|Home | About | Journals | Submit | Contact Us | Français|
Many bacterial moonlighting proteins were originally described in medically, agriculturally, and commercially important members of the low G+C Firmicutes. We show Elongation factor Tu (Ef-Tu) moonlights on the surface of the human pathogens Staphylococcus aureus (SaEf-Tu) and Mycoplasma pneumoniae (MpnEf-Tu), and the porcine pathogen Mycoplasma hyopneumoniae (MhpEf-Tu). Ef-Tu is also a target of multiple processing events on the cell surface and these were characterised using an N-terminomics pipeline. Recombinant MpnEf-Tu bound strongly to a diverse range of host molecules, and when bound to plasminogen, was able to convert plasminogen to plasmin in the presence of plasminogen activators. Fragments of Ef-Tu retain binding capabilities to host proteins. Bioinformatics and structural modelling studies indicate that the accumulation of positively charged amino acids in short linear motifs (SLiMs), and protein processing promote multifunctional behaviour. Codon bias engendered by an A+T rich genome may influence how positively-charged residues accumulate in SLiMs.
Elongation factor Thermo unstable (Ef-Tu) is one the most abundant proteins in bacteria1, 2. It functions as an essential and universally conserved GTPase that ensures translational accuracy by catalysing the reaction that adds the correct amino acid to a growing nascent polypeptide chain3. After the incoming aminoacyl-tRNA docks with the mRNA, GTPase activity induces a conformational change releasing Ef-Tu from the ribosome3–5. In Escherichia coli, Ef-Tu is comprised of three functional domains known as domain I (amino acids 1–200), domain II (amino acids 209–299) and domain III (amino acids 301–393)6. Domain I forms a helix structure with Rossmann fold topology, a structural motif found in proteins that bind nucleotides, while domains II and III are largely comprised of beta sheets3, 7. The GTP/GDP binding domains are housed in domain I, while domains I and II are needed for nucleotide exchange. Domains II and III physically adjust to form an amino acid tRNA binding site3, 5. Ef-Tu sequences derived from phylogenetically diverse species share considerable sequence identity and have been used to generate phylogenetic descriptions of the tree of life8. In eukaryotes, domain III also has a role in actin polymerisation via an actin-bundling domain9, 10.
Despite its highly conserved function in protein synthesis, non-canonical functions have been described for Ef-Tu in all kingdoms of life. Ef-Tu lacks a signal secretion motif yet the ability to execute moonlighting functions often requires the molecule to localise to the cell surface. Ef-Tu is a multifunctional protein in higher order eukaryotes11–16, parasites17–20, fungi21 and it is has been identified on the surface of a wide range of Gram positive and Gram negative pathogenic and commensal bacteria that associate with metazoan species2, 22–29. Bacterial Ef-Tu interacts with nucleolin30, 31, fibrinogen and factor H23, 26, plasminogen and several complement factors26, 27, 32, laminin33, CD2134, fibronectin2, 33, 35, 36, is immunogenic37 and adheres to the surface of Hep-2 cells33 underscoring the multifunctional adhesive characteristics that have been assigned to this molecule. Ef-Tu binds sulfated carbohydrate moieties found on glycolipids and sulfomucin and promotes the binding of Lactobacillus reuteri to mucosal surfaces indicating that Ef-Tu can interact with carbohydrates38. Notably, antibodies against Ef-Tu are induced during infections caused by Staphylococcus aureus 39, 40 Mycoplasma capricolum 41, Mycoplasma ovipneumoniae 37, Chlamydia trachomatis 42, Burkholderia pseudomallei 43 and Mycoplasma hyopneumoniae 44. Ef-Tu has been identified in six surfacome studies (excludes cell membrane and envelope isolations)45–50 performed on S. aureus and Ef-Tu is one of twelve proteins consistently identified in the exoproteome of S. aureus from patients with bacteraemia51. The major staphylococcal autolysin Alt is implicated in playing a role in secreting cytosolic proteins including Ef-Tu into the extracellular milieu24. Moonlighting proteins are likely to be exported via several mechanisms including within secreted extracellular vesicles52, during cell lysis53 and via association with proteins that are secreted by the Sec machinery54.
The ability of Ef-Tu to be secreted onto the cell surface occurred early in the evolutionary interplay between plant pathogenic bacteria and their eukaryote hosts and is a well described pathogen associated molecular pattern (PAMP) molecule55, 56. Plants have evolved pattern recognition receptors (PRR) in their cell membranes that are designed specifically to recognise PAMP molecules released by bacterial and fungal pathogens56–62. An Ef-Tu receptor (EFR) found within Brassica lineages63, 64 recognises the highly conserved N-terminal 18 amino acids (elf18) in the native Ef-Tu molecule56, 63, 64. Binding triggers signal transduction events in plant roots that ensure that pathogenic bacteria are either contained within callose deposits, destroyed by cellular apoptosis, or succumb to an oxidative burst elicited by the production of hydrogen peroxide63. A region spanning surface exposed amino acids 176–225, in Ef-Tu from the Gram-negative bacterial pathogen Acidovorax avenae, interacts with a different PRR in monocotyledonous plants (see Fig. 1)65. EFR has been transferred from the Brassica species Arabidopsis thaliana into the monocot species, rice and transgenic rice plants display enhanced innate immune responses when exposed to elf18 from Xanthomonas oryza, a major rice pathogen66. These studies show that plants have evolved sophisticated molecular machinery to identify Ef-Tu that is released onto the cell surface by diverse plant pathogenic bacteria.
