The Emergence of the Slc11 Clade MCb_gut: A Parsimonious Hypothesis for the Dawn of Lactobacillales in the Gut of Early Vertebrates

Abstract

The Lactobacillales (LB) stand apart among bacterial orders, using manganese (Mn) instead of iron to support their growth and swiftly ferment complex foods while acidifying their environment. The present work investigates whether a shift in the use of Mn could mark the origin of LB. Transmembrane carriers of the ubiquitous Slc11 family play key roles in LB physiology by catalyzing proton-dependent Mn import. In prior studies, the Slc11 clade found in LB (MntH Cb, MCb) showed both remarkable structural plasticity and highly efficient Mn uptake, and another Slc11 clade, MCg1, demonstrated divergent evolution coinciding with emergence of bacterial genera (e.g., Bordetella, Achromobacter). Herein, the Slc11 clade MCb is subdivided in sister groups: MCb_ie and MCb_gut. MCb_ie derives directly from the Slc11 clade MCa, pointing an intermediate stage in the evolution of MCb_gut. MCb_ie predominates in marine Bacillaceae, is more conserved than MCb_gut, lacks the structural plasticity that typify MCb_gut carriers, and responds differently to identical mutagenesis. Exchanging MCb_ie/MCb_gut amino acid residues at sites that distinguish these clades showed conformation-dependent effects with both MCb_ie and MCb_gut templates, and the 3D location of the targeted sites in the carrier structure together suggests that the mechanism to open the inner gate, and release Mn into the cytoplasm, differs between MCb_ie and MCb_gut. Building on the established phylogeny for Enterococcus revealed that a pair of genes encoding MCb_gut was present in the common ancestor of LB, as MCb_gu1 and MCb_gu2 templates exhibited distinct structural dynamics properties. These data are discussed when examining whether MCb_gut⁺ LB could emerge in the upper gut of early vertebrates (ca. 540 mya), through genome contraction and evolution toward Mn-centrism, as they specialized as gastric aids favoring stomach establishment in jawed vertebrates through bi-directional communication with host nervous, endocrine and immune systems.

1. Introduction

The Solute carrier 11 (Slc11) family of proton (H⁺)-dependent importers of Mn²⁺ and/or Fe²⁺ belongs to the APC superfamily and adopts the tridimensional (3D) LeuT fold [1,2,3]. This fold consists of tandem repeats of five transmembrane helices that display inverted transmembrane topology. The repeats fold intertwined and coexist in alternate 3D configurations that are exchanged as the carrier molecule switches between two main states, either outward open (OO) or inward open (IO). The absence of sequence similarity between the two halves of the Slc11 carrier hydrophobic core implies asymmetric evolution, which could favor directional substrate transport, i.e., H⁺-dependent import [4,5].

Molecular evolutionary genetic and structural analyses revealed that Slc11 synapomorphy functionally distinguishes this protein family from the rest of the APC superfamily [6]. Slc11 synapomorphy comprises 11 (quasi) invariant amino acid (aa) residues forming a 3D network that articulates Slc11 substrate-selective carrier conformational exchange between the OO state and the IO state. The functional maintenance of this 3D network amid diverging Slc11 clades (e.g., bacterial MCb and MCg1) testifies to the resilience and evolvability of the molecular mechanism embedded through this synapomorphy in Slc11 architecture.

Moreover, the Slc11 mechanism of carrier coupling to the transmembrane protonmotive force (pmf), to catalyze cytoplasmic import of divalent metals such Mn and Fe (Me²⁺), demonstrated stepwise evolution in the form of serial synapomorphies that distinguished groups of Slc11 homologs found either in prokaryotes (the proton-dependent Mn transporters, MntH) or in eukaryotes (the Natural resistance-associated macrophage proteins, Nramp, Table S1).

MntH synapomorphies demonstrate successive sets of coevolutionary rate-shifts that progressively edified an extended ‘H⁺-network’ interacting with the pmf, from anaerobic bacteria (MntH B, before aerobiosis) to aerobic bacteria (MntH A) and to hyperthermophile TACK and Asgard Archaea (MntH H). This latter evolutionary intermediate likely shared a common ancestor with the first eukaryotic representative of the Slc11 family (Nramp precursor, [7]).

The Slc11 H⁺-network further evolved in eukaryotes, during the transition from the first to last eukaryotic common ancestor (FECA to LECA), yielding two types of Nramp: prototype Nramp, mainly found in primitive eukaryotes, and archetype Nramp, predominant in multicellular organisms of both animal and vegetal realms (Figure 1A). Coupled with phagocytosis, archetype Nramp can restrict bacterial survival by depleting the acidified phagosomal milieu of Me²⁺, such as Mn and Fe [8]. Arguably, the evolution of the Slc11 mechanism of carrier coupling to the pmf resulted in functional antagonism, yielding eukaryotic Nramp carriers that import Me²⁺ more efficiently than their bacterial counterparts (MntH).

However, both types of prototype Nramp genes (pN-I and pN-II, Table S1) were independently transferred horizontally towards bacteria. Therefore, bacterial MntH are polyphyletic [9], either of prokaryotic origin (MntH B, MntH A, MntH H) or of eukaryotic ancestry (MntH C). Alphafold (AF) [10]/Colabfold (CF) [11] 3D modeling of native MntH carriers of eukaryotic origin, from the sister clades MntH Cb (MCb) and MntH Cg1 (MCg1) Figure 1), coupled with the modeling of phylogeny-informed site-directed mutants, demonstrated paths of divergent evolution which affected both the carrier shape and functional plasticity [6].

A hallmark of MCb carriers is their conformational flexibility, and protein sequences from distinct clades (>80% sequence identity) yield 3D models representing discrete conformers. Regarding MCg, a mutation altering both the H⁺-network in transmembrane helix 4 (h4) and carrier shape contributed to found MCg1 clade, and further divergent epistasis within this clade yielded an apparently functional neovariant, whose emergence coincided with novel bacterial genera (Bordetella and Achromobacter). This unexpected result suggested that, by analogy, the emergence of the MCb carrier might correspond as well with the rise of other bacterial genera.

