Elastase in Intestinal Mucus Enhances the Cytotoxicity of Shiga Toxin Type 2d*

Shiga toxin variant type 2d (Stx2d) produced by some strains of Shiga toxin-producing Escherichia coli is com-posed of an enzymatically active A subunit and a B (binding) pentamer. The cytotoxicity of Stx2d is increased (activated) 10–1000-fold for Vero cells when the toxin is incubated with mucus obtained from the small intestine of mice. In this study we isolated an Stx2d activator and identified it as a mouse elastase with strong homology to human elastase IIIB. Moreover, commercially available porcine pancreatic elastase preparations also activated Stx2d cytotoxicity although with a lower specific activity than isolated mouse elastase. Elastase directly nicked the Stx2d A subunit to A 1 and A 2 , an event that did not correlate with activation. How- ever, elastase also reduced the size and changed the isoelectric point of the A 2 peptide, as determined by SDS-polyacrylamide gel electrophoresis and two-dimen-sional electrophoresis followed by Western immunoblot analysis. This elastase-mediated size and charge shift in the A 2 peptide of Stx2d occurred concurrently with ac- tivation of the toxin. Both the reduction in size of the Stx2d A 2 peptide by incubation with elastase as well as the associated activation of Stx2d cytotoxicity were fully inhibited by elastatinal, an elastase-specific inhibitor. The Shiga toxins (Stx) 1 expressed by Stx-producing Escherichia coli (STEC) and Shigella dysenteriae

Shiga toxin variant type 2d (Stx2d) produced by some strains of Shiga toxin-producing Escherichia coli is composed of an enzymatically active A subunit and a B (binding) pentamer. The cytotoxicity of Stx2d is increased (activated) 10 -1000-fold for Vero cells when the toxin is incubated with mucus obtained from the small intestine of mice. In this study we isolated an Stx2d activator and identified it as a mouse elastase with strong homology to human elastase IIIB. Moreover, commercially available porcine pancreatic elastase preparations also activated Stx2d cytotoxicity although with a lower specific activity than isolated mouse elastase. Elastase directly nicked the Stx2d A subunit to A 1 and A 2 , an event that did not correlate with activation. However, elastase also reduced the size and changed the isoelectric point of the A 2 peptide, as determined by SDS-polyacrylamide gel electrophoresis and two-dimensional electrophoresis followed by Western immunoblot analysis. This elastase-mediated size and charge shift in the A 2 peptide of Stx2d occurred concurrently with activation of the toxin. Both the reduction in size of the Stx2d A 2 peptide by incubation with elastase as well as the associated activation of Stx2d cytotoxicity were fully inhibited by elastatinal, an elastase-specific inhibitor.
The Shiga toxins (Stx) 1 expressed by Stx-producing Escherichia coli (STEC) and Shigella dysenteriae are a family of potent cytotoxins that are comprised of a single A subunit associated with a pentamer of B subunits (see Ref. 1 for a comprehensive review of STEC). The A subunit of Stx has N-glycosidase activity that cleaves an adenine residue from the 28 S ribosomal RNA within the 60S ribosome, an event that halts protein synthesis and leads to the death of the affected cell. The A subunit of Stx is susceptible to trypsin cleavage near its C terminus. Such cleavage results in the formation of an A 1 subunit of 28 kDa that has the N-glycosidase activity and an A 2 peptide of 4 kDa that remains associated with A 1 through a single disulfide bond. The A 2 peptide links A 1 to the B pentamer (2). The B pentamer of most Stx family members binds preferentially to globotriaosylceramide on eukaryotic cells (3,4).
We have previously reported (5,6) that when a variant of Stx2, called Stx2d (formerly named SLT-IIvh) is incubated with crude mucus isolated from the small intestine or colon of mice or the colon of humans, the cytotoxicity of the mucustreated Stx2d increases 10 -1000-fold for Vero cells. We call this enhancement of cytotoxicity "activation." STEC strains that produce Stx2d are exquisitely virulent in a streptomycintreated mouse oral challenge model (7). In this mouse model, pathogenicity of STEC is due primarily to toxin production, as demonstrated by the finding that mice are protected by passive immunization with anti-Stx2 antibodies (7). The low LD 50 of certain STEC strains correlates with the capacity of these strains to express the activable Stx2d (6). For example, one virulent strain, B2F1, that has an oral LD 50 of Ͻ10 colony forming units (8), produces two activable toxins named Stx2d1 (formerly SLT-IIvha) and Stx2d2 (formerly SLT-IIvhb) (5) to distinguish them from one another. By contrast, STEC O157:H Ϫ strain E32511/HSC that replicates as well as B2F1 in mouse small intestinal mucus but produces nonactivable Stx2c has an oral LD 50 of 10 10 colony forming units (8).
