BacA(SbmA) importer of legume symbiotic NCR peptides: Protein architecture, function, and evolutionary implications

Significance

Sinorhizobium meliloti BacA_Sm and Escherichia coli SbmA_Ec are closely related proteins that function as homodimeric transporters to import peptides and other cargos through the cytoplasmic membrane into the cytoplasm. BacA is critical for S. meliloti to establish a nitrogen-fixing symbiosis with its legume hosts because of its ability to import Nodule Cysteine-Rich (NCR) plant peptides. This import protects the bacteria inside the nodule from the potentially lethal effects of these NCR peptides while also enabling NCRs to establish intracellular interactions that are essential for symbiosis. Our extensive multidisciplinary studies offer important insights into function of BacA/SbmA transporters and provide a molecular explanation for why BacA/SbmA orthologs from mammalian pathogens can replace BacA_Sm but those from other rhizobia cannot.

Abstract

Some legumes encode families of NCR (Nodule-Cysteine-Rich) peptides that cause their rhizobial partners to terminally differentiate during the development of a nitrogen-fixing symbiosis. Sinorhizobium meliloti, whose plant hosts Medicago truncatula and Medicago sativa express ca. 600 NCR peptides during root nodule development, possesses a symbiotically essential BacA_Sm protein that imports certain NCR peptides into the cytoplasm. This import permits proteolytic degradation of the NCR peptides, thereby protecting the endocytosed bacteria from their antimicrobial peptide-like lethality, while also allowing certain NCR peptides to undergo their symbiotically critical interactions with cytoplasmic components, for example heme-sequestration in the case of NCR247. Our study employed 54 S. meliloti bacA_Sm missense mutants (35 to cysteine and 19 to glycine) that we tested for protein production, ability to establish a nitrogen-fixing symbiosis, and their susceptibility to killing by higher levels of the NCR247 and the Bac7(1-35) peptides. We also used the Single Cysteine Accessibility Method to make topological inferences. Our detailed genetic, biochemical, structural, and physiological analyses have revealed that BacA_Sm and SbmAhomodimers function as finely tuned transporters, whose structures can be relatively easily disrupted by single amino acid changes. Our finding that several mutations that differentially separate nitrogen-fixation, NCR247 import, and Bac7(1-35) import map to the lining of the peptide-binding cavity suggests a molecular explanation underlying the paradoxical observation that SbmA/BacAs from pathogens can fully replace BacA_Sm, whereas BacAs from other rhizobia cannot.

Legumes can grow without nitrogen fertilizer because they are able to establish a symbiosis with rhizobia in which the bacteria convert N₂ gas into ammonia (nitrogen-fixation) in return for fixed carbon from the plant. After invading the root nodules they elicit, the rhizobia are endocytosed into membrane compartments (symbiosomes) within the cytoplasm of plant cells in the interior of the nodule where they undergo a set of physiological changes that convert the rhizobia into nitrogen-fixing bacteroids (1). In some legumes, the plant utilizes a family of defensin-like Nodule Cysteine Rich (NCR) peptides expressed specifically in the nodules to cause the endocytosed bacteria to terminally differentiate, a process thought to increase the efficiency of nitrogen fixation (2–4). In the case of the Sinorhizobium meliloti/Medicago truncatula symbiosis, the plant NCR peptide family consists of more than 600 cationic, acidic, and neutral members ranging from 24 to 65 amino acids, which are expressed in successive waves as the symbiosis is being established (2, 5–9). The effects of these NCR peptides are profound as they cause major physiological, transcriptional, and morphological changes associated with bacteroid differentiation, including endoreduplication, cell elongation or branching, and alterations of membrane permeability. A few NCR peptides have shown to be critically required to establish the symbiosis (10–13), whereas others are not (14), and still others play a role determining fixation level compatibility between the host and the symbiont (15, 16). At higher concentrations, cationic NCR peptides with pI > 9 can act as fast-acting antibiotics that can kill bacteria via cell lysis through disruption of their cytoplasmic membrane (17, 18). The detection of ca. 140 NCR peptides in the cytosol of bacteroids is consistent with some of the NCR peptides also interacting with intracellular bacterial targets (19, 20).