Indeed, MCb is the only Slc11 type found in the order Lactobacillales (LB), and the majority of known MCbs are encoded by LB genomes. In contrast, the close relatives Bacillales may harbor either mntH A, mntH Ca or mntH Cb genes [7]. The extraordinary flexibility and functional plasticity of MntH Cb 3D models may relate to high carrier efficiency, able to support LB’s mntH-dependent competitive exclusion of other organisms [12,13].

It is perhaps no coincidence that LB stand out as a bacterial order that relinquished heme-based metabolism to become Mn-centric microorganisms. As Mn excess can jeopardize iron-based metabolic functions, high efficiency Mn import seems particularly suited to cells notably lacking iron-cofactored enzymes for respiration, antioxidant defenses and the tricarboxylic acid cycle [14,15] and which rely on Mn instead of Fe for growth [12,16,17]. Based on the available data, it can therefore be hypothesized that the emergence of the MCb clade is concomitant to the dawn of LB.

The consensus on LB origin indicates they derived from a Bacillale ancestor through ~1/3 genome reduction and by adapting to some microaerated/anaerobic nutrient-rich environmental niche [14,15]. Given that LB are both notoriously proficient at predigesting (fermenting) complex food and common residents of animal gut [18,19], it seems plausible that LB could originate there. LB are demonstrated residents of the gut of various vertebrates (fish, reptiles, birds, and mammals) and a landmark study proposing the genus Enteroccoccus arose more than 425 million years ago (mya), as animals colonized land [20], suggests the LB common ancestor could have evolved earlier, during the Ediacaran–Cambrian transition for instance.

To test the hypothesis that the emergence of the (Slc11) MCb clade may coincide with LB origin, a sequence of possible evolutionary events linking MCa and MCb clades was established and their functional implications tested in silico using AF/CF 3D modeling. Several questions were addressed: Is there a group of MntH Cb seqs that display common properties identifying them as possible intermediates (hereby named MCb_ie for MCb environmental intermediate) between MntH Ca and the rest of the MCb clade (hereafter named MCb_gut)? Do MCb_ie⁺ microorganisms form a defined Bacillale taxon and/or share a common habitat? How do MCb_ie and MCb_gut in silico structural dynamics compare? Do they support a distinct mechanism of Mn import in LB? Does MCb_gut emergence predate the origin of Enterococci and what was the mntH Cb_gut gene complement of the common ancestor of LB?

The results presented define the Slc11 clades MCb_ie and MCb_gut which together form the phylogroup MCb. MCb_ie predominates in marine Bacillaceae and is closely related to both its ancestor clade, MCa, and the common precursor of MCb_gut. Regarding MCb_gut, it is in fact a pair of genes, encoding MCb_gu1 and MCb_gu2, that typifies LB. The three MCb clades MCb_ie, MCb_gu1 and MCb_gu2 vary in their evolutionary conservation and conformational properties as the opening of the carrier inner gate apparently evolved in the transition from MCb_ie to MCb_gut. These data are discussed, relating the emergence of MCb_gut⁺ LB and the development of the gastro-intestinal tract of early vertebrates.

2. In Silico Results

2.1. P. halotolerans A0A1I3CNB9 Identifies the Clade MCb_ie, Intermediate between MCa and MCb_gut

Prior work suggested that reevaluating the phylogenetic structure of the MCb clade may delineate a putative precursor lineage (MCb_ie) from LB’s MCbs (MCb_gut). NCBI sequence wp_092092601.1 from the Carnobacteriaceae Pisciglobus halotolerans (Uniprot ID: A0A1I3CNB9), representing a cluster of sequences (seqs > 70% aa id), identified the most basal branching node of MCb and demonstrated relatively little specific divergence compared to the rest of the MCb clade [7]. Detailing the phylogenetic relationships of the A0A1I3CNB9 clade, i.e., MCb_ie, amid eukaryotic prototype Nramps and other bacterial MntH C clades (Figure 1B) confirmed both its basal position relative to the rest of MCb homologs, i.e., the MCb_gut clade, and the relative sequence conservation of its members.

MCb_ie conservation contrasts with the diversity of the MCa clade, which comprises both the closest known relatives of prototype Nramp-I (pN-I) and candidate precursors of both MCb and MCg clades [7]. The phylogenetic position of MCb_ie identifies an evolutionary intermediate between MCa and MCb_gut (Figure 1 and Figure S1). MCb_ie clade topology shows two subsets (Figure S1A): a ‘crown’ formed by conserved sequences from core marine Bacillales (Bacillaceae, Paenibacillaceae) plus more divergent subsets from other (Lacto)bacillales spp. that exhibit relatively reduced genomes.

MCb_ie is most prevalent in Bacillaceae (45 seqs, 95% id cutoff, Figure S1A) and MCb_ie⁺ genera represent about 10% of the Bacillaceae family (Figure S1B). MCb_ie phylogeny segregates several established genera of Bacillaceae, a pattern seemingly compatible with the vertical inheritance of mntH Cb_ie. Yet, both Bacillaceae and Planococcaceae encode predominantly the more ancestral MntH A (MA) protein, which contrasts with Paenibacillaceae wherein MntH Ca (MCa) predominates (205 seqs, 95% id cutoff). Neither Listeriaceae [21,22,23] nor Carnobacteriaceae (LB) harbor any MA coding gene, and contrarily to Paenibacillaceae, MCb (MCb_ie or MCb_gut) is enriched in both of these families. In comparison, Lactobacillaceae possess exclusively MCb_gut (MCb_gu1 and/or MCb_gu2, detailed in Section 2.6).

The topology of the MCb_ie ‘crown’ group shows clusters of seqs from Paenibacillaceae that diverge from clusters of seqs from Bacillaceae (Figure S1A). Moreover, only spp. of Paenibacillaceae (e.g., Paenibacillus: NZ_CP019717, NZ_CP020327 and Brevibacillus: CP048799, CP139435) were found harboring both genes encoding MCa and MCb_ie, including some Paenibacillus isolates wherein MCb_ie appears inactivated (NZ_CP019794, NZ_CP020557). Together with the large genome size of MCb_ie⁺ Paenibacillaceae spp., the data support that MCb_ie could originate in Bacillaceae and subsequently spread among Paenibacillaceae spp. Similarly, mntH Cb_ie gene transfers toward spp. showing genome erosion (e.g., Listeraceae and Carnobacteriaceae) may account for the relative divergence of their encoded MCb_ie proteins.

