGO5-RNAi transgenic roots suggests that AGO5 may play a role in negatively regulating the biogenesis of the cell wall during root development.

Further KEGG pathway functional classification not only confirmed the participation of AGO5 in cell wall biogenesis, but also revealed that gene silencing of PvAGO5 increases the expression of genes related to the jasmonic acid biosynthesis and ethylene perception ( Figures 4C, D and Supplementary Table 9 ). This functional classification analysis also indicated that the downregulation of PvAGO5 diminished the expression of genes encoding diverse types of transporters, among them auxin-, cation-, carbohydrate- and amino acid transporters ( Supplementary Figure 5 ). Altogether, this transcriptional analysis implicates PvAGO5 in root development, likely through the regulation of genes involved in cell wall biogenesis and in the modulation of multiple phytohormones.

Our spatio-temporal analyses indicated that PvAGO5 is expressed in root hairs in response to rhizobia, suggesting that this AGO plays a role in the rhizobial infection process. To evaluate this hypothesis, we analyzed the transcriptional responses of PvAGO5-RNAi transgenic roots during the first day of interaction with R. tropici. Our transcriptome analyses revealed no overall drastic changes in the expression of canonical genes involved in the molecular dialogue between both symbionts and the rhizobial infection process when compared with rhizobia-inoculated transgenic roots expressing the empty vector. Instead, this transcriptome analysis revealed that the expression of genes related to plant defense, which must be modulated to allow the rhizobial infection (Berrabah et al., 2015; Wang et al., 2016; Berrabah et al., 2019), was significantly increased in PvAGO5-RNAi transgenic roots during the first day of interaction with rhizobia ( Figure 5 ). We observed that the expression of genes encoding NBC-ARC domain-containing disease resistance proteins (e.g., Phvul.008g071500.1 and Phvul.001g133400.1) or LRR Receptor-Like Serine/Threonine-Protein Kinase (e.g., Phvul.005g162000.1 and Phvul.002g187300.1) increased two-fold in response to rhizobia ( Figure 5 ). These transcriptional data indicate that the downregulation of PvAGO5 activates the expression of plant defense-related genes in response to rhizobia.

Gene silencing of AGO5 significantly reduces nodule size and the number of rhizobial-infected nodule cells (Reyero-Saavedra et al., 2017). To explore the basis of this phenotype, we analyzed the transcriptome of mature nodules expressing the PvAGO5-RNAi or the empty vector. This analysis led us to the identification of 1,006 DEGs, of which 766 were upregulated and 240 downregulated in PvAGO5-RNAi nodules.

Functional classification of the DEGs indicated that many of the up-regulated genes were related to plant defense ( Figure 5 ). For example, we found that oxidative burst-associated genes (i.e., PvRbohA: Phvul006g090200.1) or genes encoding resistance and pathogenesis-related (PR) proteins were significantly upregulated in PvAGO5-RNAi nodules ( Figure 5 ). Interestingly, we also identified a three-fold induction of an Rj4 orthologue (Phvul.002g107900) which encodes a thaumatin-like pathogenesis-protein known to restrict nodulation to specific rhizobial strains in soybean ( Supplementary Figure 6 and Supplementary Table 7 ) (Tang et al., 2016).

Silencing of PvAGO5 in mature nodules resulted in the significant upregulation of genes involved in the biosynthesis and signaling of the phytohormones auxin, ethylene, and jasmonic acid ( Figure 4 and Supplementary Table 9 ). In contrast, genes involved in cytokinin degradation were up-regulated, whereas genes participating in its perception and signaling were downregulated in mature PvAGO5-RNAi nodules ( Figure 4 and Supplementary Table 9 ).

Furthermore, we observed that the expression of genes participating in cell wall biogenesis and remodeling significantly increased ( Figure 3 and Supplementary Table 7 , 8 ). However, we also observed that some genes belonging to this category were downregulated. For instance, the expression of the gene Increasing Nodule Size1 (INS1: Phvul.003g016300.1), which encodes for a UDP-Glycosyltransferase/trehalose-phosphatase, was significantly diminished in PvAGO5-RNAi nodules ( Supplementary Figure 6 and Supplementary Table 7 ). Altogether, these transcriptional data indicate that PvAGO5 regulates nodule development and functioning by modulating the expression of genes involved in the phytohormone balance, control of the plant defense response, and cell wall biogenesis, which is determinant for this symbiosis.

Legume hosts constantly translocate different mineral nutrients to the nodules to sustain an effective symbiosis. Defects in this mineral nutrient exchange compromise the symbiosis (Liu et al., 2020; Castro-Rodríguez et al., 2021). Hence, we investigated whether PvAGO5 downregulation affects the expression of genes encoding mineral nutrient transporters in mature nodules. We found that the expression of genes encoding Fe²⁺-, Mn²⁺, and K⁺ transporters was significantly diminished in PvAGO5-RNAi mature nodules ( Supplementary Figure 5 ). We also observed that the expression of Phvul.007g275300.1 that encodes the PHO1 transporter, which is required to transfer phosphate from infected nodule cells to bacteroids (Nguyen et al., 2021), was significantly downregulated. In contrast, genes encoding different ABC and MATE transporter family members, as well as sugars- and amino acid- transporters, were upregulated. Altogether, these transcriptional data indicate that PvAGO5 regulates nodule functioning by modulating the expression of genes involved in mineral nutrient transport underpinning the symbiosis.

