Microbial Pathogenesis 83-84 (2015) 47e56

Contents lists available at ScienceDirect

Microbial Pathogenesis journal homepage: www.elsevier.com/locate/micpath

Recombinant fragilysin isoforms cause E-cadherin cleavage of intact cells and do not cleave isolated E-cadherin Daria Kharlampieva a, *, Valentin Manuvera a, b, Oleg Podgorny a, b, c, Ekaterina Grafskaia a, b, Sergey Kovalchuk a, d, Olga Pobeguts a, Ilya Altukhov a, b, Vadim Govorun a, b, d, Vassili Lazarev a, b a Scientific Research Institute of Physical-Chemical Medicine of the Federal Medical and Biological Agency, Malaya Pirogovskaya str. 1a, Moscow 119435, Russia b Moscow Institute of Physics and Technology, Institutskiy per. 9, Dolgoprudny, Moscovskaya obl 141700, Russia c Koltzov Institute of Developmental Biology of the Russian Academy of Sciences, Vavilov str. 26, Moscow 119334, Russia d Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry of the Russian Academy of Sciences, Miklukho-Maklaya str. 16/10, Moscow 117997, Russia

a r t i c l e i n f o

a b s t r a c t

Article history: Received 30 January 2015 Received in revised form 28 April 2015 Accepted 15 May 2015 Available online 18 May 2015

The fragilysin (BFT) is a protein secreted by enterotoxigenic Bacteroides fragilis strains. BFT contains zincbinding motif which was found in the metzincins family of metalloproteinases. In this study, we generated three known recombinant isoforms of BFT using Escherichia coli, tested their activity and examined whether E-cadherin is a substrate for BFTs. BFT treatment of HT-29 cells induced endogenous E-cadherin cleavage, and this BFT activity requires the native structure of zinc-binding motif. At the same time recombinant BFTs did not cleave recombinant E-cadherin or E-cadherin in isolated cell fractions. It indicates that E-cadherin may be not direct substrate for BFT. We also detected and identified proteins released into the cultural medium after HT-29 cells treatment with BFT. The role of these proteins in pathogenesis and cell response to BFT remains to be determined. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Bacteroides fragilis Metalloprotease BFT Fragilysin E-cadherin

1. Introduction Colonization of human body by microorganisms begins immediately after birth, and many of these microorganisms become really essential to the host. Bacteria play a major role in processes of nutrition, digestion and immunity. The majority of microorganisms colonizing a colon are anaerobes, and about of 25% of them are species of Bacteroides genus [1]. Bacteroides are gram-negative bileresistant non-spore-forming rods. Bacteroides fragilis normally presents in intestinal flora and participates in carbohydrate fermentation and biotransformation of bile acids [2]. The population of B. fragilis comprise 1e10% of the total intestinal population of bacteria from Bacteroides genus [2]. However, when B. fragilis escapes the gastrointestinal tract, it may cause significant pathology. The capsule of

Abbreviations: BFT, Bacteroides fragilis toxin; mBFT, mature BFT; prBFT, proform of BFT consisting of prodomain and catalytic domain; ETBF, enterotoxigenic Bacteroides fragilis; NTBF, non-toxigenic Bacteroides fragilis. * Corresponding author. E-mail address: [email protected] (D. Kharlampieva). http://dx.doi.org/10.1016/j.micpath.2015.05.003 0882-4010/© 2015 Elsevier Ltd. All rights reserved.

B. fragilis initiates an abscess formation [3]. Abscesses without treatment may expand and rupture, resulting in bacteremia. B. fragilis is the most common organism accompanying bloodstream infections and abdominal abscesses among Bacteroides species. B. fragilis attracted interest of researchers because it was revealed the association of the bacterium with acute diarrheal disease in newborn lambs, which was accompanied by inappetance, depression and a high mortality rate [4]. In their study Myers and colleagues [4] found that some isolates of B. fragilis stimulated fluid accumulation into ligated intestinal loop of lambs and calves. B. fragilis strains inducing fluid accumulation into ligated intestinal loop were termed enterotoxigenic B. fragilis (ETBF) in contrast to non-toxigenic strains (NTBF) which are lacked this ability. In 1987, enterotoxigenic B. fragilis strains were isolated from the human diarrhoea stool specimens [5]. Till the study of Myers and colleagues [4] B. fragilis had not been reported to cause fluid accumulation in the intestine or to cause diarrhea in any species of animal. Because the profuse watery diarrhea indicate enterotoxin type of mediation, the authors decided to test whether ETBF secreted protein with ability to cause fluid accumulation in ligated intestinal loop. They provided the first evidence that ETBF

48

D. Kharlampieva et al. / Microbial Pathogenesis 83-84 (2015) 47e56

secreted a heat-labile protein stimulating intestinal secretion [6]. Further, this protein (termed fragilysin, or BFT e B. fragilis toxin) was isolated. It was demonstrated that isolated BFT stimulates intestinal secretion and causes intestinal damage, neutrophilic inflammation and in some cases necrosis and hemorrhage [7]. BFT is a secreted protein encoded by gene into 6 kb pathogenicity island [8]. BFT is synthesised as a preproprotein containing a signal peptide and two domains. During maturation the signal peptide and N-terminal domain (prodomain) are cleaved. Remaining C-terminal domain (catalytic domain) is the mature form of the toxin [9]. Catalytic domain contains the HEXXHXXGXXH motif, which is a zincbinding motif found in the metzincins family of metalloproteinases [10,11]. Prodomain inhibits activity of catalytic domain [12]. Three isoforms of the BFT with several differences in their primary structures were identified [9,13e15]. Differences in the sequences of BFT-1, BFT-2 and BFT-3 were observed in 2e5 positions in N-terminal domain and in up to 25 positions in the C-terminal domain. It was demonstrated, that all isoforms of BFT destroy the zonula adherens tight junctions in intestinal epitelium by inducing E-cadherin cleavage, resulting in actin cytoskeleton rearrangements [16,17]. Loss of tight junction leads to entry of fluid into intestinal lumen. This may explain diarrhoea caused by ETBF [17]. On the other side BFT treatment of cells may induce transcription of new genes. It was demonstrated, that BFT treatment of cells leads to synthesis of proinflammatory cytokines (IL-8, MCP-1, CXCL1) [18]. Also BFTinduced E-cadherin cleavage promotes colonic permeability and access of innate mucosal immune cells to luminal bacterial antigens. This possibly promotes mucosal inflammatory and secretory responces [19,20]. Moreover, cleaved E-cadherin fragments possess the oncogenic role [21]. Moreover, in other studies it has been found that cleavage of E-cadherin, induced by BFT-2, releases b-catenin. Then b-catenin nuclear translocation leads to expression of protooncogene C-myc and cell proliferation [22]. Thus data indicating possible role of BFT in mucosal inflammation and colorectal cancer formation are accumulated. Because BFT is a secreted protein with zinc-binding motif which is specific for metzincins family of metalloproteinases and because BFT treatment of HT-29 cells induced Ecadherin cleavage, the goal of our study was to investigate whether BFT cleaves E-cadherin directly. Direct interaction between E-cadherin and BFT was not demonstrated before. To this end we generated wild-type recombinant isoforms of BFTs and BFTs with mutated HEXXHXXGXXH motif (glutamic acid residue mutated to an alanine residue and with zinc ion-chelating histidine residues substituted to tyrosine residues). We tested the recombinant proteins using HT-29 cells to reveal whether E-cadherin cleavage occurs. We also examined recombinant BFTs using azocoll, azocasein and gelatin. Proteolysis of this substrates by BFTs 1 and 3 was described previously [11,12]. Moreover, we tested activity of recombinant BFTs using recombinant thioredoxins containing linker which are potentially cleaved by BFT according to Shiryaev S.A. [23]. We produced recombinant E-cadherin in Escherichia coli and Expi293F cells and isolated an enriched fraction of HT-29 cell membrane proteins and suspension of the enriched membrane fraction. We examined whether E-cadherin isolated from these sources is cleaved by recombinant BFTs. Finally, to find new potential substrates for BFT we identified proteins released in culture medium after BFT treatment of HT-29 cells. 2. Materials and methods 2.1. Isolation of DNA fragments encoding prBFTs 1 and 3 To obtain DNA encoding proform of BFT 1 and 3, consisting of prodomain and catalytic domain (prBFT) we performed screening of DNA from faeces samples. Biological material was obtained