Phylogenomic analyses [24,25] support this interpretation, showing marine MCb_ie⁺ Bacillaceae, e.g., Heyndrickxia and Lederbergia genera, as more likely candidate precursors of LB than the MA⁺ Bacillus spp. (B. subtilis and B. anthracis, Figure 2 and Figure S2). Staphylococcaceae and Listeriaceae are known as highly derived Bacillales families of Fe-dependent bacteria [26,27] and both lack MA encoding genes.

Altogether, the data presented clarify the emergence of the Slc11 clade MCb from its precursor MCa (itself derived from eukaryotic pN-I, Figure 1A). An intermediate step in this process produced the sister clades MCb_ie and MCb_gut, respectively, predominant in marine Bacillaceae, such as Heyndrickxia and Lederbergia, which are aerobic, and in anaerobic LB. These results provide molecular insight into the hypothetical evolutionary transition relating MA⁺ Bacillales to MCb_gut⁺ LB through a marine MCb_ie⁺ Bacillaceae intermediate.

2.2. AF/CF 3D Modeling of Native MCb_ie and MCb_gut Carriers Differs Significantly

The multiple alignment of diverse sequences from the three groups MCa, MCb_ie and MCb_gut demonstrate 29 membranous sites that coevolved in the transition between MCa and MCb, i.e., coevolved mutations fixed in the common ancestor of MCb_ie and MCb_gut (Figure 3A,B). Also, sequence variations in extra membranous loops of the carrier molecule show several instances of selective conservation between MCa and MCb_ie vs. MCb_gut (e.g., l 4/5, l6/7, l7/8, l8/9 and l9/10, Figure S3), indicating that the common precursor of the MCb clade evolved directly from MCa.

A 3D display of these 29 coevolved sites between MCa and MCb, on superposed MCb_gut models representing both the OO and IO states (cracked-open representation segregating half segments of helix 5 (h5), Figure 3C), shows accumulation along h5. The tilt of h5 modeled during MCb_gut carrier conformation switch was shown to mobilize separate communities of networked residues, involving one or the other half of h5, each contributing to open the inner gate of the carrier toward the cell interior [6]. The 29 (MCa/MCb) rate-shifted sites segregate in distinct areas involving either h5C, h7C, h1b, h2N, h6a and h11C (right panel) or h5N, h4C and h8N plus additional sites scattered across the ‘hash module’, which hosts a large part of the Slc11 H⁺-network (left panel).

These data indicate that evolution of the precursor of the MCb clade directly from MCa involved a set of coevolved sites. These evolutionary coupled mutations, common to both MCb_ie and MCb_gut, may form a clade-specific (MCb) synapomorphy. Their accumulation in local areas of the 3D structure that involve one half or the other of h5, whose tilting during MCb_gut carrier cycle connects outer gate locking and inner gate opening [6], suggests these processes evolved in the transition from the MCa to MCb common ancestor.

The 3D modeling properties of MCb_gut templates differ from other MntH C clades, apparently due to their intrinsic carrier plasticity and the presence of flexible loops [6,7]. Based on the candidate MCb synapomorphy (Figure 3B,C), one could expect that AF2/CF models of native MCb_ie carriers would resemble known MCb_gut structures either modeled or solved. However, the 3D modeling of MCb_ie templates differed significantly (Figure 4).

The 3D correspondence of available Uniprot/AF2 MCb_ie models showed very similar predicted structures that did not clearly segregate with either the OO state or the IO state of solved structures of MCb_gut homologs (Figure 4A). This result contrasts with the array of functional conformers that AF2 predicted using MCb_gut sequences (Figure 4A, inset).

Examining the models obtained for four representative MCb_ie templates (~70% pairwise aa id) shows they do not represent clearly distinct OO or IO candidate conformers (Figure 4B), and their sequence relationships (cf Figure 1B, i.e., between WP_256212980.1 (NCBI proxy for Uniprot A0A143Y4E1), WP_092092601.1 (A0A1I3CNB9), WP_092652552.1 (A0A1H9T065) and WP_277619782.1 (proxy for A0A3D8TQ78)) suggests MCb_ie phylogeny does not influence AF2 or CFnt predictions. The two sequences that produced similar CFnt models, apparently closer to the MCb_gut IO state, were selected for further analysis (A0A143Y4E1 and A0A1H9T065, both from Carnobacteriaceae spp., Trichococcus palustris and Isobaculum melis, respectively).

The topology of the MCb clade (MCb_ie and MCb_gut) resembles the previously characterized MCg1 clade ([6] and Figure 1B): both subdivide in sister groups that differ in their relative extent of divergence. For the MCg1 clade, this situation was correlated with (i) founding mutations that introduced intrinsic structural constraints and (ii) reversion to a less constrained model structure as a result of extensive epistatic divergence.

Regarding MCb_gut, divergence from the MCb common ancestor produced remarkably flexible and intrinsically dynamic structures, whose predicted conformations reflected phylogenetic diversity (Figure 4A, inset). In contrast, the 3D models obtained using relatively diverse MCb_ie seem structurally constrained (Figure 4B), suggesting that MCb_gut could evolve out of epistatic divergence from (its common ancestor with) MCb_ie.

2.3. Evolution of MCb Carrier Cycling between MCb_ie and MCb_gut

To examine this, the possible functional evolution between MCb_ie and MCb_gut was investigated by comparing the AF/CF 3D modeling of native and mutant templates targeted at sites that underwent coevolutionary rate-shifts in the transition from MCb_ie to MCb_gut (putative MCb_gut synapomorphy). Meaning, 15 or 16 mutations that distinguish MCb_gut from MCb_ie were cointroduced into the sequences A0A143Y4E1 and A0A1H9T065, respectively (71.5% aa id, Figure 4C and Figure 5A and Table S2). This approach aimed at testing the impact on the MCb_ie predicted structure of 17 comutations that typify instead the MCb_gut clade.