Amino acid sequence comparison of the activable Stx2d and nonactivable Stx2c reveal very few sequence differences (5). Indeed, the B subunit of Stx2d2 and Stx2c are identical, an observation that suggests that the key to the activable phenotype lies within the A subunit. There are only two amino acid differences between the A subunits of Stx2d2 and Stx2c, and both of these amino acids are located in the A 2 peptide (9), a fact suggesting that the effect crude intestinal mucus has on Stx2d involves the A 2 peptide.
The general nature of the alterations to Stx2d that occur after incubation with mucus were examined in a previous report (5). We demonstrated that there is no apparent modification to the A 1 subunit of activated Stx2d based on electrophoretic mobility; changes to the A 2 subunit could not be discerned at the time of the initial report (5) because the polyclonal serum used in the Western immunoblot analysis did not detect the A 2 peptide. We also showed that isolated intestinal mucus nicks the A subunit of Stx2d and Stx2 to A 1 and A 2 in a manner similar to trypsin treatment, but only mucus and not trypsin activates Stx2d. In support of this, we found that trypsin-nicked Stx2d that is not activated can subsequently be activated by isolated mucus (5). From these data we concluded that nicking of the Stx2d A subunit to the A 1 and A 2 peptides is not equivalent to activation. In this study, we sought to isolate and identify the factor(s) from mouse crude small intestinal mucus that activates Stx2d and to further address the effect of this activating factor on Stx2d.

EXPERIMENTAL PROCEDURES
Stx2d Activation Assay--The Stx2d cytotoxicity-enhancing activity of various mucus preparations and other samples was determined as described previously (5). Briefly, 50 -60 ng of Stx2d, purified as reported (10), was incubated in a total volume of 30 l with samples of different mucus preparations (as listed in Table I), chromatography fractions (from below), or various commercial porcine pancreatic elastase preparations (Sigma; Roche Molecular Biochemicals; Calbiochem, San Diego, CA) for 1 h-2 h at 37°C in the appropriate buffers. The commercially available porcine pancreatic elastase preparations were of varying purity. Based on estimations of Coomassie Blue-stained SDS-PAGE gels, the elastase content of these preparations ranged from ϳ50% to over 90% (data not shown). The porcine pancreatic elastase (EC 3.4.21.36) from Calbiochem (the purest) had a specific activity of 313 units/mg protein and a chymotrypsin content of 0.1% (as per the manufacturer's labeling). The commercial elastase preparations were resuspended in 140 mM NaCl, 5 mM KCl, 2.5 mM phosphate, 10 mM HEPES, pH 7.4, 2.0 mM CaCl 2 , and 1.3 mM MgSO 4 (11). The 50% cytotoxic doses (CD 50 ) of the treated toxin samples were determined on Vero cells as described previously (5,12,13). Fold activation of Stx2d cytotoxicity by various samples was determined by measuring the increase in Stx2d CD 50 per g of activating substance protein.
Some samples were preincubated with various protease inhibitors (Sigma) (see Table II and Fig. 4 for concentrations) for 30 min at 37°C prior to the addition of Stx2d. The percentage of maximal cytotoxicityenhancing activity following protease inhibitor treatment was determined as the fold activation in the presence of an inhibitor divided by the fold activation in the absence of the inhibitor of similarly treated samples.
Isolation of an Stx2d-activating Factor from Mouse Crude Small Intestinal Mucus (Fig. 1)-Crude mucus from the small intestine of 10 -15 male CD-1 mice was isolated and pooled as described previously (5,7). The mucus pool was diluted to a final protein concentration of 10 mg/ml with 10 mM HEPES buffer, pH 7.4, and stored at Ϫ80°C until further use. Samples of mucus proteins (300 mg) were precipitated by a final concentration of 60% ammonium sulfate, pH 8.9, for 30 min at 4°C. The precipitated mucus protein was resuspended in ion exchange chromatography (IEC) buffer (50 mM diethanolamine, 0.05 M NaCl, pH 8.9). (Note, all buffers used in this and subsequent steps contained 10% glycerol and 0.1 mM EDTA, and all chromatography was carried out at 22°C.) The resuspended sample was subjected to ion exchange chromatography on a Q Sepharose High Performance resin column (Amersham Pharmacia Biotech) with a bed volume of 25 ml. Following extensive washing of the column with IEC buffer, protein bound to the ion exchange column was eluted with an increasing linear salt gradient from 0.05 to 1 M NaCl in IEC buffer. Protein elution was monitored by absorbance at 280 nM.
