Strains and plasmids used in this study are listed in Supplementary Table S1 . R. leguminosarum strains were routinely grown at 28°C in yeast mannitol broth (YMB) (Vincent, 1970), tryptone-yeast extract (TY) (Beringer, 1974), Rhizobium minimal media (Rmin) (O’Gara and Shanmugam, 1976), or universal minimal salt (UMS) medium (Wheatley et al., 2017). Escherichia coli strains were grown at 37°C in LB medium (Bertani, 1951). E. coli DH5α was used for standard cloning procedures (Hanahan, 1983), and E. coli S17.1 (Simon et al., 1983) for conjugative plasmid transfer between E. coli and R. leguminosarum. Antibiotic concentrations used were as follows (µg ml⁻¹): ampicillin, 100; kanamycin, 50; tetracycline, 5 (for R. leguminosarum) or 10 (for E. coli).

For growth curve analyses, bacterial cultures were grown in zinc-free UMS medium up to an OD₆₀₀ of 0.6 and diluted to an initial OD₆₀₀ of 0.01 in fresh zinc-free UMS or the same medium containing 50 µM EDTA with or without 50 µM ZnSO_4, MnSO_4, or FeSO₄ supplementation. Growth curves were determined with cell cultures incubated in 100-well Honeycomb plates (Growth Curves Ltd., Piscataway, NJ, USA) with continuous double-orbital shaking in a Bioscreen C Pro device (Growth Curves Ltd., Piscataway, NJ, USA) at 28°C, with OD₆₀₀ measurement intervals every 30 min for 30 h.

DNA purification, digestion with restriction enzymes, ligation, gel electrophoresis, amplification by polymerase chain reaction (PCR), and E. coli transformation were performed as described in standard protocols (Sambrook and Russell, 2001). Primers used for mutants and plasmid construction are listed in Supplementary Table S2 .

Derivatives of Rlv SPF25 incorporating zniA and zniCBA mutations were generated by homologous recombination using the suicide vector pK18mobsac (Schäfer et al., 1994). For this purpose, a first round of PCR reactions amplified two DNA fragments of ca. 1 Kb corresponding to up- and downstream regions of the zniA (rlv_3444) gene using primers P1_rlv_3444_BamHI/P2_rlv_3444 and P3_rlv_3444/P4_rlv_3444_HindIII, respectively, and genomic DNA from the Rlv UPM791 strain as a template. The two PCR products of the flanking DNA regions of the gene were linked via an overlapping GC-rich (C₅G₅C₅) sequence, previously incorporated using P2_rlv_3444 and P3_rlv_3444 primers, by a fusion PCR (Cha-aim et al., 2009) using primers P1_rlv_3444_BamHI/P4_rlv_3444_HindIII. Following the same procedure, regions covering zniC (rlv_3442) upstream and zniA (rlv_3444) downstream were amplified using primers P1_rlv_3442-44_BamHI/P2_rlv_3442–44 and P3_rlv_3444/P4_rlv_3444_HindIII, respectively. A fusion PCR with primers P1_rlv_3442-44_BamHI/P4_rlv_3444_HindIII was performed to amplify the fragment carrying the zniCBA deletion. Fusion PCR products containing zniA and zniCBA deletions were cloned separately in pBlueScript-II KS+ as BamHI-HindIII restriction fragments, sequenced, and moved to the pK18mobsac suicide vector. The resulting plasmids, pK18.ZniA and pK18.ZniCBA, were introduced independently into the SPF25 strain by conjugation, and homologous recombination by a double crossover event was selected by the sacB system in Rhizobium minimal medium (Schäfer et al., 1994). Gene deletions were verified by PCR and sequencing analysis, and the transconjugant strains carrying the mutations in zniA and zniCBA were designated as UPM1629 and UPM1631, respectively.