As a result, the CFnt modeling of the compound mutant A0A143Y4E1_15mt (Table S2) predicted a structure closer to the IO state of MCb_gut (Figure 4C). This suggested MCb_gut synapomorphy may signify functional determinants, as the corresponding residues map to 3D areas either partially overlapping with MCb synapomorphy (Figure 5A and Figure 3B,C, respectively) or specific to MCb_gut (e.g., l 7/8). In addition, while the native MCb_ie A0A143Y4E1 CFnt model cannot fully superimpose onto the MCb_gut IO CFpdb model (Q5HQ64_WFLF, [6]), the predicted structure for the compound mutant A0A143Y4E1_15mt does so almost perfectly (Figure 5B).

These data suggest that a select portion of MCb_gut carrier, which contributes to inner gate opening [6], was rearranged compared to MCb_ie. Per residue root mean square deviation (pRMSD) from the MCb_gut IO CFpdb model (Q5HQ64_WFLF) shows the segments rearranged in the MCb_ie compound mutant include l1/2 & h2, h4-h6, l 7/8 & h8, and l 9/10 & h10 (Figure 5C). These structural rearrangements extend beyond the areas targeted for mutagenesis (Figure 5A), implying the targeted sites may directly mediate MCb carrier conformation change.

Investigating this further, the reciprocal mutational approach was applied to MCb_gut homologs from distinct clades (>80% aa id) that produce discrete native conformers, either OO or IO or in between ([6] and Figure 6A), to determine how MCb_ie-specifying residues affect the MCb_gut predicted structure.

This yielded two types of results: (i) for templates whose predicted native conformers adopt the IO state, the 17 mutations introduced resulted in the prediction of OO conformers, a result similar to that obtained by targeting instead five sites of Slc11 synapomorphy ([6] and Figure 6A,B); (ii) considering templates whose predicted native conformers adopt the OO state or in-between state, identical compound mutagenesis had comparatively little or no effect.

The conformational effects of the reciprocal mutageneses performed were both asymmetric and specific, favoring an MCb_gut-like IO state for the MCb_ie A0A143Y4E1_15mt mutant while triggering MCb_gut conformation switch from the IO state to the OO state but showing marginal effects on alternate MCb_gut carrier conformational states (OO and in between). Together these results demonstrate the functional impact of exchanging a set of coevolved residues that distinguish MCb_gut from MCb_ie, showing a conformation-dependent effect on MCb_gut carrier 3D modeling.

2.4. Mutagenesis of Slc11 Synapomorphy Supports Functional Divergence of MCb_gut and MCb_ie

A comparative analysis of CF modeling sensitivity to mutagenesis targeting Slc11 synapomorphy has proven useful to decipher the functional evolution between MCb and MCg1 clades and within the MCg1 clade [6]. The introduction of the canonical VNT mutation (h6 VY, h3 YN, and h6 T) into templates from both clades resulted in the conversion of native IO conformers into OO mutant conformers. However, the respective sensitivity to either h3 YN and/or h6 VY ⁺/₋ h6 T mutations varied between clades.

Applying the VNT mutation to MCb_ie A0A143Y4E1 produced the expected CFnt modeling result, which predicted an OO conformer like the reference MCb_gut Q5HQ64-VNT [6], indicated by a comparable broad scale rearrangement (Figure 7A) and quasi-superimposable structures (Figure 7B). In contrast, none of native A0A143Y4E1 predicted structures (using CFnt or AF2) nor the CFnt model for A0A143Y4E1_15mt fitted alike the MCb_gut OO model conformer. Hence, mutating five sites of Slc11 synapomorphy induced a conformational response of MCb_ie A0A143Y4E1 like those observed before with the MCb_gut and MCg1 clades.

However, detailing the effect of h3 YN and/or h6 VY ⁺/₋ h6 T mutations showed that, unlike MCb_gut, MCb_ie A0A143Y4E1 predicted conformation was fully switched in response to the h3 YN mutation only (Figure 8A,B), as previously observed using MCg1 A0A149PND7 [6]. Yet, contrary to MCg1 A0A149PND7, additional mutations targeting potential suppressor sites of a YN-induced conformation switch had no impact on the modeling of A0A143Y4E1 YN (Figure 8C).

In addition, A0A143Y4E1_15mt modeling was insensitive to h3 YN mutagenesis and neither the A0A143Y4E1 YN, A0A143Y4E1_15mt YN nor A0A143Y4E1_15mt model were affected by adding two other h3 mutations (Figure 8D). Furthermore, reversing three coevolutionary rate-shifts that distinguish MCb_gut in l 7/8 (DPA sites #582, #589, #596, Figure 5A) restored a degree of conformational response to h3 YN mutagenesis, observed with the A0A143Y4E1_15mt DPAYN model, akin to the A0A143Y4E1 YN modeling response (Figure 8E).

These results indicate MCb_ie modeling operated the conformation switch toward the OO state in response to the Slc11 synapomorphy VNT mutation, as expected from prior modeling studies of MCb_gut and MCg1. But MCb_ie A0A143Y4E1 modeling displays unique sensitivity to h3 YN mutation, which is lost after introducing 15 evolutionary coupled mutations that specify the clade MCb_gut. These data support divergent structure/function relationships between MCb_gut and MCb_ie and suggest a conformational role for intervening loops that are distal from the Mn²⁺ binding site.

2.5. Conformation-Dependent Impact of Mutations Targeting MCb l 7/8 and l 8/9

Reversing MCb_ie/MCb_gut rate-shifts in l 7/8 (DPA) was guided by observing that they map to a specific area of the MCb_gut structure (Figure 5A), i.e., not part of the evolutionary transition from MCa to MCb (Figure 3B,C) and polymorphic (Figure S3). And preserving this area as is in MCb_ie retained some conformational flexibility in the MCb_ie mutant A0A143Y4E1_15mt DPAYN (Figure 8E). Moreover, the functional importance of these DPA residues for MCb_ie conformational response to h3 YN mutagenesis is underscored by their conservation in the MCa clade (Figure S3, sites #582, #589, #596).