according to guidelines of the ethics committee of Scientific Research Institute of Physical-Chemical Medicine. We performed DNA isolation from faeces as it was described previously [24]. We used nested PCR to identify sequences encoding BFTs 1 and 3 in faecal DNA. For the first round of nested PCR, we used primers u for 601 e u rev and for the second round of nested PCR, we used primers u for 632 e u rev 1000 (Table S1). The resulting PCR product was isolated by preparative electrophoresis. Nucleotide sequences of the fragments were identified by sequencing On AbiPrism 3730xl platform (Applied Biosystems, USA) using the BigDye Terminator Cycle Sequencing Kit (v. 3.1) and AbiPrism 3730xl (Applied Biosystems, USA). DNA samples containing sequences encoding BFT isoforms 1 and 3 were used to obtain fulllength fragments encoding prBFTs. The first round of nested PCR was performed with primers nestF and nestR and for the second round we used primers pBft-Bgl e С-Bft-Sal1iso for prBFT 1 and pBft-Bgl e С-Bft-Sal for prBFT 3 (Table S1). Detailed methods are described in the Supplementary (see Sections 1e4). 2.2. 6xHisTag prBFTs-1, 2, 3 plasmid construction We generated the recombinant plasmids encoding full-length prBFT 1, 2 and 3 (without signal peptide) fused with the signal peptide of fd phage GIII protein and a C-terminal 6xHisTag (pBAD/ GIII-prBft1, pBAD/GIII-prBft, pBAD/GIII-prBft3). Generation of the recombinant plasmid pBAD/GIII-prBft encoding full-length prBFT was described previously [25]. Construction of the plasmids encoding prBFT 1 and prBFT 3 was performed as described previously for pBAD/GIII-prBft [25]. Isolated DNA fragments encoding prBFTs 1 and 3 (see 2.6. Isolation of DNA fragments encoding prBFTs 1 and 3) were used. The resulting plasmids were named pBAD/GIIIprBft1 and pBAD/GIII-prBft3, respectively. We also generated plasmid pETmin/prBFT-His coding for fulllength prBFT-2 (without signal peptide) fused with C-terminal 6xHis Tag. We performed PCR using the plasmid pBAD/GIII-prbft as a template and BAD F and BAD R as primers. Amplicons were cloned into pETmin vector. Plasmid pETmin was derived from the commercial plasmid pET-22b(þ) (Novagen, USA). Detailed methods and the plasmid maps (Figs. S1, S2, S3) are represented in the Supplementary (Section 5). 2.3. Untagged prBFT-1, 2, 3 plasmid construction To investigate the effect of the polyhistidine sequence on protein activity and compare isoforms' digestion activity, we generated recombinant wild-type BFTs (three known isoforms) without a 6xHis Tag. We generated plasmids pBAD/GIII-prBft1-min, pBAD/ GIII-prBft2-min, pBAD/GIII-prBft3-min encoding untagged prBFTs1, 2, 3 fused to the fd phage GIII protein signal peptide. To this end we performed PCR using pBAD/GIII-prBft and pBAD/GIII-prBft3 as templates and pBft-Bgl and C BFT min as primers. In the case when pBAD/GIII-prBft1 was used as template for PCR, we changed C BFT min primer to C BFT m1 primer. These primers allowed to introduce stop codon before the sequence encoding 6xHis Tag. The remaining preparation steps were as listed in 2.2. 2.4. Site-directed mutagenesis To obtain mutant BFTs we used primers listed in Table S2. As a result of site-directed mutagenesis we obtained the plasmids encoding the following mutant prBFTs: (i) the catalytic glutamic acid residue mutated to alanine (E349A): pBAD/GIII-prBft1-E349A, pBAD/GIII-prBft2-E349A, pBAD/GIII-prBft3-E349A; (ii) the zincchelating histidine residues mutated to tyrosine residues: pBAD/ GIII-prBft2-HY (H348Y, H352Y, H358Y).

D. Kharlampieva et al. / Microbial Pathogenesis 83-84 (2015) 47e56

Detailed method is described in the Supplementary (Section 6). 2.5. Generation of plasmids encoding potential BFT substrates Sequence encoding full-length E-cadherin with a signal peptide and C-terminal 6xHis Tag was cloned into pcDNA3.4 (Life Technologies, USA). Further the resulting vector will be referenced as pcDNA3.4-cad. We also generated plasmids encoding thioredoxins with linker encoding potential BFT cleavage sites. Linkers encoding potential cleavage sites were selected according to previously published data [23]. The oligonucleotides Ln1F-Ln1R, Ln2F-Ln2R, Ln3F-Ln3R, Ln4FLn4R, Ln5F-Ln5R were annealed pairwise and cloned into the pET32a(þ) vector (Novagen, USA) treated with BglII and KpnI (Life Technologies, USA). The resulting constructs were pET-Trx-Ln1, pET-Trx-Ln2, pET-Trx-Ln3, pET-Trx-Ln4, and pET-Trx-Ln5. 2.6. Generation and processing of recombinant BFTs E. coli Top10 cells (Life Technologies, USA) were transformed with plasmids encoding wild-type and mutant BFTs, followed by plating on LB dish (150 mg/ml ampicillin) and incubation overnight at 37  C. Then, 10 ml LB medium containing 150 mg/ml ampicillin was inoculated with a single colony and grown at 37  C in a shaker at 180 rpm for 16 h. The culture was added to 1 L LB medium (150 mg/ml ampicillin) and grown to OD600 ~ 0.8. Expression was induced by adding arabinose to a final concentration of 2 g/L. The cells were cultured additionally for 4 h at 37  C, spun down (3000 g, 15 min), and resuspended in 50 ml TE buffer (10 mM TrisHCl and 1 mM EDTA at рН 8.0). The cells were sonicated using a Branson Sonifier 250 (VWR Scientific, USA) sonicator according to the manufacturer's guidelines. The lysate was purified by centrifugation (15,000 g, 25 min) and the pellet was washed twice with 50 ml 1% (v/v) Triton X-100. The pellet was stored at 20  C until need. To isolate prBFTs we centrifuged an aliquot of the inclusion body suspension from 1 L of the culture (4500 g, 15 min) and dissolved in 20 ml buffer A (20 mM Naþ-phosphate buffer, 8 M urea, 0.5 М NaCl and 10 mM imidazole, рН 7.4) with 0.1% (v/v) b-mercaptoethanol. The solution was purified by centrifugation (50,000 g, 15 min) and dialysed against 500 ml PBS with 0.1% (v/v) b-mercaptoethanol at 10  C for 12 h. Subsequently, proteins were undergone to partial tryptic digestion to process proform of BFT to mature form. Trypsin (Life Technologies, USA) was added to dialysed inclusion body samples to a final concentration of 1e2 mg/ml, and the samples were incubated at 37  C for 1e1.5 h. Digestion conditions were individually tested for each sample. PMSF (final concentration 2 mM) was added to inactivate the trypsin. The sample was applied to a column filled with 5 ml Ni Sepharose High Performance media (GE Healthcare, USA) equilibrated with buffer B (20 mM Naþphosphate buffer, 0.5 М NaCl and 10 mM imidazole, рН 7.4) The column was washed with 50 ml buffer B at a flow rate of 1 ml/min, and the binding polypeptides were eluted with buffer D (20 mM Naþ-phosphate buffer, 0.5 М NaCl and 50 mM EDTA). The purified protein was mature BFT, consisting of catalytic domain (mBFT). mBFT was applied to a HiTrap Desalting 5 ml Column (GE Healthcare, USA) and transferred to PBS according to the manufacturer's guidelines. The E. coli BL21(DE3) gold (Life Technologies, USA) were transformed with the pETmin/prBFT-His vector. 100 ml LB medium containing 150 mg/ml ampicillin was inoculated with a single colony and grown at 37  C in a shaker at 180 rpm for 4e5 h. Then 3 L of 2xLB medium (NaCl e 5 g/L), containing 10 mM lactose was inoculated with pre-grown culture and grown overnight at 25  C. The cells were spun down (3000 g, 15 min), washed with PBS and resuspended in 40 ml of distilled water. The cells were disrupted by