To generate an in-frame deletion in the znuA gene, regions up- and downstream of the znuA gene were PCR amplified using primers P1_ZnuA_BamHI/P2_ZnuA and P3_ZnuA/P4_ZnuA_HindIII, respectively, and UPM791 genomic DNA as a template. The two PCRs were linked via an overlapping GC-rich sequence by a fusion PCR using primers P1_ZnuA_BamHI/P4_ZnuA_HindIII to amplify the fragment containing the deletion of the znuA gene. The resulting DNA fragment was digested with BamHI-HindIII restriction enzymes and cloned in pBlueScript-II KS+, sequenced, and then moved to the pK18mobsac suicide vector. The new plasmid generated (pK18.ZnuA) was introduced into SPF25 strain and into derivative strains UPM1629 (ΔzniA) and UPM1631 (ΔzniCBA) by conjugation, and homologous recombinations were selected by the sacB system. Gene deletion was verified by PCR and sequencing analysis, and the transconjugant strains carrying mutations in znuA, znuA/zniA, and znuA/zniCBA were designated as UPM1628, UPM1630, and UPM1632, respectively.

To generate mutant derivatives bearing a deletion in the zur gene, two DNA regions up- and downstream of the gene were amplified by PCR using primer pairs P1_Zur/P2_Zur and P3_Zur/P4_Zur, respectively. The resulting DNA fragments were fused by an overlapping GC sequence by PCR using P1_Zur/P4_Zur primers. The PCR product carrying the zur deletion was cloned in plasmid pCR2.1TOPO, sequenced, and moved to the pK18mobsac vector as an XbaI-BamHI restriction fragment, rendering the pK18.Zur plasmid. This construction was introduced into SPF25 and into its derivative strain UPM1630 (ΔznuAΔzniA), and homologous recombination was selected by the sacB system, giving rise to UPM1633 (Δzur) and UPM1634 (ΔzurΔznuAΔzniA). Gene deletion was checked by PCR and sequencing analysis.

For analysis of promoter expression, transcriptional gene fusions were generated with the promoterless gusA gene in plasmid pLMB51 (Tett et al., 2012). DNA fragments containing the upstream regions of zniC and znuA genes were PCR amplified with primers Prom1_rlv_3442-44_BamHI_F/Prom5_rlv_3442-44_XbaI_R and Prom_ZnuA_BamHI_F/Prom_ZnuA_XbaI_R, respectively, and UPM791 genomic DNA as a template. PCR products were cloned separately into pBlueScript-II KS+ plasmid using BamHI-XbaI restriction sites, sequenced, and moved to pLMB51 plasmid, thus generating pLMBZniCBA and pLMBZnuA plasmids, respectively (Tett et al., 2012). For regulatory analysis of the expression of the zniCBA operon, truncated forms of the DNA region upstream of the operon containing 537, 258, and 88 bp from the ATG codon were amplified with primers Prom2_rlv_3442-44_BamHI_F/Prom5_rlv_3442-44_XbaI_R, Prom3_rlv_3442-44_BamHI_F/Prom5_rlv_3442-44_XbaI_R, and Prom4_rlv_3442-44_BamHI_F/Prom5_rlv_3442-44_XbaI_R, respectively. PCR products were cloned separately into pBlueScript-II KS+ plasmid, sequenced, and moved to pLMB51 plasmid as a BamHI-XbaI restriction fragment, rendering pLMB537.ZniCBA, pLMB258.ZniCBA, and pLMB88.ZniCBA, respectively.

