The commensal bacterium Enterococcus faecalis is a common cause of nosocomial infections worldwide. The increasing prevalence of multi-antibiotic resistant E. faecalis strains reinforces this public health concern. Despite numerous studies highlighting several pathology-related genetic traits, the molecular mechanisms of E. faecalis virulence remain poorly understood. In this work, we studied 23 bacterial proteins that could be considered as virulence factors or involved in the Enterococcus interaction with the host. We systematically tested their interactions with human proteins using the Human ORFeome library, a set of 12,212 human ORFs, in yeast. Among the thousands of tested interactions, one involving the E. faecalis virulence factor ElrA and the human protein FHL2 was evidenced by yeast two-hybrid and biochemically confirmed. Further molecular characterizations allowed defining an FHL2-interacting domain (FID) of ElrA. Deletion of the FID led to an attenuated in vivo phenotype of the mutated strain clearly indicating that this interaction is likely to contribute to the multifactorial virulence of this opportunistic pathogen. Altogether, our results show that FHL2 is the first host cellular protein directly targeted by an E. faecalis virulence factor and that this interaction is involved in Enterococcus pathogenicity.
Enterococcus faecalis belongs to the human core intestinal microbiota, i.e the limited number of species that are shared by more than 50% of the population1. Although it is a common inhabitant of our intestinal tract, E. faecalis is also a major opportunistic pathogen for critically ill or immuno-compromised patients2, 3. Indeed, E. faecalis is ranked among the most common nosocomial pathogens worldwide infecting the bloodstream and urinary tract. Understanding the molecular mechanisms of E. faecalis pathogenicity is therefore crucial to design new therapeutic approaches to treat enterococcal infections even when strains resistant to antibiotics are involved. E. faecalis does not harbour virulence factors such as effectors and toxins identified in Gram-positive pathogens. Instead, the pathogenicity of E. faecalis relies on a multifactorial process that involves biofilm formation, antibiotic4, 5 and stress resistances6,7,8, adhesion to host cells and intracellular survival3.
Several E. faecalis surface exposed proteins involved in different processes are likely to contribute to bacterial pathogenicity e.g., adhesion to host cells, host cell invasion and bacterial protection against host defenses. Except for the enterococcal fibronectin-binding protein A (EfbA)9, most of these surface exposed proteins possess a surface-exposure motif (anchoring or retention motif) such as the WxL domain like of the enterococcal leucine-rich protein A (ElrA)10 or a LPxTG-anchored domain as for the aggregation substance (AS) proteins11, adhesin to collagen (Ace)12, Ebp pilus13, 14, the FSS15, and PrgB16 proteins.
These E. faecalis surface exposed proteins were reported as targeting different host extracellular matrix proteins. Among the members of the AS (aggregation substances) proteins, Asa1 interacts with intestinal epithelial cell lines, macrophages and neutrophils via its binding to CD11b/CD18 integrin, fibronectin and thrombospondin17, 18. Ace is characterized by adjacent immunoglobulin-like folds that define the family of Microbial Surface Components Recognizing Adhesive Matrix Molecules (MSCRAMMs) and it interacts with collagen type I and laminin12, 19. The anchorless EfbA protein binds fibronectin9. Finally, Ebp pili adhere to fibrinogen, collagen, and to platelets14, 20. Overall, these proteins or complexes seem to have multifunctional roles, suggesting that they may interact with different host-cell unknown components through different functional domains.
To extend our knowledge on the mechanisms developed by E. faecalis to interact with its host, we used a systematic high-throughput yeast two-hybrid (HT-Y2H) approach to identify protein interactions between selected surface protein of this bacterium and the human proteome. Such screening approaches were successfully used with pathogenic microbes, including bacteria such as Brucella 21 or Pseudomonas syringae 22 and viruses such as Epstein-Barr23, Hepatitis C24, HTLVs25, or dengue fever virus26. Here, we report the systematic screening for interactions between 23 selected E. faecalis proteins and the human ORFeome v3.1 (hORFeome), and the identification of a specific interaction between the E. faecalis ElrA and the human FHL2, a member of the Four-and-a-Half LIM Domains Protein family. Disruption of this interaction led to an impairment of the in vivo pathogenic phenotype strongly suggesting a role of this newly identified interaction in the E. faecalis-host interplay.
Selection of 23 E. faecalis ORFs
We defined three criteria based on previous publications in the field of host-bacteria interactions to select 23 proteins expressed by E. faecalis and that could promote interaction with the host namely proteins harbouring a putative surface-exposure motif (anchoring or retention motif) and/or a eukaryotic-like domain and proteins already described as involved in E. faecalis virulence (Table 1 and Fig. 1).
Based on sequence analysis, we first selected 12 surface exposed proteins known to be involved in E. faecalis virulence: 10 containing an LPxTG anchoring domain (EbpA, EbpB, EbpC, Ace, Fss1, Fss2, Fss3, EF3314, Asa1 and PrgB), 1 containing a WxL retention domain (ElrA) and 1 anchorless protein (EfbA). It is well established that cell surface anchored proteins are involved in bacterial-host interactions and that in pathogens, they are often required for the first steps of infection by promoting the interaction with the host targeted cells27,28,29. Moreover, these surface exposed virulence factors have either prokaryotic or eukaryotic-like domains, which can be involved in interaction with the host components (Table 2). Based on these data, we selected 11 additional E. faecalis proteins that have surface anchors and/or eukaryotic-like domains but which are not integral membrane proteins that were excluded to avoid false positive hit due to hydrophobic regions in the Y2H screen30, 31. The additional selected E. faecalis proteins were: EF2525, EF2713, EF2250, EbsG, Prg-B-like, EF0485, EF2224, EF0592, EF2090, EF0377, EF0286.