The smallest and best characterized NCR peptide is 24 amino acid NCR247 from M. truncatula, which is essential for symbiosis (13). We have previously shown that NCR247 binds heme with nanomolar affinity and sequesters it, first into hexamers and then higher-order complexes that render the heme biologically unavailable (13). Once imported into the cytoplasm of developing bacteroids, NCR247 creates a physiological state of heme deprivation, which in turn induces the import of the high levels of iron necessary for nitrogenase synthesis (13, 21, 22). However, the biological action of NCR247 is complex, as NCR247 also interacts with multiple proteins and causes substantial transcriptional changes affecting ∼15% of the genome (21, 22) Moreover, we have recently shown that, besides acting in the cytoplasm, NCR247 acts in the periplasm to induce the ChvI/ExoS- and FeuP/FeuQ-controlled regulons (23).

The S. meliloti BacA_Sm protein, an inner membrane peptide transporter, is required for the development of a nitrogen-fixing symbiosis with its Medicago hosts. BacA’s essentiality was revealed by the finding that S. meliloti bacA null mutants elicit ineffective nodules containing bacteria within the infection threads but no mature bacteroids because the rhizobia are killed upon release from the infection thread (24, 25). It has subsequently been learned that BacA’s ability to import NCR peptides is essential in two different ways for the development of a nitrogen-fixing symbiosis. The first symbiotically essential role played by BacA is to protect the rhizobia from the antimicrobial effects of the NCR peptides by importing them into cytoplasm where they can be proteolytically degraded, thus lowering the concentration of the NCR peptides in the periplasm and membrane (26, 27). Consequently, an S. meliloti bacA mutant is more sensitive to killing by cationic peptides such as NCR247 than the corresponding bacA⁺ strain (13). This mechanism of antimicrobial peptide resistance is conceptually similar to that used by various pathogens. For example, Haemophilus and Salmonella use the SapABCDF system to import antimicrobial peptides into their cytoplasm for degradation by cytoplasmic proteases such as Lon and ClpXP (28, 29). In the case of S. meliloti, the cytoplasmic protease SapA can also contribute to the degradation of internalized NCR peptides such as NCR247 (30). The second way that BacA plays an essential role in symbiosis is by importing certain NCRs peptide, for example NCR247, that need to be in the cytoplasm to exert their critical symbiotic effect(s) (13).

BacA is an ortholog of the Escherichia coli SbmA (SbmA_Ec) protein (25), which has been principally studied for its role in importing structurally unrelated peptides such as microcin B17, Bac7(1-35), various antimicrobial peptides, and aminoglycosides (31). A particularly well studied peptide cargo of SbmA_Ec is the bovine polyproline peptide Bac7, which acts intracellularly to inhibit translation by binding to the peptide exit channel of the 50S ribosome (32). The Bac7(1-35) and [Bac7(1-16) N-terminal fragments of Bac7 have been widely used in studies of both SbmA and BacA. For example, in each case, a fluorescently labeled Bac7 derivative was used to provide the first experimental evidence that E. coli SbmA and S. meliloti BacA are peptide importers (33, 34). To reach its intracellular target, the Bac7 peptide employs the same “Trojan-horse” strategy employed by certain other antimicrobial peptides such as microcin B17 of using the SbmA/BacA to cross the bacterial inner membrane so it can reach its cytoplasmic target(s). Thus, in contrast to the situation with NCR peptides, sbmA and bacA mutants are more resistant to the Bac7(1-35) peptide (33).

Despite E. coli not being a plant symbiont, SbmA_Ec proved to be isofunctional with S. meliloti BacA_Sm (35), completely suppressing the symbiotic deficiencies of an S. meliloti bacA mutant, as did the BacA_Ba ortholog of the intracellular mammalian pathogen Brucella abortus (36). Paradoxically, the bacA⁺ genes from a variety of rhizobia failed to restore nitrogen fixation to an S. meliloti ΔbacA mutant (37–40), although they did allow some limited developmental progression toward the bacteroid state. DiCenzo et al. have hypothesized that this difference is due to the BacA_Sm-SbmA_Ec-BacA_Ba group being more proficient at importing highly cationic peptides than the BacAs from other rhizobia (37, 38, 41).

