Identification and Characterization of Fusolisin, the Fusobacterium nucleatum Autotransporter Serine Protease

Fusobacterium nucleatum is an oral anaerobe associated with periodontal disease, adverse pregnancy outcomes and colorectal carcinoma. A serine endopeptidase of 61–65 kDa capable of damaging host tissue and of inactivating immune effectors was detected previously in F. nucleatum. Here we describe the identification of this serine protease, named fusolisin, in three oral F. nucleatum sub-species. Gel zymogram revealed fusobacterial proteolytic activity with molecular masses ranging from 55–101 kDa. All of the detected proteases were inhibited by the serine protease inhibitor PMSF. analysis revealed that all of the detected proteases are encoded by genes encoding an open reading frame (ORF) with a calculated mass of approximately 115 kDa. Bioinformatics analysis of the identified ORFs demonstrated that they consist of three domains characteristic of autotransporters of the type Va secretion system. Our results suggest that the F. nucleatum fusolisins are derived from a precursor of approximately 115 kDa. After crossing the cytoplasmic membrane and cleavage of the leader sequence, the C-terminal autotransporter domain of the remaining 96–113 kDa protein is embedded in the outer membrane and delivers the N-terminal S8 serine protease passenger domain to the outer cell surface. In most strains the N-terminal catalytic 55–65 kDa domain self cleaves and liberates itself from the autotransporter domain after its transfer across the outer cell membrane. In F. nucleatum ATCC 25586 this autocatalytic activity is less efficient resulting in a full length membrane-anchored serine protease. The mature serine protease was found to cleave after Thr, Gly, Ala and Leu residues at the P1 position. Growth of F. nucleatum in complex medium was inhibited when serine protease inhibitors were used. Additional experiments are needed to determine whether fusolisin might be used as a target for controlling fusobacterial infections.


Introduction
Fusobacterium nucleatum is a ubiquitous oral anaerobic rod classified into five subspecies nucleatum, polymorphum, vincentii, animalis, and fusiforme [1]. Development of periodontal disease has been correlated with a sharp increase in the numbers of F. nucleatum [2,3]. F. nucleatum has a remarkable ability to attach to a range of early and late colonizing oral species [4,5,6,7,8,9] in a process termed coaggregation or coadherence, and has therefore been suggested as a bridging organism that contributes to the structural formation of the multi-species dental biofilm [6,10].
Virulence mechanisms of F. nucleatum include adhesion to and invasion of host cells [11] and induction of proinflammatory cytokines [12,13]. F. nucleatum is also the periopathogen most commonly found in systemic infections [2]. It is strongly implicated in preterm deliveries [14,15], and was also found to be dominant in the microenvironment of colorectal carcinoma [16,17] and to promote its acceleration [18,19].
Bacterial pathogens have developed strategies to enable their survival and growth within their specific hosts. Surface and secreted proteases are common virulence factors employed by microorganisms for colonization of new sites within the host, acquisition of growth nutrients and evasion of the host defenses [20]. Serine proteases are the most abundant and functionally diverse group of proteolytic enzymes in eukaryotic and prokaryotic organisms [21]. A family of extracellular serine proteases secreted through the Type V autotransporter secretion pathway, has been described in pathogenic Gram-negative species of Neisseria, Shigella, Escherichia coli, Citrobacter rodentium, Salmonella and Edwarsiella species [22]. These bacterial serine proteases hydrolyze host intracellular and extracellular protein substrates leading to cytoskeleton destruction [23,24], induction of autophagy [25,26] or impaired immunity [27].
Oral bacteria found in the subgingival plaque are predominantly anaerobic and rely on the utilization of peptides and amino acids for energy [28,29]. The proteases of these oral microorganisms are implicated in the degradation of host periodontal tissues while supplying the bacteria's nutritional requirements [30,31].
Amino acids and peptides are the preferred substrates for F. nucleatum's growth [32,33,34] and growth of fusobacteria depends on the availability of free glutamate, histidine, serine and lysine [35]. Under natural conditions, the above amino acids are not found in free form but are incorporated in proteins that have to be degraded for the desired amino acids to become accessible.
Previous studies reported a fusobacterial serine protease activity associated with a molecular mass of 65 kDa [36,37,38,39]. This protease was shown to be capable of degrading components of periodontal tissues, and to inactivate host defense effectors [39]. The aim of this study was to identify and characterize the F. nucleatum 65 kDa serine protease which we named fusolisin.