49

sonication with a Branson Sonifier 250 (VWR Scientific, USA) according to the manufacturer's guidelines and PMSF was added to a final concentration 2 mM. The lysate was purified by centrifugation (50,000 g, 25 min). 8x buffer B was added to the supernatant up to 1x and the sample was applied to a column filled with 5 ml Ni Sepharose High Performance (GE Healthcare, USA). The column was washed with 50 ml buffer B at a flow rate of 1 ml/min, and the binding polypeptides were eluted with buffer D. The sample was dialysed against 500 ml PBS at 10  C for 12 h. The following stages (processing to mature form, subsequent metal-chelate chromatography and transfer to PBS) were the same as for prBFT from refolded inclusion bodies. Further isolated protein will be referenced as mBFT2-His-Sol. To determine an ability of BFT to form oligomers we used size exclusion chromatography (SEC). SEC was performed with a Tricorn 10/300 column filled with Superdex 200 (GE Healthcare, USA). The column was equilibrated with buffer containing 20 mM sodium phosphate and 8 g/L NaCl, pH 7.4. The flow rate was 1 ml/min, the sample volume was 0.1 ml containing 1 mg protein of interest. The column was calibrated with a Gel Filtration Calibration Kit LMW (GE Healthcare, USA) according to the manufacturer's guidelines. All manipulations were performed with an AKTA FPLC chromatograph (GE Healthcare, USA). 2.7. Generation of recombinant E-cadherin in Expi293F™cells The Expi293F™ cells (Life Technologies, USA) were cultivated and transfected with the pcDNA3.4-cad vector according to the manufacturer's guidelines. After the addition of enhancers, the cells were grown for 48 h and collected for the analysis of E-cadherin expression by western blot. The remaining cell suspension (30 ml, 2  106 cells/ml) was divided into 2 tubes and centrifuged at 200 g for 5 min. The pellets were frozen at 70  C. To isolate E-cadherin, one of the tubes was thawed on ice, and cells were dissolved in 10 ml buffer A. The solution was purified by centrifugation (50,000 g, 15 min) and applied to a column filled with 5 ml Ni Sepharose High Performance media (GE Healthcare, USA) The column was washed with 50 ml buffer A at a flow rate of 1 ml/min, and the binding polypeptides were eluted with buffer C (20 mM Naþ-phosphate buffer, 8 M Urea, 0.5 М NaCl and 500 mM imidazole, pH 7.4). The resulting fractions from metal-chelate chromatography were analysed by western blot for E-cadherin. The E-cadherin-containing fraction was dialysed against PBS, pH 7.4 at 10  C for 16e20 h. After dialysis, recombinant E-cadherin became soluble. Immediately after dialysis, the sample containing recombinant E-cadherin was used to determine BFT activity. 2.8. Generation of thioredoxins containing potential BFT cleavage sites The E. coli BL21(DE3) strain (Life Technologies, USA) was transformed with plasmids encoding thioredoxin with different linkers and with the original plasmid. 10 ml LB medium containing 150 mg/ml ampicillin was inoculated with an overnight culture, and cells were grown at 37  C in a shaker at 180 rpm for 2 h to an OD600 ~ 0.8. Expression was induced by adding IPTG to a final concentration of 1.0 mM. The cells were cultured additionally for 4 h at 37  C, spun down (3000 g, 15 min) and resuspended in 2.5 ml buffer B. The cells were disrupted by sonication with a Branson Sonifier 250 (VWR Scientific, USA) according to the manufacturer's guidelines. The lysate was purified by centrifugation (50,000 g, 25 min). The supernatant was applied to a column filled with 5 ml Ni Sepharose High Performance (GE Healthcare, USA). The column was washed with 50 ml buffer B at a flow rate of 1 ml/min, and the binding polypeptides were eluted with buffer E (20 mM Naþ-

50

D. Kharlampieva et al. / Microbial Pathogenesis 83-84 (2015) 47e56

phosphate buffer, 0.5 М NaCl and 500 mM imidazole, pH 7.4). Eluted proteins were applied to a HiTrap Desalting 5 ml Column (GE Healthcare, USA) and transferred to PBS according to the manufacturer's guidelines. 2.9. Enriched E-cadherin fraction isolation from HT-29 cells

analysed by SDS-PAGE (recombinant E-cadherin from E. coli) and Western blot (the other samples). Recombinant thioredoxins (50 mg/ml) containing potential BFT cleavage sites were incubated with mBFT2-His at 5 mg/ml for 24 h at 37  C. Mutant BFT mBFT2-E349A and trypsin were used as controls. The samples were analysed by SDS-PAGE.

HT-29 cells (ATCC, USA) were seeded in 25 cm2 flasks at 37  C under 5% CO2 in DMEM (Life Technologies, USA) containing 10% fetal bovine serum (FBS, Life Technologies, USA), and 2 mM GlutaMax (Life Technologies, USA). After 48 h, НТ-29 cells were washed three times in Hanks' Balanced Salt Solution (HBSS, Life Technologies, USA), and PBS plus 5 mM EDTA was added to each flask and placed in a CO2 incubator for 15 min. The cells were collected and centrifuged at 200 g for 5 min 1 ml PBS with 2 mM PMSF were added to the cell pellet, and the sample was frozen at 20  C and subsequently thawed. The sample was centrifuged at 200 g for 5 min, and the supernatant was discarded. Freeze/thaw/ centrifuge cycles were repeated twice. The resulting pellet was processed as follows:

2.13. Western blotting analysis

1) The cell pellet from 1.5 flasks (25 cm2) was resuspended in 300 ml Triton X-100 and centrifuged at 12,000 g for 10 min. The supernatant was collected and analysed for E-cadherin presence by western blot. A portion of the supernatant was dialysed against PBS, pH 7.4 for 12 h at þ10  C; 2) The cell pellet from 1.5 flasks (25 cm2) was resuspended in 450 ml PBS, pH 7.4, and sonicated with a Branson Sonifier 250 (VWR Scientific, USA) for 1 min on ice (output control e 3, duty cycle e 50).