Recent structural studies have offered major insights into the mechanism of peptide import by both SbmA and BacA (31, 42). Cryo-EM structures of both the SbmA_Ec and BacA_Sm transporters in the outward-open conformation revealed a homodimeric structure with a novel fold termed the SbmA-Like Peptide Transporter fold (SLiPT) (31, 42). The structure consists of a core transmembrane domain (TMD), comprised of 12 TMs (six from each protomer), while two additional TM0 domains (formed of TM0a and TM0b helices) flank the TMD, forming two peripheral domains (31). Although the overall TMD structure resembles that of type IV ABC transporters, SbmA_Ec and BacA_Sm are instead energized by the proton gradient with the proton translocation pathway defined by a glutamate ladder that is formed by conserved glutamates from both protomers within the TMD (31, 42); the structures revealed a central gate that isolates the peptide-binding cavity from the glutamate ladder in the outward-open conformation. The structure of both the SbmA_Ec and BacA_Sm was determined in an outward-open conformation, which has the cavity open to the periplasm (31). Subsequently, we solved the structure of SbmA_Ec in the inward-open conformation that has the cavity facing the cytoplasm (42). Opening of the central gate (Y372 in SbmA_Ec; Y368 in BacA_Sm) and movement of the proton along the glutamate ladder changes the shape of the cavity from an hourglass in the outward-open conformation to a cone shape in the inward-open conformation.

In this work, we used genetic, biochemical, physiological, and structural analysis to gain insights into how BacA’s structure and conformational changes enable it to carry out its critical complex roles in the Rhizobium–legume symbiosis. In addition, our observations have allowed us to propose a model to explain how rhizobial BacAs can maintain their ability to import a wide variety of NCR peptides yet undergo evolutionary tuning to match the specific family of NCRs made by their legume host. This model also suggests a mechanistic explanation for the paradoxical observation that BacA orthologs from the mammalian pathogens B. abortus and E. coli are isofunctional with S. meliloti BacA, yet BacA orthologs from various other rhizobia are not.

Results

Summary of the Study.

Our study employed 54 S. meliloti bacA_Sm missense mutants (Fig. 1, Top): 35 newly constructed single cysteine bacA_Sm mutants and 19 single glycine bacA_Sm mutants that had been partially characterized in an earlier publication prior to our awareness of NCR peptides (43). Since BacA_Sm lacks cysteines despite being 420 amino acids in length, we chose to make monocysteine mutants so we could employ the Single Cysteine Accessibility Method (SCAM) to make inferences about the topology of the BacA protein in living cells under different conditions (44). The SbmA/BacA structures adopt outward and inward conformations along the transport cycle (Fig. 1, Left Middle). The 54 missense mutants have been mapped onto one subunit of the dimeric protein in the outward-facing conformation (Fig. 1, Right).

Fig. 1.