Ion exchange chromatography fractions that contained Stx2d cytotoxicity-enhancing activity were pooled and adjusted to 4 M NaCl and to a pH of 7. The adjusted fractions were then subjected to hydrophobic interaction chromatography (HIC) on Phenyl Sepharose High Performance resin (Amersham Pharmacia Biotech) in a column with a bed volume of 25 ml. Following sample loading, the column was washed extensively with 50 mM diethanolamine, 4 M NaCl, pH 7, buffer (HIC buffer). Bound protein was eluted by a decreasing linear salt gradient of 4 to 0.05 M NaCl and a concurrent increasing pH gradient of pH 7 to 8.9 in HIC buffer.
Hydrophobic interaction chromatography fractions that contained Stx2d cytotoxicity-enhancing activity were pooled and n-octyl ␤-D-glucopyranoside (Sigma) was added to a final concentration of 0.1%. The pooled fractions were filtered through a YM 100 disc ultrafilter in an Amicon pressurized, stirred cell (Millipore, Bedford, MA). The ultrafiltration flow-through material was desalted to a final NaCl concentration of less than 10 mM by repeated dilution with 0.1% octyl ␤-Dglucopyranoside in water and ultrafiltration using a YM 10 filter (Millipore).
The desalted sample was separated by isoelectric focusing in a mini Rotofor cell (Bio-Rad) using 2% BioLyte ® ampholytes, pH range 4/6 (Bio-Rad) at a constant power of 10 W. Harvested Rotofor fractions were tested for Stx2d cytotoxicity-enhancing activity, and the peak fractions were pooled. Samples of some of these pools were subjected to a second round of isoelectric focusing in the absence of additional ampholytes to determine the isoelectric point of the isolated Stx2d-activating factor.
Electrophoresis-Fractions from the various steps in the Stx2d activator isolation scheme were analyzed by SDS-PAGE (14). Samples of 250 -500 l from each fraction were precipitated with 10% cold trichloroacetic acid. Precipitated proteins from each sample were washed with cold acetone and then resuspended in 1 M Tris-HCl, pH 8, and subjected to SDS-PAGE on continuous 10 or 12% gels. The polyacrylamide gels were stained with Silver Stain Plus (Bio-Rad).
Gelatin gel electrophoresis was used to identify potential Stx2d activators with proteolytic activity. The procedure was conducted as described previously (15) with some variation. Briefly, gelatin (Bio-Rad) was co-polymerized with acrylamide to a final concentration of 0.2% gelatin in an SDS-PAGE gel. Samples to be subjected to this analysis were concentrated by microcentrifugation on a Microcon YM-10 filter (Millipore), and the concentrated samples were applied directly to the gel. After separation of the proteins in the samples by electrophoresis, the gel was incubated in 2.5% Triton X-100 (Sigma) for 1 h at room temperature to remove SDS. Triton X-100 was removed by repeated washes in deionized water. The washed gel was then incubated overnight at 37°C in 10 mM HEPES buffer, pH 7.4. The gel was washed again with deionized water, fixed for 30 min in 10% acetic acid, 40% methanol, and then stained with colloidal Coomassie Blue (16).
Stx2d (0.6 -6 g as indicated in figure legends) or Stx2 (5 g) were treated with either mucus, trypsin (Promega, Madison, WI) or various samples from the Stx2d activator isolation scheme (as indicated in Figs. 5 and 6) and then analyzed by continuous 10% SDS-PAGE, two-dimensional electrophoresis, or 4 -12% NuPAGE Bis-Tris gels (as per the manufacturer's instructions; Novex, San Diego, CA). Two-dimensional electrophoresis was done by the method of O'Farrell (17) with a Bio-Rad Protean II xi 2-D Cell. Proteins were separated in the first dimension based on isoelectric point focused in a combination of pH range 3/10 and 4/6 ampholytes (4:1 ratio, respectively). The second dimension separated proteins based on size on continuous 15% SDS-PAGE. Separated proteins from these three gel systems were transferred to BAS-NC nitrocellulose (Schleicher & Schuell) by electroblotting (18). Western immunoblot analysis of the nitrocellulose filter was conducted as described previously (5) with a 1:1000 dilution of either polyclonal rabbit anti-Stx2 or affinity-purified anti-Stx2d A 2 (preparation of both described below) as the primary antibody.