Protein cleavage is emerging as an important post-translational modification that can expand protein function67–70. This is evident in the genome reduced Mollicutes where species specific Mycoplasmal adhesins and lipoproteins are targets of complex processing events67, 71–86. Cleavage fragments are retained on the bacterial cell surface and function as adhesins that bind heparin-like glycosaminoglycans67, 73–75, 77, 79, 80, fibronectin67, 76, 78, 84 and circulatory molecules such a plasmin(ogen) that regulate the fibrinolytic system67, 76, 78, 79, 81. Cleavage motifs have been chemically defined in M. hyopneumoniae using mass spectrometry and occur at phenylalanine residues in the motif S/T-X-F↓-X-D/E, within stretches of hydrophobic amino acids, and at trypsin-like sites in diverse molecules including adhesins, lipoproteins and in metabolic enzymes that traffic to the cell surface77, 79, 82, 83, 85. Cleavage fragments are known to be further processed by aminopeptidases83, 85 that also localise on the cell surface87, 88. We propose that protein processing represents another layer by which proteins can expand and modify protein function and is under recognised as a post-translational modification in prokaryotes.
In this study we identified Ef-Tu, and an extensive repertoire of processed cleavage fragments of Ef-Tu, on the surface of human pathogens S. aureus and Mycoplasma pneumoniae, and the porcine pathogen M. hyopneumoniae. Protein cleavage events were mapped using a systems wide dimethyl labelling protocol that allows for the identification of modified N-terminal peptides (neo-N-termini) by liquid chromatography tandem mass spectrometry (LC-MS/MS) and enabled us to determine how Ef-Tu is processed and presented on the cell surfaces of these pathogens. We further characterised the non-canonical functions of Ef-Tu from M. pneumoniae (MpnEf-Tu) and show that it is a multifunctional protein that can not only bind to and activate plasminogen in the presence of host activators, but is also capable of binding to structurally and chemically diverse host molecules.
The amino acid sequences of Ef-Tu from M. pneumoniae (MpnEf-Tu), M. hyopneumoniae (MhpEf-Tu), and S. aureus (SaEf-Tu) share 60.7% sequence identity. MpnEf-Tu resides on the cell surface of M. pneumoniae and binds fibronectin2. The fibronectin-binding regions have been mapped and are located at the end of domain I and at the beginning of domain II89, 90 and most of domain III is also involved in binding fibronectin89. It is not known if sequence conservation in fibronectin-binding regions of MhpEf-Tu and SaEf-Tu is sufficient to afford these Ef-Tu homologs the ability to bind fibronectin. Several Mycoplasma species73, 91 and S. aureus 92–94 are known to interact with heparin. Putative heparin-binding domains were computationally predicted and mapped onto each of the Ef-Tu molecules (Fig. 1). Several of these were conserved in all three Ef-Tu sequences in domains I, II and III.
LC-MS/MS analysis of tryptic peptides released from the cell surface of S. aureus, M. pneumoniae and M. hyopneumoniae were separately mapped to SaEf-Tu, MpnEf-Tu and MhpEf-Tu respectively. In other experiments, tryptic peptides generated by digesting biotinylated cell surface proteins that were captured by avidin agarose chromatography were also separately mapped to SaEf-Tu, MpnEf-Tu and MhpEf-Tu. Peptides identified by mass spectrometry from both techniques spanned the entire length of Ef-Tu, (Figure S1) consistent with the hypothesis that a sub-population of Ef-Tu molecules are exposed on the cell surface of the three pathogens (Fig. 2) while the remainder perform an essential function in the cytosol. Tryptic peptides spanning the length of SaEf-Tu, MpnEf-Tu and MhpEf-Tu were also characterised when LC-MS/MS analysis was performed on tryptic digests of high salt (>500mM) eluents of proteins that were retained on heparin agarose (Fig. 2).
As part of a larger study that sought to identify the repertoire of proteins in M. pneumoniae, M. hyopneumoniae and S. aureus that are targets of proteolytic processing events, we employed a dimethyl labelling protocol to tag N-terminal peptides and identify precise endoproteolytic cleavage sites (Table 1). Further evidence that SaEf-Tu, MpnEf-Tu and MhpEf-Tu are targets of protein cleavage events was obtained by LC-MS/MS analysis of i) SDS-PAGE gel slices separately loaded with biotinylated M. pneumoniae, M. hyopneumoniae and S. aureus surface proteins captured by avidin chromatography, ii) bacterial proteins that eluted from heparin agarose using high salt (>500mM NaCl), iii) protein spots representing bacterial whole cell lysates and surface biotinylated proteins separated by 2D-PAGE, and iv) size fractioned whole cell lysate proteins resolved by SDS-PAGE.