Properties and phenotypes of the 54 Gly and Cys missense bacA_Sm mutants of S. meliloti displayed at the corresponding positions of the open-outward and open-inward conformations of its homologous isofunctional E. coli ortholog, SbmA_Ec. The Top panel shows the properties and phenotypes of the BacA_Sm missense mutants with respect to: i) whether they made BacA_Sm protein detectable in Western blots protein or not, ii) their ability to establish a nitrogen-fixing symbiosis, iii) their susceptibility to killing by higher levels of the NCR247 peptide, iv) their susceptibility to killing by the Bac7(1-35) peptide, and v) the topological location of certain BacA monocysteine mutants with a wild type phenotypes inferred by SCAM methodology. Red indicates bacA mutant phenotype, yellow indicates wild-type phenotype and orange indicates an intermediate phenotype. The Middle Left panel shows the overall conformational changes undergone by BacA and SbmA homodimers during the transport cycle. The Middle Right panel shows the SbmA_Ec equivalents of the 54 BacA_Sm mutants mapped on one of the SbmA_Ec protomers shown in pink (the equivalent positions in the other protomer are shown in black). The E. coli SbmA numbering is shown with the corresponding S. meliloti BacA amino acid numbering subscripted. (A and B). 14 of the 54 bacA_Sm mutants exhibited a null (Fix⁻ NCR247⁻ Bac7⁻) or virtually null phenotype. The locations of these 14 altered amino acids in the mutant BacA_Sm proteins are shown at the corresponding locations on the outward-open (A) and inward-open SbmA_Ec structures (B). The positions of the 9 mutated amino acids that result in no protein detectable in Western blots are shown in red. The positions of the 5 mutated amino acids that do make protein are shown in cyan. (C and D). The locations of the 45 missense BacA mutants that do make protein with respect to their Fix⁺ phenotype (Fix⁺ dark green, Fix^+/− light green, Fix⁻ red, mixed blue) are shown at the corresponding locations on the outward-open (C) and inward-open (D) SbmA_Ec structures. (E and F) The location of bacA mutants that make protein but exhibit a partial-loss-of-function phenotype (“split phenotype”) are shown at the corresponding locations on the outward-open (E) and inward-open (F) SbmA_Ec structures. (G and H). The location of the 31 missense BacA_Sm Cys mutants that make protein are shown at the corresponding locations on the outward-open (G) and inward-open (H) SbmA_Ec structures. The color indicates the results of efforts to determine their in vivo topological position using the SCAM procedure: cytoplasmic (green), periplasmic (blue), anomalous (noncanonical) SCAM result (red).

We used S. meliloti ΔbacA derivatives carrying plasmids with the various bacA alleles to test: i) whether the mutants made BacA_Sm protein or not as assessed from Western blot analyses using a BacA-specific antibody, ii) their ability to establish a nitrogen-fixing symbiosis as inferred from easily measured characteristics such plant height (SI Appendix, Table S1), iii) their susceptibility to killing by higher levels of the NCR247 peptide (SI Appendix, Fig. S1), and iv) their susceptibility to killing by the Bac7(1-35) peptide (SI Appendix, Fig. S2). The data from the SCAM analyses of the single cysteine bacA mutants that made BacA_Sm protein that allowed topological information to be inferred are presented in SI Appendix, Fig. S3. The topology SCAM data of BacA in vivo are all in complete agreement with the published BacA and SbmA structures (31, 42).

As summarized in (Fig. 1 A–H), we are now able to interpret the phenotypic effects of the amino acid changes in these 54 missense BacA_Sm proteins in terms of their structure. Since the resolution of the BacA_Sm structure is limited to 6 Å, we have chosen to display the positions of the 54 BacA changes at the corresponding positions of the more highly resolved SbmA_Ec outward-open and inward-open structures (31, 42) (Fig. 1, Middle Right and SI Appendix, Fig. S4). Although SbmA_Ec and BacA_Sm are closely related (SI Appendix, Fig. S4) and isofunctional (35), most of our structural inferences are likely to be correct but we note this limitation. In principle, the amino acid changes we introduced could perturb BacA_Sm membrane insertion, folding, dimerization, binding to the peptide cargo, or the conformational changes necessary for peptide transport. Collectively, our analyses indicate that BacA_Sm and SbmA_Ec function as finely tuned import machines, whose structures can be relatively easily disrupted by single amino acid changes to cysteines or glycines in ways that either prevent a stable protein from being made or result in an intact but nonfunctional protein. Other single amino acid mutations discussed below had more subtle effects on BacA function that proved to be informative. We corroborate the phenotypic behavior of a few distinct mutants in light of our published cryo-EM structures (31, 42).

Missense Mutations That Result in a Null Phenotype and No Protein.