HT-29 cells were treated with mBFT2-His, the culture media was collected and the protein sample underwent gel-free digestion of the protein sample with surfactant DTT and RapiGest SF or DTT only and LC-MS analysis. Detailed methods are described in the Supplementary (Sections 7e9). Proteins were identified with the Mascot search engine v2.2.07 against the UniProt sequence database (The UniProt Consortium, ftp.uniprot.org/pub/databases/uniprot/current_release/ knowledgebase/complete/, downloaded on April 6, 2012, which contains 535,248 amino acid sequences). To perform identification of proteins the raw LC/MSeMS datasets were converted to Mascot Generic Format using AB SCIEX MS Data Converter v1.3. The Mascot searches were conducted with the following parameters: trypticspecific peptides, maximum of one missed cleavages, a peptide charge state limited to 1þ,2þ and 3þ, a peptide mass tolerance of 10 ppm, a fragment mass tolerance of 0.5 Da, variable modifications caused by Oxidation(M) and Carbomidomethylation(C), the taxonomy was selected as Homo sapiens. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium [26] via the PRIDE partner repository with the dataset identifier PXD001271 and DOI 10.6019/PXD001271. For comparative analysis of protein amount we calculated Exponentially Modified Protein Abundance Indexes (emPAIs) [27]. The statistical significance of changes in emPAIs was evaluated by an unpaired two-tailed Student's t-test. P-values less than 0.05 were considered significant. emPAIs were normalised by median emPAI per protein. A heat map was constructed using the gplots library for R programming language.

These steps allowed us to collect (i) an enriched fraction of membrane proteins, dissolved in Triton X-100, (ii) the same fraction, transferred to PBS by dialysis, and (iii) suspension of the enriched membrane fraction. The resulting probes were used to determine BFT activity. 2.10. HT-29 biological assay НТ-29 biological assay was performed as previously described [25]. BFT biological activity was detected by observing morphological changes (cell rounding) of HT-29 cells with an Nikon Eclipse E800 microscope (Nikon, Japan) or using an Olympus Live Cell Imaging System (Olympus IX51, Japan). E-cadherin cleavage after treatment of HT-29 cells with BFTs was detected by western blot. 2.11. Determination of prBFT autoactivation Recombinant wild-type prBFTs 1, 2, 3 (with and without 6xHis Tag) were incubated in PBS at pH 7.4 for 24 h at 37  C with Zn2þ(1 mM) and without. In all cases protein concentration was concentrations were 0.5 mg/ml. The samples were analysed by SDS-PAGE. 2.12. Determination of BFT 1, 2, 3 proteolytic activity Gelatin, azocoll and azocasein (Sigma, USA) were used to determine the proteolytic activity of recombinant BFTs in vitro. Proteolytic activity assay was performed as previously described [25]. Recombinant E-cadherin (isolated from E. coli [25] and Expi293F™ cells and dialysed against PBS, pH 7.4) and enriched Ecadherin fractions from HT-29 cells (see 2.8. and 2.10) was incubated with recombinant wild-type BFTs (mBFT1-His, mBFT2-His, mBFT3-His) at 5 mg/ml in DMEM for 3 h at 37  C. The samples were

Western blotting analysis was performed as described previously [25]. We used E-cadherin monoclonal mouse antibody (dilution 1:1,000, Invitrogen, USA) and horseradish peroxidaselinked anti-mouse IgG (from sheep, dilution 1:10,000, GE Healthcare, USA). The membranes were processed with ECL Plus Western blotting detection reagents (GE Healthcare, USA) according to the manufacturer's guidelines. The signals were detected on ChemiDoc MP (BioRad, USA). 2.14. Comparative analysis of protein concentration in culture medium after mBFT2-His treatment of HT-29 cells

3. Results 3.1. Isolation of recombinant BFTs and HT-29 biological assay Sequences encoding BFTs were identified in 18 of 220 DNA samples isolated from faeces. Eight samples contained sequences coding for BFT-1, eight for BFT-2 and two for BFT-3. These samples were used to amplify DNA fragments of interest followed by subsequent construction of expression vectors. The resulting expression vectors we used to produce recombinant BFTs. When constructions based on pBAD/GIII-B plasmid were used (for map see Fig. S1), we did not determine any significant differences in isolation, renaturation and purification of any recombinant BFTs. Analysis of the producer strains' cell lysates revealed a large accumulation of recombinant proteins at the expected molecular weights. Upon

D. Kharlampieva et al. / Microbial Pathogenesis 83-84 (2015) 47e56

fractionation, the recombinant proteins localised to insoluble fractions. We obtained soluble proproteins by refolding inclusion bodies. After proprotein processing by limited tryptic digestion, we obtained mature proteins with the expected electrophoretic mobility of 20 kDa. To increase the yield of the protein of the interest and simplify the procedure we significantly modified our previously published purification scheme [25]. We excluded the metal-chelate chromatography of the inclusion bodies dissolved in urea. We induced refolding of the inclusion bodies by dialysis and processed them by limited tryptic digestion. By subsequent metal-chelate chromatography we purified mature proteins of the interest (mBFTs). Addition of EDTA into elution buffer significantly increased the yield of the proteins of the interest. We used the same purification protocol for 6xHis tagged and untagged proteins. While purifying untagged recombinant BFTs, we found that all proproteins and mature proteins maintained the ability to bind a metal-chelate sorbent at salt concentrations that exclude binding via electrostatic interactions (Fig. 1). Some researchers reported that recombinant prBFT generated in E. coli had an ability to autoactivation [12] and others that it had not [23,28]. We tested autoactivation of prBFTs and found that it did not occur. When we used pETmin/prBFT-His (for map see Fig. S2) coding for full-length prBFT-2 (without signal peptide) fused with C-terminal 6xHis Tag we obtained accumulation of recombinant protein with expected molecular weight both in soluble E. coli fraction and in inclusion bodies. After metal-chelate chromatography and dialysis soluble prBFT-2 was processed by limited tryptic digestion. In this case we also obtained mature protein (Fig. S4) with the expected electrophoretic mobility of 20 kDa (mBFT2-His-Sol). All recombinant proteins generated in this study are listed in Table 1. To determine whether BFTs form oligomers we used SEC. We found that mBFT2-His as well as mBFT2-His-Sol forms a complex with high molecular weight which is not retained by Superdex 200, i.e. not less than several hundred kDa (Fig. 2). Further we tested biological activity of the recombinant proteins. We evaluated changes in HT-29 cell morphology (Fig. 3A) and E-cadherin degradation after treatment with recombinant proteins (Fig. 3 B). We found that recombinant mature wild-type BFT isoforms (with or without the 6xHis Tag) and mBFT2His-Sol caused HT-29 cell rounding and induced E-cadherin cleavage (Fig. 3). On the another side recombinant wild-type prBFT isoforms (with or without the 6xHis Tag), mature mutant BFT isoforms with mutation E349A, mature mutant BFT 2 with zinc-chelating histidine residues substituted to tyrosine residues (H348Y, H352Y, H358Y) did not cause HT-29 cell rounding and induce E-cadherin cleavage. The results are summarised in Table 1. Thus, recombinant wild-type mBFTs generated in our study possesses the same activity which

Fig. 1. Analysis of prBFT1-min and prBFT2-min binding to Ni-sepharose. prBFTs after metal-chelate chromatography were analysed by SDS-PAGE. prBFT1-min: 1 e peak; 2 e flowthrough; 3 einput sample; prBFT2-min: 4 e peak; 5 e flowthrough; 6 e input sample. Staining with Coomassie Blue.