Culture supernatant and outer membrane vesicle preparation
Four-day-old F. nucleatum cultures were harvested by centrifugation at 10,0006g for 20 min at 4uC. Culture supernatants were collected and filtered through a 0.2 mm filter (Whatman Schleicher & Schuell, Germany). Supernatants were either concentrated610 using a Centricon microconcentrator (50,000molecular-weight cutoff; Amicon) or used for outer membrane vesicle preparation.
For vesicle preparation cell-free supernatants were centrifuged at 100,0006g for 2 hrs. The supernatant was discarded and the pellet containing the vesicles was washed twice with TBS by centrifugation at 100,0006g. The pellet was stored at 220uC until further use.

Gel Electrophoresis
For zymogram analysis, samples were dissolved at room temperature in sample buffer (192 mM Tris-HCl [pH 6.8], 30% glycerol, 9% SDS) without b-mercaptoethanol and subjected to SDS-PAGE using 7.5% gels containing 240 mg/ml human fibrinogen (Sigma-Aldrich, Germany). Following electrophoresis, the gels were incubated for 30 min at room temperature in Trisbuffered saline (TBS, 0.05 M Tris-HCl [pH 7.8], 0.1 M NaCl), containing 2% Triton X-100 and then washed three times with TBS. Gels were incubated overnight at 37uC. Proteolytic activity was visualized as a clear band against a blue background after staining with Coomassie brilliant blue R-250 as described before [39].
For denaturing SDS-PAGE, samples were dissolved, boiled at 100uC for 5 min in sample buffer containing 2% b-mercaptoethanol and the gels were stained with Coomassie brilliant blue. Molecular masses of protein bands were calculated by linear regression analysis of molecular mass standards.

Mass spectrometry (MS) identification and database searching
Bands were excised from denaturizing gels and subjected to Qtof2 (Micromass, Manchester, UK) equipped with a nanospray capillary [40], analyzed by electrospray ionization tandem mass spectrometry (ESI-MS/MS) and peptides were identified as described before [40,41].

DNA isolation
Chromosomal DNA was isolated from F. nucleatum ATCC 25586 using the mini GenElute Bacterial Genomic DNA kit (Sigma-Aldrich, Germany) according to the manufacturer's instructions. Plasmid DNA was isolated using the Qiagen spin miniprep kit (Qiagen, Germany).

Expression of Fsp25586 in F. nucleatum ATCC 23726
The DNA fragment containing fsp25586 and 556 bp of its upstream region was amplified using the F-25586-SP90  and R-25586-SP90 (59-CCgagctcACTAGTGTTAGTGACGCAA-39) primers that include a SacI restriction site (small case letters). The 3.9 kb PCR product was restricted with SacI (New England Biolabs Inc. USA), and inserted into the SacI site of the pHS30 E. coli-F. nucleatum shuttle vector [42,43] to generate pHSPROT. Plasmid electroporation into F. nucleatum ATCC 23726 was performed as described previously [43]. Clones were selected on Columbia agar plates supplemented with 5% sheep blood (Hylabs, Israel) and 5 mg/ml thiamphenicol (Sigma-Alderich, Germany).  Sequencing of FN1426 DNA was isolated from F. nucleatum ATCC 25586 as described above. The following primers were used to amplify and sequence the FN1426 gene: The 16S rRNA gene of F. nucleatum FDC 364 was amplified by PCR using universal primers 4F (CCA GAG TTT GAT YMT GGC) and 1541R (GAA GGA GGT GWT CCA DCC). The resulting product was sequenced (gene bank accession number KM023647) and blasted against the National Center for     When added, PMSF and AEBSF were used at a final concentration of 1 mM and 2 mM respectively. P. gingivalis supernatants were prepared by centrifugation of four day cultures at 10,0006g for 10 min at 4uC., collection of the supernatants and filtration through a 0.2 mm filter (Whatman Schleicher & Schuell, Germany). When added, P. gingivalis supernatants were diluted 1:10 in the tested reaction. The final reaction contained 180 ml of diluted bacteria in a total volume of 220 ml. Bacterial growth (anaerobic for F. nucleatum and aerobic for E. coli) was monitored using microplate real-time kinetic measurements as described by us in detail previously [44]. Results represent mean and standard deviation of triplicate of an independent representative experiment repeated three times.