For the construction of pBBRZniA, the DNA region upstream of the zniC gene and the zniA gene from its 16 bp upstream sequence were amplified with primers Prom3_rlv_3442-44_BamHI_F/P2_Com_rlv_3444_R and P3_Com_rlv_3444_F/P4_Com_rlv_3444_XbaI_R, respectively, and UPM791 genomic DNA as a template. The two PCRs were linked by an overlapping GC-rich sequence through a fusion PCR using primers Prom3_rlv_3442-44_BamHI_F/P4_Com_rlv_3444_R. The same procedure was performed for the construction of pBBRZniA_ST, but the P4_Com_rlv_3444Strep_XbaI_R, including the sequence coding for StrepTag II peptide (WSHPQFEK) for in-frame fusion of the tag sequence to the 3´end of the gene, was used. To generate the pBBRZnuA plasmid, DNA covering the znuA gene and its promoter region was PCR-amplified with Prom_ZnuA_BamHI_F/Com_ZnuA_XbaI_R primers. The PCR products obtained were cloned separately in the pBlueScript-II KS+ vector as BamHI-XbaI restriction fragments, sequenced, and moved to the broad host range vector pBBR1MCS-2 (Kovach et al., 1995). Site-directed mutagenesis of the zniA_ST gene was performed in pBBRZniA_ST using complementary oligonucleotides with adequate centered nucleotide substitutions ( Supplementary Table S2 ) and DpnI digestion to eliminate the native template. Substitution of target nucleotides as well as the absence of additional mutations was verified by sequencing. The pLMBcZni plasmid was generated by amplifying by PCR the zniCBA operon and its promoter region using primers Prom1_rlv_3442-44_BamHI_F and rlv_3442-44_XbaI_R, and UMP791 genomic DNA as template. The PCR product was cloned in the pBluescript-II KS+ plasmid, sequenced, and then moved as a BamHI-XbaI fragment to the pLMB51 plasmid.

Pea (Pisum sativum L. cv. Frisson) and lentil (Lens culinaris cv. Magda) were surface-sterilized, pregerminated, planted in Leonard jar assemblies under bacteriologically controlled conditions, and inoculated with 1 ml of early stationary phase bacterial cultures grown in YMB as previously described (Sánchez-Cañizares et al., 2011). Plants were grown in a greenhouse with a standard nitrogen-free plant nutrient solution (Brito et al., 1994). When required, the nutrient solution was supplemented with ZnSO₄ at the desired concentration. Plants were grown at 18/25°C night/day with 16/8 h of light/dark photoperiod and harvested 21 days (pea) or 28 days (lentil) after inoculation. Shoots were collected and dried (60°C for 48 h), and shoot dry weight was determined. Shoot total nitrogen content was measured using a TruMac C/N analyzer (Leco Corporation).

β-Glucuronidase activity assays were performed with free-living cells and bacteroids as described in Lodwig et al. (2004) using X-GlcA (5-bromo-4-chloro-3-indolyl-β-D-glucuronide) as substrate. Values of activity were calculated as OD₄₂₀ min⁻¹ (mg protein)⁻¹ and expressed as Miller units. Bacteroids were obtained from the nodules as previously described (Cacho et al., 2010). Protein content was obtained following the bicinchoninic acid method (Smith et al., 1985) after alkaline digestion of cells at 90°C in 2 N NaOH for 10 min using bovine serum albumin as the standard.

RNA was extracted from nodule samples as described in Ballesteros-Gutiérrez et al. (2023). RNA concentration was quantified with a NanoDrop spectrophotometer and tested for possible DNA contamination by PCR using the RNA samples as templates and primers specific for rpoD (rpoD_qPCR_F/rpoD_qPCR_R). RNA integrity was confirmed by electrophoresis in a 1% agarose gel. cDNA was synthesized from 500 ng of DNA-free RNA using the PrimeScript RT reagent kit (Takara, Saint-Germain-en-Laye, France), supplemented with RNase out (Invitrogen, Thermo Fisher Scientific, Massachusetts, USA) following the manufacturer’s specifications. Quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR) was performed with Power SyBR Green master mix (Applied Biosystems) and primers rlv_3444_qPCR_F/rlv_3444_qPCR_R for zniA, znuA_qPCR_F/znuA_qPCR_R for znuA, hupL_qPCR_F/hupL_qPCR_R for hupL, and rpoD_qPCR_F/rpoD_qPCR_R for rpoD genes. Gene expression levels were normalized to the rpoD reference gene (Ballesteros-Gutiérrez et al., 2023). Determinations were performed with RNA extracted from three independent biological samples, with the threshold cycle (C _T) determined in triplicate. The relative levels of transcription were calculated using the 2^{–ΔΔ C T} method (Livak and Schmittgen, 2001).