Of the 15 cleavage fragments of MpnEf-Tu identified in this study, 11 were identified in the biotinylated 1D and 2D SDS-PAGE. Notably, three of the four cleavage fragments derived from MhpEf-Tu and two of six fragments of SaEf-Tu that were enriched during heparin affinity chromatography were also identified in biotinylation experiments (Figures S4, S5 and S6). Ten, four and six cleavage fragments that span different regions of MpnEf-Tu, MhpEf-Tu and SaEf-Tu respectively were recovered from a heparin agarose chromatography using salt concentrations well above the physiological concentration of 150mM. All the fragments recovered from heparin affinity chromatography across all three pathogens contained at least one of the predicted heparin-binding domains that reside within MpnEf-Tu, MhpEf-Tu and SaEf-Tu. These data suggest that the processing events that generate Ef-Tu cleavage fragments, occur on the surface of each of these pathogens and that the fragments may retain an ability to interact with high sulfated glycosaminoglycans such as heparin. To ascertain the nature of the protease(s) responsible for Ef-Tu surface cleavage, the MEROPs database was used to search 56 cleavage events. However, no strong predictions could be made after searching both P4-P3-P2-P1↓P1′-P2′-P3′-P4′ and P2-P1↓P1′-P2′ cleavage motifs.
A single heparin-binding consensus motif (XBBBXXBX, where B is a basic residue) with the sequence DKRHYAHV is found within the amino acid sequences of SaEf-Tu, MpnEf-Tu, and MhpEf-Tu, yet we found several Ef-Tu fragments that were retained during heparin agarose chromatography that did not span this motif. SaEf-Tu, MpnEf-Tu, and MhpEf-Tu sequences were examined for additional motifs enriched with clustered basic residues. In MpnEf-Tu we identified 12 putative heparin-binding motifs dispersed throughout the protein (Table S1). Many of these putative heparin-binding motifs, particularly sequences 37aKegKsaatRy47, 183pKweaKiHd191 and 248lRpiRKa254 were localised to non-essential regions defined here as evolutionary unconserved regions (See S9 - Supplementary Materials: Bioinformatics and Table S1). Using ISIS95, which predicts protein-protein interaction (PPI) sites from sequence information, MpnEf-Tu is predicted to have eight surface exposed PPI sites that are capable of binding macromolecules (Table S2A) such as glycosaminoglycans including four that reside within putative heparin-binding motifs 2aReKfdRsKpHv13, 73 dKRHyaHv80 and 370eKgsKfsiReggRt383 (Table S1). Notably, the key residues (underlined and in bold) in the four binding sites were all unconserved residues as determined by ConSurf96. Putative heparin-binding fragments derived from MpnEf-Tu typically displayed more putative PPI sites and were more intrinsically disordered than the parent molecule and some fragments displayed putative nucleic acid interaction sites (Fig. 2), which are absent in the unprocessed, parent molecule. Additionally, three short linear motifs located in unconserved regions of Ef-Tu that were not predicted binding sites in the parent molecule were predicted to be exposed in MpnEf-Tu fragment 3 (37 aKegKsaatRy 47), fragment 4 (183 pKweaKiHd191), fragment 5 (37 aKegKsaatRy47), fragment 6 (248lRpiRKa254), fragment 7 (37aKegKsaatRy47 and 183 pKweaKiHd191), fragment 10 (183 pKweaKiHd191), fragment 12 (183 pKweaKiHd191) and fragment 13 (248 lRpiRKa254) (Table S1).
The prediction tool MODELLER97 was used to predict the structures of Ef-Tu for all three pathogens based on Ef-Tu from E. coli. For the M. pneumoniae prediction, the E. coli Ef-Tu (PDB: 4G5G_A) had a structure ID percentage of 70.5% and a zDOPE score of −0.93. For M. hyopneumoniae, the E. coli Ef-Tu (PDB: 1DG1_H) had a structure ID percentage of 68.6% and a zDOPE score of −0.72. For S. aureus, the E. coli Ef-Tu (PDB: 1DG1_H) had a structure ID percentage of 75.1% and a zDOPE score of −0.88. All nine distinct cleavage sites for M. pneumoniae and S. aureus and four sites for M. hyopneumoniae have all been mapped in the ribbon structures (Figure S2). Cleavage sites located in regions that are predicted to release the three domains are mostly surface accessible within the molecule. The location and accessibility of the heparin-binding domains in MpnEf-Tu, MhpEf-Tu and SaEf-Tu and the two published fibronectin-binding domains in MpnEf-Tu are depicted in Figure S3.
