The leaderless communication peptide (LCP) class of quorum-sensing peptides is broadly distributed among Firmicutes

Leaderless communication peptide (LCP) system is broadly distributed

To assess the distribution of SIP-like LCP-based qs systems across bacteria, we employed a large-scale search strategy that includes: (i) search for RopB homologs across a dataset of 129,001 bacterial genomes and 9,421 reference metagenomics-assembled genomes (MAGs), (ii) probing the genomic vicinity of RopB homologs for small ORFs with a preceding Shine-Dalgarno Ribosome-Binding-Site (RBS) motif that is indicative of their likely translation, (iii) identification of clans of RopB homologs for which the most likely translated adjacent sORF is ultrasmall and encodes a SIP-like LCP, and (iv) functional validation by assessing the regulatory activity of chosen subset of candidate LCPs.

A Blastp search of RopB against the target dataset resulted in 19,280 hits (sequence identity ≥25%, mutual length coverage ≥70%, Supplementary Data 1) encoded in 15,776 genomes and 39 MAGs distributed across 468 taxa. These 19280 hits correspond to 975 unique protein sequences that are predicted to harbor C-terminal tetratricopeptide repeat domain, a hallmark structural element responsible for peptide recognition by the RRNPPA family of receptors⁷ (Supplementary Data 1). Importantly, 478 out of 975 RopB homologs (~49%) are flanked by at least one candidate sORF with a high confidence upstream RBS motif (Supplementary Data 1)^13,14.

To determine whether clans of RopB homologs are enriched for their association with LCPs, we inferred the phylogeny of RopB-like receptors and mapped the amino acid sequence of flanking sORF with the strongest RBS motif on its corresponding receptor leaf of the tree (Fig. 1b, Supplementary Data 1). The two previously well-characterized clans of RopB homologs correspond to streptococcal Rgg regulators, which recognize adjacently encoded canonical propeptides that are post-translationally processed into mature short hydrophobic peptides (SHPs)^15,16. Accordingly, the two Rgg clans appeared as two well-delineated clans in the RopB tree and were correctly mapped to SHP propeptides, which lends validation to our methodology to identify clans of qs systems (Fig. 1b and S1, Supplementary Data 1 and 2). Remarkably, RopB and its evolutionarily closest relatives represent a distinct clan of 183 receptors that were frequently associated with an ultrasmall peptide with amino acid composition similar to SIP (Fig. 1b, Supplementary Data 1 and 2). We term the new clan, RopB clan, and the receptors in RopB clan represent a new class of qs system that uses LCPs as signaling molecules.

The refined manual assessment of the candidate cognate LCP of each receptor within the RopB clan (Fig. 2 and Supplementary Data 2) revealed that the LCP system is broadly distributed in bacterial genomes of a large taxonomic diversity, spanning over Streptococcaceae, Lactobacillaceae, Enterococcaceae, Carnobacteriaceae, Bacillaceae and Staphylococcaceae families (Fig. 1c). Most of the candidate receptor-LCP pairs are encoded in the bacterial genomes, and a non-negligible number of LCP systems are present in either plasmids or integrative and conjugative elements (ICEs) (Fig. 3). However, receptor-LCP pairs were not detected in prophages or phages. The genetic elements encoding LCP-receptor pairs are found in various host-associated microbiomes, waste waters, and fermented food products. The LCP systems are prevalent in the genomes of several clinically relevant human pathogens such as S. pyogenes or Enterococcus casseliflavus, and in animal pathogens such as S. porcinus and S. pseudoporcinus (Fig. 3).

Most putative LCPs are 8 to 10 amino acids long with few exceptions such as LCPs from Lysinibacillus sphaericus, Staphylococcus delphini, Enterococcus durans, and Granulicatella balaenopterae. The predicted LCPs are highly hydrophobic and are mostly comprised of aliphatic and aromatic amino acids (Fig. 2 and S1). The peptides encode distinct communication codes with unique LCP amino acid sequences, suggesting that LCP-mediated communication is specific among closely related bacterial species or strains. Importantly, the characterized RopB-SIP system of S. pyogenes represents only a small subset of the identified LCP systems (Fig. 2), indicating the broader distribution of LCP systems and their potential unappreciated roles in bacterial pathophysiology.