Identification of the fusolisin restriction site
Fusolisin was purified from extracellular vesicles by preparative SDS-PAGE followed by electroelution as described before [39]. Briefly, four-day-old F. nucleatum cultures were sedimented by centrifugation at 90006g for 20 min. The supernatant was collected and filtered through a 0.2 mm filter. The filtrate was aliquoted and outer membrane vesicles were sedimented by centrifugation at 100,0006g for 2 h. The supernatant was discarded, and the precipitate containing the extracellular vesicles was washed twice with 50 mM Tris-HCl (pH 7.8) by centrifuga-tion as described above. For fusolisin purification, the vesicles were subjected to electroelution after separation by SDS-PAGE as follows: the extracellular vesicles were dissolved in sample buffer (without b-mercaptoethanol, see above), centrifuged for 2 min at 10,0006g and submitted to SDS-PAGE (7.5% acrylamide). The protease was electroeluted from the gel using a Bio Trap 1000 electroeluter (Schleicher and Schuell, Germany) with Tris-glycine buffer (25-192 mM) without SDS for 2 h at 200 volts followed by 10 h at 100 volts. The fusolisin enzyme was then stored at -20uC.
Identification of the fusolisin substrate specificity was determined by hydrolysis of fibrinogen and identification of the resulting peptides by mass spectrometry. The purified enzyme (0.25 mg) was incubated with 2.5 mg of fibrinogen in 40 ml TBS pH 8.0 at 37uC for 16 h. A similar reaction mixture with heat inactivated protease (3 min at 100uC) served as control.
The reaction mixture of fusolisin-mediated hydrolysis of fibrinogen was submitted to peptide mapping after N-terminus labeling by reductive dimethylation as follows: the protein sample in 8 M Urea and 50 mM Hepes (pH 8) was modified with 20 mM formaldehyde in the presence of 100 mM NaCBH 3 (60uC for 15 min). Neutralization was performed with 500 mM ammonium bicarbonate (final concentration). The protein sample was reduced with 2.8 mM DTT (60uC for 30 min), modified with 9.4 mM iodoacetamide in 100 mM ammonium bicarbonate (room temperature for 30 min in the dark) diluted 4 fold and digested with modified trypsin (Promega) overnight at 37uC in a 1:50 enzymeto-substrate ratio.
The resulting peptides were desalted on a stage tip (C18) and resolved by reverse-phase chromatography on 0.0756200-mm fused silica capillaries (J&W) packed with Reprosil reversed phase Cleavage of FRETS-25 Thr was analyzed as described above but without the N-terminus labeling.
Fusolisin's restriction specificity was verified using the FRET substrate CPQ2-Gly-Phe-Ile-Thr-Ala-Phe-Pro-Lys-(5FAM)-Arg-Arg-NH2 that was custom synthesized by CPC scientific (Sunnyvale, CA, USA). The peptide was dissolved in DMSO to a concentration of 1 mM and further diluted with TBS to the desired concentration. The reaction was performed and monitored as described above with the lex = 485 nm and the lem = 535 nm.