51

was reported for BFT secreted by ETBF. Moreover, the native structure of zinc-binding motif is necessary for E-cadherin release from HT-29 cells. 3.2. Proteolytic activity assay using gelatine, chromogenic substrates and thioredoxins containing peptide linker To characterise generated recombinant proteins we used some of previously described substrates. Surprisingly we found that all wild-type recombinant BFT isoforms with or without 6xHis Tag (proteins No. 1e6 in Table 1) did not cleave gelatin, azocoll or azocasein both with and without Zn2þ ions. Also, we did not detect gelatin and azocoll cleavage by mBFT2-His-Sol. Probably our method of purification allows to obtain protein samples without impurities which can lead to cleavage of nonspecific substrates. Then we characterised recombinant BFTs using protein with linker which was reported as specific motif for BFT-3 cleavage by Shiryaev S.A. et al. [23]. Based on this motif (Pro-X-X-Leu-(Arg/Ala/Leu)Y with the most preferred amino acid residues at positions marked X [23] we inserted several linkers (PRPLRA, PRGLRA, PRPLAA, PAPLRA, PAGLAA) into the protein encoded by the plasmid pET32a (þ), which is a thioredoxin with an additional “tail” with a total molecular weight of approximately 20 kDa. In the case of linker cleavage, two fragments with masses of 16 kDa and 4 kDa are produced. The molecular weight difference of the original protein and the expected cleavage product mass of 16 kDa is easily detected by SDS-PAGE. We treated thioredoxins plus linkers with recombinant BFTs (Fig. 4). However, we did not detect linker cleavage after addition of the recombinant BFTs. Perhaps substrate tertiary structure is also important for cleavage. Thus recombinant wildtype BFTs generated in our study did not cleave nonspecific substrates (azocoll, azocasein and gelatin) and thioredoxin with potential cleavage sites for BFT. 3.3. Proteolytic activity assay using E-cadherin isolated from different sources E-cadherin may be a BFT substrate, because treatment of HT-29 cells with BFTs 1, 2, or 3 induces E-cadherin cleavage. However, we did not observe cleavage of recombinant E-cadherin isolated from E. coli after incubation with recombinant 6xHis tagged wild-type BFTs (proteins No.No. 1e3 in Table 1). Perhaps postranslational modifications are important for BFT recognition of E-cadherin and subsequent cleavage. For this reason we obtained recombinant Ecadherin from Expi293F™ cells using metal-chelate chromatography. Because in the case of the biological assay on HT-29 cells we performed treatment of cells in DMEM medium, we incubated recombinant E-cadherin from Expi293F™ cells with BFTs in DMEM too. However, we did not observe E-cadherin cleavage by any tested BFT isoforms (Fig. 5A). Additionally, we isolated enriched fraction of membrane proteins from HT-29 cells using Triton X-100 but we did not detect the cleavage of E-cadherin in this case too (Fig. 5 B, C). We hypothesized that BFT cleavage activity requires its contact with membrane lipids. Thus, we isolated the enriched membrane fraction from HT-29 cells. However, BFT isoforms still did not cleave Ecadherin (Fig. 5 D). At the same time, when whole HT-29 cells were treated with the same BFT samples, E-cadherin was cleaved. Thus, recombinant BFT isoforms cause E-cadherin cleavage of intact cells and do not cleave isolated E-cadherin. 3.4. Identification of proteins released into culture medium after BFT treatment of HT-29 cells We found that BFTs did not cleave E-cadherin isolated from different sources. By this reason we searched other proteins which

52

D. Kharlampieva et al. / Microbial Pathogenesis 83-84 (2015) 47e56

Table 1 The recombinant BFTs generated in this study. No.

Protein name

Isoform

6xHis

Mutations

E-cadherin cleavage (НТ-29 cells) and changes in HT-29 cell morphology

1. 2. 3. 4. 5. 6. 7. 8 9. 10. 11.

mBFT1-His mBFT2-His mBFT3-His mBFT1-min mBFT2-min mBFT3-min mBFT1-E349A mBFT2-E349A mBFT3-E349A mBFT2-HY mBFT2-His-Sol

1 2 3 1 2 3 1 2 3 2 2

þ þ þ e e e e e e e þ

e e e e e e E349A

þ þ(see þ þ(see þ(see þ(see (see (see (see (see þ

may be potential substrates for BFTs. To this end we performed LCMS analysis to detect proteins released into the culture medium after treatment of HT-29 cells with mBFT2-His. We used two probeprocessing methods: gel-free digestion of the protein sample with surfactant RapiGest SF and DTT, or DTT only. The number of identified proteins are listed in Table 2 for each of the three replicates. The number of identified peptides per protein are listed in Table S3. We calculated Exponentially Modified Protein Abundance Indexes (emPAIs) (Table S4) and constructed heat maps (Fig. 6). As shown on Fig. 6 the amount of proteins such as Kunitz-type protease inhibitor 1 (Swiss-Prot acc. O43278), lymphocyte antigen 75 (SwissProt acc. O60449), carbonic anhydrase 9 (Swiss-Prot acc. Q16790) and carbonic anhydrase 12 (Swiss-Prot acc. O43570), V-set and immunoglobulin domain-containing protein 10-like (Swiss-Prot acc. Q86VR7), carcinoembryonic antigen-related cell adhesion molecule 1 (P13688) increased in HT-29 cell culture medium after BFT treatment. In addition, cadherin superfamily members were detected, including cadherin-1 (E-cadherin Swiss-Prot acc. P12830), protocadherin-1 (Swiss-Prot acc. Q08174) and protocadherin Fat 1 (Swiss-Prot acc. Q14517). Thus for the first time we identified

Fig. 2. SEC profiles for mBFT2-His and mBFT2-His-Sol. Recombinant mBFT2-His and mBFT2-His-Sol form complexes with high molecular weight which are not retained by Superdex 200, i.e. not less than several hundred kDa. Mw of markers are 75.0, 29.0 and 6.5 kDa.

H348Y, H352Y, H358Y e

Fig. 3 A) Fig. Fig. Fig. Fig. Fig. Fig. Fig.

3 3 3 3 3 3 3

B) B) B) B) B) B) B)

proteins which are released after BFT treatment of HT-29 cells. 4. Discussion In our study we generated recombinant BFT isoforms fused to a 6xHis Tag and in untagged form because the C-terminal 6xHis Tag could potentially change protein activity. We did not detect any differences among the isoforms as well as among tagged and untagged proteins in level of accumulation, renaturation and processing of proproteins to mature form. We found that all untagged proproteins and mature proteins bound to metalchelate sorbent at salt concentrations that exclude binding via electrostatic interactions. We cannot explain prBFT binding via their zinc-binding motif, because this region is covered by the prodomain that was revealed by X-ray analysis [12]. Moreover, we found that the mutated protein in which three histidine residues at the zinc-binding motif were substituted with tyrosine residues also bound to metal-chelate sorbent. Thus, we can exclude participation of the zinc-binding motif in binding of the metal-chelate sorbent. To test activity of the recombinant BFTs we used the method developed by Weikel C.S. and colleagues [29]. We observed rounding of the HT-29 cells and detected E-cadherin cleavage after treatment of the HT-29 cells with recombinant tagged and untagged BFT isoforms that confirms previous investigations [13,17]. Different non-specific substrates for BFT were described previously [11,12]. It was reported, that isoform 1 cleaves gelatin and azocoll but not azocasein [11], and isoform 3 cleaves azocasein and azocoll [12]. We have shown that, in contrast to previous reports, none of the three recombinant BFT isoforms (6xHis tagged or untagged) hydrolysed nonspecific substrates. We suggest two possible explanations: (i) incorrect folding of the target proteins or (ii) our method of purification of the recombinant BFTs allows to obtain protein samples without impurities which could lead to cleavage of nonspecific substrates. There are two major arguments against the first suggestion: (i) purified BFTs demonstrate the normal activity on HT-29 cells (i.e. induce cell rounding and E-cadherin cleavage); (ii) recombinant protein catalytic domain has significant resistance to trypsin digestion that was described for BFT produced by ETBF [30,31]. To exclude suggestion about incorrect folding, we generated prBFT-2, accumulated in E. coli in soluble fraction. We purified it and processed into mature form (mBFT2-His-Sol). We found that mBFT2-His-Sol caused rounding of HT-29 cells and induce E-cadherin cleavage as well as mBFT2-His from the inclusion bodies, but at the same time it did not cleave azocoll and gelatine. However, soluble prBFT2-His was accumulated in E. coli in less quantities than in the inclusion bodies. The inclusion bodies consist of the protein of the interest at about 90% whereas soluble fraction contains a lot of other proteins. Because the proteins obtained in