It was notable that MpnEf-Tu was recovered from M. pneumoniae native cell lysates that were loaded onto affinity columns coupled with A549 epithelial cell surface proteins, fetuin, fibronectin, actin or plasminogen (Figure S4). Consistent with these data, rMpnEf-Tu bound to immobilized A594 cells in microtitre plate binding assays (Fig. 3a). Proteins that bind (recombinant pyruvate dehydrogenase subunit B) and that do not bind (P08 fragment of P1 adhesin) to A594 cells were used positive and negative controls respectively98. Binding of rMpnEf-Tu to A594 cells was partially inhibited when anti-rMpnEf-Tu antibodies, but not pre-immune antiserum, was present (Fig. 3b). MhpEf-Tu was recovered from native cell lysates of M. hyopneumoniae that were loaded onto affinity columns coupled with PK15 epithelial cell surface proteins, fibronectin, actin, or plasminogen (Figure S4). MpnEf-Tu has previously been shown to bind fibronectin2 and we independently confirmed this in microtitre plate binding assays. Furthermore, our binding assay suggests that M. pneumoniae encodes fibronectin-binding proteins other than Ef-Tu (Fig. 4). MpnEf-Tu, and nine of the fifteen cleavage fragments of MpnEf-Tu, were recovered from affinity columns loaded with fibronectin (Figure S3). Of the nine cleavage fragments, seven spanned the known fibronectin-binding regions described previously (see Fig. Fig.11)89, 90. We also identified fragments from columns coupled to fibronectin that spanned the N-terminus of MpnEf-Tu suggesting that other fibronectin-binding domains are yet to be identified in this molecule. MhpEf-Tu and six cleavage fragments of MhpEf-Tu were retained by columns coupled with fibronectin (Figure S5). The cleavage fragments spanned the N- and C-terminal ends, as well as the central region of MhpEf-Tu suggesting that it may contain fibronectin-binding domains.
Ten fragments spanning different regions of MpnEf-Tu (Figure S4) and one MhpEf-Tu fragment (Figure S5) were identified from affinity columns coupled with biotinylated surface proteins derived from A549 and PK-15 cells, respectively. MpnEf-Tu and MhpEf-Tu, and fragments derived from them, were recovered from actin-coupled columns (Figures S4 and S5). Five fragments of MpnEf-Tu were recovered during affinity chromatography using fetuin as bait (Figure S4).
M. pneumoniae 99–101 and M. hyopneumoniae 78, 81 have both been shown to bind plasminogen onto their cell surface and assist with its conversion to plasmin. In the current study, MpnEf-Tu and MhpEf-Tu were both recovered during plasminogen agarose chromatography. Fragments spanning different regions of MpnEf-Tu (Figure S4) and MhpEf-Tu (Figure S5) were recovered from plasminogen coupled agarose beads.
Antibodies raised against rMpnEf-Tu were used to show that MpnEf-Tu resides on the surface of colonies of M. pneumoniae (Figure S7). Our surfaceome studies (unpublished data) identified candidate proteins that could be used as a negative control for these studies and antibodies raised against recombinant 1-phosphofructokinase (FruK) from M. pneumoniae were used for this purpose (Figure S7). To further investigate the binding capabilities of rMpnEf-Tu, we examined the ability of the molecule to interact with a range of host molecules. rMpnEf-Tu bound to fetuin (KD=53±14nM), actin (KD=19±3nM) and heparin (KD=42.5±1.5nM) in the nanomolar range and to plasminogen (KD=933±388nM) in the micromolar range, using microscale thermophoresis (Figure S8). We extended these studies using microtitre plate binding assays to confirm that rMpnEf-Tu binds plasminogen and fibronectin, and also show that rMpnEf-Tu binds fibrinogen, vitronectin, lactoferrin and laminin in a dose dependent manner (Fig. 4). Binding of rMpnEf-Tu to plasminogen was significantly reduced by the addition of an increasing concentration of NaCl and ε-aminocaproic acid (Fig. 5a). Notably, ε-aminocaproic acid was effective at blocking interactions between M. pneumoniae and plasminogen while high concentrations of NaCl were less effective (Fig. 5a). These data suggest that lysine residues play a significant role in binding interactions between Ef-Tu and plasminogen, and M. pneumoniae cells and plasminogen.
In the presence of plasminogen activators tPA and uPA plasminogen bound to rMpnEf-Tu is converted to plasmin and can degrade fibrinogen and vitronectin (Fig. 5b). Collectively these studies highlight the widespread multifunctional capabilities of Ef-Tu and the cleavage fragments derived from it.
Ef-Tu moonlights on the cell surface of S. aureus, M. pneumoniae and M. hyopneumoniae, three phylogenetically diverse, pathogenic bacteria that belong to the low G+C Firmicutes. Using a combination of microscale thermophoresis and microtitre plate binding assays we show that rMpnEf-Tu binds strongly to heparin (KD=42.5±1.5nM), fetuin (KD=53±14nM) and actin (KD=19±3nM), as well as to laminin, plasminogen, vitronectin, lactoferrin, fibronectin, and fibrinogen. Plasminogen bound to rMpnEf-Tu can be converted to plasmin in the presence of plasminogen activators tPA and uPA (Fig. 5). We also extend these finding by showing that SaEf-Tu, MpnEf-Tu and MhpEf-Tu are targets of processing events on the cell surface of these bacterial pathogens but the biological significance of this warrants further investigation (see below). Molecules are not strictly confined to compartments in the bacterial cell and can perform novel functions at different cellular locations24, 25, 54, 102–105. Much remains to be learnt about how proteins, especially those lacking signal motifs, localise on bacterial cell surfaces.