Bioinformatics analysis
Public databases were searched for similar sequences with the BLASTN, BLASTP, and BLASTP/PSI algorithms using default parameters. The features of the predicted proteins were examined by the Pfam programs (http://www.sanger.ac.uk/Software/ Pfam/search.shtml). The ExPASy server was used to predict the proteins' molecular weights [45]. Multiple alignment was performed using CLUSTAL W [46,47]. Structure prediction was generated using the Protein Homology/analogY Recognition Engine (Phyre) [48]. Rare Codon Caltor (http://www.doe-mbi. ucla.edu/,sumchan/caltor.html) was used before cloning in E. coli.

Identification of fusolisin
Gel zymograms using human fibrinogen as a substrate, revealed proteolytic activity in the growth media supernatant (Fig. 1) and in outer membrane vesicles prepared from all of the tested F. nucleatum strains that represent three F. nucleatum subspecies: nucleatum, polymorphum and vincentii ( Table 1). The molecular weight of the detected proteases varied from 55 to 101 kDa as estimated by gel migration (Fig. 1, Table 1). All the detected proteases were inhibited by the serine protease inhibitor PMSF (presented for F. nucleatum strains FDC 364, ATCC 25586, 12230 and ATCC 23726 in Fig. 2).
Gel-purified proteases of outer membrane vesicles prepared from the genome-sequenced F. nucleatum strains ATCC 25586 and ATCC 49256 were identified using mass spectrometry (MS). Tryptic fragments of the 99 kDa proteolytic protein of F. nucleatum ATCC 25586 matched those of the entire putative 115 kDa serine proteases designated FN1426 [Genbank Index number (GI):19704758]. Tryptic peptides of both the 55 kDa and 101 kDa serine endopeptidases partially purified from F. nucleatum ATCC 49256 were found to match those of the putative 108 kDa serine protease designated FNV0835 (GI:34763535). However, while the peptide sequences generated from the 99 kDa proteolytic protein extracted from F. nucleatum ATCC 25586 corresponded to the entire FN1426 protein sequence (Fig. 3A), those generated from the 55 kDa proteolytic band of F. nucleatum ATCC 49256 matched only the N-terminal domain of FNV0835 (Fig. 3B) suggesting that the 55 kDa protease of F. nucleatum ATCC 49256 also originated from a larger precursor. The 62 kDa protease of the non-sequenced strain FDC 364 [39] was found to be most homologous to the N-terminal domain of FN1426 of F. nucleatum ATCC 25586 (data not shown).

Sequence analysis of fusolisin
Previous annotation of the FN1426 sequence of F. nucleatum ATCC 25586 [49], showed that the start codon proposed earlier by Kapatral and colleagues [50] has been misannotated. Upon detailed analysis, we found that the reading frame proposed by Kapatral and colleagues [50] was indeed truncated. However, we found that the reading frame extends even further than corrected by Desvaux and colleagues [49] which is missing 291 base pairs at the beginning of the gene. The FN1426 gene bank sequence begins with a TTG codon (an alternate start codon in bacteria), our proposed new open reading frame begins with ATG which is located 291 base pairs upstream. Furthermore, analysis of the sequence taken from the gene bank using the SignalP 4.0 server [51], found no signal peptide while analysis of our proposed sequence revealed a putative signal peptide cleavage site between amino acids 58 and 59. This observation was verified by PCR and sequencing of the proposed ORF and the 400 bases upstream to it. The corrected predicted 1,058 amino acid ORF, was named Fsp25586 (for Fusobacterial Serine Protease of F. nucleatum 25586) and was deposited in the gene bank accession number KJ634469. Henceforth, all of the following analysis will refer to the new sequence of Fsp25586. The orthologous sequences FNV0835 will be referred to as Fsp49256, FNP_2077 will be referred to as Fsp10953 and HMPREF0397_0469 will be referred to as Fsp23726.
Amino acid sequence alignment (Fig. 4) of Fsp25586 revealed a high homology (71% similarity and 61% identity) with that of Fsp49256, 71% similarity and 60% identity with that of Fsp10953 and 63% similarity and 57% identity with the available partial sequence of the homologous serine protease Fsp23726. Previous annotation of the FN1426 (Fsp25586) and the FNV0835 (Fsp49256) open reading frames revealed a signal peptide and three other functional domains [49]. The N-terminal, peptidase domain [amino acids 131-471 in Fsp25586 and 1-406 in Fsp49256] were found to belong to the peptidase S8 domain family. The C-terminal domain (amino acids 788-1047 in Fsp25586 and 690-995 in Fsp49256) belong to the autotransporter superfamily. While the C-terminal autotransporter domain of Fsp25586 and Fsp49256 were highly conserved (93% identity and 98% similarity), a higher divergence was found between the catalytic domains of the proteases of the two species (37% identity, 47% similarity). As a member of the S8 family of subtilisins, the amino acid sequence analysis of fusolisin revealed that the arrangement of the active site catalytic triad is Asp-His-Ser [52,53] that was identified using NCBI's conserved domain database (CDD) [54] in the amino acid sequences of Fsp49256, Fsp10953, Fsp23726 and Fsp25586, and can be seen in Fig. 4.