Production of a Rabbit Anti-Stx2 Polyclonal Serum and Affinity Purified Rabbit Anti-Stx2d A 2 Antibodies-Antiserum against Stx2 was prepared by inactivating Stx2, purified as previously reported (10), with a 0.1% final concentration of formaldehyde at pH 7.5 for 6 weeks at ambient temperature. New Zealand White rabbits (1 kg) were given intraperitoneal injections of this Stx2 toxoid in Titermax (as per the manufacturer's instructions, CytRx, Norcross, GA) every 2 weeks on the following schedule: injections 1-3, 25 g of Stx2 toxoid; injection 4, 25 g of Stx2 toxoid plus 100 ng of native Stx2; injection 5, 25 g of Stx2 toxoid plus 5 g of native Stx2; and injections 6 -9, 25 g of Stx2 toxoid plus 25 g of native Stx2. Sera collected from test bleeds were inactivated at 55°C for 30 min and tested for Stx2 neutralization activity on Vero cells and for specificity by Western immunoblot analysis (data not shown).
Antiserum specific for the A 2 fragment of Stx2d was generated at Genemed Biotechnologies Inc. (San Francisco, CA) by injecting rabbits with a synthetic peptide, NEESQPECQITGDRP, conjugated to keyhole limpet hemocyanin. The antiserum was purified by affinity chromatography with the peptide as the absorbent. The specificity of the purified antiserum was tested by immunoblot analysis with native or denatured toxin preparations (data not shown).
Amino Acid Sequencing-To identify the isolated intestinal mucus factor with Stx2d activating activity, amino acid sequence information was obtained. The eluant with Stx2d cytotoxicity-enhancing activity from a hydrophobic interaction chromatography column (total volume, 10 ml) was concentrated to 1 ml by Amicon ultrafiltration with a YM 10 filter. This sample was further concentrated by precipitation with trichloroacetic acid and subjected to SDS-PAGE electrophoresis as described above. The separated proteins were stained by colloidal Coomassie Blue, and the stained gel was sent to the W. M. Keck Biomedical Mass Spectrometry Laboratory at the University of Virginia for sequencing of the activator protein by capillary liquid chromatographymass spectrometry and capillary liquid chromatography-tandem mass spectrometry. Peptide sequences obtained by these methods were compared with the nonredundant data base of GenBank TM mouse EST entries by BLAST (19) and Sequest algorithm searches.
Protein Concentration Determination-Protein concentrations were measured with a bicinchoninic acid protein assay kit (Pierce) with bovine serum albumin as a standard.

Survey of Cytotoxicity-enhancing Activity by Crude Intestinal
Mucus from Various Sources--Previous work (5) showed that crude mucus from the small intestine and colon of mice and the colon of humans activates (increases) the cytotoxicity of Stx2d for Vero cells. To begin to identify the Stx2d-activating factor in mucus, the cytotoxicity-enhancing activity of crude intestinal mucus preparations from several readily available sources was investigated. Mouse crude mucus preparations from all sources examined had activating activity similar to CD-1 mouse small intestinal mucus (the mucus used in the initial study (5)) (Table I). These preparations included crude mucus isolated from the small intestine of gnotobiotic, germ-free BALB/c mice, i.e. mice raised in a sterile environment without bacterial colonization. Porcine crude small intestinal and colonic mucus preparations from cesarean-derived/colostrum-deprived piglets (20) also exhibited Stx2d cytotoxicity-enhancing activity but at lower specific activities than mouse mucus (Table I). Intestinal mucus from ferrets and weaned piglets displayed higher cytotoxicity-enhancing activity than mucus from cesarean-derived, colostrum-deprived piglets, but none of these samples had a specific activity as high as that of intestinal mucus from mice. Weaned piglet mucus depleted of protein and ferret colonic mucus did not show Stx2d cytotoxicity-enhancing activity, nor did mucus isolated from the small intestine of rabbits. CD-1 mice were chosen as the source for crude small intestinal mu-cus for isolation of the Stx2d-activating factor based on specific activity and availability.

