Origin and Evolution of Bacterial Periplasmic Force Transducers

Abstract

In double-membraned bacteria, non-equilibrium processes that occur at the outer membrane are typically coupled to the chemiosmotically energized inner membrane. TolA and TonB are homologous proteins which energetically couple inner membrane motor proteins to the essential processes of outer membrane stabilization and substrate import, respectively. The evolutionary trajectories of these proteins have been difficult to elucidate due to low-sequence conservation, yet they are thought to transduce force similarly. Here, this problem was addressed using structural prediction approaches to identify and annotate force transduction operons to trace their distribution and evolutionary origins. In the process, we identify a novel outer membrane-tethering system and a previously unknown family of monomeric force transducers. This approach revealed putative tolA genes, and thus the core organizational principles of the tol-pal operon throughout diverse bacterial taxa. We discovered that the α-helical structure of the periplasm-spanning domain II of TolA previously thought its hallmark, is anomalous amongst most Tol-Pal systems. This structure is mainly prevalent in γ-proteobacteria, likely in adaptation to their lifestyle. Comparison of Tol-Pal and Ton system distribution suggests that TolA emerged from a TonB paralogue and co-emerged with Pal, the outer membrane-tethering lipoprotein that functionalizes the Tol-Pal system. We also determined that TolB, the Pal-mobilizing protein, likely emerged from a family of outer membrane proteins; and CpoB, a periplasmic factor that coordinates peptidoglycan remodeling with cell division, was originally a lipoprotein present in the ancestral Tol-Pal system. The extensive conservation of the Tol-Pal system throughout Gracilicutes highlights its significance in bacterial cell biology.

Keywords: Tol-Pal system; Ton system; bacteria; bacterial cell envelope; force transduction; molecular motors.

Graphical Abstract

Fig. 1.

The tol-pal operon is widely conserved in Proteobacteria, but TolAII secondary structure varies by clade. a) TolA examples typically have a three-domain structure, where the TMH is followed by a disordered linker, an α-helical/proline-rich region, followed by the globular C-terminus (final 100 residues). TolA PsiPred predictions often show a mixture of α-helical and proline-rich regions, not defined by a specific ratio or relative position within domain II. The vast majority of TolA species feature a predicted β-strand (motor box) within a disordered region always found after the TMH, which is strongly suggested to interact with the dimeric component of the motor protein (Loll et al. 2024; Zinke et al. 2024). b) The tol-pal loci from a range of species were curated by identifying a contiguous region containing a motor, force transducer, tolB and pal. Guide tree (top right) adapted from Sharma et al. (Sharma et al. 2022) (supplementary fig. S1.2, Supplementary Material online). “A” denotes Acidothiobacillia, while “O” denotes Oligoflexia. TolA secondary structures were predicted using PsiPred, and the α-helix:proline ratios calculated (McGuffin et al. 2000) (supplementary SI S1, Supplementary Material online). These ratios are represented in colored outer circles. Circumferential numbers indicate respective length of domain II. Some species have putative “split” tol-pal systems (denoted by “//”), where different components are not proximal on the genome. Evidence for loci and TolAII composition analysis indicated in supplementary SI S1, Supplementary Material online.

Fig. 2.

Quantification of TolAII/TonBII structural diversity. a) The total length of domain II for selected TolA and TonB sequences is plotted here by species (supplementary SI S1 to S2, Supplementary Material online). Each point is colored by proteobacterial class (“AB”= Acidothiobacillia, “OF”= Oligoflexia). A Pearson's correlation test suggests that the length of TolAII and TonBII is weakly positively correlated in each species (Score = 0.28). These data show that lengths of both TolAII and TonBII also vary greatly within proteobacterial classes, but the majority of TolAII sequences are longer than TonBII in any given species (below grey line). Species of note are indicated, to illustrate the variation in TolAII/TonBII length. Two TonB medians were observed (i/ii). b) For each selected TolAII and TonBII sequence, the total number of α-helical residues is plotted against the total number of proline residues. Many TolA sequences are α-helix dominant (e.g. E. coli), but many mixed (e.g. B. pertussis) and proline-dominant sequences (R. centenum) are also observed. For TonB sequences, the majority are proline-dominant or mixed, but some α-helix dominant species are present (e.g. A. avenae). Note that proline residues are a minimal proxy for the number of PPII-helical residues present.

Fig. 3.

tol-pal gene organization and α-helix:proline ratios of selected gracilicute species. a) Gracilicutes with identified tol-pal loci display similar gene organizations (arrows) and TolAII structural composition analysis (outer circles). This alignment is derived from portion of the phylogenetic tree published by Witwinowski et al., and annotated with those taxa (black branches of guide tree, top left) (Witwinowski et al. 2022). The emergence of Pal is indicated on the guide tree (green point). The α-helix:proline ratio of representative species TolA are indicated, with full tol-pal loci and analysis of those species analyzed in supplementary SI S1, Supplementary Material online. Fusion genes indicated by multi-colored arrows. Circumferential figures indicate the number of residues found in TolAII. “PPII-α Transition” refers to the increased representation of α-helix-rich TolA sequences within Proteobacterial taxa. Uncultured taxa are indicated by Candidatus (“C.”). b) The distribution of Gracilicute TolAII lengths more closely resembles proteobacterial TonBII than TolAII. The median lengths (solid bars) and upper/lower quartiles (dashed bars) are indicated. A Mann–Whitney U test suggests that proteobacterial TonB and Gracilicute TolA domain II length distributions do not differ significantly (P = 0.1487), while proteobacterial TolAII length distributions differ significantly from both (P < 0.0001). c) The distribution of Gracilicute TolAII sequences features a significantly lower proportion of proline residues than proteobacterial TonBII sequences (P = 0.0003), but greater than proteobacterial TolAII sequences (P = 0.013). d) The distribution of Gracilicute TolAII sequences indicates a similar proportion of α-helical residues to proteobacterial TonBII sequences (P = 0.2704), both of which are significantly lower in α-helical residues than proteobacterial TolAII sequences (P < 0.0001). For both b) and c), median values are labeled. For panels b–d), values for E. coli TolA (green) and TonB (pink) are shown.