Altogether, we selected a set of 19 potentially surface exposed proteins (Fig. 1 and Table 1): 16 proteins contain an LPxTG anchoring domain, 2 proteins have a WxL retention domain and 1 is an anchorless protein. LPxTG-proteins EF2525 and EF2713 exhibiting neither predicted domain nor homology with other bacterial proteins were selected and can be considered as E. faecalis specific proteins. The 17 other proteins contained either prokaryotic or eukaryotic-like domains likely to promote protein-protein interactions. The prokaryotic domains are: (i) the Glucan binding domain and Antigen_C (PrgB, Asa1, PrgB-like and EF0485), (ii) a collagen- or fibronectin-binding domain (Fss1, Ace and EfbA) and (iii) the Leucine rich repeat (LRR) domain (ElrA and EF2250) interaction. Regarding the eukaryotic-associated motif, we found the von Willebrand factor type A (vWA) domain that was originally found in the blood coagulation protein von Willebrand factor (vWF), a signature of adhesion in eukaryotes present in EbsG and EbpA, two Ig-like fold domains (i) the bacterial immunoglobulin-like BIG3 domain detected in EF2505 and (ii) the Cna_B repeat detected in 6 E. faecalis proteins (EbpA, EbpB, EbpC, Fss1, Fss2 and Fss3). Note that the Fss proteins 1 to 3 have been shown to bind fibrinogen15. Three proteins exhibiting the ankyrin repeat, fibronectin binding or collagen eukaryotic-like domains were also selected. EF0377 displays 4 ankyrin domains. The ankyrin domains have been shown to promote specific interactions with host proteins leading to the impairment of the targeted molecule activity. Many examples demonstrated the role of ankyrin repeat bacteria effectors in the hijacking of eukaryotic functions by the pathogen during the infection process32. EF0286 and EF1249 have a fibronectin-binding domain (pfam07299 that enables adhesion of the bacteria to the host tissues. EF2090 exhibits collagen domains involved in protein interactions of fiber structures.
Identification of ElrA FHL interactions using HT-Y2H
The interactions between the 23 selected E. faecalis putative virulence factors and the Human proteome were tested using the yeast two-hybrid approach33. The 23 E. faecalis ORFs were amplified by nested PCR amplification34, cloned in the pDONR223 using the Gateway system and transferred in the pDestDB-GW vector (see Material and Methods) prior to interactomic assays. Each selected E. faecalis ORFs was tested with the human ORFeome v3.1 containing 12,212 human ORFs35, 36. Among the 280,876 tested interactions, a single protein out of the 23 tested gave positive hits. The candidates interacting pairs of proteins were identified, revealing ElrA10, interacted with 3 human ORFs: FHL2 (31042@A08), GOPC (11063@H10) and CDC23 (11088@H06). The interactions were individually validated by mating the yeast strains carrying ElrA fused to the Gal4 DNA binding domain (BD) and the human ORFs flanked by the Gal4 Activating Domain (AD) and growing them on medium selecting for strong interactions. The interaction of ElrA with FHL2 was the only one confirmed. FHL2 is a member of the four-and-a-half-LIM protein family, known to participate in many cellular processes37 and including FHL1, FHL2, FHL3, FHL4 and FHL5/ACT in humans. The LIM domain is a highly conserved double zinc-finger motif38.
The hORFeome contains 5 different FHL sequences: FHL1 843 ncl (11050@D04), FHL2 840 ncl (31042@A08), FHL3 843 ncl (11031@C01), FHL3 519 ncl (11018@E03) a truncated variant of FHL3 and FHL5 855 ncl (11050@F02). In the same experimental set-up we tested the individual interaction of each clone with the ElrA protein. In stringent selective conditions, the interaction was only detected for FHL2 (Fig. 2). As no interaction was detected between ElrA and the other FHL family members, we concluded that in the Y2H assay, ElrA interacts specifically with FHL2.
Identification of the minimal ElrA domain interacting with FHL2
To identify the domain of ElrA involved in the interaction with FHL2, several deletions of the ElrA protein (Fig. 3A) were generated using primers located at the limits of the 5 domains previously described10: the signal peptide (aa 1 to 26), the cap domain (aa 61 to 141), the Leucine Rich Repeat domain (aa 142 to 387), the Immunoglobulin-like domain (aa 388 to 473) and the WxL domain (aa 474 to 723). The 21 ElrA-fragments resulting from the amplifications were fused to the Gal4-DB in the pDestDB-GW vector (Table S1) and tested for their ability to interact with FHL2 fused to the Gal4-AD in yeasts. Interestingly, all the fragments containing the WxL domain interacted with FHL2 whereas no interaction was observed with the other 15 ElrA truncated mutants (Fig. 3B). The WxL domain contains 2 WxL motifs at position 628–630 (WxL1) and 712–714 (WxL2) and is responsible for the interaction with the peptidoglycan and the bacterial cell surface retention of ElrA10.
We further investigated whether the two WxL motifs were responsible for the interaction with the FHL2 human protein. Sixteen sub-fragments of the WxL domain (aa 474 to 723) were amplified (Fig. 3A) and cloned in fusion with Gal4-DB. We tested the interaction with FHL2 fused to the Gal4-AD; the region spanning from the residues 474 to 606 interacted with FHL2 defining the minimal FHL2 interaction domain (FID) of ElrA (Fig. 3C and D). Interestingly the 2 WxL motifs located between the aa 628 to 630 and the aa 712 to 714 are not involved in the interaction with FHL2. Consequently, the ElrA region that was previously called the WxL domain (aa 474 to 723) indeed corresponds to two different functional domains: (i) FID (residues 474 to 606) and the WxL domain (residues 607 to 723).
GST pull-down and immunofluorescence microscopy were used as independent methods to confirm the FID/FHL2 interaction showed by Y2H. Three GST-ElrA fusions corresponding to (i) ElrA27–723 i.e., the full-length protein without its signal-peptide, (ii) ElrA27–473 i.e., the N-terminal part of ElrA without the FID and the WxL domains and (iii) ElrA474–606 corresponding to the FID were tested with a 3Flag tagged FHL2 in a pull-down assay (Fig. 4A). As expected, the interaction between FHL2 and ElrA27–723 was observed. Interaction of FHL2 also occurred to a minor extend with ElrA474–606 that contained only the FID, were observed (Fig. 4B).
HeLa cells were transfected with plasmids encoding a HA-FHL2 and a myc-ElrA constructs coding for ElrA27–723 (containing FID, FID + ) or ElrA27–473 fragments (deleted for the FID, FID-). By confocal microscopy, a clear co-localisation was observed between FHL2 and the wild-type ElrA27–723 fusion protein, but not with the FID-truncated ElrA27–473 fragment. These data are consistent with an interaction of FHL2 with ElrA through the C-terminal part of ElrA in a cellular context (Fig. 5). Altogether these results are in agreement with a role of FID (aa 474 to aa 606) in the ElrA/FHL2 interaction.