Processing of fusobacterial serine protease
Autocatalytic processing is common in type Va secretion systems [55,56,57] and in subtilisins [58]. Although F. nucleatum ATCC 25586 remains refractory to plasmid transformation, others and us were previously successful with plasmid expression in F. nucleatum ATCC 23726 [43,59]. As can be seen in figures 1E and 5A, the serine protease detected in the growth medium of F. nucleatum ATCC 23726 (Fsp23726) is approximately 56 kDa. Zymogram analysis of culture supernatants prepared from F. nucleatum ATCC 23726 expressing Fsp25586 of strain ATCC 25586 revealed the presence of the 99 kDa Fsp25586 protease in addition to the typical 56 kDa protease of ATCC 23726 (Fig. 5B). The fact that Fsp25586 was not cleaved when expressed in F. nucleatum ATCC 23726 suggests that the processing of Fsp25586 is not efficient compared to that in Fsp23726 (and the orthologs in the other tested F. nucleatum strains, Table 1). It is possible that Fsp25586 lacks the restriction site that is cleaved to release the catalytic domain from the autotransporter domain, or that this cleavage site is not exposed for cleavage.

Fusolisin's cleavage site
Previous work in our laboratory failed to determine the substrate specificity of the fusobacterial proteolytic activity using a large variety of synthetic chromogenic substrates [39]. We now determined the cleavage sites of fusolisin by MS analysis of the peptides resulting from hydrolysis of fibrinogen. Fusolisin cleaved fibrinogen preferentially at the C-terminal side of small residues (Thr, Gly, Ala and Ser), though Leu and Asp were also cleaved ( Table 2). However, the major peak of fusolisin mediated fibrinogen hydrolysis resulted from cleavage of Thr at the P1 position. To verify that fusolisin preferentially cleaves Thr at the P1 position, hydrolysis of the FRETS-25 Thr substrate (PeptaNova) by fusolisin was examined. FRETS-25 Thr, is a protease substrate library that contains a highly fluorescent 2-(N-methylamino)benzoyl (Nma) group linked to the side chain of the amino-terminal D-2,3-diamino propionic acid (D-A2pr) residue, along with a 2,4dinitrophenyl (Dnp) group that acts as a quencher, linked to the eamino group of a Lys residue. The fluorophore and quencher are connected by the peptide Gly-[Phe/Ala/Val/Glu/Arg] -[Pro/ Tyr/Lys/Ile/Asp]-Thr-Ala-Phe-Pro. Mass spectrometry analysis of the Fsp23726-mediated FRETS-25 Thr hydrolysis products revealed that similar to fibrinogen cleavage, the major peaks obtained resulted from Thr in the P1 position (Table 2B). Preference for the presence of Ile in the P2 position was also observed (Table 2B).
Fu-S-P was found to be a useful biomarker for detecting F. nucleatum ATCC 25586 (not shown) and ATCC 23726. As shown in Fig. 7, 5610 5 -1610 6 fusobacterial cells were sufficient to produce measurable activity after 2 hrs.
Fusolisin is essential for growth of F. nucleatum in a complex medium The genes coding for Fsp25586 and Fsp49256 were located in genomic loci involved in metabolic functions suggesting a nutritional role for fusolisin. Indeed, growth of F. nucleatum in a complex medium was inhibited by the AEBSF (not shown) and PMSF serine protease inhibitors (Fig. 8). Both inhibitors did not affect growth of E. coli used as control (Fig. 8), ruling out that growth attenuation of F. nucleatum by both serine-protease inhibitors resulted from non specific toxicity. P. gingivalis is a proteolytic anaerobic periodontal pathogen frequently isolated together with F. nucleatum [3]. The gingipain proteases produced by P. gingivalis are cysteine proteases which are not inhibited by PMSF. Addition of filter-sterilized gingipain-containing supernatant collected from a P. gingivalis culture [60], relieved the PMSF inhibitory effect on F. nucleatum's growth (Fig. 8). Addition of the P. gingivalis supernatant did not affect the pH of the reaction mixture and did not reduce the inhibitory activity of PMSF as tested on trypsin under similar conditions (data not shown). These results suggest that P. gingivalis can enable fusolisin-independent growth of F. nucleatum.