D. Kharlampieva et al. / Microbial Pathogenesis 83-84 (2015) 47e56

53

Fig. 3. HT-29 cells after treatment with recombinant proteins. A e Morphological changes of HT-29 cells after treatment with mBFT2-His (2 mg/ml) and prBFT2-His (5 mg/ml). In control sample PBS was added. The images were obtained using Olympus Live Cell Imaging System. Morphological changes of HT-29 cells after treatment with mBFT1-His, mBFT3His, mBFT1-min, mBFT2-min, mBFT3-min and mBFT2-His-Sol were the same as for mBFT2-His. B e E-cadherin western blot of HT-29 lysates. Control e НТ-29 cells treated with PBS for 1 h; 1e7 e HT-29 cells treated for 1 h with recombinant proteins: 1 e mBFT1-min; 2 e mBFT1-E349A; 3 e mBFT2-min; 4 e mBFT2-E349A; 5 e mBFT2-HY; 6 - mBFT3-min; 7 e mBFT3-E349A. All recombinant proteins were used at a concentration of 1 mg/ml. After treatment with recombinant proteins or PBS, HT-29 cells were lysed in Laemmli sample buffer, sonicated and applied to 10% SDS-PAGE. The molecular weight of E-cadherin is 120 kDa.

Fig. 4. BFT treatment of thioredoxin with PRPLRA linker (Trx-Ln1). 1 e Input sample, 2e8 e Trx-Ln1 after incubation for 24 h at 37  C with: 2 e mBFT2-His; 3 e mBFT2-His and 5 mM EDTA; 4 e mBFT2-His and 2 mM PMSF; 5 e mBFT2-E349A; 6 e mBFT2E349A and 5 mM EDTA; 7 e mBFT2-E349A and 2 mM PMSF, 8 etrypsin. Recombinant BFTs were used at 5 mg/ml. SDS-PAGE, staining with Coomassie Blue.

soluble E. coli fraction and alternatively refolded from inclusion bodies possessed the same activity, we performed subsequent experiments with the refolded proteins. Further we characterised recombinant wild-type BFTs using thioredoxin with inserted linker containing potential site for BFT

Fig. 5. Proteolytic activity of recombinant mBFT1-His, mBFT2-His, mBFT3-His. Western blot for E-cadherin. The following substrates were used: A. recombinant E-cadherin from Expi293F™ (1e4, 6 e incubation with mBFT1-His, mBFT2-His, mBFT3-His, protein storage buffer, trypsin, respectively, 5 e input sample); B. Enriched fraction of HT-29 cell membrane proteins, dissolved in Triton X-100 (1e4 e incubation with protein storage buffer, mBFT1-His, mBFT2-His, mBFT3-His, respectively); C. the same fraction as B transferred to PBS by dialysis (1e5 incubation with trypsin, protein storage buffer, mBFT1-His, mBFT2-His, mBFT3-His, respectively); D. E-cadherin in enriched membrane fraction from HT-29 (1e5: incubation with protein storage buffer, trypsin, mBFT1-His, mBFT2-His, mBFT3-His, respectively).

cleavage. Specific motif for BFT-3 cleavage e Pro-X-X-Leu-(Arg/ Ala/Leu)Y was identified by Shiryaev et al. [23] using a combinatorial peptide library. We found that none of the three recombinant BFT isoforms hydrolysed thioredoxins containing peptide linkers (see 2.9, 3.2. and Fig. 4). In contrast to study by Shiryaev et al. [23] in which the authors performed their experiments using peptides we used the protein with inserted linker containing potential cleavage site for BFT. We suggest that BFT possesses a conformational specificity, i.e. the conformation of the sites adjacent to the cleavage site in substrate is important for recognition and binding of BFTs. Based on data that the recombinant wild-type 6xHis tagged and untagged BFTs produced in our study demonstrated activity in the HT-29 cells assay and induced E-cadherin cleavage we tested whether E-cadherin is direct substrate for BFTs. We generated Ecadherin in E. coli and Expi293F™ expression systems and isolated E-cadherin containing fractions from the HT-29 cells and found that none of BFT isoforms cleaved E-cadherin from these sources. We previously reported that mBFT2-His did not cleave recombinant Ecadherin isolated from E. coli [25]. Similar to the other studies we investigated whether proteolytic activity of BFT is necessary for its biologic effect in HT-29 cells. To this end we introduced mutations into BFT's HEXXHXXGXXH motif, which is a zinc-binding motif found in the metzincin family of metalloproteinases [10,11]. Histidine residues of this motif chelate zinc ion which are essential for catalysis as well as glutamic acid residue. It is known that mutations in zinc-binding motif of metalloproteinase lead to loss of proteolytic activity and as a consequence, function. For example, antrax toxin lethal factor which is a metalloprotease loss its lethal activity after substitution of H686 and H690 implicated in zinc binding to alanine residue as well as after substitution of catalytic glutamic acid residue to cysteine residue [32]. For pallilysin, metalloprotease secreted by Treponema pallidum it was demonstrated that mutations in zinc-binding motif abolished host component proteolysis [33]. In our study we generated mutant BFTs (Table 1, proteins No.7e10) with the catalytic glutamic acid residue mutated to alanine and with the zinc-chelating histidine residues mutated

54

D. Kharlampieva et al. / Microbial Pathogenesis 83-84 (2015) 47e56

Table 2 Number of identified proteins in the culture medium after mBFT2-His treatment of HT-29 cells.

Control e replicate 1 Experiment e replicate 1 Control e replicate 2 Experiment e replicate 2 Control e replicate 3 Experiment e replicate 3

Number of identified proteins (probe preparation with DTT only)

Number of identified proteins (probe preparation with DTT and RapiGest)

1027 1310 1165 1151 1521 1560

1433 1462 1489 1648 1543 1345

to tyrosine residues. We found that mutated proteins did not alter morphology of the HT-29 cells and did not induce E-cadherin cleavage in contrast to recombinant wild-type BFTs at the same time intervals (see Fig. 3 B). Thus, the native structure of the zincbinding motif is necessary for BFT biological activity. This conclusion was also supported by data that the addition of PMSF (serine proteinase inhibitor) had no effect on mBFT2-His activity in HT-29 cells while addition of the divalent cation chelator EDTA inhibited E-cadherin cleavage after HT-29 cells treatment with mBFT2-His [25]. The ability of some of proteases to induce cleavage of E-cadherin was described previously (for review see [34]). However for some of these proteases E-cadherin cleavage was detected only in whole cells or cell lysates while for others direct cleavage of isolated Ecadherin was demonstrated. For example, the ability of eukaryotic metalloproteases MMP-7 and meprinb as well as for cysteine protease Kgp (one of the gingipains) from Porphyromonas gingivalis to cleave directly immunoprecipitated E-cadherin was revealed [35,36]. On another side ADAM-10 processed E-cadherin in cell lysates but not immunoprecipitated E-cadherin which indicates possible participation of downstream proteases in ADAM-10 induced E-cadherin cleavage [35]. The study of E-cadherin cleavage is complicated by existence of proteolytic cascades [34]. For example, ability of MMPs to activate other family members was reported [37]. By this reason it is difficult to distinguish effect due to catalytic activity on a substrate of the upstream metalloprotease versus a catalytic activation of a downstream mediator and subsequent substrate cleavage. In the case of BFT treatment of cells we cannot exclude participation of cell proteases in BFT-induced Ecadherin cleavage.