Fourteen of the 54 missense mutants had a null (Fix⁻ NCR247⁻ Bac7⁻) or virtually null phenotype. Of these, nine make no or virtually no protein (BacA_Sm mutant/SbmA_Ec: T59C/L55, S164C/S160, H165G/H161, F168C/F164, W182G/W178, R194G/R190, F223G/F219, R284G/R280, V366C/362) presumably because the amino acid substitution makes the mutant BacA susceptible to proteolytic degradation by preventing proper membrane insertion, proper folding, or dimerization. Consideration of the location of these amino acid changes in the outward facing and inward facing structures offers insights into why the mutant proteins may result in unstable or partially folded proteins that would be susceptible to proteolysis. For example, as shown in Fig. 2A, the T59C/L55 and F223G/F219 mutations are within the TM0 domain and phosphatidylglycerol (PG) lipid interface vicinity that are likely to interfere with the correct insertion to the membrane; TM0 does not undergo conformational changes during the transport cycle and it has been suggested to anchor the TMD to the inner membrane, therefore small changes in sequence are not tolerated (31, 42). As shown in Fig. 2B, F168/F164, located in the TM2 helix of one protomer, interacts with an aromatic rich pocket formed by Y313’/Y309’, F314’/F310’, and Y309’/Y313’of the TM5’ in the opposite protomer. This cluster of amino acids is present in both the outward- and inward-open conformations and likely helps to stabilize the dimer interface during the various conformational changes associated with transport. Mutation of the S164/S160and H165/H161 to cysteine and glycine, respectively, which are in close proximity to TM2, also interferes with stable dimer formation (Fig. 2B).

Fig. 2.

Structure-based analysis of the missense mutations. (A and B) Mutants that result in null phenotype and no protein are likely due to the destabilization of the protein. (A) The T59C/L55 and F223G/F219 mutants are close to the TM0 domain that may interfere with correct folding. (B). Mutations close to the aromatic rich pocket, Y313’/Y309’, F314’/F310’, and Y309’/Y313’, can disrupt the dimer interface. (C and D) Mutations that show null phenotype but make protein are likely interfering with conformational changes or proper folding. (C) Mutations at the kinks of TM4 and TM5 result in misfolded protein according to the SCAM data as they are important for conformational changes and structural integrity. (D) D198G/D194, which is found close to the glutamate ladder and cytoplasmic gate, is being shown to be essential for transport activity and it is likely that the proton cannot be translocated thus rendering the transporter inactive.

Missense Mutations That Result in a Null Phenotype but Production of a Stable Protein.

Of the 14 mutants with a null (Fix⁻ NCR247⁻ Bac7⁻) phenotype, 5 make normal or substantial amounts of protein (BacA_Sm mutant/SbmA_Ec: N113C/N109, D198G/D194, L269C/L265, Y320C/Y316, and L336C/V332).

Interestingly, three of these mutants (L269C_Sm/L265_Ec, Y320C_Sm/Y316_Ec, and L336C_Sm/V332_Ec) affect amino acids conserved between BacA and SbmA located in distinct positions that likely interfere with either conformational changes or correct folding (Fig. 2C). The SCAM data for both the L269C/L265 and L336C/V332 suggests that the protein is misfolded (SI Appendix, Fig. S3). L269C/L265 is found in the middle of TM4 that displays a kinked conformation. TM4 undergoes conformational changes from outward- to inward-open states and the introduction of a cysteine may be interfering with correct folding. Similarly, the L336C/V332 mutant is located at the top of TM5 where a kinked helix exists. The null phenotype in the D198G/D194 mutant is consistent with our previous work where D194A resulted in reduced to no transport of Microcin B17 and Microcin J25, respectively (31). D198G/D194 is located at the interface of the glutamate ladder and the cytoplasmic gate (Fig. 2D), and we proposed that the proton might bind to D194 during the transport cycle.

Missense Mutations That Result in the Production of a Stable Protein but Differentially Affect Symbiotic Nitrogen Fixation and Transport of the NCR247 and Bac7 Peptides.