Discussion
Obtaining energy by the fermentation of a small number of peptide-derived amino acids was shown to be essential for the growth of F. nucleatum [32,34,61]. To date, the only detected endopeptidase activity in F. nucleatum was that of an un-identified serine protease with a molecular weight of 61-65 kDa [36,37,38,39]. In the present work this protease, now named fusolisin, has been identified and characterized at the genetic level. The theoretic isoelectric point of the 55 kDa derivative of Fsp49256 is 5. This calculated isoelectric point is in agreement with that (pH 5) determined for the 65 kDa diisopropylfluorophosphate-binding outer membrane protein of F. nucleatum Fev1 [37]. Mass spectrometry analysis and identification of the high (99 kDa) and the low (55 kDa) molecular weight F. nucleatum serine proteases partially purified from strains ATCC 25586 and ATCC 49256 demonstrated (Fig. 3) that both originate from a precursor with a calculated molecular weight of approximately 115 kDa. Sequence analysis suggested that this precursor contains a signal peptide, a serine protease domain, a linker, and an autotransporter domain. Structure prediction of the serine protease and of the autotransporter domains was generated using the Protein Homology/analogY Recognition Engine (Phyre) [48] and can be seen in Figure 9 and in Movies S1 and S2.
Our experimental results and in silico predictions suggest that the fusobacterial serine protease belongs to the autotransporter proteins superfamily of the type Va secretion pathway [55,56,57]. This conclusion is in agreement with a previous in silico analysis of the F. nucleatum genome [49].
Our bioinformatics results (Fig. 4) suggest that the 101-55 kDa fusolisin proteases (Fig. 1, Table 1) all derive from a precursor of approximately 115 kDa. This precursor crosses the cytoplasmic membrane presumably via the Sec (Secretion) pathway where the signal peptide is removed. The C-terminal domain of the remaining approximately 96-101 kDa poly-peptide then forms a b-barrel pore structure in the outer bacterial membrane (Fig. 9A, Movie S2). The N-terminal serine endopeptidase domain is then most likely threaded through the autotransporter and transported across the outer membrane to the cell's exterior surface. An alternative model suggests that several autotransporter domains oligomerize and form a wide channel that enables the transfer of the catalytic passenger domain [56].
In the case of Fsp25586, the protease remains intact (99 kDa, Table 1, Fig. 1). In the case of Fsp49256 and its homologs in strains ATCC 10953, FDC 364 and ATCC 23726 the protease can remain intact and cell bound, or the catalytic domain can self cleave the peptide bridge connecting both domains and release itself to form the detected 55-62 kDa protease (Table 1, Fig. 1). Mass spectrometry analysis of the self cleavage product of Fsp49256 suggests that the restriction site is located after amino acid 572 (Fig. 3A). Furthermore, the amino acid sequence GYIT (amino acids 578 to 581 in Fig. 3B) is present in the peptide bridge connecting the catalytic domain to the autotransporter in Fsp49256 and absent in the peptide bridge in Fsp25586 (Fig. 3A). This sequence is similar to the fusolisin P4-P1 cleavage site GFIT in Fu-S-P, indicating that GYIT may be the self cleavage restriction site of Fsp49256.