Further we supposed that BFT cleaves substrates other than Ecadherin. To this end we used mass spectrometry and analysed the proteins released into the culture medium after BFT treatment of HT-29 cells. Previously, Wu et al. revealed that treatment of HT29/C1 cells with BFT-2 induced protein release from the membrane surface, termed “shedding” [38]. The proteases that induce membrane protein shedding have been named “sheddases”. These proteases induce cleavage of the extracellular domains of type I integral membrane proteins and GPI-anchored proteins close to the membrane surface [39]. The A Disintegrin and Metalloproteases (ADAMs) family is one of the major families of proteases that induce the release of membrane protein extracellular domains [39]. Goulas et al. performed X-ray analysis of prBFT-3 and found that its catalytic domain is structurally similar to ADAM family members [12]. We found that mBFT2-His also induces the release of extracellular domains from type I integral membrane proteins. Thus, BFT may be classified as a sheddase. Furthermore, we identified proteins released in culture medium after BFT treatment of HT-29 cells, which are members of the cadherin superfamily and potential substrates for ADAMs (e.g., ADAM10, ADAM15) [40e43]. It is known that many of sheddases are misregulated or overexpressed in disease [34]. For example, ADAM 15 overexpressed in prostate, breast and lung cancer. Another member of ADAM family, ADAM10 upregulated in metastatic melanoma compared to primary melanoma. Members of MMP (matrix metalloprotease) family are reported to over-express in cancer of different organs. Enterotoxigenic B. fragilis producing BFT (which demonstrates sheddase activity) is associated with colorectal cancer [44]. The oncogenic role of Ecadherin fragments released by sheddases is widely discussed in [21,34]. The role of other proteins released after BFT treatment in

Fig. 6. Comparative analysis of protein amount in culture medium after HT-29 cells treatment with mBFT2-His. The heat map was constructed using gplots library for R programming language. emPAI e Exponentially Modified Protein Abundance Indexes. K e control, BFT e experiment, 1 efirst biological replicate, 2 esecond biological replicate, 3 ethird biological replicate. A e heat map for sample preparation with dithiothreitol only (DTT), B e heat map for sample preparation with RapiGest and DTT.

D. Kharlampieva et al. / Microbial Pathogenesis 83-84 (2015) 47e56

cellular response and development of pathology remains a subject for further investigation. Thus, all generated in our study recombinant wild-type BFT isoforms (6xHis tagged and untagged) possessed the same biological activity in HT-29 cells, i.e. caused cell rounding and induce E-cadherin cleavage. E-cadherin cleavage in HT-29 cells required the native structure of zinc-binding motif which is specific for metzincin family of metalloproteases. None of generated in our study the recombinant wild-type BFT isoforms cleaved gelatin, azocasein, azocoll, which is in contrast to previously published data. We found that wild-type BFTs did not cleave directly recombinant E-cadherin obtained in E. coli and eukaryotic cells and Ecadherin which is contained in enriched fraction of membrane proteins isolated from HT-29 cells. This led us to conclusion that Ecadherin is not a direct BFT substrate. Finally BFT like ADAM family proteases caused the release of type I membrane proteins. Acknowledgements Microscopy analysis was performed at Optical Research Group, Koltzov Institute of Developmental Biology, Russian Academy of Sciences. This work was supported by Ministry of Education and Science of the Russian Federation (RFMEFI57514X0075). Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.micpath.2015.05.003. References [1] A.A. Salyers, Bacteroides of the human lower intestinal tract, Annu Rev. Microbiol. 38 (1984) 293e313. [2] H.M. Wexler, Bacteroides: the good, the bad, and the nitty-gritty, Clin. Microbiol. Rev. 20 (2007) 593e621. [3] A.O. Tzianabos, A.B. Onderdonk, B. Rosner, R.L. Cisneros, D.L. Kasper, Structural features of polysaccharides that induce intra-abdominal abscesses, Science 262 (1993) 416e419. [4] L.L. Myers, B.D. Firehammer, D.S. Shoop, M.M. Border, Bacteroides fragilis: a possible cause of acute diarrheal disease in newborn lambs, Infect. Immun. 44 (1984) 241e244. [5] L.L. Myers, D.S. Shoop, L.L. Stackhouse, F.S. Newman, R.J. Flaherty, G.W. Letson, et al., Isolation of enterotoxigenic Bacteroides fragilis from humans with diarrhea, J. Clin. Microbiol. 25 (1987) 2330e2333. [6] L.L. Myers, D.S. Shoop, B.D. Firehammer, M.M. Border, Association of enterotoxigenic Bacteroides fragilis with diarrheal disease in calves, J. Infect. Dis. 152 (1985) 1344e1347. [7] R.J. Obiso Jr., D.M. Lyerly, R.L. Van Tassell, T.D. Wilkins, Proteolytic activity of the Bacteroides fragilis enterotoxin causes fluid secretion and intestinal damage in vivo, Infect. Immun. 63 (1995) 3820e3826. [8] J.S. Moncrief, A.J. Duncan, R.L. Wright, L.A. Barroso, T.D. Wilkins, Molecular characterization of the fragilysin pathogenicity islet of enterotoxigenic Bacteroides fragilis, Infect. Immun. 66 (1998) 1735e1739. [9] A.A. Franco, L.M. Mundy, M. Trucksis, S. Wu, J.B. Kaper, C.L. Sears, Cloning and characterization of the Bacteroides fragilis metalloprotease toxin gene, Infect. Immun. 65 (1997) 1007e1013. [10] W. Bode, F.X. Gomis-Ruth, W. Stockler, Astacins, serralysins, snake venom and matrix metalloproteinases exhibit identical zinc-binding environments (HEXXHXXGXXH and Met-turn) and topologies and should be grouped into a common family, the ‘metzincins’, FEBS Lett. 331 (1993) 134e140. [11] J.S. Moncrief, R. Obiso Jr., L.A. Barroso, J.J. Kling, R.L. Wright, R.L. Van Tassell, et al., The enterotoxin of Bacteroides fragilis is a metalloprotease, Infect. Immun. 63 (1995) 175e181. [12] T. Goulas, J.L. Arolas, F.X. Gomis-Ruth, Structure, function and latency regulation of a bacterial enterotoxin potentially derived from a mammalian adamalysin/ADAM xenolog, Proc. Natl. Acad. Sci. U. S. A. 108 (2011) 1856e1861. [13] G.T. Chung, A.A. Franco, S. Wu, G.E. Rhie, R. Cheng, H.B. Oh, et al., Identification of a third metalloprotease toxin gene in extraintestinal isolates of Bacteroides fragilis, Infect. Immun. 67 (1999) 4945e4949. [14] N. Kato, C.X. Liu, H. Kato, K. Watanabe, Y. Tanaka, T. Yamamoto, et al., A new subtype of the metalloprotease toxin gene and the incidence of the three bft subtypes among Bacteroides fragilis isolates in Japan, FEMS Microbiol. Lett. 182 (2000) 171e176.