In vivo importance of the ElrA FHL2 Interacting Domain
Identification of an ElrA domain that interacts with FHL2, led us to hypothesize that deletion of the FID domain may affect virulence as it would impair ElrA interaction with FHL2. To test this hypothesis and considering a potential tissue-specific interaction between ElrA and FHL2, we compared the virulence of a mutant strain deleted for the FID domain to the parental strain (OG1RF) and the ΔelrA mutant strain in two different mouse models: urinary tract (UTI) and systemic infection (SI) models. We constructed and tested two independent isogenic elrA mutants deleted in-frame of residues 474 to 606 that encompass the FHL2 interacting domain (elrAΔFID). In the UTI model (Fig. 6A), the ΔelrA strain exhibited a decrease of 3 log units (P = 0.0003) in kidney and of 1 log unit decrease (P = 0.00304) in bladder compared to the WT strain, revealing that elrA contributes to pathogenesis of E. faecalis urinary tract infection. Interestingly, the organ bacterial loads upon infection with elrAΔFID were intermediate between the values observed with the WT and those obtained with the ΔelrA-mutated strain, indicating that ElrA-FHL2 interaction is implicated in the ElrA-dependent pathogenesis in the urinary track. We then examined the virulence of the three strains in a systemic infection model post-infection intravenously (Fig. 6B). Strain ΔelrA showed significantly decreased burden of the liver (~ 1 log units) and kidneys (1.5 log unit) compared to the WT strain. In contrast, strain elrAΔFID showed an intermediate phenotype and appeared to colonize the liver at a comparable level as the WT while it colonized kidneys at a significant lower level (~0.4 log units of reduction for each strain; P = 0.02 and P = 0.049 for the two tested strains) than the WT, supporting that elrA contributes to E. faecalis systemic infection and partially relies on the interaction of ElrA with FHL2. Altogether, these results indicated that the interaction between ElrA and FHL2 have a role in systemic and urinary infections.
Several bacterial pathogen effectors bind host proteins to hijack the host functions for their own benefits39. In Gram-positive bacteria, several bacterial virulence factors are located at the pathogen surface such as the internalins of L. monocytogenes 27. In this study, we selected and cloned 23 putative virulence factors of the E. faecalis opportunistic pathogen and tested their ability to interact with host proteins using the hORFeome v3.135, 36. We identified a strong interaction between the virulence factor ElrA of E. faecalis and the human protein FHL2, and determined the minimal interacting domain of ElrA, named FID (FHL2-Interacting Domain). In addition, we showed that FID is partially involved in the ElrA-dependent virulence in vivo. To our knowledge, besides interactions with extracellular matrix proteins, FHL2 is the first intracellular host protein targeted by a virulence factor of E. faecalis.
In our Y2H experimental set-up using stringent conditions33, only strong interactions are detectable, this explains the low range of positive hits. In the selected surface proteins, ElrA is one of the proteins that is likely to have a transient interaction with the bacterial surface as its WxL domain promotes a non covalent interaction with the peptidoglycan10. Then ElrA could be considered both as a surface and as a secreted protein. For the proteins covalently bound to the bacterial surface (e.g. those with an LPxTG motif), the Y2H have shown limits in the discovery of surface-surface protein interactions. This could be due to the low level of posttranslational modifications in yeast while they are important for many cell surface proteins (e.g., O- or N-glycosylation for eukaryotic cell receptors40) or the absence of multimerisation in Y2H experiments while it is required for the activity of many eukaryotic surface receptors. For example, the detection of an interaction between the bacterial flagellin and the TLR5 homodimer receptor is not detectable using Y2H (A. Jamet personal data). However, the set of entry vectors encoding E. faecalis ORFs generated in this study, provides an interesting tool of easily transferable inserts to Gateway compatible vectors for other applications, such as the expression of an E. faecalis protein fused with tags that can be used for purification or tandem affinity purification allowing the purification of stable complexes to identify the partners involved41 or with reporters such as the GFP, which can be used to localize the protein42.
ElrA is an internalin-like protein that contains a signal-peptide, a leucine-rich repeat domain followed by an Ig-like domain and at the C-terminal end, a WxL domain, which interacts with the bacteria peptidoglycan10. It was demonstrated that: i) the inactivation of elrA significantly reduced E. faecalis virulence in a peritonitis model and that ElrA contributed to the bacterial intracellular persistence in macrophages and increased the host inflammatory response by stimulating in vivo the production of interleukin-610. Overexpression of ElrA protects from phagocytosis in vitro and increases E. faecalis virulence a mouse peritonitis model43. It was also shown that elrA is the first gene of an operon poorly transcribed under laboratory growth conditions but induced in mice during infection10. However, so far no ElrA ligand was identified. In this study, we identified a physical interaction between ElrA and FHL2, a member of the four-and-a-half-LIM-only family proteins. This interaction happens in vitro and impacts the in vivo infectious process.
The members of the four-and-a-half-LIM-only protein family are involved in many physiological processes probably through their ability to interact specifically with various proteins. Due to their interacting ability, the LIM-only proteins and by extension all proteins containing LIM domain(s), are considered as adaptors or scaffolding proteins44. To date in the host-microbe interactions context, a specific interaction has been shown between Tax1 a protein of the human T-cell leukemia virus 1 (HTLV-1) and FHL345. Our study is the first report describing an interaction between a bacterial protein and a specific member of the FHL family. FHL2 has a strong homology with the other members of the four-and-a-half-LIM-only protein family, which includes FHL1, FHL3, FHL4 and FHL5. The human FHL2 amino acid sequence is 48.2% identical with FHL1, 53.4% with FHL3, 48.4% with FHL4, and 59.1% with FHL5. Although all FHLs could promote protein-protein interactions37, we demonstrated that ElrA interacts specifically with FHL2. FHL2 is involved in heart physiology, bone formation, muscular function and placental development. In addition, FHL2 is involved in a wide range of various cellular processes such as cytoarchitecture, cell adhesion, cell survival, cell mobility, signal transduction and other processes through the regulation of gene expression46. FHL2 enhances the transcription of the tumor necrosis factor (TNF) receptor-associated factor 6 (TRAF6) through the stabilization of NF-κB after activation47. FHL2 is enzymatically inactive but has been shown to interact with more than 100 proteins through its LIM domains48 although, the comparison of various FHL2 partners do not allow the identification of a consensus interacting sequence. The FHL2 partners are components of crucial signaling pathways (protein kinases, transcription factors, receptors or receptors interacting proteins), proteins known to modulate RNA splicing, DNA replication and repair. The broad range of cellular partners and involvement in cellular processes make FHL2 an interesting target for a pathogenic bacterium to ensure its survival. However, the clear picture of the cellular mechanisms underlying and rising from the FHL2/ElrA interaction remains to be found.