In addition to the 5 bacA_Sm missense mutations discussed above that result in production of a stable protein and cause a null phenotype, 6 missense mutations caused “split phenotypes” that strongly differentially affect symbiotic nitrogen fixation and transport of the NCR247 and Bac7 peptides (Fig. 1). In the absence of a peptide bound SbmA or BacA structures, these mutants provide insights on peptide selectivity and recognition. These partial loss of function mutants fell into the three different classes: i) Fix⁻ NCR247⁻ Bac7⁺ (N335C/N331). This phenotype could result from a defect in importing NCR247 and possibly some other symbiotically essential NCR peptide(s), while retaining an ability to import the symbiotically irrelevant Bac7 peptide. ii) Fix⁻ NCR247⁺ Bac7⁺ (Y166C/Y162, Q332C/Q328, R389G/Q385, and the central gate residue Y372C/Y368). This phenotype could result from a defect in importing at least one symbiotically essential NCR peptide other than NCR247 while retaining an ability to import the symbiotically irrelevant Bac7 peptide. iii) A Fix⁻ NCR247⁺ Bac7⁻ (F363G/F359), which could result from a defect in importing at least one symbiotically essential NCR peptide(s) other than NCR247 and a defect in importing the symbiotically irrelevant Bac7 peptide. An additional 7 missense mutations resulted in milder split phenotypes (Q193G/Q189, T199C/T195, V206C/L202, G241C/G237, G338C/G334, G349C/G345, and Q365C/Q361).

The structural locations of the 13 missense mutations that caused strong and mild split phenotypes proved to be informative. For 5 of these (Q332C, N335C, F363G, Q365C, and Y372C), the residues that are altered are amino acids that line the cavity in its outward-open conformation. Y372/Y368, the central gate, is at the bottom of the cavity (Fig. 3 A–C). In addition, the Fix⁻ NCR247⁻ Bac7⁻ null missense mutant N113C also affects a residue lining the cavity (Fig. 3 A–C). Taken together, these observations demonstrate that even a simple missense mutation affecting the peptide binding cavity in the open-outward conformation can affect BacA’s ability to discriminate between different substrates.

Fig. 3.

Amino acids lining the BacA cavity in the outward-open conformation that affect the bacA phenotype. (A–C) Outward-open conformation: Front, Periplasmic, and Expanded Periplasmic views. (D–F) Inward-open conformation: Front, Cytoplasmic, and Expanded Cytoplasmic views. BacA number and mutation followed by corresponding SbmA number. Residues whose mutation results in a split phenotype: Fix⁻ NCR247⁺ Bac7⁺ (Q332C_Sm/Q328_Ec and Q365C_Sm/Q361_Ec teal; Y372C_Sm/Y368_Ec central gate red); Fix⁻ NCR247⁻ Bac7⁺ (N335C_Sm/N331_Ec yellow); Fix⁻ NCR247⁺ Bac7⁻ (F363G_Sm/F35_Ec forest green). Mutation of another residue lining the cavity, N113C_Sm/(N109_Ec (orange), results in a null phenotype. In addition, the G241C_Sm/G23_Ec mutation (in the loop between TM3 and TM4) and the G349C_Sm/G345_Ec mutation (in the loop between TM5 and TM6) (magenta) result in mild split phenotypes. Residues M356_Sm/M352_Ec, N361_Sm/N357_Ec, and R367_Sm/R363_Ec (lime) differ between the BacA_Sm, SbmA_Ec, BacA_Ba group and the Rhizobium BacA group. (G) A comparison of portions of the amino acid sequence logos of the two BacA subsets developed by diCenzo et al. (37) showing that three amino acids (M356/M352, N361/N357, R367/R363 that line the outward-open peptide-binding cavity are not conserved between the BacA_Sm, SbmA_Ec, BacA_Ba group and the Rhizobium BacA group. (H–K) Amino acids located in the cytoplasmic part of BacA whose mutation results in a strong split phenotype (Y166C_Sm/Y162_Ec and R389G_Sm C/385_Ec magenta) or a mild split phenotype (Q193_SmG/189G_Ec,T199C_Sm /T195_Ec, and V206C_Sm /L202_Ec pink). (H and I) Outward-open conformation: Front, Cytoplasmic views. J-K. Inward-open conformation: Front, Cytoplasmic views.