We sought to gain a better understanding of how Ef-Tu has evolved to be a multifunctional binding protein. SaEf-Tu, MpnEf-Tu and MhpEf-Tu all putatively bind heparin, each sharing the consensus heparin-binding motif XBBBXXBX (sequence: dKRHyaHv) as well as a number of other heparin-binding motifs (see Fig. 1 and Table S1). It is notable that while this motif (dKRHyaHv) is conserved in the Ef-Tu from M. pneumoniae, M. hyopneumoniae and S. aureus only part of the motif, with the sequence RHyaHv, is conserved in Ef-Tu from other bacterial sources. The addition of DK residues is predicted to impart a putative PPI site. Twelve putative heparin-binding motifs identified in MpnEf-Tu (Table S1) were predicted to predominantly localise to non-essential, unconserved regions of the molecule that do not unduly influence its ability to function as an elongation factor. Short linear motifs (SLiMs) typically ranging from three to ten amino acids play crucial roles in mediating PPIs106–108. In eukaryotes, these motifs are typically located in intrinsically unstructured, disordered regions of proteins that impart plasticity and are reported to favour transient, low affinity and reversible interactions106, 109. Notably, MpnEf-Tu formed strong interactions with fetuin, heparin, and actin suggesting that the accumulation of SLiMs may be sufficient to form high affinity interactions.
Positively charged amino acids in SLiMs play a crucial role in interactions between proteins and highly sulphated glycosaminoglycans such as heparin110, and other molecules such as actin111, plasminogen112, DNA113, 114 and fibronectin69, 84, 115. Here we identified SLiMs enriched in positively charged amino acids in different regions of MpnEf-Tu, including sequences 37aKegKsaatRy47, 183pKweaKiHd191, and 248lRpiRKa254, and identified eight surface exposed PPI sites, including three that reside within putative heparin-binding motifs 2aReKfdRsKpHv13, 73 dKRHyaHv80, and 370eKgsKfsiReggRt383. It is notable that the lysine analog, ε-amino caproic acid, was shown to be a potent inhibitor of interactions between MpnEf-Tu and plasminogen, and M129 whole cells and plasminogen, underscoring the important role played by positively charged amino acids in binding interactions with host molecules (Fig. 5a). Overlapping SLiMs are frequently identified in multifunctional proteins106, 116. In M. hyopneumoniae, the C-terminal sequence 1070KKsslKvKitvK1081 in the multifunctional cilium adhesin, P97 binds both heparin and fibronectin84 and overlapping peptides from a region within phosphoglycerate kinase from group B streptococcus strain NCS13 with sequence 203sKvsdKigvienlleKadKv222 and 213enlleKadKvligggmtytf232 bind both actin and plasminogen112. Similarly, we were able to identify SLiMs enriched in positively charged amino acids in SaEf-Tu and MhpEf-Tu. The accumulation of positively charged residues in SLiMs, possibly as a consequence of an A+T rich genome, facilitates binding to a wide range of host molecules in the low G+C Firmicutes. Our data is consistent with the proposition that the accumulation of surface exposed SLiMs represents a mechanism to generate protein multifunctionality in bacterial proteins.
S. aureus 92 and M. hyopneumoniae 73–75, 82, 84, 87, 88, display cell surface, heparin-binding proteins that are important to the pathogenic potential of these species. Interactions between heparin-binding proteins and target receptors in host cell membrane allow microbes to colonise a wide range of niche sites, traverse tissue barriers and disseminate from their initial point of contact and form biofilms117. S. aureus 118, 119, M. pneumoniae 120, 121 and M. hyopneumoniae (our unpublished data) are all capable of forming biofilms. The extracellular matrix of S. aureus biofilms is derived from a mixture of eDNA and cytoplasmic proteins118, 122–127 and electrostatic interactions between cytoplasmic proteins and eDNA is thought to tether cells together in S. aureus and mixed species biofilms127. In S. aureus, the addition of heparin increases biofilm production in a protein dependant manner which implies that heparin-binding proteins are important for biofilm development92. Notably, Ef-Tu has been identified on the surface of S. aureus under biofilm inducing conditions122. These observations lend weight to the hypothesis that the accumulation of positively charged amino acids in SLiMS represents a powerful mechanism to promote PPIs that underpin essential biological processes such as the formation and maintenance of biofilms.
Bacterial pathogens including Campylobacter jejuni 69, Mycoplasma gallisepticum 86, and Chlamydia trachomatis 128 process molecules that are secreted to the cell surface. In M. hyopneumoniae, processing of cilium adhesin families has been reported extensively and cleavage motifs have been mapped73, 77, 80, 83. Recently we showed that lactate dehydrogenase is cleaved on the surface of M. hyopneumoniae generating fragments with putative multifunctional binding capabilities68. In M. pneumoniae, cleavage fragments of the major adhesin P1 and DnaK have been shown to comprise part of the cytoskeletal attachment organelle complex129 and Mycoplasma derived lipoproteins are targets of processing events that release powerful immunomodulatory peptides71, 130–132. These observations prompted us to utilise a systems wide, protein dimethyl labelling strategy to investigate protein processing. Here we identified and characterised numerous processing sites in Ef-Tu derived from all three bacterial pathogens. Furthermore, our surface biotinylation studies indicate MpnEf-Tu, MhpEf-Tu and SaEf-Tu, were a target of multiple processing events on the surfaces of M. pneumoniae, M. hyopneumoniae and S. aureus, respectively. Our work strongly suggests that the accumulation of positively charged residues in the SLiMs found in Ef-Tu facilitates binding to a wide range of host molecules, and potentially to eDNA and that protein cleavage events expand the functional complexity of proteins that moonlight on the cell surface. We propose that processing is a mechanism that has evolved to promote multifunctional behaviour more broadly and lends itself to the creation of novel binding sites in moonlighting proteins that retain a strict conformational structure needed to execute their canonical function.