Some Serine Protease Inhibitors Blocked the Stx2d Cytotoxicity-enhancing Activity of CD-1 Mouse Crude Small Intestinal
Mucus-Mouse small intestinal mucus contains proteolytic activity as evidenced by the capacity of crude mucus to nick the Stx2d A subunit to the A 1 and A 2 peptides in a manner similar to purified trypsin (5). This nicking activity alone is not equivalent to activation (5), however. In this study we found that a general protease inhibitor mix blocked the Stx2d cytotoxicityenhancing activity of crude mucus (Table II). This observation taken with our findings that protein-depleted mucus lost activity (Table I) and that the cytotoxicity-enhancing activity in   mucus was sensitive to boiling and freeze-thawing (data not shown) suggested that this activator was a protein with proteolytic activity. To determine the kind of protease the activator might represent, more specific protease inhibitors were tested for the capacity to prevent Stx2d activation (Table II). Some protease inhibitors specific for serine proteases, e.g. soybean trypsin inhibitor, blocked the cytotoxicity-enhancing activity of mouse crude small intestinal mucus, whereas others did not. The pattern of the inhibition of Stx2d cytotoxicityenhancing activity of crude mucus by various serine protease inhibitors suggested that the Stx2d activator had serine protease activity (Table II) but did not immediately suggest the identity of the activator.

Isolation of a Factor from Mouse Crude Small Intestinal Mucus That Activated Stx2d
Activity-The Stx2d activator isolation scheme presented in Fig. 1 and detailed under "Experimental Procedures" was developed empirically. Stx2d cytotoxicity-enhancing activity generally eluted at 0.3 M NaCl from the Q Sepharose column and 1 M NaCl from the Phenyl Sepharose HP column (data not shown). Table III shows the increase in specific cytotoxicity-enhancing activity achieved at each step in the isolation. Each step of the isolation procedure also reduced the number of protein bands in the fractions that contained Stx2d cytotoxicity-enhancing activity (Fig. 2). Following the final isolation step, a single dominant ϳ32-kDa protein consistently remained in the Rotofor fractions that had Stx2d cytotoxicity-enhancing activity (Fig. 2). This protein was found to have an isoelectric point of ϳ5.2 (data not shown). This 32-kDa protein also had proteolytic activity when examined on gelatincontaining SDS-PAGE (Fig. 2).
Various Serine Protease Inhibitors Block the Stx2d Cytotoxicity-enhancing Activity in HIC Fractions-The ϳ32-kDa protein isolated as described above and suspected of having Stx2d cytotoxicity-enhancing activity frequently stuck to the filtration membranes used in the later steps of the activator isolation scheme (data not shown). Because of this loss of material in the ultrafiltration and desalting steps, hydrophobic interaction chromatography eluant fractions that contained Stx2d cytotoxicity-enhancing activity were used to conduct some further functional assays. HIC eluant was incubated with a series of protease inhibitors prior to incubation with Stx2d. Many of these inhibitors eliminated or reduced the Stx2d cytotoxicityenhancing activity of HIC eluant (Table II). Although the pattern of activator inhibition still did not definitively identify the Stx2d activator, the results supported the theory that the activator was a serine protease.
Identification of the Isolated Mucus Protein with Stx2d Cytotoxicity-enhancing Activity-To determine the identity of the dominant ϳ32-kDa protein in isolated fractions that contained Stx2d cytotoxicity-enhancing activity, some amino acid sequence data were determined. Material eluted from the HIC column was separated by SDS-PAGE, and the ϳ32-kDa protein believed to be the Stx2d-activating factor was eluted from the gel, digested with trypsin, and subjected to mass spectrometry.  2. A composite of SDS-PAGE of pooled peak Stx2d cytotoxicity-enhancing activity fractions from various steps in the isolation scheme of the Stx2d-activating factor and a lane from a gelatin gel on which the isolated activator was tested for proteolytic activity. Samples from the various steps were concentrated by precipitation with trichloroacetic acid where appropriate and subjected to electrophoresis on continuous 10% acrylamide gels. All of the lanes were stained with silver stain except the gelatin gel, which was stained with colloidal Coomassie Blue. Samples included: mouse crude small intestinal mucus (Crude Mucus, 300 g); mucus proteins resuspended in IEC buffer following ammonium sulfate precipitation ((NH 4 ) 2 SO 4 Precipitated, 300 g); eluant from a Q Sepharose ion exchange column (IEC Eluant, 27.5 g); eluant from a Phenyl Sepharose HP hydrophobic interaction column (HIC Eluant, 2.5 g); Rotofor fraction with Stx2d cytotoxicity-enhancing activity (Rotofor Fraction, 0.5 g; note that the presence of ampholytes in this sample interferes with protein concentration assays, so the protein concentration in this sample was estimated based on densitometric comparisons); and HIC eluant subjected to electrophoresis on a 0.2% gelatin gel and then processed as detailed under "Experimental Procedures" (Gelatin Gel, 2.5 g concentrated by microspin column). The numbers on the left indicate the positions of molecular mass markers in kilodaltons. The arrow on the right indicates the position of the isolated Stx2d-activating factor that was sequenced. a The presence of ampholytes in Rotofor samples interferes with protein concentration assays, so the concentration of Stx2d activator in Rotofor peak samples was estimated based densitometric comparisons of silver-stained SDS-PAGE (data not shown).