Enterococcus are traditionally considered as commensal extracellular bacteria, during its infectious process it can be found intracellular within urothelial cells shed from the bladder of patients49 or surviving for long periods into macrophages50,51,52,53. Moreover, several laboratories showed that E. faecalis escapes the phagocytic vacuole to avoid lysosomal elimination51,52,53. At late stages of infections, intracellular E. faecalis activates phosphatidylinositol 3-kinase signaling to block apoptotic cell death in macrophages52, 53. As FHL2 is detected in the cytoplasm, we could anticipate that FHL2/ElrA interaction occurs during bacterial intracellular survival. However, as ElrA is non-covalently attached to the bacterial peptidoglycan10, we cannot rule out the possible release of ElrA directly into the cytoplasm of the host cell during the early interaction steps. What are the incidences or signalling pathways targeted by the ElrA/FHL2 interaction remain unknown. The interaction with ElrA could even compete with an FHL2 interacting protein or create a new complex with additional partner(s) as it is often the case in the host-pathogens protein-protein interactions54. Interestingly, ElrA is not the only surface-exposed protein that is interacting with host cytoplasmic partners. Indeed, Listeria express at its surface 2 virulence factors implicated in its cytoplasmic survival: ActA (actin assembly-inducing protein) and InlK (Internalin-like protein K) that interact with Arp2/3 and the Major Vault Protein (MVP), respectively55. The parallel between ElrA and InlK is interesting, as both proteins are surface-exposed virulence factors and only expressed in vivo 55. Like FHL2, MVP is implicated in multiple cellular processes but its precise role remains unknown. MVP is recruited at Listeria surface impairing autophagic recognition and consequently leading to Listeria survival in the cytoplasm56. It is tempting to speculate a similar role for ElrA/FHL2 interaction as E. faecalis has been recently described to resist autophagy52. However the role of ElrA in E. faecalis autophagy resistance remains to be investigated.
The FHL2 Interacting Domain (FID) of ElrA is 100% conserved in all the sequenced strains of E. faecalis suggesting a crucial functional role of this domain. Among the 24,788 prokaryotic registered genome57, apart the Enterococcus genus, the FID was detected in an internalin-like protein of Weissella halotolerans exhibiting 35% of identity with the FID of ElrA. This W. halotolerans protein could be an ElrA-like protein as it shares 34% of identity with ElrA and contains in addition to the FID, an LRR repeat domain and a WxL domain. It is noteworthy that W. halotolerans is an opportunistic pathogen for the rainbow trout58.
In contrast to the ElrA-deficient strain, we did not find any difference between the wild-type and elrAΔFID strain in the peritoneal infection model (data not shown) suggesting that the FID is not involved in the ElrA-dependent virulence via this infectious entry and that probably other domains of the ElrA protein are important for virulence; but it also demonstrated that the FID-deleted protein is stable and functional. On the opposite, in the UTI model, the impairment of the interaction of ElrA with FHL2 led to a systemic infection but with intermediate E. faecalis burdens in mice organs compared to wild-type strain (higher counts) and the fully deleted elrA strain (lower counts), revealing that ElrA contributes to virulence in ascending urinary tract infection. Differences in levels of attenuation of elrAΔFID in the models of UTI and SI suggest that the interaction between ElrA and FHL2 is tissue-specific and that other domains of ElrA may be at least as important as the FID to ensure a complete virulence. This observation may be related to the expression pattern of FHL2. It is expressed at basal level in most of the human and mice tissues (in human:<20 fragments per kilobase of exon per million fragments mapped (FPKM)) but shows a higher level of expression in given tissues such as the heart muscle (516 FPKM), the ovary (456 FPKM), the placenta (74 FPKM) and the urinary bladder (90 FPKM)59).
Altogether, our study evidenced a strong interaction between ElrA (from E. faecalis) and FHL2 (from human) with consequences in 2 infectious models giving new hints to decipher the mode of action of this bacterial effector. This work underlined that the involvement of ElrA in Enterococcus virulence could depend on the route of infection. Understanding the functional role and the significance of the interaction of ElrA of E. faecalis with FHL2 could provide new insights into the mechanisms of virulence of this poorly understood human pathogen.
Materials and Methods
Bacterial and yeast strains, growth conditions, plasmids and oligonucleotides
The strains used in this work are summarized in Table 2. Escherichia coli strains were grown at 37 °C in Luria-Bertani (LB) broth or agar medium supplemented with the required antibiotics: ampicillin (100 µg.mL−1), spectinomycin (100 µg.mL−1), erythromycin (150 µg.mL−1). E. faecalis strains were grown at 37 °C in M17 broth or agar supplemented with 0.5% of glucose (M17G) and the required antibiotics: tetracycline (5 µg.mL−1), erythromycin (30 µg.mL−1). Saccharomyces cerevisiae strains were grown in rich YEPD medium (yeast extract 10 g.L−1, peptone 20 g.L−1 and dextrose 20 g.L−1) or in synthetic complete medium (SC) lacking the amino acids (Leu, Trp and His) or nucleobase (Ade) corresponding to its auxotrophies. The vectors used for the constructs generated in this work are listed in Table 2. The plasmids and primers generated by the cloning of the 23 genes of E. faecalis strain V583 and the 39 deletants of the elrA gene in the pDONR and pDestDB-GW vectors are listed in Table S1.
Gateway cloning of full-length E. faecalis ORFs and partial elrA fragments
The full-length E. faecalis selected ORFs (EfORFs) were cloned using the Gateway system (Invitrogen) as described in Dobrijevic et al.34. For the Y2H experiments, all the EfORFs were transferred from the pDONR223 into the pDestDB-GW60 using the LR clonase II (Invitrogen) to generate the yeast vectors expressing the DB-EfORF fusion proteins. As we used the Gateway recombination method (Invitrogen), the attB1.1 (GGGGACAACTTTGTACAAAAAAGTTGGC) and attB2.1 (GGGGACAACTTTGTACAAGAAAGTTGGGTC) sequences were added to the 5′ end of each forward and reverse primers, respectively. An amber stop codon in frame with the elrA sequence was incorporated in all reverse primers. The PCR products were then cloned in the pDONR223 as previously described34.