Along similar lines, mutation at site #632 (Figure 5A) in l 8/9 (Figure S3) was reevaluated. To mimic more precisely the MCb_ie/MCb_gut residue exchange vs. the MCb_ie/MCa exchange, the two sites #632 and #633 (cf Figure S3) were initially mutated in A0A143Y4E1_15mt (counted as one mutation). To verify whether this dual mutation also introduced a 3D constraint, it was reverted as well, either alone or combined to l 7/8 DPA reversion mutation (Figure 9A). In both instances, restoring MCb_ie native sequence in l 8/9 recovered some conformational flexibility. The modeling of the amended mutant A0A143Y4E1_15mt DPA-KP responded to YN-induced conformation exchange like native A0A143Y4E1 (Figure 9C). However, in the absence of the YN mutation, the amended mutant A0A143Y4E1_15mt DPA-KP deviated locally from the reference MCb_gut Q5HQ64 IO conformer (Figure 9B) and did not appear superimposable anymore (Figure 9C,D).

In sum, introducing 15 coevolutionary rate-shifts that distinguish MCb_gut from MCb_ie into MCb_ie recipient A0A143Y4E1 allows for the modeling of an IO conformer that is superimposable onto the MCb_gut reference IO conformer (A0A143Y4E1_15mt). Conversely, the reciprocal mutagenesis of MCb_gut members demonstrated native conformation-dependent effects, underscoring the functional significance of the targeted sites (cf Section 2.3).

However, A0A143Y4E1_15mt lost 3D flexibility and responsiveness to the YN-induced conformation exchange from the IO state to the OO state. This impact was alleviated by reverting five of the mutations that targeted extramembranous loops distal from the substrate binding site (l 7/8 & l 8/9). Yet, these same reverted sites abrogated the 3D resemblance observed between the MCb_ie compound mutant and reference MCb_gut structure in the IO state. These five sites therefore demonstrate a conformation-dependent effect in the MCb_ie template A0A143Y4E1, and similar results were obtained using another MCb_ie template, A0A285P4W2 (from the Bacillaceae Terribacillus aidingensis, Figure S4).

To conclude, reciprocal mutations of coevolutionary rate-shifts that took place as MCb_ie and MCb_gut diverged exert distinct conformation-dependent effects on the 3D modeling of members of each clade. This implies the targeted sites contribute to the mechanism of MCb carrier cycling, but their respective role may differ within the MCb_ie or MCb_gut model structure. Together, these data relate MCb_gut-specific functional divergence to the IO state of the carrier, which seems consistent with a possible shift in Mn import capacity in LB.

2.6. MCb_gut Phylogeny Reveals a Pair of Distinct Carriers in the Common Ancestor of LB

To approximate a possible period for this functional transition, from marine MCb_ie⁺ Bacillaceae toward MCb_gut⁺ LB, MCb_gut phylogeny was further explored, first, among spp. of the Enterococcaceae family because its origin was dated (>425 million years ago [20]) and its structure comprises four clades, one of which diverged basally [30]. MCb_gut phylogeny is complex due to pervasive horizontal gene transfers [31] and multiple copies per genome (e.g., six in the Levilactobacillus suantsaii strain CBA3634 chromosome, NZ_CP059603.1). This may explain why the phylogenies of enterococcacal MCb_gut and Enterococci spp. vary (Figure S5). But the general topology of the tree and the profound dichotomy of the Enterococcus clade A vs. clades B-D were recovered, implying a significant phylogenetic signal encoded by MCb_gut. The separate grouping of MCb_gut seqs resulting from gene transfer to bacteria of different phyla, either enterobacterial or actinobacterial spp., indicates distinct events. The position of the Enterococcaceae P. termitis seq suggested it may demarcate two types of MCb_gut, meaning that perhaps the common ancestor of Enterococcaceae possessed a pair of distinct mntH Cb_gut genes.

To address this possibility, MCb_gut phylogeny was determined across the LB order using a set of representative seqs from each LB family (Figure 10). The results show the presence of two clusters of MCb_gut, labeled MCb_gu1 and MCb_gu2, found both in Enterococcaceae and Lactobacillaceae. In addition, the data suggest two other clades could diverge independently from MCb_gu2.

The present tree thus supports the possible existence of two MCb_gut copies in the common ancestor of LB. Several spp. of Lactobacillaceae harbor both genes encoding MCb_gu1 and MCb_gu2 on their chromosome (e.g., NZ_LT854705.1, NZ_CP059603.1), implying perhaps some LB retained both genes since their origin.

The presence of both MCb_gu1 and MCb_gu2 in the common ancestor of LB suggests they are not functionally equivalent. To evaluate this proposal, three individual sequences from each type were subjected to AF/CF modeling using either native sequences or compound mutants. Native AF2 models showed that, both in Lactobacillaceae and Enterococcaceae, MCb_gu1 and MCb_gu2 yielded distinct conformers in the OO state (Figure 11A and Figure 11B, respectively). The MCb_gu2 type appeared closer to the fully OO conformation and less structurally conserved among Enterococcaceae. The introduction of a set of (reciprocal) mutations corresponding to the MCb_ie/MCb_gut evolutionary transition (cf Section 2.3) confirmed it had little impact on the OO conformers of the MCb_gu1 and MCb_gu2 types. These data are in accordance with previous observations showing a strong phylogenetic component in the AF/CF modeling of MCb_gut [6].

To compare MCb_gu1 and MCb_gu2 structural dynamics a combination of mutations able to mimic the carrier forward transition, i.e., exchanging the OO state modeled by CF pdb using the native sequence for the IO state of a mutated carrier, was necessary. Along sites previously identified as conformationally active, either as part of Slc11 synapomorphy or contributing to inner gate opening, additional sites were targeted in helices 3, 4, 8 and 9, which together constitute the Slc11 carrier ‘hash module’ known to interact with the pmf [2,3]. All targeted sites are evolutionary conserved so that identical mutations were introduced among both the MCb_gu1 and MCb_gu2 templates (Table S2).

Introducing a series of 15 phylogenetically aware mutations into the Paucilactobacillus wasatchensis template PLB_gu1 enabled us to model an IO conformer like the MCb_gut Q5HQ64 reference CF model (Figure 12A,C). Applying this combination of mutations to other MCb_gu1 had a similar yet less pronounced effect, producing models more advanced toward, albeit not switched to, the IO state (Figure 12E and Figure S6A,C). In comparison, the MCb_gu2 mutant response appeared more sequence-specific with two templates (Ece_gu2, Lbr_gu2) progressing toward the IO state (Figure 12F and Figure S6B,D).