Protein sequences were obtained from the GenBank database. Multiple sequence alignment was created with SnapGene 6.2 software using the Clustal Omega multiple alignment algorithm v.1.2.4 (Chenna et al., 2003). The phylogenetic tree of the protein alignment was analyzed with MEGA 12 (Tamura et al., 2021) using a neighbor-joining algorithm method (Saitou and Nei, 1987) with bootstrapping (1,000 iterations). The Uniprot database (https://www.uniprot.org/) was used to perform protein BLAST and amino acid sequence identity analysis. Zur box prediction from DNA upstream of zniCBA and znuA genes were generated based on the conserved sequence motifs using the bioinformatics tool Multiple EM for Motif Elicitation (MEME) (Bailey et al., 2006). The analysis included the 42 Zur-binding motifs from Rhizobiales obtained from the Regprecise database of the curated inferences of regulons in prokaryotic genomes (Novichkov et al., 2009). Structural prediction of the mature RLV_3444 protein (25–299 residues) was carried out using the AlphaFold 3 server (Abramson et al., 2024). Structural comparison with the PDB database was performed at the DALI server (http://ekhidna2.biocenter.helsinki.fi/dali/) (Holm et al., 2023). The program was run with the default server parameters. Alignments for further data analysis and figure preparation were done in Pymol (v.3.1.3.1, Schrödinger). Published protein structures were obtained from the PDB database.

Statistical analysis was performed by analysis of variance (ANOVA) linear model test following completely random design. Multiple comparisons of means were analyzed as indicated in the corresponding legend figure. Statistical analysis and graph generation were performed with GraphPad Prism 8 (GraphPad Software).

Inspection of the Rlv UPM791 genome reveals the presence of a chromosomic operon (rlv_5427-5425) encoding the putative high-affinity zinc transporter system ZnuABC. The znuA (rlv_5427) gene is oriented in the opposite direction to the znuC (rlv_5426) and znuB (rlv_5425) genes ( Figure 1A ). Rlv ZnuA shares a high amino acid sequence identity with its homologues in S. fredii, A. tumefaciens and S. meliloti (identities of 69.1%, 67.31%, and 66.56%, respectively) ( Supplementary Figure S1 ). A Zur-like transcriptional regulator is encoded by gene rlv_ 5424 located immediately downstream of the znuB gene ( Figure 1A ). RLV_5424 is very similar to Zur from A. tumefaciens C58, S. fredii CCBAU 45436 and S. meliloti 1021 strains (74.24%, 72.52%, and 70.23% sequence identities, respectively). A Zur box containing a 15 bp sequence was identified upstream znuA gene, very similar to the conserved Zur motifs of Rhizobiales, localized 41 bp upstream of the translational start site of the gene ( Figure 1B ).

The Rlv UPM791 genome encodes a metal ABC transporter system containing the host-dependent metal-binding protein RLV_3444 (Durán et al., 2021). This protein is encoded by the chromosomic rlv_3444 gene, which is part of an operon with a typical organization of bacterial ABC permeases and also encodes an ATP-binding protein (rlv_3442) and a permease (rlv_3443) ( Figure 1A ). In addition, a conserved Zur motif was also identified 153 bp upstream of the rlv_3442 gene. A comparison of the RLV_3444 amino acid sequence with protein databases revealed a highly significant level of conservation (over 75% identity) with proteins annotated as metal, Zn, or Zn/Mn ABC transporter SBPs in the genomes of several members of Rhizobiaceae. Furthermore, RLV_3444 exhibits a high amino acid sequence identity with the functionally characterized SBPs from K. pneumoniae ZniA (53.42% sequence identity), A. tumefaciens TroA (41.61% identity) and Paracoccus denitrificans AztC (40.88% identity), previously shown to participate in zinc uptake (Chaoprasid et al., 2016; Handali et al., 2015; Maunders et al., 2024) ( Supplementary Figure S1 ). Sequence alignments revealed that RLV_3444 contains the three fully conserved histidine residues (H62, H127, and H193, RLV_3444 numbering) required for the coordination of a zinc atom in the crystal structures of K. pneumoniae ZniA and P. denitrificans AztC ( Supplementary Figure S2 ). In RLV_3444, a glutamate at position 268, conserved in K. pneumoniae ZniA, where it participates in zinc coordination (Maunders et al., 2024), was also identified. In contrast, the short central histidine-rich domain present in P. denitrificans AztC and A. tumefaciens TroA and associated to Zn²⁺ specificity (Blindauer, 2015) is not present either in RLV_3444 or in K. pneumoniae ZniA.