Construction of the 3Flag tag-FHL fusions
The FHL CDS were transferred from pDONR223 to the Gateway destination vector pV189961 using the LR recombination cloning, generating the p3flag-FHL2, p3flag-FHL3, p3flag-FHL1 and p3flag-FHL5 vectors.
High-throughput yeast two-hybrid
Yeast two-hybrid screening was performed as previously described33. Yeast matings were performed with Y8800 and Y8930 strains62. The pDestDB vectors carrying the EfORFs were transformed into the Y8930 MATα yeast and selected on solid synthetic complete (SC) medium lacking leucine (SC-Leu). The autoactivating baits were identified by testing the growth of the yeasts containing the pDESTDB-EfORFs on solid SC medium lacking leucine and histidine (SC-Leu-His) and containing 2 mM of 3-amino-triazole (3-AT); all growing clones were not selected.
Each DB-EfORF MATα yeast strain was mated with the 12,212 human AD-ORFs pooled into 130 mini-libraries as described by Rual et al.33 on YEPD agar. After overnight growth at 30 °C, the diploids expressing the GAL1::HIS3 yeast two-hybrid reporter were selected on SC-Leu-Trp-His plates containing 2 mM 3-AT63. De novo autoactivators were eliminated using the counter-selectable marker CYH2 60. Positive colonies were picked for PCR amplification and identification of the interacting proteins by sequencing of the AD- and DB-ORFs. The Human proteins found to interact with the E. faecalis proteins were then individually tested against each EfORF. For this purpose, we mated MATα Y8930 containing individual DB fused to Ef ORF and MATa Y8800 containing the vector expressing the AD fused to the interacting human candidates. The plates were incubated overnight at 30 °C on appropriate SC minimal media to select for the expression of GAL1::HIS3 or GAL2::ADE2. To eliminate the yeast background due to residual growth on the minimal media, the plates were twice velvet-cleaned as described63. The same procedure was used to test the interaction of ElrA fragments with FHL2. At each step of the interaction screening and validation procedures the following yeast control strains were spotted on the plates: diploids yeasts containing  pPC97-CYH (DB-empty) and pPC86 (AD-empty),  DB-RB1 and AD-E2F1,  DB-fos and AD-jun,  pPC97 (DB-empty) and pPC86 (AD-empty),  DB-DP1 and AD-E2F1,  DB-FRM-3 and AD-PKc_like.
HEK293T cells (ATCC CRL-3216) and Hela (ATCC CCL-2) were maintained in RPMI 1640 supplemented with 10% heat-inactivated foetal calf serum, 2 mM glutamine, 1X Non essential Amino acid, penicillin (50 U/ml) and streptomycin (50 U/ml) in a humidified 5% CO2 atmosphere at 37 °C. All culture media and supplements were supplied by Life technologies.
GST-ErlA fusion proteins (ElrA(27–473), ElrA(27–723) or ElrA(474–606)) were produced in E. coli ER266 and purified as previously described21. HEK293T cells were transfected with the p3flag-FHL1, p3flag-FHL2, p3flag-FHL3 and p3flag-FHL5 vectors using the jetPrime reagent (Polyplus) following the manufacturer’s instructions. The transfected cells were lysed 10 min on ice in PBS-1% Triton X100-Complete anti-protease cocktail (lysis buffer). Cell debris and nucleus were removed by centrifugation (at 12,000 g, 15 min). Then, the cell extracts (200–400 μg of proteins) were incubated overnight at 4 °C with the various GST-ElrA proteins bound to the beads in the lysis buffer. After incubation, the beads were washed six times in the lysis buffer supplemented with 0.5 M NaCl. The original lysates (inputs) and pulled-down proteins were analyzed by Western blotting using the monoclonal mouse anti-FLAG M2 (Sigma-Aldrich 1:2000) or the monoclonal mouse anti-GST (Sigma, 1:2000) as primary antibodies and the HRP-conjugated mouse anti-human IgG as secondary antibody. The Amersham ECLTM Prime Western blotting detection (GE Healthcare) was used for the revelation step.
Immunofluorescence and confocal microscopy
The FHL2 CDS was transferred by LR from the pDONR223 to the pCMVHA-GW vector21 generating the pCMVHA-FHL2 plasmid. Similarly, the myc-ElrA(27–723) and myc-ElrA(27–473) were produced from the pCMVmyc-ElrA vectors generated by LR with pCMVmyc-GW destination vector64. The transfection in HeLa cells were performed as described21. HeLa cells were grown on coverslips and fixed in 3% paraformaldehyde in PBS, for 15 min. Fixed cells were washed in PBS and permeabilized with 0.5% Triton for 5 min. Coverslips were incubated with primary antibodies in PBS 5% horse serum (i.e., mouse anti-c-myc IgG1κ (clone 9E10, Roche), or rat anti-HA IgG1 (3F10, Roche)) for 30 min at room temperature, washed in PBS 5% horse serum and then incubated with the secondary antibodies (i.e., donkey anti-mouse IgG conjugated to Alexa Fluor 488 (Invitrogen)), and goat anti-rat IgG conjugated to Alexa Fluor 594 (Invitrogen). Coverslips were mounted onto glass slides using Mowiol (Polysciences). Images were acquired using a Zeiss LSM510 laser scanning confocal microscope and analysed using the Zeiss Zen software (Mima2 facilities, Jouy-en-Josas France).
Construction of the E. faecalis elrAΔFID mutants
Molecular cloning techniques and gel electrophoresis were performed as described previously65. To delete FID domain of the elrA gene, upstream and downstream regions flanking the domain were amplified by PCR using E. faecalis OG1RF genomic DNA as template, and the following primer pairs: (i) FIDup1 (5′-TTAGTCCGTTAGGTCAAGC-3′) and FIDup2 (5′-TTCTTTGACAGGTTGGACG-3′); (ii) FIDdown1 (5′-CGTCCAACCTGTCAAAGAAAGTAGCACGACCCAATATC-3′) and FIDdown2 (5′-AATAGCCACGCCTAAAGTG-3′). The two fragments (803 and 815 pb) were purified and a second PCR using FIDup1 and FIDdown2 allowed the association of the two fragments in a single PCR product that was cloned in pGEM-T Easy (Promega Corporation) to generate pGEM-T-elrAΔFID. After EcoRI restriction of the pGEM-T-elrAΔFID, the ΔFIDelrA fragment was cloned into EcoRI-restricted pG+host966, generating pG+host9-elrAΔFID. The resulting plasmid was then introduced into OG1RF. An in-frame elrAΔFID mutant was selected as previously described10 and the deletion of the FID domain was confirmed by sequencing of the chromosomal locus. The procedure was repeated a second time with another pG+host9-elrAΔFID vector to obtain an independent elrAΔFID mutant. The mutants were named VE18355–3 and VE18355–4 (Table 2).