Five peptide sequences of various lengths were obtained from this procedure. Three of these peptide sequences were found by a Sequest search of the EST data base (data not shown) to have homology with various elastases (21). A BLAST search of the nonredundant data base of the GenBank TM mouse EST entries revealed that the amino acid sequence of these three peptides closely matched the translated amino acid sequence of a predicted mouse gene (gb:AA771563) that has homology to the human elastase IIIB gene (gb:M18692; Ref. 22 and Fig. 3). The other two peptide sequences (LASPVTLNAR and ATI-TLTSSAGK) had homology to mouse trypsin (23) and an E. coli outer membrane protein (24), respectively. Based on previous data that showed that trypsin does not activate Stx2d cytotoxicity (5) and the current finding that mucus from gnotobiotic animals contained cytotoxicity-enhancing activity (Table I), we concluded that the Stx2d activator was most likely a mouse elastase.
Elastatinal Inhibited the Cytotoxicity-enhancing Activity of Crude Mucus and Isolated Stx2d Activator-To provide further evidence in support of elastase as the Stx2d-activating factor found in crude mucus, the elastase-specific inhibitor elastatinal (25) was preincubated with either crude mucus from the small intestine, HIC eluant, or the isolated 32-kDa protein from the Rotofor fractions. As shown in Fig. 4, elastatinal inhibited the cytotoxicity-enhancing activity of all three samples, a finding that substantiates our hypothesis that the Stx2d activator is an elastase.
Commercially Available Porcine Pancreatic Elastase Preparations Activated Stx2d Cytotoxicity-To test whether elastase isolated from other animal sources can activate the cytotoxicity of Stx2d, various preparations of porcine pancreatic elastase from several suppliers were tested. All of these elastase prep-arations increased the cytotoxicity of Stx2d 10 -100-fold (data not shown). Furthermore, the Stx2d cytotoxicity-enhancing activity found in these commercial preparations was completely inhibited by elastatinal (data not shown). However, the porcine pancreatic elastase preparations were not as effective at enhancing Stx2d cytotoxicity as the isolated mouse Stx2d-activating factor. The specific activity of the porcine elastase prepa-  a dose-dependent manner (B). A, the elastase-specific inhibitor elastatinal prevents activation of Stx2d by mouse crude small intestinal mucus and isolated activator (mouse elastase). Crude mucus (50 g), hydrophobic interaction chromatography eluant (200 ng), or pooled peak activating activity fractions from Rotofor isoelectric focusing (30 ng, estimated by densitometry) were preincubated for 30 min in the presence or absence of elastatinal (1 g, 30 ng, or 30 ng, respectively) at 37°C. Stx2d (60 ng) was then added to each sample, and the samples were incubated an additional 1 h at 37°C. 10-fold serial dilutions of each sample on Vero cells were used to determine the CD 50 /mg protein. Samples include: Stx2d incubated with mouse crude small intestinal mucus ("Mucus"), in the presence (ϩ) or absence (Ϫ) of elastatinal; Stx2d incubated with cytotoxicity-enhancing activity-containing fractions eluted from a Phenyl Sepharose HP column (HIC Eluant), in the presence (ϩ) or absence (Ϫ) of elastatinal; and Stx2d incubated with pooled peak cytotoxicity-enhancing activity-containing, fractions from isoelectric focusing (Rotofor Peak) in the presence (ϩ) or absence (Ϫ) of elastatinal. The data represent geometric means of CD 50 /mg protein from at least three experiments. The error bars indicate one standard error of the mean. Note that the cytotoxicity of untreated Stx2d was similar to the cytotoxicity of elastatinal-treated samples, and elastatinal alone does not affect the cytotoxicity of Stx2d (data not shown). B, purified Stx2d (60 ng) was incubated with increasing concentrations of isolated mouse elastase (as indicated on the figure) for 1 h at 37°C. 10-fold serial dilutions of each mixture was transferred to Vero cells to determine the CD 50 /sample. Fold activation was calculated as the CD 50 for each elastase-treated sample divided by the CD 50 for buffer-treated Stx2d control. Each datum square represents the mean of three experiments. The error bars indicate one standard error of the means. Squares without error bars indicate errors too small to depict. rations ranged from 0.8-to 4.7-fold activation/g protein, whereas the isolated mouse elastase had an average specific activity of 170-fold activation/g protein.