Independent evidence that subtle amino acid changes in the outward-facing peptide-binding cavity play an important role in fine-tuning the specificity of a rhizobial BacA protein for the particular set of NCR peptides present in its legume host was obtained from a comparison of sequence differences between the BacA_Sm/SbmA_Ec/BacA_Ba group (restore nitrogen fixation to an S. meliloti ΔbacA mutant) and the Rhizobium BacA group (do not restore nitrogen fixation to an S. meliloti ΔbacA mutant) (37).Three residues that stand out as being different between the two classes, (BacA_Sm /SbmA_Ec M356/M352, N361/N357, and R367/R36,) (Fig. 3G) also line the cavity in the outward-open conformation. In addition, N335, whose mutation results in a split phenotype, often differs between the two groups. Fig. 3 A–F shows the location of all these residues in the cavity in outward- and inward-open conformations. Taken together, these observations demonstrate that even a simple missense mutation affecting the peptide binding cavity in the open-outward conformation can affect BacA’s ability to discriminate between different substrates.

Despite the striking number of missense mutations causing a split phenotype affecting the cavity in the outward-open conformation, we note that amino acid alterations elsewhere in the protein can also result in split phenotypes. Y166C/Y162 (Fix⁻ NCR247⁺ Bac7⁺) and R389G/385 (Fix⁻ NCR247⁺ Bac7⁺) gave very strong split phenotypes, while Q193/Q189, T199/T195, V206/L202 resulted in mild split phenotypes. Interestingly, all these mutants are located in close proximity to the glutamate ladder, and it is unclear how they would influence the specificity of BacA (Fig. 3 H–K); upon peptide binding the cavity will adopt an occluded conformation from outward-open to inward-open conformations, and we speculate that BacA adopts slightly different occluded conformations in the presence of the different peptides that could allow it to distinguish them. The G241C (in the periplasmic loop between TM3 and TM4) and G349C (in the periplasmic loop between TM5 and TM6) mutants result in mild split phenotypes (Fig. 3 A and B). As these mutants are located away from the cavity, we speculate that these sites may be acting as the initial site for peptide binding prior to translocation into the cavity.

Discussion

Our previous structural work has provided a basis of understanding the molecular mechanism of SLiPT transporters. Our detailed genetic, biochemical, structural, and physiological analyses have now revealed that BacA_Sm and SbmAhomodimers function as finely tuned import machines, whose structures can be relatively easily disrupted by single amino acid changes that prevent a stable protein from being made or result in an intact but nonfunctional protein. Moreover, our efforts to relate the location of the missense mutations within the BacA structure of the BacA homodimer to our knowledge of BacA physiology and its evolution led us to suggest a model that offers a plausible molecular explanation for one of the most striking but perplexing results concerning BacA’s role in symbiosis—bacA orthologs from two mammalian pathogens (SbmA_Ec and BacA_Ba) can fully restore the ability of an S. meliloti ΔbacA mutant to establish a normal nitrogen-fixing symbiosis with its Medicago hosts M. truncatula and Medicago sativa (35, 36), but bacA orthologs from rhizobial strains (e.g., S. fredii NGR234, R. leguminosarum bv. viciae and sativum) that interact with different legume hosts cannot (37, 38). These rhizobial bacA orthologs do, however, allow a S. meliloti ΔbacA mutant to proceed to a somewhat more developmentally advanced stage than is observed in ΔbacA nodules although many of the cells eventually die (37, 38, 41).

Although it has been proposed that this difference is due to the BacA_Sm, SbmA_Ec, and BacA_Ba being more proficient at importing highly cationic NCR peptides (37, 38, 41) it has not been clear type of evolutionary changes in their amino acid sequence could have enabled these BacA/SbmA transport proteins to discriminate between different types of NCR peptides while retaining their abilities in import a variety of peptide cargoes. It should be noted that selecting for NCR peptide specificity is more complicated than just selecting for cationic character because multiple chemical characteristics of an individual peptide may be required for its biological effects. For example, although the 24 amino NCR247 is strongly cationic peptide with seven basic amino acids and a pI of 10.5, it also has the strongest hydropathy index and protein-interaction index of any NCR peptide (22) and its cysteines are used as axial ligands on each side of the iron when it sequesters heme (13).