In vivo infections
The mouse urinary tract infections (UTIs) experiments were performed by following the protocol described by Lebreton et al.67. Briefly, groups of 15 female BALB/c mice (10 weeks old; Harlan Italy Srl, San Pietro al Natisone, Udine, Italy) were transurethrally challenged with 1 × 104 colony-forming units (CFUs) of wild-type OG1RF and its derivatives mutants ΔelrA and two independent elrAΔFID mutants (Table 2). 48 hours after infection, bladder and kidney pair cultures were prepared from mice that had been humanely killed to determine the recovered CFUs from organ homogenates.
We used a previously well-established systemic intravenous infection (SI) model68, the strains were cultured as in the UTI model. A 100 μL aliquot from each strain suspension (1 × 109 bacteria/mL) was injected into the tail vein of each of five female BALB/c mice. Seven days after infection, mice were sacrificed, and their organs (kidneys and livers) were removed aseptically, weighed, and homogenized, to determine the recovered CFUs. The experiments for each animal model were made in triplicate, and comparison among the organ counts was performed by use of an unpaired t test.
Mice were observed twice daily for morbidity and mortality, and any mice that exhibited noticeable morbidity (i.e., lethargy, scruffy coats) were euthanized immediately by CO2 inhalation. In addition, to minimize animal suffering and distress during the intraurethral catheterization, mice were anesthetized by isoflurane inhalation.
All methods were performed in accordance with the relevant guidelines and regulations.
The mouse experiments were approved by the Institutional Animal Use and Care Committee at the Università Cattolica del Sacro Cuore, Rome, Italy (permit number BB21, 04/07/2012), and authorized by the Italian Ministry of Health, according to the Legislative Decree 116/92, which implemented the European Directive 86/609/EEC on laboratory animal protection in Italy. Animal welfare was routinely checked by veterinarians of the Service for Animal Welfare.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Qin, J. et al. A human gut microbial gene catalogue established by metagenomic sequencing. Nature 464, 59–65, doi:10.1038/nature08821 (2010).
Bar, K., Wisplinghoff, H., Wenzel, R. P., Bearman, G. M. & Edmond, M. B. Systemic inflammatory response syndrome in adult patients with nosocomial bloodstream infections due to enterococci. BMC infectious diseases 6, 145, doi:10.1186/1471-2334-6-145 (2006).
Agudelo Higuita, N. I. & Huycke, M. M. In Enterococci: From Commensals to Leading Causes of Drug Resistant Infection (eds M. S. Gilmore, D. B. Clewell, Y. Ike & N. Shankar) (Massachusetts Eye and Ear Infirmary, 2014).
Arias, C. A. & Murray, B. E. The rise of the Enterococcus: beyond vancomycin resistance. Nature reviews. Microbiology 10, 266–278, doi:10.1038/nrmicro2761 (2012).
Hollenbeck, B. L. & Rice, L. B. Intrinsic and acquired resistance mechanisms in enterococcus. Virulence 3, 421–433, doi:10.4161/viru.21282 (2012).
Riboulet, E. et al. Relationships between oxidative stress response and virulence in Enterococcus faecalis. Journal of molecular microbiology and biotechnology 13, 140–146, doi:10.1159/000103605 (2007).
Fisher, K. & Phillips, C. The ecology, epidemiology and virulence of Enterococcus. Microbiology (Reading, England) 155, 1749–1757, doi:10.1099/mic.0.026385-0 (2009).
Rince, A. et al. Physiological and molecular aspects of bile salt response in Enterococcus faecalis. International journal of food microbiology 88, 207–213 (2003).
Torelli, R. et al. The PavA-like fibronectin-binding protein of Enterococcus faecalis, EfbA, is important for virulence in a mouse model of ascending urinary tract infection. The Journal of infectious diseases 206, 952–960, doi:10.1093/infdis/jis440 (2012).
Brinster, S. et al. Enterococcal leucine-rich repeat-containing protein involved in virulence and host inflammatory response. Infection and immunity 75, 4463–4471, doi:10.1128/iai.00279-07 (2007).
Schlievert, P. M. et al. Aggregation and binding substances enhance pathogenicity in rabbit models of Enterococcus faecalis endocarditis. Infection and immunity 66, 218–223 (1998).
Rich, R. L. et al. Ace is a collagen-binding MSCRAMM from Enterococcus faecalis. The Journal of biological chemistry 274, 26939–26945 (1999).
Schluter, S. et al. The high biofilm-encoding Bee locus: a second pilus gene cluster in Enterococcus faecalis? Current microbiology 59, 206–211, doi:10.1007/s00284-009-9422-y (2009).
Nallapareddy, S. R. et al. Conservation of Ebp-type pilus genes among Enterococci and demonstration of their role in adherence of Enterococcus faecalis to human platelets. Infection and immunity 79, 2911–2920, doi:10.1128/iai.00039-11 (2011).
Sillanpaa, J. et al. A family of fibrinogen-binding MSCRAMMs from Enterococcus faecalis. Microbiology (Reading, England) 155, 2390–2400, doi:10.1099/mic.0.027821-0 (2009).
Bhatty, M. et al. Enterococcus faecalis pCF10-encoded surface proteins PrgA, PrgB (aggregation substance) and PrgC contribute to plasmid transfer, biofilm formation and virulence. Molecular microbiology 95, 660–677, doi:10.1111/mmi.12893 (2015).
Sussmuth, S. D. et al. Aggregation substance promotes adherence, phagocytosis, and intracellular survival of Enterococcus faecalis within human macrophages and suppresses respiratory burst. Infection and immunity 68, 4900–4906 (2000).
Rozdzinski, E., Marre, R., Susa, M., Wirth, R. & Muscholl-Silberhorn, A. Aggregation substance-mediated adherence of Enterococcus faecalis to immobilized extracellular matrix proteins. Microbial pathogenesis 30, 211–220, doi:10.1006/mpat.2000.0429 (2001).