Structural prediction of mature RLV_3444 protein (25–299 residues) was carried out using the AlphaFold 3 server, and the model obtained was compared to the available structures in the PDB database using the Dali server. Proteins showing the best structural homologies with RLV_3444 were zinc-, manganese-, and iron-binding proteins, with the highest scores for K. pneumoniae ZniA (PDB id: 8SVC; Maunders et al., 2024) and P. denitrificans AztC (PDB id: 5W57; Neupane et al., 2017) ( Supplementary Table S3 ), suggesting similar functions for these proteins. Structural alignment of RLV_3444 with AztC and ZniA has revealed that it shares the two independent domains that interact to create a metal-binding pocket between them connected by a rigid α-helix characteristic of the cluster A SBPs-associated family (Berntsson et al., 2010) ( Figure 2A ). Structural alignments have also shown a highly conserved metal binding site ( Figure 2B ).

Taken together, these results suggest that RLV_3444 is a potential zinc-binding protein of an ABC transporter system. Based on the bioinformatic analysis and the functional and regulatory role of this protein (see below), the corresponding gene was designated as zniA (zinc import), and we also designated the rlv_3442–3444 operon as zniCBA ( Figure 1 ).

In order to investigate the functional role of ZniA under free-living conditions, growth curves of SPF25 and its derivatives UPM1628 (ΔznuA), UPM1629 (ΔzniA), and UPM1630 (ΔznuAΔzniA) strains were determined in UMS medium with no zinc added. The results indicated that all bacterial strains exhibited similar growth under these zinc-depleted conditions ( Figure 3A ). Atomic absorption spectromety analysis of zinc in the “Zn-free” medium revealed the presence of ca. 1 µM Zn²⁺ traces, likely too high to induce real zinc deprivation conditions (Outten and O’Halloran, 2001), so we added EDTA (50 μM) to chelate zinc cations. In the presence of EDTA, the growth of znuA- and zniA-deleted single mutants was similar to that observed in the wild-type strain. Conversely, the bacterial strain carrying deletions in both znuA and zniA genes (UPM1630) showed a clearly defective growth phenotype ( Figure 3A ). This phenotype was fully reverted by the expression of znuA or zniA genes in pBBRZnuA or pBBRZniA plasmids, respectively. Furthermore, the phenotype of the double mutant was also reverted by supplementing the medium with Zn²⁺ (50 µM ZnSO₄) indicating that the defective phenotype of UPM1630 was due to its inability to take up zinc ( Figure 3B ). Supplementation of EDTA-containing medium with other metals (Mn²⁺ and Fe²⁺) did not restore the growth of the znuA/zniA-deficient derivative UPM1630 ( Figure 3B ), indicating that the growth defect was specifically related to a zinc deficiency. Altogether, these data indicate that ZniA plays a functional role similar to that of ZnuA, both proteins being able to support wild-type levels of growth under zinc-limiting conditions in R. leguminosarum free-living cells.