To further distinguish the modeling properties of MCb_gu1 and MCb_gu2 templates, a set of discriminatory type ii evolutionary rate-shifts was targeted to perform reciprocal residue exchanges between MCb_gu1 and MCb_gu2 seqs (Figure 12D). Again, all three MCb_gu1 behaved similarly and were modeled into more OO conformers, implying some antagonism between the two series of mutations tested altogether (Figure 12G and Figure S6E). Reciprocal mutagenesis had distinct effects on MCb_gu2 templates, which appeared either unchanged or further progressing toward the IO state (Figure 12H and Figure S6F).

Lastly, evolutionary conservation among representative sets of MCb_gu1 and MCb_gu2 seqs suggests these clades experienced distinct selective pressures (Figure 12I,J and Figure S7).

From these results, it is concluded that MCb_gut phylogeny replicates the evolutionary dichotomy previously established among Enteroccoccus spp. It appeared the common ancestor of LB likely possessed a pair of genes encoding seemingly distinct types of MCb_gut. The transition from marine Bacillaceae MCb_ie toward LB’s MCb_gut could proceed sequentially with the functional evolution of a substrate-dependent step, such as inner gate opening, followed by gene copying and a separate evolution of distinct carriers.

3. Discussion

This in silico study correlates the emergence of LB with the functional divergence of a pair of (Slc11) MCb_gut carriers. MCb_gut carriers share a common ancestor with their sister clade MCb_ie, and together, they form the group MCb, which evolved from the group MCa. MCb_ie predominates in marine Bacillaceae and demonstrates evolutionary conservation; it relates directly MCa to MCb_gut because MCb_ie seqs display non-overlapping patterns of identical residues with relatives from each group. Comparative studies between the sister clades MCb_ie and MCb_gut indicate functional innovation to open the inner gate of MCb_gut carriers as well as MCb_gut-specific structural dynamics plasticity. Apparently, MCb_gut could evolve from MCb_ie in response to selective pressure that fosters functional divergence. Assuming this was the case, such evolutionary pressure indicated novel environmental conditions. It is hypothesized that the gastro-intestinal tract of early marine vertebrates provided a novel niche conducive for the progressive evolution of an MCb_ie⁺ Bacillaceae into the ancestor of LB, which carried a pair of genes coding for distinct MCb_gut.

To what extent did the data gathered support the possibility that Mn-dependent, MCb_gut⁺ LB evolved from some Bacillaceae precursor adapted to marine environment? First, several bacterial orders enriched in MCa⁺ spp. (e.g., Cyanobacteria, Pseudomonadales (gamma proteobacteria, GPB), Burkholderiales (BPB), Rhodobacterales, Sphingomonadales, Hyphomicrobiales (Rhizobiales, APB)) [7] might have formed Precambrian subaerial microbial mats that could include Bacillus spp. [32,33]. Such a terrestrial setting would facilitate the acquisition of the mntH Cb_ie gene precursor (mntH Ca) by some Bacillus spp. Bacillaceae represent the primary source of the mntH Cb_ie gene, consistent with evidence of recent gene transfers toward different spp. of Paenibacillaceae (Paenibacillus and Brevibacillus spp.) in which mntH Ca dominates. And many MCb_ie⁺ genera are marine Bacillales found in deep-sea sediments (e.g., Bacillus, Paenibacillus, Lysinibacillus and Terribacillus spp.), while others populate various environments, including the gut of animals [34,35].

Second, while the MntH A (MA) type, which is ancestral to all mntH C genes that were derived from eukaryotic cells (Figure 1A, Table S1), predominates in Bacillaceae, a fraction of Bacillaceae spp. acquainted with marine environment adopted instead MCb_ie. MCb_ie phylogeny indicates genera such as Lederbergia, Siminovitchia, Heyndrickxia, and Neobacillus (Figure S1B, [36]) may have inherited mntH Cb_ie vertically. Their phylogeny suggests these species as more likely candidate precursors of LB than MA⁺ Bacillaceae, such as B. subtilis and B. cereus. Also, these marine Bacillaceae grow aerobically, which implies Fe-based metabolism, and some were recovered as well from a dairy environment or found in the gut of various animals. The emergence of MCb_ie from MCa in marine Bacillaceae could thus mark a first step in the adaptation to the gut of animals such as invertebrates.

A consolidated hypothesis may thus be viewed in successive steps (Figure 13) with (i) the emergence of MCa⁺ Bacillaceae within subaerial/terrestrial microbial mats, which also colonized marine habitat, (ii) the adaptation of marine MCb_ie⁺ Bacillaceae to the gut of invertebrates, and (iii) the selection of Mn-dependent LB carrying a pair of mntH Cb_gut genes in the gut of early vertebrates. Previous studies suggest an exchange of MA for MCa affected the coupling of the pmf to Mn²⁺ import [7], which could improve resistance to environmental stresses (e.g., dessication, temperature, osmotic stress) and provide a path toward adaption to an acidic environment through the evolution of MCa to MCb_ie and then to MCb_gut—a step shown to the impact opening of the inner gate to carry Mn²⁺ ions inside cells. Prior works on MCb_gut homologs revealed both the plasticity of their functional dynamics [6] and their efficient use by LB to pump Mn²⁺ in competitive environments [12].

Whether this LB evolution toward Mn-based metabolism matched the final design of the gastro-intestinal (GI) tract of early vertebrates is a matter of conjecture. But this possibility may provide an interesting angle to approach our intimate relationships with LB because vertebrate development owed to successive innovations following the differentiation of the pharynx at the entry of the digestive tract in chordates, including, for instance, the development of sensory organs and vertebrate novelties such as the spleen, individualized pancreas, heart innervation and the eye’s own muscles [37,38], which might have fostered mutualistic interactions between emerging LB and their host.