Nallapareddy, S. R., Qin, X., Weinstock, G. M., Hook, M. & Murray, B. E. Enterococcus faecalis adhesin, ace, mediates attachment to extracellular matrix proteins collagen type IV and laminin as well as collagen type I. Infection and immunity 68, 5218–5224 (2000).
Nielsen, H. V. et al. Pilin and sortase residues critical for endocarditis- and biofilm-associated pilus biogenesis in Enterococcus faecalis. Journal of bacteriology 195, 4484–4495, doi:10.1128/jb.00451-13 (2013).
de Barsy, M. et al. Identification of a Brucella spp. secreted effector specifically interacting with human small GTPase Rab2. Cellular microbiology 13, 1044–1058, doi:10.1111/j.1462-5822.2011.01601.x (2011).
Mukhtar, M. S. et al. Independently evolved virulence effectors converge onto hubs in a plant immune system network. Science (New York, N.Y.) 333, 596–601, doi:10.1126/science.1203659 (2011).
Calderwood, M. A. et al. Epstein-Barr virus and virus human protein interaction maps. Proceedings of the National Academy of Sciences of the United States of America 104, 7606–7611, doi:10.1073/pnas.0702332104 (2007).
de Chassey, B. et al. Hepatitis C virus infection protein network. Molecular systems biology 4, 230, doi:10.1038/msb.2008.66 (2008).
Simonis, N. et al. Host-pathogen interactome mapping for HTLV-1 and −2 retroviruses. Retrovirology 9, 26, doi:10.1186/1742-4690-9-26 (2012).
Mairiang, D. et al. Identification of new protein interactions between dengue fever virus and its hosts, human and mosquito. PloS one 8, e53535, doi:10.1371/journal.pone.0053535 (2013).
Bierne, H. & Cossart, P. Listeria monocytogenes surface proteins: from genome predictions to function. Microbiology and molecular biology reviews: MMBR 71, 377–397, doi:10.1128/mmbr.00039-06 (2007).
Pizarro-Cerda, J., Kuhbacher, A. & Cossart, P. Entry of Listeria monocytogenes in mammalian epithelial cells: an updated view. Cold Spring Harbor perspectives in medicine 2, doi:10.1101/cshperspect.a010009 (2012).
Yamaguchi, M., Terao, Y. & Kawabata, S. Pleiotropic virulence factor - Streptococcus pyogenes fibronectin-binding proteins. Cellular microbiology 15, 503–511, doi:10.1111/cmi.12083 (2013).
Keskin, O., Gursoy, A., Ma, B. & Nussinov, R. Principles of protein-protein interactions: what are the preferred ways for proteins to interact? Chemical reviews 108, 1225–1244, doi:10.1021/cr040409x (2008).
Stynen, B., Tournu, H., Tavernier, J. & Van Dijck, P. Diversity in genetic in vivo methods for protein-protein interaction studies: from the yeast two-hybrid system to the mammalian split-luciferase system. Microbiology and molecular biology reviews: MMBR 76, 331–382, doi:10.1128/mmbr.05021-11 (2012).
Voth, D. E. ThANKs for the repeat: Intracellular pathogens exploit a common eukaryotic domain. Cellular logistics 1, 128–132, doi:10.4161/cl.1.4.18738 (2011).
Rual, J. F. et al. Towards a proteome-scale map of the human protein-protein interaction network. Nature 437, 1173–1178, doi:10.1038/nature04209 (2005).
Dobrijevic, D. et al. High-throughput system for the presentation of secreted and surface-exposed proteins from Gram-positive bacteria in functional metagenomics studies. PloS one 8, e65956, doi:10.1371/journal.pone.0065956 (2013).
Rual, J. F. et al. Human ORFeome version 1.1: a platform for reverse proteomics. Genome research 14, 2128–2135, doi:10.1101/gr.2973604 (2004).
Lamesch, P. et al. hORFeome v3.1: a resource of human open reading frames representing over 10,000 human genes. Genomics 89, 307–315, doi:10.1016/j.ygeno.2006.11.012 (2007).
Shathasivam, T., Kislinger, T. & Gramolini, A. O. Genes, proteins and complexes: the multifaceted nature of FHL family proteins in diverse tissues. Journal of cellular and molecular medicine 14, 2702–2720, doi:10.1111/j.1582-4934.2010.01176.x (2010).
Morgan, M. J. & Madgwick, A. J. The LIM proteins FHL1 and FHL3 are expressed differently in skeletal muscle. Biochemical and biophysical research communications 255, 245–250, doi:10.1006/bbrc.1999.0179 (1999).
Hicks, S. W. & Galan, J. E. Exploitation of eukaryotic subcellular targeting mechanisms by bacterial effectors. Nature reviews. Microbiology 11, 316–326, doi:10.1038/nrmicro3009 (2013).
Walsh, C. Posttranslational Modification of Proteins: Expanding Nature’s Inventory: Roberts and Compagny Publishers. (2006).
Rigaut, G. et al. A generic protein purification method for protein complex characterization and proteome exploration. Nature biotechnology 17, 1030–1032, doi:10.1038/13732 (1999).
Dotreppe, D., Mullier, C., Letesson, J. J. & De Bolle, X. The alkylation response protein AidB is localized at the new poles and constriction sites in Brucella abortus. BMC microbiology 11, 257, doi:10.1186/1471-2180-11-257 (2011).
Cortes-Perez, N. G. et al. Overexpression of Enterococcus faecalis elr operon protects from phagocytosis. BMC microbiology 15, 112, doi:10.1186/s12866-015-0448-y (2015).
Kadrmas, J. L. & Beckerle, M. C. The LIM domain: from the cytoskeleton to the nucleus. Nature reviews. Molecular cell biology 5, 920–931, doi:10.1038/nrm1499 (2004).
McCabe, A., Hashimoto, K., Hall, W. W. & Sheehy, N. The four and a half LIM family members are novel interactants of the human T-cell leukemia virus type 1 Tax oncoprotein. Journal of virology 87, 7435–7444, doi:10.1128/jvi.00070-13 (2013).
Kleiber, K., Strebhardt, K. & Martin, B. T. The biological relevance of FHL2 in tumour cells and its role as a putative cancer target. Anticancer research 27, 55–61 (2007).
Dahan, J. et al. LIM-only protein FHL2 activates NF-kappaB signaling in the control of liver regeneration and hepatocarcinogenesis. Molecular and cellular biology 33, 3299–3308, doi:10.1128/mcb.00105-13 (2013).