Fifteen cleavage fragments of MpnEf-Tu were identified in this study of which eleven reside on the cell surface. Unlike full length MpnEf-Tu, none of the fragments were retained in all six affinity chromatography columns, but five were identified in at least five affinity columns (fragments 5, 6, 7, 8, and 10 in Figure S4). Fragments 5, 8, and 10 were retained in columns coupled with: A549 surface proteins, fetuin, fibronectin, actin, and heparin. Fragments 6 and 7 were retained in columns coupled with: A549 surface proteins, fibronectin, actin, heparin, and plasminogen. Fragment 4 was identified in eluents from columns coupled with A549 surface proteins and heparin while Fragment 9 was identified in eluents from columns coupled with fetuin and actin (see Figure S4). These data indicate that retention of the fragments during affinity chromatography is dependent on the host molecule that is coupled to the agarose beads and the sequence of the Ef-Tu fragment. Further studies are needed to quantify the binding characteristics of fragments of Ef-Tu with host molecules.
Cleavage fragments of cytosolic proteins that moonlight on the cell surface add another layer of complexity to the concept of multifunctional proteins. We show that processing exposes SLiMs that would otherwise be inaccessible for interactions with potential binding partners. Recently, a peptidome study of a protease deficient strain of Lactococcus lactis identified 1800 distinct peptide fragments in spent growth medium that were derived from proteolytic activity targeting both surface accessible and cytosolically derived proteins133. Similar studies by the same group indicated that surface accessible proteins in other Firmicute species including Listeria monocytogenes, Enterococcus faecalis and Streptococcus thermophilus were also targeted by complex processing events133. Previously we have shown that processing events play an important role in the maturation of key adhesin families in pathogenic mycoplasma species67, 72–86. Here we extend these findings to show that surface proteolysis is critical in shaping the surface proteome more broadly and that processing represents a novel and under recognised mechanism to expand protein function.
In summary, Ef-Tu moonlights on the surface of bacteria where it is a target of proteolytic processing events. Computational analysis of fragments of MpnEf-Tu suggest they are inherently more disordered and display putative PPI sites that are inaccessible in the parent molecule, generating unprecedented functional diversity on the cell surface. Further studies, using systems wide methodologies, are needed to determine how processing generates biologically important effector molecules and if protein processing is fundamental to the expansion of protein function in bacteria belonging to different phylogenetic clades.
A full description of the experimental section is listed in the S10 - Supplementary Materials 3.
M. pneumoniae (M129 strain; ATCC 29342) was cultured in modified Hayflick’s medium at 37°C in tissue culture flasks as described previously134.
S. aureus (SH 1000 strain) was cultured in TSB (Oxoid, Hampshire, UK) at 37°C with shaking and harvested during early stationary phase. Protease inhibitors (Roche Diagnostics®, North Ryde, Australia) in PBS were added to the cells during harvest and washes with PBS. For S. aureus lysis, cell pellets were freeze-dried overnight before added to pre-cooled metal milling canisters with 12 small metal beads. The canister was cooled in liquid nitrogen and milled at a maximum frequency of 30Hz for 1minute for 15 rounds; cooling in liquid nitrogen between rounds. Proteins were than solubilised in 7M urea, 2M thiourea, 50mM LiCl, 50mM Tris-HCl (pH 8.8), 1% (w/v) C7bZ0 with protease inhibitors followed by sonication at maximum intensity for 30seconds for 20 rounds resting on ice in between.
Human lung carcinoma cells (A549; ATCC CCL-185) were cultured in RPMI 1640 medium (Invitrogen, Carlsbad, CA) supplemented with 10% heat inactivated fetal bovine serum at 37°C with 5% CO2 in tissue culture flasks.
Porcine kidney epithelial (PK-15) cells were cultured in DMEM medium (Invitrogen) supplemented with 10% heat inactivated fetal bovine serum at 37°C with 5% CO2 in tissue culture flasks.
Details about host proteins and human proteins used in this article are supplied in supplementary materials (S10.1).
Biotinylation of the M. pneumoniae cell surface was carried out as described in67. M. hyopneumoniae and S. aureus cells were washed with PBS after centrifugation before the resuspending in EZ-link sulfo-NHS-biotin (Thermo Fisher Scientific, North Ryde, Australia). M. hyopneumoniae and S. aureus cells were biotinylated for 30seconds and 1minute, respectively. Quenching, lysis (for M. hyopneumoniae), avidin purification and western blotting were the same as for M. pneumoniae. Lysis for S. aureus cells is described above in section ‘Strains and cultures and reagents’.