Vertebrate inventions took place around the 1R whole genome duplication (WGD) that occurred in the early Cambrian and which produced, in particular, the archetype Nramp ‘Ohnologs’ Nramp1 and Dmt1 ([39,40,41] and Figure S8). NRAMP1 (SLC11A1) expression depends on C/EBP [42,43] and dominates in the spleen, blood and tissue phagocytes, contributing to host resistance to infection [8,44]. Dmt1/Nramp2 (Slc11a2) catalyzes H⁺-dependent Fe²⁺ import both as part of the transferrin endocytic cycle that delivers iron through the body and at the apical membrane of duodenal enterocytes [45,46]. This context may suggest the Ediacaran transition to early Cambrian as a possible evolutionary interval for LB emergence while the GI tractus of early vertebrates was developing into its final form.

High efficiency Mn intracellular accumulation under acidic conditions makes physiological sense for LB because Mn as a micronutrient accelerates their growth under lactate stress [47]. The lactic acidification of an LB environment has important effects on the GI tract: inhibiting the growth of potential microbial pathogens while providing a nutrient for potentially beneficial bacteria, such as the spore-forming component of the colonic microbiota [48], which, by consuming lactate that is transformed into short-chain fatty acids (SCFAs), may stimulate LB growth in return [49,50]. These effects, added to LB’s ability to predigest complex foods could possibly help develop the GI tract of early vertebrates by favoring the differentiation of the gastric organ (stomach).

The GI tracts of urochordates and cephalochordates comprise organs homologous to the liver and pancreas of vertebrates, and chordates share the molecular basis of extracellular digestion with vertebrates. Yet, their genomes lack the pepsinogen gene encoding a gastric digestive enzyme and cephalochordates demonstrate evidence of intracellular digestion in the hepatic cecum. This suggests that extracellular digestion systems based on both stomach and pancreatic digestive enzymes emerged specifically in the common ancestor of vertebrates ([51,52] and based on the homologs of GTEx stomach-specific encoded polypeptides, [53], detected in Cephalochordates: Lipase F, gastric type >50% aa id; in Chondrichthyes: Pepsinogen >60% aa id; in Tetrapods: Trefoil factor 1 & 2, >67% aa id).

LB strains isolated from mammalian source (mainly rodents) colonize the gastric epithelium by forming biofilms [18,54,55,56]. To resist the harsh environment of the stomach lumen, LB use glutamate and aromatic L-amino acid decarboxylases which help in sustaining the transmembrane pmf to energize metabolism and transport [57,58]. The fact that these bacterial enzymes produce amino acid-derived compounds used as neurotransmitters (NTs) in bilaterian animals [59], such as serotonine, dopamine and GABA [55], both in the central nervous system (CNS) and the enteric nervous system (ENS) [60], suggests a possible feed forward loop [61] where proto-LB residing in the upper gut of some vertebrate ancestor would respond to food intake by generating an acidic environment, facilitating digestion and contributing to gut–brain signaling.

An acidic environment in the duodenum of vertebrates is key for absorbing the growth supporting micronutrient iron in the form of Fe²⁺ [46]. As it seems, selection of duodenal Slc11a2- and H⁺-dependent Fe²⁺ uptake may thus correspond to the period of the establishment of the vertebrate stomach. And the prospect that these processes were supported by Mn²⁺-dependent acid-resistant LB expressing MCb_gut parologs may bear significance in terms of distinct substrate selectivity among Slc11 homologs and the partition of micronutrient resources for mutual benefits [62] to the vertebrate host (Fe) and to the LB residing in its gut (Mn). Arguably, an example of ecological–evolutionary interaction [61,63,64] that provided a rich foundation to diversify vertebrate holobionts.

Previous in silico data obtained with MCb_gut demonstrated highly plastic structural dynamics that may increase Mn import efficiency, hereby sustaining LB resistance to acidic and osmotic stress. Herein, a further study of MCb_gut revealed a parology that predated LB radiation, while the sister carrier MCb_ie showed structural differences likely altering Mn intracellular release. These data may suggest that the evolution of LB MCb_gut could foster a novel, Mn-dependent way of life in an acidic environment stocked up with food while relinquishing the use of Fe to their host. This process was likely progressive, contracting LB’s ancestral MCb_ie⁺ Bacillaceae genome and going along with the differentiation of the stomachal pouch as a specialized organ in the neo gastro-intestinal tract of the ancestor of jawed vertebrates.

Gastric LB could impact host behavior and food habits through the synthesis of NTs that act directly on leukocytes within the GI tract lamina propria and affect nerve ganglia of the ENS, producing signals relayed through afferent vagal connections to the CNS [65,66]. These instructions could be modulated by intestinal SCFA-producing bacteria, which may act directly on vagal intestinal terminals and stimulate NT signals [67]. In return, efferent vagal influx from the CNS toward the spleen produces NTs that directly regulate the secretion of proinflammatory cytokines by leukocytes [68]. Similarly, enteric bacteria could also modulate local steroid secretion, contributing to reduce stress hormone levels and inform the CNS to adjust behavioral output [60]. It would seem possible that the emergence of LB is owed at least partly to their ability to communicate with their host via the gut–brain axis.

Vertebrate transition to extracellular intestinal digestion presented inherent risks that required immunological surveillance because luminal microbes opting for a pathogenic lifestyle may weaken the intestinal barrier’s integrity and create a portal for host infection [69,70]. Viewing regulatory lymphocytes as metabolic sensors, the fact that Tregs mediate suppressive functions when food is scarce, whereas the inflammatory functions of Th17 are fueled by glycolysis [71,72], suggests food digestion potentially incurs intestinal inflammation that depends on subsequent resolution to maintain homeostasis and develop immune tolerance to food-derived determinants [73]. Also, lactic acid sensing has an anti-inflammatory effect by modulating macrophage phenotype polarization [74,75]. According to these views, gut LB and SCFA producers and their metabolites could directly instruct the emerging adaptive immune system while sustaining systemic innate immunity due to Firmicutes intrinsic lysozyme sensitivity [76].

Also, both lymphocyte and erythrocyte cell types emerged as vertebrate specificities [77]. Duodenal Fe absorption boosted with gastric acidification likely helped develop the erythron (e.g., hetero-tetrameric hemoglobin in jawed vertebrates, [78]) to distribute oxygen within tissues and sustain their metabolic activity. Apparently, LB Mn-based metabolism and gastric acidification could thus enhance host Fe- and oxygen-based metabolism. And combined with their behavioral and immunological influence, LB’s role in stimulating host metabolism could elevate energetic efficiency and help diversify early vertebrates. The divergence of LB families may have been contemporaneous, perhaps reflecting various mutualist interactions implemented by diverse hosts.