Our previous phylogenetic analyses (42) indicated that SbmA/BacA likely evolved from the TMD of the ABC transporter YddA, which in some organisms is expressed from a single gene but in others from two separate genes encoding TMD and ABC subunits. The proposed evolutionary mechanism is that a gene encoding only the TMD was horizontally transferred to a new organism, where it subsequently evolved a glutamate ladder and other structural adaptations that allowed it to be powered by the membrane potential instead of ATP hydrolysis. On the basis of their analyses, Smith et al. (41) suggest that the SbmA/BacA family originated in a bacterium in the order Rhizobiales over 500 Mya prior to spreading to other Proteobacteria by horizontal gene transfer. Since legumes did not evolve until about 60 Mya (45), the initial spread of SbmA/BacA family members may have been driven in part by their abilities to protect against membrane-damaging antimicrobial peptides, a trait beneficial to animal and plant pathogens, and that they were only subsequently co-opted to support legume symbiosis in rhizobia. Since NCR peptides appear to have rapidly evolved and diversified (46, 47), there has been a corresponding rapid evolution in the substrate specificity of the BacA proteins (37).

Our analyses suggest that one important evolutionary mechanism that has allowed BacA to retain its relatively promiscuous import ability to import a variety of NCR peptides yet still finely adapt to the cocktail of NCR peptides made by its particular legume hosts (4), is the acquisition of subtle amino acid changes in outward-facing peptide binding cavity of BacA and in the internal cavity of the occluded conformation that the peptide experiences during import (42). The apparent requirement for S. meliloti’s BacA to be especially adept at importing the highly cationic NCR peptides in its Medicago hosts (38) can explain the incomplete complementation following expression of BacAs in rhizobia whose legume hosts have a lower proportion of highly cationic NCR peptides (38). Since NCR peptides are expressed in consecutive waves during infection of wild-type nodules and can be separated into early and late stage NCR peptides (4). It is possible that the ability of these rhizobial bacA orthologs to nevertheless allow a S. meliloti ΔbacA mutant to proceed to a somewhat more developmentally advanced stage than is observed in ΔbacA nodules could be due them being able to import various NCR peptides expressed early in nodule development, but not certain cationic NCR peptides such as NCR247 that play a role during later stages of nodule development such as nitrogen-fixation.

Our model can also explain the underlying molecular mechanisms responsible for S. meliloti BacA having undergone rapid, convergent evolution to the BacA/SbmA proteins of the pathogenic genera Brucella, Escherichia, and Klebsiella (37). It seems plausible that this evolutionary convergence could have been driven by the exposure of the BacA/SbmA orthologs in these pathogens being similarly exposed to highly cationic peptides in their mammalian hosts.

Materials and Methods

BacA protein production, BacA antibody generation, the protocols for the Substituted Cysteine Accessibility Method (SCAM), and Protein Analysis of BacA are described in SI Appendix, Materials and Methods.

Bacterial Strains.

The bacterial strains used in this study are described in SI Appendix, Table S2 and their growth conditions in SI Appendix, Materials and Methods.

BacA Cloning and Site Directed Mutagenesis.

To create defined site directed mutations (SDM), the S. meliloti bacA gene was cloned into the EcoRI and BamHI sites of pUC19. On this construct, Site directed mutagenesis of this cloned gene was performed using a QuikChange II site directed mutagenesis kit (Agilent), according to the manufacturer’s instructions. Primers used for SDM creation of the defined SDMs are shown in SI Appendix, Table S3.

BacA Complementation Constructs.

All the bacA complementation constructs that were generated for this study are shown in SI Appendix, Table S4. Their construction is described in SI Appendix, Materials and Methods.

Assays to Determine Phenotype of bacA Mutants.

M. sativa plant symbiosis experiments were conducted exactly as previously described (48).

NCR247 sensitivity survival assays were performed in MOPS buffer supplemented with casamino acids as described previously using the defined concentrations of NCR247 (21). In some cases, the incubation time was extended to 10 h to observe better separation between wild type/complemented and ΔbacA mutant bacteria. Bac7(1-35) sensitivity survival assays were performed as described previously (34).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Study data are included in the article and/or SI Appendix.