Elastase Can Nick the Stx2d A Subunit to the A 1 and A 2 Peptides in a Trypsin-like Manner-When Stx2d was treated with isolated mouse elastase (from Rotofor fractions, Fig. 5) or commercially available porcine pancreatic elastase (data not shown), activation occurred and the Stx2d A subunit was nicked to A 1 and A 2 . The nicking of the Stx2d A subunit to A 1 and A 2 by elastase was inhibited by elastatinal, whereas nicking by trypsin was not (Fig. 5). Conversely, trypsin nicking of the Stx2d A subunit was inhibited by soybean trypsin inhibitor, whereas elastase nicking of the A subunit to A 1 and A 2 was not (Fig. 5). These findings, taken with the activation studies above, suggest that elastase has Stx2d nicking as well Stx2d cytotoxicity-enhancing activity. The nicking of the Stx2d A subunit to A 1 and A 2 by elastase appeared to be a separate event from the enhancement of Stx2d cytotoxicity by elastase because complete nicking occurred at concentrations of elastase below the minimum required to activate Stx2d cytotoxicity (Figs. 4B and 5). Furthermore, Stx2d was completely nicked by concentrations of porcine pancreatic elastase too low to activate the toxin, and the porcine elastase-nicked Stx2d could be activated by the addition of higher concentrations of elastase (data not shown).
Crude Mucus from Mouse Small Intestine and Isolated Mouse Elastase Affected the Mobility and Isoelectric Point of the Stx2d A 2 Peptide-Crude small intestinal mucus from mice has no apparent effect on the size of the A 1 peptide (5). To assess whether such crude mucus alters the size of the A 2 peptide, we generated an antiserum specific for the Stx2d A 2 peptide and used gel systems capable of resolving small differences between the molecular weights and changes in the isoelectric point of peptides. These analyses determined that there was a slight but reproducible difference in the mobility of the A 2 peptide of Stx2d treated with trypsin alone versus Stx2d treated with trypsin plus crude mucus or isolated mouse elastase (Fig. 6). The increase in mobility of the A 2 peptide was only seen in activated Stx2d samples and was completely inhibited by the addition of elastatinal (Fig. 6).
To further assess possible changes in elastase-treated Stx2d, two-dimensional electrophoresis was conducted. This analysis revealed that there was a change in the isoelectric point of the A 2 peptide of elastase-treated Stx2d (activated) versus buffertreated Stx2d (nonactivated). The A 2 peptide from elastasetreated Stx2d migrated to a more basic pH, 82 mm from the cathode end of the gel in the first dimension of the electrophoresis, whereas the A 2 peptide from buffer-treated Stx2d migrated to a more acidic pH, 91 mm from the cathode end of the gel (Fig.  7). There was no apparent difference in the isoelectric points of the A 1 subunits of elastase-treated versus buffer-treated Stx2d (Fig. 7). When buffer-treated versus elastase-treated Stx2 was analyzed by two-dimensional electrophoresis, there was no apparent change in the isoelectric point of the A 1 or A 2 peptides (data not shown). Elastase treatment also did not enhance the cytotoxicity of Stx2 for Vero cells (data not shown). DISCUSSION We have identified an Stx2d-cytotoxicity enhancing factor from mouse crude small intestinal mucus as an elastase. Al- though elastases have not previously been associated with activation or enhancement of toxic activity of bacterial toxins, a number of such toxins are expressed as inactive protoxins that require activation by proteolytic digestion to become toxic. This type of activation can occur either extracellularly (e.g. Clostridium septicum ␣ toxin (26)) or intracellularly (e.g. Pseudomonas exotoxin (27), anthrax toxin protective antigen (28), and others (29)). A second kind of toxin activation is more unusual and involves enhancement of toxicity of an already active toxin by exogenous proteases (e.g. Clostridium perfringens enterotoxin (30 -32)). Stx2d appears to fall into both of these categories. To become fully cytotoxic (33), Stx2d is nicked by proteases in or near a loop containing two Cys residues to separate the A subunit into the A 1 and A 2 peptides (2,29), and then, as we report here, this fully cytotoxic Stx2d appears to undergo an additional proteolytic cleavage event that enhances its cytotoxicity. The proteases that activate many toxins are often trypsin or chymotrypsin, in the case of extracellular activation (26,30,31), or furin, in the case of some intracellular activation (28). This is the first report of which we are aware that links elastase to toxin activation.