According to recent views, the Ediacaran–Cambrian transition, which produced animal radiation, could be due to a combination of causes, including transient catastrophic events (snowball earths and hypoxic environments) and progressive ‘biotic replacement’ through successive pulses of evolutionary innovations and ecosystem engineering [79,80,81,82,83,84]. The 1R genome duplication that occurred prior to the divergence of jawed vertebrates may indeed testify to disruptive environmental changes during this period. On the other hand, the ability of extant LB probiotics to regulate mammalian brain, digestive and immune activity [55,60,66] suggests that emerging Mn-dependent MCb_gut⁺ LB colonizers, adapted to the upper gut of ancestral vertebrates, could impact the feeding habits and behavior of their host, thereby influencing its evolution [85,86].

Hence, a LB-driven feed forward loop may have contributed to diversify early vertebrate body plans, and turning to jawless vertebrates may be instructive in this regard. The sea lamprey has a life cycle comprising three distinct stages: blind larvae burrowing into the sand feeding on microbes and detritus by filtering surrounding water; the first metamorphosis rearranges sensory organs, forms a new esophagus and modifies the digestive system of free living juvenile lampreys that display high diet conversion efficiency and temperature-dependent growth; the second metamorphosis atrophies the intestine of fasting adults that die after spawning [87,88]. Apparently, the microbiota associated with growing wild juveniles is selectively enriched in LB, Streptococcus spp., in particular. Regarding hagfish gut microbiota, a close relative of LB Weissella cibaria (MCb_gut⁺) was detected in wild Eptatretus, while in captive specimens, MntH A⁺ enterobacteria dominate (Figure S9). These data support the notion that gut LB could directly contribute to the growth and development of early vertebrates. However, continuous host development, perhaps tributary of further innovations such as jaws, heterodimeric hemoglobin and myelinization as well as the 2R genome allotetraploidization [39,40,78,89], was likely required to further vertebrate radiation and inflate marine biodiversity by fueling an evolutionary arms race toward ever more efficient predators [81,90].

In sum, it is proposed that MCb_gut⁺ LB emerged in the upper gut of early vertebrates (ca. 540 mya, i.e., prior to vertebrate 1R WGD). LB colonization formed a novel holobiont that prefigured the final design of the gastro-intestinal tract of jawed vertebrates, whose development could benefit from boosted nutritional efficiency procured through LB–host interactions. Establishing a feed forward loop for food access and tolerance could incentivize LB genome reduction and adaptation as Mn-dependent gastric aids, while their pro-gastric functions stimulated vertebrate Fe-dependent growth and development. Accordingly, LB evolution involved successive phases: (i) reductive, during marine MCb_ie⁺ Bacillaceae adaptation to early vertebrate upper gut, leading to Mn-centric, acid-resistant fermentative cells (MCb_gut⁺ LB); and (ii) expansive, through the subsequent dissemination of MCb_gut⁺ LB adapting to novel niches, including terrestrial environments. Molecular communication between extant LB and the mammalian gut–brain axis and immune system may include remnants of early vertebrate pro-gastric functions and subsequent interactions between more evolved organisms.

4. Materials and Methods

Protein sequences were collected using NCBI and Uniprot repositories and BLAST programs (PHI-BLAST, TBLASTN and BLASTP, [91]). Sequences were curated using the MPI toolkit and the MMseqs2 program to retrieve sets of diverse seqs with reduced complexity [92]. ClustalX [93] seqs clustering and graphic display was used to unambiguously classify seqs into established phylogroups, using reference seqs [7].

Each phylogenetic analysis was performed using a balanced set of groups of seqs to be analyzed plus a defined outgroup. Multiple seqs were aligned using Muscle [94] and the resulting alignments were manually edited using Seaview [95]. As needed, a Phylo-mLogo [96] display of the multiple sequence alignment was generated by writing a simple tree, aggregating clade-specific sets of seqs. Multiple sequence alignments were trimmed to eliminate sites with gaps or judged poorly informative in the Phylo-mLogo display, using Treecon [97]. Parsimony-informative sites were selected using Mega [98]. Tree calculations with IQ-Tree [99] used a number of categories of the free-rate heterogeneity model that was commensurate with the length of the parsimony-informative site alignment to estimate variations in substitution rate among sites using proportions of site categories and mixture substitution matrices [100]. Trees were displayed using FigTree (Version—v1.4.4; http://tree.bio.ed.ac.uk/software/figtree/ (accessed on 1 August 2024)) to appreciate the statistical robustness of tree nodes.

Site selection for mutagenesis used defined phylogroups of aligned seqs to identify candidate type ii evolutionary rate-shifts, distinguishing the groups under study [4] based on the Sequence Harmony and Multi-Relief approaches [101]. Sites producing highly significant scores with both approaches were selected as sets of coevolved sites discriminating sister phylogroups. Reciprocal mutagenesis was performed by exchanging sets of phylogroup-specific residues between templates from each group. Additional mutations targeted sites defining the family of Slc11 carriers (Slc11 synapomorphy) as well as sites that characterize clades part of the Slc11 family. Every point mutation introduced was verified a posteriori with the returned seq of the predicted model. Differences in evolutionary rates between sister clades were also visualized using Consurf [102].

3D modeling was performed using CF [11] with or without template (CFpdb and CFnt, respectively) and native seqs to identify the training model producing the best model, which was used to analyze corresponding mutant seqs [6]. AF2 models were retrieved from the Uniprot repository [103]. Molecular analyses using 2StrucCompare [104] allowed us to determine per residue the root mean square deviation of the Ca trace between pairs of aligned structures, which was visualized using spreadsheet software. The All Against All program (Dali EBI, [105]) was used to establish the 3D correspondence of groups of structures based on pairwise comparisons. POSA [106] was used to obtain multiple flexibly superimposed structures. Structures were visualized using Pymol (DeLano, W.L. The PyMOL Molecular Graphics System; Delano Scientific: San Carlos, CA, USA, 2002).