Trypsin shaving of M. pneumoniae cells was carried out as described previously75 with modifications. Trypsin was added to adherent M. pneumoniae cells within tissue culture flasks, and M. hyopneumoniae and S. aureus cells were resuspended in trypsin.
M. pneumoniae and M. hyopneumoniae whole cell lysates were prepared as previously described75. Lysis for S. aureus cells is described in section 'Strains and cultures and reagents' above. Proteins were reduced and alkylated with 5mM tributylphosphine and 20mM acrylamide monomers for 90min at room temperature. Insoluble material was removed by centrifugation and five volumes of acetone added to precipitate protein. After centrifugation, the protein pellet was solubilized in 7M urea, 2M thiourea, 1% (w/v) C7BzO for one- and two-dimensional gel electrophoresis.
In-gel trypsin digestion was performed as described in77. After digestion, tryptic peptides were stored at 4°C until needed for liquid chromatography tandem mass spectrometry.
Affinity purification of heparin-binding proteins for M. pneumoniae was performed as described in67. M. hyopneumoniae cells were and lysed in 10mM sodium phosphate, pH 7 with three 30second rounds of sonication. S. aureus cells were lysed as described in section 1.1 except that protein was solubilised in 10mM sodium phosphate, pH 7 with protease inhibitors followed by sonication at maximum intensity for 30seconds for 4 rounds, resting on ice in between. After centrifugation, ~300µg of soluble protein from both M. hyopneumoniae and S. aureus lysates were treated exactly the same as M. pneumoniae.
Purified fibronectin (Merck Millipore, Darmstadt, Germany), plasminogen (Merck Millipore), actin (Sigma, St. Louis, MO) and fetuin (Sigma) used in this section are described in supplementary section S10.1. Avidin purification of these host-binding M. pneumoniae proteins was carried out as described in67. Avidin purification of M. pneumoniae proteins that bind A549 surface proteins was performed as described in67.
Purified fibronectin (Merck Millipore), plasminogen (Sigma) and actin (Sigma) used in this section are described in supplementary section S10.1. Avidin purification of these host-binding M. hyopneumoniae proteins was performed as described in84. Avidin purification of M. hyopneumoniae proteins that bind PK-15 surface proteins was performed as described in82.
For this experiment and all subsequent experiments, animal experiments were approved by the ethical board of Landesdirektion Sachsen, Dresden, Germany (with the permit no. permit 24-9168.25-1/2011-1). ELISA experiments were carried out as described in98. Guinea pig rMpnEf-Tu antiserum (1:750) followed by anti-guinea pig IgG (1:1,000, Dako, Glostrup, Denmark) dilutions were used. Tetramethylbenzidine (Sigma) was added followed by 1M HCl and absorbance was measured at 450nm (620nm as reference).
Freshly grown A549 cells were used to coat wells in 96-well microtitre plates for 2h at 37°C as described in above in ‘Binding assays’. rMpnEf-Tu (10µg/ml) was incubated with guinea pig rMpnEf-Tu antiserum or pre-immune serum (1:100) concentrations were used.
Purified human proteins used were supplied by Sigma and described in supplementary section S10.1. Binding of rMpnEf-Tu (15µg/ml) to extracellular matrix proteins was performed as described previously98. The dilutions for the appropriate antisera are: (Sigma) anti-plasminogen: 1:2,500; anti-lactoferrin 1:5,000; anti-laminin 1:750; anti-vitronectin 1:5,000; anti-fibrinogen 1:3,000; anti-fibronectin 1:1,000. Followed by anti-rabbit IgG (Dako, Glostrup, Denmark) or anti-goat IgG (both 1:2,000).
Microscale thermophoresis to determine the binding affinities between Ef-Tu and a fluorescently labelled host protein was performed as described in84. Time for Microscale thermophoresis was set to 30s with fluorescence set to 5s before and 30s after each run. Each sample was scanned with 40%, 60% and 80% MST Power. Dissociation curves were plotted with hot/cold, jump or thermophoresis settings to determine dissociation constant.
Briefly, 96-well microtitre plates were coated with rMpnEf-Tu as described. Plasminogen (2.5µg) together with increasing concentrations of NaCl were added to the wells and incubated for 1.5h at 37°C. Wells were incubated with rabbit anti-plasminogen (1:3,000) followed by anti-rabbit IgG (1:2,000). Detection was done as described above in ‘Binding assays’.
ELISA was carried out as reported in98. In brief, the wells of ELISA plates were coated with rMpnEf-Tu. 2.5µg of plasminogen and increasing concentrations of ε-aminocaproic acid were added to the wells and incubated for 1.5h at 37°C. Wells were incubated with rabbit anti-plasminogen (1:3,000) followed by anti-rabbit IgG (1:2,000) and OD420nm was measured.