The structure-guided multiple amino acid sequence alignment (MSA) analyses of RopB clan receptors indicate that the LCP-binding C-terminal domain (CTD) diversified faster than the N-terminal DNA-binding domain (DBD) (Figs. S2 and S3a)^7,11,17. Among the structural elements of CTD, the α6 helix is highly conserved relative to the rest of CTD (Figs. S2 and S3a). The α6 helix constitutes a critical structural element for LCP binding as it forms the floor of the LCP binding pocket as well as engages in intramolecular interactions that provide the scaffold for the LCP-binding pocket of RopB (Supplementary Fig. 3b)^7,11,17. In accordance with the suggested functional constraint on α6 helix to bind LCPs, the site-wise dN/dS ratio analyses indicate that α6 helix of candidate LCP receptors are under strong purifying selection against amino acid substitutions (Supplementary Fig. 3a). Similarly, the MSA analyses of 12 LCP-contacting amino acids of RopB with corresponding amino acids from candidate RopB-clan receptors suggest that these amino acids evolve slower and face stronger purifying selection compared to the reminder of CTD (Supplementary Fig. 4a, b). These observations are suggestive of faster diversification of TPR motifs likely due to their innate degeneracy^18,19 relative to LCP-receptor specificity diversification. In accordance with this, pairwise comparisons of evolutionary distances between LCP receptors, LCP-contacting residues in receptors, and candidate LCPs suggest that the amino acid sequences of LCP receptors diverged faster than the LCP-contacting residues and amino acid sequences of putative LCPs, and both LCPs and LCP-contacting amino acids in RopB clan receptors diverge at similar evolutionary rate (Supplementary Fig. 5). Collectively, these observations are suggestive of the function of RopB clan receptors and candidate LCPs as qs receptor-signal pairs.

To understand the co-evolution of receptors and candidate LCPs, we aligned the 12 LCP-contacting amino acids of RopB with similarly located amino acids from RopB-clan receptors and compared them with the physicochemical characteristics of the corresponding LCPs (Supplementary Fig. 4a, b). Consistent with the preponderance of aliphatic and aromatic amino acids in majority of candidate LCPs (Supplementary Fig. 1d), the peptide-contacting amino acids are relatively well conserved among RopB clan receptors (Supplementary Fig. 4). However, compared to RopB, the peptide-contacting residues of RopB clan receptors from Ligilactobacillus muralis, Pediococcus acidilactici, Ligilactobacillus animalis, and Enterococcus casseliflavus are distinct (Supplementary Fig. 4). Accordingly, the physicochemical properties of corresponding LCPs also deviate from the typical signature of LCPs as they contain charged, polar, and proline residues (Supplementary Fig. 4). These observations suggest a tropism of RopB clan receptors for SIP-like LCPs, however, divergence exists to achieve alternative LCP specificities through receptor-LCP co-evolution. Finally, analyses of the RopB clan receptor-LCP pairs from Bacillus cereus and Lysinibacillus sphaericus revealed an evolutionary feature that is suggestive of peptidase-mediated processing of some LCPs similar to canonical RRNPP propeptides (Supplementary Fig. 4). The predicted LCP binding sites and the corresponding candidate LCPs of B. cereus and L. sphaericus have identical amino acid composition. However, the LCP from L. sphaericus have an additional eight amino acids in their C-terminus compared to B. cereus LCP (Supplementary Fig. 4). This observation suggests that LCPs may exist in longer precursor forms and the C-terminal appendages may be involved in peptidase-mediated cleavage of precursor form to release mature LCPs.

Analyses of the genomic context of representative putative LCP systems showed that the candidate genes regulated by LCPs are predominantly in a divergent context relative to the receptor (Fig. 2) and belong to 3 major categories: biosynthetic gene clusters (BGCs) predicted to produce ribosomally synthesized and post-translationally modified peptide (RiPPs) as well as non-ribosomally synthesized antimicrobials, ABC-type transporters, and type VII secretion systems that are typically involved in the translocation of virulence factors (Fig. 3). These observations suggest a broader and more diverse role for LCP systems in bacterial pathogenesis, physiology, and microbial ecology.

LCP in S. salivarius mediates gene regulation

To investigate whether the LCPs other than SIP also act as intercellular signals, we characterized the putative cytosolic receptor-LCP pair from S. salivarius (RopB_ss-LCP_ss). The LCP_ss is encoded in a megaplasmid and located downstream of ropB_ss and transcribed divergently (Fig. 4a). The LCP_ss encodes an eight amino acid hydrophobic peptide with a predicted amino acid sequence of MWLILLFL with no additional amino acids at either end (Fig. 4b). The genetic proximity of a 14-gene operon encoding a putative non-ribosomal peptide synthase biosynthesis gene cluster (NRPS-BGC) located immediately downstream of LCP_ss and transcribed convergently (Fig. 4a) suggest that NRPS-BGC is the regulatory target of the RopB_ss-LCP_ss pathway.