With the aim to consider the possibility of cross-talk between Zni and Znu systems via their respective SBPs, we constructed a mutant (UPM1631) carrying a deletion of the whole zniCBA operon. This mutant showed normal growth under all conditions tested, as expected from the presence of an active Znu system ( Figure 4 ). These results indicate that ZniCBA and ZnuABC were functionally redundant under metal-limitation conditions in the presence of EDTA (0 µM ZnSO₄, 50 µM EDTA). In order to check whether ZniA could be providing Zn²⁺ to the bacterial cell via the Znu permease in the absence of ZnuA, a znuA deletion was generated in this zniCBA mutant, thus resulting in the UPM1632 strain. The growth of the UPM1632 double mutant was similar to that observed in the wild-type strain in UMS under zinc-depleted conditions (0 µM ZnSO₄ added), but it was impaired in the presence of EDTA. The defective phenotype was reverted back to wild-type levels by supplementing the medium with ZnSO₄ (50 µM) and also by the expression of the znuA gene from the pBBRZnuA plasmid. In contrast, such complementation was not induced by the expression of zniA from pBBRZniA. As a control, we checked that the UPM1632 defective growth phenotype under metal-chelated conditions was rescued to wild-type levels with the expression of ZniCBA under the control of its own promoter in the pLMBcZni plasmid ( Supplementary Figure S3 ). These data demonstrate that ZnuA function cannot be substituted by ZniA when Zn provision depends on Znu permease.

The potential involvement of ZniA histidine residues (H62, H127, and H193) in ZniA function has been investigated. To carry out these studies, plasmids expressing ZniA_ST, incorporating a C-terminal Strep-Tag affinity tail for future monitoring and purification purposes, and variants harboring H62A, H127A, and H193A changes were generated and transferred separately by conjugation into the znuA/zniA-deleted strain UPM1630. Bacterial cultures of the corresponding transconjugant strains were analyzed for growth in UMS medium under zinc-depleted (0 µM ZnSO₄), metal-chelated (0 µM ZnSO₄, 50 µM EDTA), and zinc-repleted (50 µM ZnSO₄, 50 µM EDTA) conditions. The growth of all strains was similar to that observed in the SPF25 strain under zinc-depleted conditions ( Figure 5 ). The expression of ZniA_ST in pBBRZniA_ST complemented the defective growth phenotype of the UPM1630 strain up to wild-type levels under zinc-chelated conditions, indicating that the incorporation of the affinity tail did not affect the function of the protein. In these experiments, we observed that zniA_ST carrying the mutations H62A, H127A, or H193A did not complement the ΔzniAΔznuA mutant for growth in the medium containing EDTA (0 µM ZnSO₄, 50 µM EDTA). The defective growth associated with the mutation in the histidine residues was restored with the addition of zinc to the EDTA-chelated medium. These results suggest that these histidine residues are critical for ZniA function, likely by contributing to metal coordination along the transport process.

The metal-binding protein ZniA was shown to be overexpressed in pea bacteroids, with protein expression levels ca. twofold higher in pea than in lentil bacteroids (Durán et al., 2021). Our proteomic analysis also revealed that ZnuA was detected in bacteroids from pea but not from lentil (Durán et al., 2021). Furthermore, previous studies revealed a host-dependent contribution of ZnuA in S. fredii-soybean system (Jiao et al., 2018; Zhang et al., 2020). This prompted us to study the potential requirement for ZniA and ZnuA for efficient symbiosis of Rlv with pea and lentil plants. To carry out these studies, plants inoculated with Rlv SPF25 and its derivatives UPM1628 (ΔznuA), UPM1629 (ΔzniA), and UPM1630 (ΔznuAΔzniA) strains were grown in a standard N-free nutrient solution (0.8 µM ZnSO₄) or in the same solution containing 80 µM ZnSO₄. Symbiotic performance of znuA- and zniA-deleted strains was similar to that of the wild-type strain in both legumes under either standard- or high-zinc concentrations ( Figure 6 ). In contrast, pea and lentil plants inoculated with znuA/zniA double mutant UPM1630 accumulated significantly less shoot dry weight and less nitrogen in comparison to the wild-type strain at both zinc concentrations tested. This result indicates that high-affinity zinc transport is essential for optimal development of symbiosis of R. leguminosarum with legume plants. The impaired phenotype of the UPM1630 strain was complemented by the overexpression of wild-type copies of ZniA or ZnuA in pBBRZniA or pBBRZnuA plasmids, respectively ( Figure 6 ). In contrast, the observed impairment of plant growth in the double mutant was not corrected by the addition of additional Zn²⁺ to the nutrient solution (80 μM ZnSO₄, 100-fold higher than the original concentration). This observation indicates that, in the absence of functional Znu and Zni systems, potential alternative metal transporters do not provide the bacteroids with enough zinc for the symbiotic performance of this strain. These results indicate that both ZnuABC and ZniCBA transporter systems play interchangeable, essential functional roles in the provision of Zn²⁺ to the bacteroids under symbiotic conditions.