Degradation of human fibrinogen and vitronectin by activated plasminogen was carried out as described in98. 10µg/ml of human plasminogen was added to the wells which were then incubated with fibrinogen or vitronectin (each 15µg/ml) and urinary plasminogen activator (uPA; Sigma) or tissue plasminogen activator (tPA; each 75 ng/ml; Sigma).
Freshly grown M. pneumoniae cells were harvested and used to coat wells in 96-well microtitre plate for 2h at 37°C as described previously100. Wells were blocked before adding guinea pig rMpnEf-Tu antisera (1:500) followed by anti-guinea pig IgG (1:1,000). As a control wells were incubated with guinea pig antisera raised against total M. pneumoniae proteins.
M. pneumoniae colonies were grown on PPLO agar plates and blotted onto nitrocellulose as described previously100. Antisera to PdhB and 1-phosphofructokinase (FruK) were used as positive and negative controls, respectively.
Immunofluorescence experiments were carried out as described in100. Again guinea pig antisera to PdhB and FruK were used as positive and negative controls, respectively.
Dimethyl labelled proteins were analysed by two mass spectrometers; the Sciex 5600 and the Thermo Scientific Q Exactive™. For full technical set up and method details see supplementary materials (section S10.4).
Bioinformatic analysis of Ef-Tu used the online resources: ProtParam138, Clustal Omega139, SignalP 4.1 Server140, SecretomeP 2.0 Server141, TMpred142 and COILS (Addition of ‘yes’ to 2.5 fold weighting of positions a,d)143. The amino acid sequences of MpnEf-Tu (Uniprot#: P23568), MhpEf-Tu (Uniprot#: Q4A9G1) and SaEf-Tu (Uniprot#: Q2G0N0) were analysed using a variety of bioinformatics tools. Conservation of amino acid positions in each protein were detected using The ConSurf server96. Putative heparin-binding sites were identified using the search patterns X-[HKR]-X(0,2)-[HKR]-X(0,2)-[HKR]-X and X-[HKR]-X(1,4)-[HKR]-X(1,4)-[HKR]-X via ScanProsite144. Putative protein-protein and protein-nucleic acid interaction sites were identified using ISIS95. Intrinsically disordered regions were predicted by Meta-Disorder145, 146, which combines the outputs from original prediction methods NORSnet, DISOPRED2, PROFbval and Ucon. Solvent accessibility of each amino acid position was ascertained using evolutionary information from multiple sequence alignments and a multi-level system147. Nucleotide, DNA and RNA binding regions were predicted by SomeNA148.
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.
M.W., K.L.H. and V.M.J. are recipients of the ‘Australian Postgraduate Award’ scholarship from the University of Technology Sydney. I.J.B. is a recipient of the ‘Doctoral Scholarship’ from the University of Technology Sydney. The authors would like to thank Mark Raftery and the Bioanalytical Mass Spectrometry Facility (BMSF) for access to the Sciex 5600 and Thermo Scientific Q Exactive™ Plus mass spectrometers purchased with the ARC grant LE130100096 entitled ‘Advanced high resolution mass spectrometer for collaborative proteomic and lipidomics research’. The authors would also like to thank Jerran Santos for assisting in designing the 3D models. The authors would like to thank both University of Technology Sydney and the Deutsche Forschungsgemeinschaft (DU 1280/1-1) for funding this research. This work was partly funded by Ausgem. Ausgem is a collaborative partnership between the New South Wales Department of Primary Industries’ Elizabeth Macarthur Agricultural Institute (EMAI) and the ithree institute at the University of Technology Sydney (UTS).
M.W. acquired data and analysed it for M. pneumoniae except those listed for I.J.B. and L.H. K.L.H. acquired data and analysed it for S. aureus except data for the N-terminome that was acquired by J.R.S. M.W. and K.H. prepared all figures and tables except those listed for L.H. and A.G. M.W., K.L.H., R.D. and I.G.C. assisted with drafting the manuscript. A.G. produced the recombinant Ef-Tu of M. pneumoniae and the guinea pig antiserum. L.H. acquired binding data for A549 cells and most host proteins, performed plasminogen binding and activation studies, conducted experiments with recombinant antisera, and prepared figures with A.G. I.J.B. acquired N-terminome data for M. pneumoniae and M. hyopneumoniae, and I.J.B. and M.W. analysed it. I.J.B. assisted with the preparation of cleavage maps. V.M.J. performed the SLiM analysis and prepared figures. B.B.A.R. and J.L.T. acquired and analysed surfaceome data for M. hyopneumoniae. M.P.P. oversaw the acquisition of mass spectrometry data and assisted with data interpretation. S.P.D. initiated this study, wrote most of the manuscript, and secured funding. R.D. supervised the binding studies performed by L.H. and A.G. and secured funding. I.G.C. provided intellectual input and reviewed drafts of the manuscript. All authors reviewed and approved the manuscript.
The authors declare that they have no competing interests.
Michael Widjaja, Kate Louise Harvey and Lisa Hagemann contributed equally to this work.
Roger Dumke and Steven Philip Djordjevic jointly supervised this work.
Electronic supplementary material
Supplementary information accompanies this paper at doi:10.1038/s41598-017-10644-z
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.