In accordance with this, inactivation of ropB_ss or LCP_ss abrogated NRPS-BGC expression and cis-complementation of ∆ropB_ss and LCP_ss* mutants with ropB_ss and LCP_ss, respectively, restored NRPS-BGC expression (Fig. 4c). Similarly, the addition of synthetic LCP_ss containing the predicted amino acid sequence in native order (LCP_ss), not in scrambled order (SCRA), restored WT-like NRPS-BGC expression in the LCP_ss* mutant (Fig. 4d). However, supplementation with synthetic LCP_ss containing staggering truncations at either N-terminal or C-terminal ends (Fig. 4d) failed to activate NRPS-BGC expression in the LCP_ss* mutant, demonstrating that LCP_ss encodes a mature LCP and lacks the hallmarks of canonical bacterial peptide signals (Fig. 4d). Furthermore, supplementation with even 20X molar excess of synthetic LCP_ss failed to activate NRPS-BGC expression in ∆ropB_ss (Fig. 4d), indicating that LCP_ss activity requires its cognate receptor RopB_ss. However, despite the absence of the secretion signal sequence, the LCP_ss is secreted and reinternalized into the cytosol and acts as an intercellular signal. This was demonstrated by the presence of LCP_ss associated regulatory activity only in the secreted component of peptide-producing strains (WT, LCP_ss*::LCP_ss, and ∆ropB_ss::ropB_ss) and internalization of the exogenously added FITC-labeled synthetic LCP_ss (Fig. 4e, f and Supplementary Fig. 6). Furthermore, inactivation of the canonical bacterial peptide import machinery oligopeptide permease (∆opp) did not affect LCP_ss import (Fig. 4f)²⁰, suggesting that unknown reimport mechanisms are involved in LCP_ss import.

To elucidate the molecular mechanism of LCP_ss-mediated signaling, we investigated the sequence-specific recognition of LCP_ss by cytosolic RopB_ss by fluorescence polarization (FP) assay using FITC-labeled LCP_ss. RopB_ss binds LCP_ss with high affinity (K_d ~8 nM) (Fig. 4g) and the pre-formed RopB_ss-FITC-LCP_ss complex was disrupted only by unlabeled LCP_ss, not by non-specific SCRA (Fig. 4h), indicating that RopB_ss recognition of LCP_ss is sequence-specific. To explain the downstream consequences of RopB_ss-LCP_ss interactions, we hypothesized that LCP_ss facilitates RopB_ss interactions with target promoters and promotes RopB_ss-dependent activation of NRPS-BGC expression. To map the operator sequences for RopB_ss in LCP_ss and NRPS-BGC promoters, we performed electrophoretic mobility shift assays (EMSA) using different DNA fragments that span LCP_ss and LCP_ss-NPRS-BGC intergenic region (Supplementary Fig. 7). RopB_ss bound only to a 43-bp fragment located immediately upstream of the putative −35 hexamer of the LCP_ss promoter and did not interact with the NRPS-BGC promoter (Fig. 4i and Supplementary Fig. 7). These results indicate that LCP_ss and NRPS-BGC are likely expressed as a polycistronic transcript and RopB_ss binding site is located in the LCP_ss promoter (Supplementary Fig. 7). We further probed the 43-bp fragment for the presence of putative palindromes and found an inverted repeat with a 12 bp half site –4 bp spacer –12 bp half site motif that likely constitutes RopB_ss binding site (Supplementary Fig. 7e–f). However, the RopB_ss binding site differs from RopB-GAS binding site in several aspects including the motif arrangement, length, and nucleotide composition. The RopB-GAS binds to a palindrome with a 9 bp half site –7 bp spacer –9 bp half site motif (25 bp long)¹⁰ compared to the 12 bp half site – 4 bp spacer – 12 bp half site motif (26 bp long) of RopB_ss. The half site of the palindrome in RopB-GAS binding site has a nucleotide composition of GTTACGTNT¹⁰, which varies from RopB_ss binding site that has nucleotide composition of ATGTAACATATT (Supplementary Fig. 7f). These findings indicate that the receptors recognize operator sequences of different length and nucleotide composition in the target promoters. However, consistent with the role of RopB_ss and RopB-GAS as transcription activators¹⁰ and their likely role in the recruitment of RNA polymerase to defective promoters²¹, the binding sites for both receptors are located upstream of and around the −35 region of LCP promoters.