55

[15] J.J. Kling, R.L. Wright, J.S. Moncrief, T.D. Wilkins, Cloning and characterization of the gene for the metalloprotease enterotoxin of Bacteroides fragilis, FEMS Microbiol. Lett. 146 (1997) 279e284. [16] S.S. Koshy, M.H. Montrose, C.L. Sears, Human intestinal epithelial cells swell and demonstrate actin rearrangement in response to the metalloprotease toxin of Bacteroides fragilis, Infect. Immun. 64 (1996) 5022e5028. [17] S. Wu, K.C. Lim, J. Huang, R.F. Saidi, C.L. Sears, Bacteroides fragilis enterotoxin cleaves the zonula adherens protein, E-cadherin, Proc. Natl. Acad. Sci. U. S. A. 95 (1998) 14979e14984. [18] J.M. Kim, S.J. Cho, Y.K. Oh, H.Y. Jung, Y.J. Kim, N. Kim, Nuclear factor-kappa B activation pathway in intestinal epithelial cells is a major regulator of chemokine gene expression and neutrophil migration induced by Bacteroides fragilis enterotoxin, Clin. Exp. Immunol. 130 (2002) 59e66. [19] M. Riegler, M. Lotz, C. Sears, C. Pothoulakis, I. Castagliuolo, C.C. Wang, et al., Bacteroides fragilis toxin 2 damages human colonic mucosa in vitro, Gut 44 (1999) 504e510. [20] C.L. Sears, Enterotoxigenic Bacteroides fragilis: a rogue among symbiotes, Clin. Microbiol. Rev. 22 (2009) 349e369. [21] J.M. David, A.K. Rajasekaran, Dishonorable discharge: the oncogenic roles of cleaved E-cadherin fragments, Cancer Res. 72 (2012) 2917e2923. [22] S. Wu, P.J. Morin, D. Maouyo, C.L. Sears, Bacteroides fragilis enterotoxin induces c-Myc expression and cellular proliferation, Gastroenterology 124 (2003) 392e400. [23] S.A. Shiryaev, A.G. Remacle, A.V. Chernov, V.S. Golubkov, K. Motamedchaboki, N. Muranaka, et al., Substrate cleavage profiling suggests a distinct function of Bacteroides fragilis metalloproteinases (fragilysin and metalloproteinase II) at the microbiome-inflammation-cancer interface, J. Biol. Chem. 288 (2013) 34956e34967. [24] J. Qin, R. Li, J. Raes, M. Arumugam, K.S. Burgdorf, C. Manichanh, et al., A human gut microbial gene catalogue established by metagenomic sequencing, Nature 464 (2010) 59e65. [25] D.D. Kharlampieva, V.A. Manuvera, O.V. Podgorny, S.I. Kovalchuk, O.V. Pobeguts, I.A. Altukhov, et al., Purification and characterisation of recombinant Bacteroides fragilis toxin-2, Biochimie 95 (2013) 2123e2131. [26] J.A. Vizcaino, E.W. Deutsch, R. Wang, A. Csordas, F. Reisinger, D. Rios, et al., ProteomeXchange provides globally coordinated proteomics data submission and dissemination 32 (2014) 223e226. [27] Y. Ishihama, Y. Oda, T. Tabata, T. Sato, T. Nagasu, J. Rappsilber, et al., Exponentially modified protein abundance index (emPAI) for estimation of absolute protein amount in proteomics by the number of sequenced peptides per protein, Mol. Cell Proteomics 4 (2005) 1265e1272. [28] S.A. Shiryaev, A.E. Aleshin, N. Muranaka, M. Kukreja, D.A. Routenberg, A.G. Remacle, et al., Structural and functional diversity of metalloproteinases encoded by the Bacteroides fragilis pathogenicity island, Febs J. 281 (2014) 2487e2502. [29] C.S. Weikel, F.D. Grieco, J. Reuben, L.L. Myers, R.B. Sack, Human colonic epithelial cells, HT29/C1, treated with crude Bacteroides fragilis enterotoxin dramatically alter their morphology, Infect. Immun. 60 (1992) 321e327. [30] R.L. Van Tassell, D.M. Lyerly, T.D. Wilkins, Purification and characterization of an enterotoxin from Bacteroides fragilis, Infect. Immun. 60 (1992) 1343e1350. [31] S. Wu, L.A. Dreyfus, A.O. Tzianabos, C. Hayashi, C.L. Sears, Diversity of the metalloprotease toxin produced by enterotoxigenic Bacteroides fragilis, Infect. Immun. 70 (2002) 2463e2471. [32] K.R. Klimpel, N. Arora, S.H. Leppla, Anthrax toxin lethal factor contains a zinc metalloprotease consensus sequence which is required for lethal toxin activity, Mol. Microbiol. 13 (1994) 1093e1100. [33] S. Houston, R. Hof, L. Honeyman, J. Hassler, C.E. Cameron, Activation and proteolytic activity of the Treponema pallidum metalloprotease, pallilysin, PLoS Pathog. 8 (2012) e1002822. [34] M.M. Grabowska, M.L. Day, Soluble E-cadherin: more than a symptom of disease, Front. Biosci. (Landmark Ed.) 17 (2012) 1948e1964. [35] M. Huguenin, E.J. Muller, S. Trachsel-Rosmann, B. Oneda, D. Ambort, E.E. Sterchi, et al., The metalloprotease meprinbeta processes E-cadherin and weakens intercellular adhesion, PLoS One 3 (2008) e2153. [36] J. Katz, Q.B. Yang, P. Zhang, J. Potempa, J. Travis, S.M. Michalek, et al., Hydrolysis of epithelial junctional proteins by porphyromonas gingivalis gingipains, Infect. Immun. 70 (2002) 2512e2518. [37] M. Egeblad, Z. Werb, New functions for the matrix metalloproteinases in cancer progression, Nat. Rev. Cancer 2 (2002) 161e174. [38] S. Wu, K.J. Rhee, M. Zhang, A. Franco, C.L. Sears, Bacteroides fragilis toxin stimulates intestinal epithelial cell shedding and gamma-secretasedependent E-cadherin cleavage, J. Cell Sci. 120 (2007) 1944e1952. [39] K. Reiss, P. Saftig, The “a disintegrin and metalloprotease” (ADAM) family of sheddases: physiological and cellular functions, Semin. Cell Dev. Biol. 20 (2009) 126e137. [40] J.J. Bech-Serra, B. Santiago-Josefat, C. Esselens, P. Saftig, J. Baselga, J. Arribas, et al., Proteomic identification of desmoglein 2 and activated leukocyte cell adhesion molecule as substrates of ADAM17 and ADAM10 by difference gel electrophoresis, Mol. Cell Biol. 26 (2006) 5086e5095. [41] T. Maretzky, K. Reiss, A. Ludwig, J. Buchholz, F. Scholz, E. Proksch, et al., ADAM10 mediates E-cadherin shedding and regulates epithelial cell-cell adhesion, migration, and beta-catenin translocation, Proc. Natl. Acad. Sci. U. S. A. 102 (2005) 9182e9187. [42] A.J. Najy, K.C. Day, M.L. Day, The ectodomain shedding of E-cadherin by

56

D. Kharlampieva et al. / Microbial Pathogenesis 83-84 (2015) 47e56

ADAM15 supports ErbB receptor activation, J. Biol. Chem. 283 (2008) 18393e18401. [43] K. Reiss, T. Maretzky, A. Ludwig, T. Tousseyn, B. de Strooper, D. Hartmann, et al., ADAM10 cleavage of N-cadherin and regulation of cell-cell adhesion and

beta-catenin nuclear signalling, Embo J. 24 (2005) 742e752. [44] N.U. Toprak, A. Yagci, B.M. Gulluoglu, M.L. Akin, P. Demirkalem, T. Celenk, et al., A possible role of Bacteroides fragilis enterotoxin in the aetiology of colorectal cancer, Clin. Microbiol. Infect. 12 (2006) 782e786.