In order to investigate the regulatory mechanisms governing the expression of zni and znu genes, independent transcriptional fusions to DNA upstream from the zniCBA operon and from the znuA gene to a promoterless gusA reporter gene in the pLMB51 plasmid were constructed, thus generating plasmids pLMBZniCBA (zniCBA´-gusA) and pLMBZnuA (znuA´-gusA). These plasmids were transferred separately by conjugation to SPF25 and to its derivatives UPM1628 (ΔznuA), UPM1629 (ΔzniA), and UPM1630 (ΔznuAΔzniA) strains.

We first analyzed zni gene expression under free-living conditions. To this aim, reporter activity was determined in bacterial cultures grown in UMS medium or in the same medium supplemented with EDTA (50 µM). In SPF25 and UPM1629 (ΔzniA) strains harboring the zniCBA´-gusA reporter fusion, only basal levels of enzymatic activity (ca. 20 units) were observed when cells were grown in UMS medium ( Figure 7 ). In contrast, significant levels of activity were reported in both strains under metal-chelated conditions. Interestingly, under the same conditions, reporter activity increased in the znuA-deficient UPM1628 strain (ca. 8-fold over basal levels); finally, the expression of the reporter enzyme was highly induced (ca. 60-fold over basal levels) in the znuA/zniA-deficient UPM1630 strain. A similar pattern of results was observed when analyzing expression of the znuA gene through the use of znuA´-gusA fusion ( Figure 7 ). In this case, the level of reporter activity was always higher than in the case of zniCBA´-gusA fusion, suggesting that ZnuA is the major transporter supporting Zn provision to the bacterium. In all cases, enzymatic activities associated with either reporter fusion were reduced to basal levels when the metal-chelated medium was supplemented with zinc (50 µM ZnSO₄) in all the genetic backgrounds. These results confirm that the expression of both transporter systems is negatively regulated by zinc and that ZniCBA functions as an auxiliary transporter system in the absence of ZnuABC.

The identification of potential Zur-binding sites in the promoter region of the zniCBA operon and znuA gene ( Figure 1 ) prompted us to investigate whether the expression of both systems is regulated by the transcriptional regulator Zur. To carry out these studies, pLMBZniCBA and pLMBZnuA plasmids were transferred into UPM1633 and UPM1634 strains (Δzur and ΔznuAΔzniAΔzur derivatives of SPF25 and UPM1630, respectively), and β-glucuronidase activity was determined following incubation in the different media. In these experiments, the levels of reporter activity were highly induced in all zur-deleted strains carrying pLMBZniCBA or pLMBZnuA plasmids, regardless the levels of ZnSO₄ in the medium ( Figure 8 ). These findings indicate that the zniCBA operon and znuA are negatively regulated by Zur.

The metal specificity of responses of zniCBA and znuA was investigated in SPF25 and UPM1630 (ΔznuAΔzniA) strains harboring pLMBZniCBA or pLMBZnuA plasmids through experiments similar to those described above, but adding other metal cations (Mn²⁺, Fe²⁺) instead of Zn²⁺ to EDTA-supplemented media ( Figure 9 ). There was no effect of the addition of these metals in the case of double mutant UPM1630, indicating that the observed effect of Zn blocking was specific for this element. Interestingly, a significant repression of β-glucuronidase activity was shown when the same experiment was carried out in the wild-type background. This observation might be associated with a displacement of zinc-chelated to EDTA by these metals, thus decreasing the effective metal chelator concentration for the complexation of zinc in the medium, as previously suggested in E. coli znu regulation studies (Patzer and Hantke, 2000). This effect was not observed in UPM1630, likely because not enough zinc to repress the expression of the system enters into the cell in the absence of both transporter systems. Furthermore, a low but significant increment of the β-glucuronidase activity associated with the pLMBZnuA reporter plasmid was observed in UPM1630 strain grown in EDTA-chelated medium supplemented with manganese and iron. These results suggest that Rlv ZnuABC and ZniCBA are inducible, specifically by zinc starvation conditions, in a Zur-dependent manner.