To delineate the influence of LCP_ss on RopB_ss-promoter interactions, we assessed RopB_ss-DNA interactions in the presence and absence of LCP_ss by FP assay using FITC-labeled oligoduplex containing the identified RopB_ss binding site (Fig. 4i). The addition of LCP_ss resulted in high affinity interactions between RopB_ss and the cognate DNA sequences (K_d ~ 90 nM) compared to that of apo- or SCRA-bound RopB_ss (K_d ~ > 500 nM)(Fig. 4j and Supplementary Fig. 7g, h), indicating that LCP_ss binding promotes high affinity interactions between RopB_ss and LCP_ss promoter. Based on these observations, we proposed a model for LCP_ss signaling and LCP_ss-dependent transcription activation of NRPS-BGC by RopB_ss (Fig. 4k).

LCP system regulates streptococcal virulence factor production

To assess the functionality of LCPs in other bacteria, we first assessed the regulatory activity of LCP from the swine pathogen S. porcinus. The amino acid sequence of S. porcinus LCP (LCP_sp) is identical to that of SIP (Fig. 5a, b). The coding region of LCP_sp is flanked upstream by ropB in the divergent direction and downstream by a gene encoding cysteine protease (speB_sp) that is transcribed convergently (Fig. 5a). Consistent with the role of LCP_sp as an intercellular signal that controls speB_sp expression, supplementation of S. porcinus with synthetic LCP_sp triggered early induction of speB_sp expression, while the non-cognate SCRA_sp had no effect on gene regulation (Fig. 5c). Since the secreted cysteine protease SpeB is critical for the virulence of S. pyogenes^22,23, we reason that LCP_sp-mediated activation of speB_sp expression may impact the pathogenic traits of S. porcinus.

LCPs from different clades of the RopB clan receptors phylogeny mediate intercellular communication and gene regulation

To test whether LCP from non-streptococcal genus is functional, we characterized the regulatory activity of LCP from Enterococcus malodoratus (LCP_em) (Fig. 2). The LCP_em is 9 amino acid long and its amino acid sequence is distinct from SIP (Fig. 5d, e). The ropB_em is divergently transcribed from LCP_em. However, unlike the other characterized LCPs (above), there are no convergently transcribed genes downstream of LCP_em (Fig. 5d). Instead, there are two genes encoding T7 secretion system-effector pair (T7SS) located downstream of ropB_em and transcribed convergently from ropB_em. Additionally, there are two hypothetical genes located downstream of LCP_em but transcribed divergently from LCP_em (Fig. 5d). Since the gene arrangement is distinct from other characterized LCP systems and regulatory influence of LCP_em on these genes is unknown, we investigated the effect of synthetic LCP_em on the expression of genes in both directions. Supplementation of E. malodoratus with LCP_em induced only the expression of genes encoding T7SS and its effector and the induction was specific for LCP_em (Fig. 5f). Contrarily, the LCP_em had no influence on the expression profile of the two hypothetical genes located downstream of LCP_em (data not shown). These findings demonstrate that the LCP_em that is dissimilar to SIP acts as an intercellular signal and controls the production of an E. malodoratus T7 secretion system/effector system.

To investigate the functionality of a LCP system from a more distant LCP system, we assessed the regulatory activity of LCP from Limosilactobacillus reuteri (LCP_lr) (Figs. 2, 6). Unlike other L. reuteri strains, the L. reuteri DSM32035 strain has a naturally occurring stop codon at amino acid position 4 of the putative LCP_lr (Fig. 6b). The predicted untruncated full length LCP_lr is 8 amino acid long (Fig. 6b, c) with the characteristic aliphatic and aromatic amino acid composition (Fig. 6a–c). The ropB_lr is divergently transcribed from LCP_lr (Fig. 6a). Two hypothetical genes encoding a putative ABC-type transporter are located downstream of LCP_lr and transcribed convergently from LCP_lr (Fig. 6a). Supplementation of synthetic LCP_lr to the exponential growth of L. reuteri activated the naturally silent LCP pathway and induced the expression of genes encoding the ABC-type transporter. The induction was specific for LCP_lr as the SCRA_lr failed to activate the expression of ABC transporter (Fig. 6c). These results indicate that LCP from a distant species, LCP_lr, functions effectively as a qs signal and mediates gene regulation.