To obtain further insights into the roles of PvAGO5 in the root nodule symbiosis and to determine changes in sRNA associations during this symbiosis, we isolated, sequenced, and compared PvAGO5-bound sRNA pools from nodule primordia and mature nodules. We choose mature nodules because of the strong transcriptional activity of the PvAGO5 promoter during this stage of development. We also included nodule primordia and non-inoculated roots to capture those sRNAs that accumulate during nodule development. For each experimental condition, we included three biological replicates. For downstream analyses, we focused only on those sRNAs present in the three libraries from each of the three experimental conditions.

PvAGO5-associated sRNAs consisted of 18-22 nt sRNAs, with a 5’ C and U bias, and an enrichment and a depletion of the sequence shorter and longer than 21-nt, respectively, compared to the starting dataset ( Figures 6A, B ). We identified 76 sRNAs, 72 of them were miRNAs and the rest rhizobial-derived tRFs. Sixty-nine of the identified miRNAs were grouped into 26 known miRNA families. The other three miRNAs were classified as P. vulgaris-specific miRNAs. 50% of these 72 miRNAs were grouped into only seven families, corresponding to miR156, miR159, miR166, miR167, miR168, miR319, and miR396. ( Supplementary Table 10 ).

A comparison between the PvAGO5-bound sRNAs from the three experimental conditions revealed that 23 out of the 76 sRNAs were present in the three tested biological samples (uninoculated roots, root bearing nodule primordia, and mature nodules), whereas the rest were present in at least two biological conditions (i.e., present in roots and in nodule primordia) or unique for a particular condition (i.e., specific to mature nodules) ( Figure 6C ). Among the 23 miRNAs present in the three experimental conditions, Pvu-miR482a-5p, Pvu-miR166a-3p, Gmax-miR1511, gma-miR6300, Pvu-miR166h-3p, Pvu-miR396c-5p, Pvi-miR156a-5p, Gmax-miR156a, Pvu-miR396b-5p, Pvu-miR1511-3p, Cme-miR168, hbr-miR156, and Pvu-miRN95, were the most abundant miRNAs ( Figure 6 and Supplementary Table 10 ). Most of these microRNA families are known to be involved in the regulation of pathways that are found altered in our transcriptomic data analysis between PvAGO5-RNAi and EV plants and could therefore explain the observed phenotypic. Some of these families are directly related to nodulation such as miR482, miR166 and miR156 (Boualem et al., 2008; Li et al., 2010; Wang et al., 2015). In the case of miR156 and miR396 families, they are involved in flavonoid biosynthesis (Yang et al., 2022) which could partly explain that, when we alter the production of AGO5, these microRNAs being associated, the biosynthesis of flavonoids is altered. Finally, an important candidate for the explanation of nodulation alteration is miR1511, that has been shown to regulate root growth under abiotic stresses, via iron homeostasis and ROS accumulation in roots (Martin-Rodriguez et al., 2021). As iron and ROS are components regulating rhizobial symbiosis at different stages of its establishment, it is likely that, by altering AGO5, we alter the function of miR1511, the corresponding iron and ROS accumulations and, ultimately, nodulation.

We found eleven, zero, and twenty-six miRNAs that were uniquely bound to PvAGO5 from uninoculated roots, roots bearing nodule primordia, and mature nodules, respectively ( Supplementary Table 11 ). Among the 26 unique sRNAs bound to PvAGO5 from mature nodules, four of them were tRFs from R. tropici, which represents a 33-times enrichment compared to the original dataset ( Supplementary Table 11 ). Interestingly, this type of sRNAs was recently shown to play a role in the root nodule symbiosis in soybean (Ren et al., 2019).

Next, we predicted putative target genes of AGO5-bound sRNAs by using psRNAtarget V2 using default parameters (Dai et al., 2018). Sixty-four out of 76 identified sRNA were predicted to target at least one mRNA ( Supplementary Table 12 ). Although no GO terms nor any pathway were significantly enriched in the identified target gene set, analysis of protein domain enrichment reveals that AGO5-bound sRNAs preferentially target genes containing domains related to defense (e.g., NB-ARC, Leucine-rich repeat and Toll/interleukin-1 receptor), growth and development process (e.g., ZPR1-type TFs, SPB-box TFs, and Growth-regulating factor) or nodulation (e.g., Nodulin) ( Supplementary Table 13 ). Next, we analyzed the expression of the predicted target genes in our transcriptome data set. Interestingly, the majority of the predicted target genes were up-regulated instead of the canonical down-regulation expected for sRNA target genes ( Supplementary Table 12 ).

In summary, these data indicate that PvAGO5 can bind sRNA, both plant- and rhizobia-derived, that may play a role in the nodule development and functioning. The fact that the predicted target genes were up-regulated suggests that AGO5 may play a role in negatively regulating the protein production of sRNA targets, as well as the principal miRNA effector protein AGO1. However, further investigation is required to confirm this hypothesis.