The 55-62 kDa form of fusolisin was found in the growth media but also in outer membrane vesicles (Table 1). This indicates that the autotransporter domain is not essential for adherence of the catalytic domain to the bacteria's outer membrane surface. Hydrophobic interactions between hydrophobic sub-domains (not shown) in the catalytic domain and the membrane, or noncovalent interactions between the catalytic and autotransporter domains can enable such association [62]. In strains ATCC 23726, ATCC 10953 and FDC 364 the full-length (,96 kDa) protease was detected mainly in outer membrane vesicle preparations but not in growth media where the 56-62 kDa protease was found. This indicates that unlike the secreted mature 55-62 kDa protease, the full length protease in these strains is mostly membrane bound. The release of the mature serine protease that can act on extracellular targets [39] can enable its diffusion to distant locations and presumably increase its effectiveness [63]. Regulation of cleavage and detachment of the passenger from the autotransporter domain has been demonstrated in IcsA of Shigella flexneri [64], and the Hap autotransporter of Haemophilus influenzae [65].
The self cleavage mechanism that determines F. nucleatum serine protease secretion is yet unknown. Possible regulation of this secretion mechanism by the bacteria (in response to nutritional needs or proteinaceous host defense challenges) remains to be determined.
The fusolisin's autotransporter domain was found to be much more conserved than the catalytic one (Fig. 4). This can result from strict structural-functional requirements. An additional possibility is that being extracellular, the catalytic domain is under much greater immunological stress than the non-exposed intramembrane autotransporter domain. This constant immunological selective pressure might accelerate alterations in the catalytic domain's amino acid sequence.
Previous attempts to find the substrate specificity of fusolisin using a large number of chromogenic substrates have failed [39]. In the present study we synthesized a fusolisin substrate that allows the detection of low numbers of fusobacterial cells (Fig. 7). Such a substrate might be made useful for detection of fusobacteria associated with colorectal cancer [17,18].
In order to survive in the host, fusobacteria need to overcome the host immune system and to acquire nutrients. Our previous results demonstrated that the F. nucleatum serine protease is capable of inactivating host immune mediators [39]. Our current study demonstrates the essential nutritional role of fusolisin. This was concluded from the fact that fusolisin inhibition restricted growth of F. nucleatum (Fig. 8A). In the oral plaque, F. nucleatum is always found in a multispecies environment that contains proteolytic members [3] that are likely to relieve the dependency on fusolisin for fusobacterial growth. Indeed, addition of filtersterilized supernatant collected from a P. gingivalis culture that contains the gingipain cysteine proteases (which are not inhibited by PMSF) relieved the PMSF inhibitory effect on F. nucleatum's growth (Fig. 8A).
F. nucleatum is frequently isolated or identified by molecular methods in the amniotic fluid of preterm births [15,66,67]. In some of these isolations F. nucleatum was found as a sole bacterium [68]. In this extraoral environment, and without the proteolytic oral bacterial neighbors, it is plausible that fusobacterial proliferation in the placenta is fusolisin dependent. It is therefore tempting to speculate that inhibition of fusolisin might affect the ability of F. nucleatum to translocate from the multispecies oral cavity and colonize the placenta.

Supporting Information
Movie S1 Structure prediction of serine protease catalytic domain of Fsp25586 (catalytic triad marked) generated using the Protein Homology/analogY Recognition Engine (Phyre).

(MOV)
Movie S2 Structure prediction of the autotransporter domain of Fsp25586 generated using the Protein Homology/analogY Recognition Engine (Phyre).

Author Contributions
Conceived and designed the experiments: GB LD SCG GR AE.