We also wanted to better define the regulatory region controlling the expression of the zniCBA operon. To this aim, a deletion analysis of its promoter region was carried out using transcriptional fusions to DNA regions of the promoter truncated 537, 258, and 88 bp upstream of the operon (plasmids pLMB537.ZniCBA, pLMB258.ZniCBA, and pLMB88.ZniCBA, respectively). The new plasmids, together with pLMBZniCBA, were independently transferred to wild-type (SPF25) and zniA/znuA double mutant (UPM1630) strains. Reporter activities were determined in bacterial cultures grown in UMS and in the same medium supplemented with the metal chelator EDTA with or without zinc (50 µM ZnSO₄). In these experiments, fusions including the 537- and 258-bp fragments of the promoter region induced reporter activities that were similar to those associated with the original fusion spanning 775 bp ( Figure 10 ). In contrast, strains carrying the shortest fragment (88 bp) induced only basal levels of reporter activity under all the conditions assayed. These observations indicate that a relevant regulatory region of the ZniCBA transporter system is located between positions 258 and 88 pb upstream of the first ATG codon of the operon, which is consistent with the presence of the Zur box in this region ( Figure 1B ).

Expression analysis of the zniCBA operon and znuA gene was also carried out under symbiotic conditions. We first analyzed by qRT-PCR whether the control of the host-dependent expression of zniA was regulated at the transcriptional level. This analysis was also extended to the znuA gene, as our proteomic analysis revealed that its encoded protein was detected in pea bacteroids but not in those developed in lentil plants (Durán et al., 2021). Quantification of the levels of transcripts revealed that both zniA and znuA genes were significantly upregulated in bacteroids from both pea and lentil (ca. 10-fold and 3-fold, respectively, for zniA and znuA) ( Figure 11A ). In this analysis, transcripts of the gene encoding the [NiFe] hydrogenase large structural subunit HupL, governed by a strong host-dependent control (Brito et al., 2008), exhibited expression increments of 100-fold in pea versus lentil bacteroids (20.2 ± 3.0 and 0.18 ± 0.02 relative expression levels in pea and lentil bacteroids, respectively). These data indicate that the host-dependent expression of zniA and znuA is controlled at the transcriptional level. In addition, these observations suggest that the metal provision to bacteroids might be more restrictive in pea than in lentil.

The symbiotic induction of zniA and znuA genes was also analyzed by using transcriptional fusions of zniCBA and znuA promoter regions to the gusA reporter gene in the pLMB51 plasmid. To this aim, we analyzed reporter activities in bacteroids prepared from plants grown in standard nutrient solution (0.8 µM ZnSO₄) inoculated with SPF25 (wild type) and its derivatives UPM1628 (ΔznuA), UPM1629 (ΔzniA), and UPM1630 (ΔznuAΔzniA) bearing transcriptional fusions for zniCBA, znuA, or just empty plasmid. Data revealed that expression levels of the reporter gene associated with znuA were higher than those related to zniCBA in all genetic backgrounds and in both legume hosts, which is consistent with qRT-PCR data ( Figure 11B ). Also, the symbiotic expression of the β-glucuronidase enzyme associated with both reporter fusion plasmids was higher in strains carrying the deletion in znuA (UPM1628) than in the wild type, and even higher levels of reporter activity were scored in the double mutant UPM1630, similar to what was previously observed in free-living cells. In addition, the activity profile recorded in a zniA-deleted mutant harboring pLMBZnuA plasmid increased in comparison to that in SPF25 in pea-bacteroids, suggesting compensatory levels of expression of this system in the absence of ZniA for the potential restriction to zinc in this legume. Enzymatic activities derived from both reporter fusion plasmids increased in lentil bacteroids induced by the UPM1630 strain, suggesting a higher level of zinc deprivation in this host in the absence of both transport systems.