Johannessen, M., Moller, S., Hansen, T., Moens, U. & Van Ghelue, M. The multifunctional roles of the four-and-a-half-LIM only protein FHL2. Cellular and molecular life sciences: CMLS 63, 268–284, doi:10.1007/s00018-005-5438-z (2006).
Horsley, H. et al. Enterococcus faecalis subverts and invades the host urothelium in patients with chronic urinary tract infection. PloS one 8, e83637, doi:10.1371/journal.pone.0083637 (2013).
Baldassarri, L. et al. Glycosaminoglycans mediate invasion and survival of Enterococcus faecalis into macrophages. The Journal of infectious diseases 191, 1253–1262, doi:10.1086/428778 (2005).
Gentry-Weeks, C. R., Karkhoff-Schweizer, R., Pikis, A., Estay, M. & Keith, J. M. Survival of Enterococcus faecalis in mouse peritoneal macrophages. Infection and immunity 67, 2160–2165 (1999).
Zou, J. & Shankar, N. The opportunistic pathogen Enterococcus faecalis resists phagosome acidification and autophagy to promote intracellular survival in macrophages. Cellular microbiology 18, 831–843, doi:10.1111/cmi.12556 (2016).
Zou, J. & Shankar, N. Enterococcus faecalis infection activates phosphatidylinositol 3-kinase signaling to block apoptotic cell death in macrophages. Infection and immunity 82, 5132–5142, doi:10.1128/iai.02426-14 (2014).
LaRock, D. L., Chaudhary, A. & Miller, S. I. Salmonellae interactions with host processes. Nature reviews. Microbiology 13, 191–205, doi:10.1038/nrmicro3420 (2015).
Dortet, L., Mostowy, S. & Cossart, P. Listeria and autophagy escape: involvement of InlK, an internalin-like protein. Autophagy 8, 132–134, doi:10.4161/auto.8.1.18218 (2012).
Dortet, L. et al. Recruitment of the major vault protein by InlK: a Listeria monocytogenes strategy to avoid autophagy. PLoS pathogens 7, e1002168, doi:10.1371/journal.ppat.1002168 (2011).
Tatusova, T., Ciufo, S., Fedorov, B., O’Neill, K. & Tolstoy, I. RefSeq microbial genomes database: new representation and annotation strategy. Nucleic acids research 42, D553–559, doi:10.1093/nar/gkt1274 (2014).
Liu, J. Y., Li, A. H., Ji, C. & Yang, W. M. First description of a novel Weissella species as an opportunistic pathogen for rainbow trout Oncorhynchus mykiss (Walbaum) in China. Veterinary microbiology 136, 314–320, doi:10.1016/j.vetmic.2008.11.027 (2009).
Uhlen, M. et al. Proteomics. Tissue-based map of the human proteome. Science (New York, N.Y.) 347, 1260419, doi:10.1126/science.1260419 (2015).
Vidalain, P. O., Boxem, M., Ge, H., Li, S. & Vidal, M. Increasing specificity in high-throughput yeast two-hybrid experiments. Methods (San Diego, Calif.) 32, 363–370, doi:10.1016/j.ymeth.2003.10.001 (2004).
Braun, P. et al. An experimentally derived confidence score for binary protein-protein interactions. Nature methods 6, 91–97, doi:10.1038/nmeth.1281 (2009).
Yu, H. et al. High-quality binary protein interaction map of the yeast interactome network. Science (New York, N.Y.) 322, 104–110, doi:10.1126/science.1158684 (2008).
Walhout, A. J. & Vidal, M. High-throughput yeast two-hybrid assays for large-scale protein interaction mapping. Methods (San Diego, Calif.) 24, 297–306, doi:10.1006/meth.2001.1190 (2001).
Henry, T. et al. The Salmonella effector protein PipB2 is a linker for kinesin-1. Proceedings of the National Academy of Sciences of the United States of America 103, 13497–13502, doi:10.1073/pnas.0605443103 (2006).
Sambrook, J. & Russell, D. W. Molecular cloning: a laboratory manual. 3rd ed. ed. Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory (2001).
Maguin, E., Duwat, P., Hege, T., Ehrlich, D. & Gruss, A. New thermosensitive plasmid for gram-positive bacteria. Journal of bacteriology 174, 5633–5638 (1992).
Lebreton, F. et al. ace, Which encodes an adhesin in Enterococcus faecalis, is regulated by Ers and is involved in virulence. Infection and immunity 77, 2832–2839, doi:10.1128/iai.01218-08 (2009).
Michaux, C. et al. SlyA is a transcriptional regulator involved in the virulence of Enterococcus faecalis. Infection and immunity 79, 2638–2645, doi:10.1128/iai.01132-10 (2011).
Nallapareddy, S. R. et al. Endocarditis and biofilm-associated pili of Enterococcus faecalis. The Journal of clinical investigation 116, 2799–2807, doi:10.1172/jci29021 (2006).
Singh, K. V., La Rosa, S. L., Somarajan, S. R., Roh, J. H. & Murray, B. E. The fibronectin-binding protein EfbA contributes to pathogenesis and protects against infective endocarditis caused by Enterococcus faecalis. Infection and immunity 83, 4487–4494, doi:10.1128/iai.00884-15 (2015).
Nallapareddy, S. R., Singh, K. V., Duh, R. W., Weinstock, G. M. & Murray, B. E. Diversity of ace, a gene encoding a microbial surface component recognizing adhesive matrix molecules, from different strains of Enterococcus faecalis and evidence for production of Ace during human infections. Infection and immunity 68, 5210–5217 (2000).
Creti, R. et al. Surface protein EF3314 contributes to virulence properties of Enterococcus faecalis. The International journal of artificial organs 32, 611–620 (2009).
Sahm, D. F. et al. In vitro susceptibility studies of vancomycin-resistant Enterococcus faecalis. Antimicrobial agents and chemotherapy 33, 1588–1591 (1989).
Dunny, G. M., Brown, B. L. & Clewell, D. B. Induced cell aggregation and mating in Streptococcus faecalis: evidence for a bacterial sex pheromone. Proceedings of the National Academy of Sciences of the United States of America 75, 3479–3483 (1978).
We thank the confocal microscope platforms MIMA2 (Microscopie et Imagerie des Micro-organismes, Animaux et Aliments). P. Serror thanks the French-Brazilian USP COFECUB project Uc Me 147/13.
Electronic supplementary material
About this article
Scientific Reports (2018)