Fig. 4.

vWA proteins may transduce force via motor coupling. a) Species featuring homologous vWA genes adjacent to tolQR-like motor genes, detected using WebFlags by homology to the Tichowtungia vWA transducer family (Saha et al. 2020). An uncharacterized tetratricopeptide repeat (TPR) gene was found to co-occur with the motor-vWA genes, suggesting interaction. b) AF3 prediction of vWA proteins with a TolA/TonB-like topology suggest they may transduce force in a similar manner to TolA/TonB, by binding the edge of the motor complex. Notably, these vWA transducer protein TMHs possess the conserved SHLS motif and histidine (His₃₅) that is essential to motor interactions formed by TolA and TonB. c) Species featuring vWA genes adjacent to tolQ-like motor genes only. Gene clusters shown sample the top 200 homologs of Idiomarina iohensis as detected by WebFlags (Saha et al. 2020). No tolR/exbD/motB-like genes were detected. d) This family of vWA proteins may form an obligate complex with TolQ-like pentamers via a TPH (blue). Of note, a conserved pair of aspartate residues (one per vWA monomer) are predicted to be coordinated by a conserved ring of threonine residues, an essential interaction conserved in MotA-MotB, ExbB-ExbD and TolQ-TolR. The TPH may rotate within the pentamer to directly couple PMF to the application of tensile force at the OM. AF3 models with plDDT and PAE scores are indicated in supplementary fig. S1.7, Supplementary Material online.

Fig. 5.

Structural phylogeny of TolB-related protein models suggests TolB emerged from an OMP-TolB-like family. a) Unrooted Foldtree tree reveals clades comprising peptidases, mixed enzymatic functions, small β-propeller proteins featuring a single α-helix, OMPs with a TolB-like periplasmic domain and TolB, and many proteins comprising only a β-propeller. Decorating AF models are colored by plDDT. Annotated TolB query results and UniProt/AF accessions are available in SS1. The phylogenic distribution of TolB-like OMPs is wider than TolB (supplementary fig. S1.6, Supplementary Material online). FoldTree outputs (featuring UniProt/AF accessions), and TolB sequences/alignment/hmm profile are available in SI 3. b) Structural similarities suggest a route for TolB emergence. The OMP-TolB-like ancestor of TolB may resemble the extant example shown from Bdellovibrionales (A0A1F3SUA3), with loss of the OM-barrel leading to an ancestor resembling an extant TolB-like protein (Balneolaceae shown; A0A356X2S7), followed by domain loss to the TolB common ancestor, to modern (“neontic”) TolB (E. coli shown; P0A855).

Fig. 6.

PorE is a putative multi-domain OM-lipoprotein comprising CpoB, TolB, carboxypeptidase and pal subdomains. a) AF3 model of P. gingivalis PorE, with labeled subdomains. b) Three of the core six Tol-Pal proteins exhibit subdomains with structural homology to PorE (Research UCJNa 2019; Jumper et al. 2021). Crucially, PorE resembles Pal with the inclusion of TolB/CpoB/CPE resident proteins, suggesting a Pal-adjacent OM-PG tethering role.

Fig. 7.

Distribution of tol and ton systems suggest ancestral TolA was a TonB paralogue. Taxa featuring curated Tol-Pal systems are found within Gracilicutes only, whereas most didermal species across both Terrabacteria and Gracilicutes feature Ton systems. This suggests that the LBCA possessed a functional Ton system (pink arrow). The most likely point of TolA emergence (green arrow) occurs near the origin of Pal proposed previously (Witwinowski et al. 2022). Phylogenetic tree adapted from Witwinowski et al. (Witwinowski et al. 2022). Taxa are predominantly didermal unless noted as monoderms (“M”). Of the Firmicutes, Negativicutes are the only diderms where Ton systems were detected. In Actinobacteria, a few species feature TBDT genes, but only in monodermal clades where they are likely non-functional (Beaud Benyahia et al. 2025). Species flagged for Ton systems are noted in SS1, as well as those identified from GenBank accessions that include metagenome assembled genomes (MAGs). Uncultured taxa are indicated by Candidatus (“C.”).

Fig. 8.

A hypothetical route for TonB-to-TolA speciation. TonB duplication generates paralogues with varied affinities for different TBDT TonB-boxes (black arrows). Structural phylogeny revealed a family of FhaC-TolB-like proteins that present plausible extant homologs of an OMP from which TolB emerged, by loss of its β-barrel and subsequent domain reduction (Fig. 5b). Emergence of Pal and the TolB-Pal interaction likely coevolved with the emergence of TolA (right). The β-strand addition interaction of TonBIII remains conserved in TolAIII. It is unclear whether TolB or Pal was first-recruited to the division site, or how the Tol-Pal system achieves septal trafficking.