Protein sequencing based on mass spectrometry yielded five peptide sequences for the isolated activator. Three of these peptide sequences matched the predicted amino acid sequence of a translated mouse gene identified in the nonredundant data base of GenBank TM mouse EST entries (Fig. 3). This mouse gene is similar to the human elastase IIIB gene (22). The other two peptides had homology to mouse trypsin (24) and E. coli outer membrane protein 1 (24), respectively. Possible explanations for the presence of these two peptides in the sequencing reaction include: (i) the peptides were artifacts of mass spectrometry-based sequencing; (ii) one or both of these peptides originated from proteins that co-purified with the mouse elastase, e.g. mouse trypsin has a similar molecular weight and isoelectric point to mouse elastase (23); and (iii) one or both of these peptides may actually be part of the mouse elastase IIIB protein. The translated amino acid sequence for the mouse elastase IIIB gene (as presented in GenBank TM ) appears to be shorter than would be predicted from comparison with the human elastase IIIB protein sequence (Fig. 3). Therefore, the two unidentified peptides may have originated from the putative missing C-terminal amino acid sequence of this mouse elastase.
Elastatinal is a peptide derived from Actinomyces species that specifically inhibits elastase activity (25). This peptide completely inhibited the cytotoxicity-enhancing activity in both HIC peak fractions (Fig. 4) and the isolated 32-kDa protein in Rotofor fractions (Figs. 4 and 6). Higher concentrations of elastatinal also inhibited most, if not all, of the Stx2d cytotoxicity-enhancing activity of crude intestinal mucus (Figs. 4 and  6). These results are consistent with our conclusion that the 32-kDa protein is a mouse elastase with Stx2d cytotoxicityenhancing activity.
Examination of Table II reveals some differences between the effect of various protease inhibitors on the cytotoxicityenhancing activity of mouse crude small intestinal mucus versus partially purified Stx2d activating factor. Notably, although aprotinin, antipain, chymostatin and N ␣ -p-tosyl-Llysine chloromethyl ketone did not have any effect on the cytotoxicity-enhancing activity of crude mucus when these inhibitors were used at the maximum recommended concentration (see Table II for concentrations), these protease inhibitors blocked some of the cytotoxicity-enhancing activity of the partially purified activator. Two possible explanations for these apparently contradictory observations are as follows. First, the possibility exists that there are proteases other than elastase in intestinal mucus that can enhance the cytotoxicity of Stx2d, and these proteases are not inhibited by any of these four protease inhibitors. Second, and an explanation we consider more likely, is that the maximum recommended concentration for these protease inhibitors is insufficient to block the cytotoxicity-enhancing activity of elastase in crude mucus because elastase is only one of many serine proteases found in intestinal mucus. The many serine proteases in crude mucus may compete for inhibitors that are not specific or as efficient for elastase and overwhelm their capacity to block elastase-mediated enhancement of Stx2d. This theory is supported by examination of the inhibition activities of elastatinal and 3,4-dichloroisocourmarin. Elastatinal is an elastase-specific inhibitor and almost completely blocked the cytotoxicity-enhancing activity of crude mucus (Fig. 4), and 3,4-dichloroisocourmarin is described by the manufacturer as "inhibiting serine proteases like elastase" and completely blocked the cytotoxicity-enhancing activity of crude mucus (Table II).
Although commercially available porcine pancreatic elastase preparations activated Stx2d cytotoxicity, the specific activity of these preparations was significantly lower than that of isolated mouse elastase (see text). Although Tani et al. (22) have demonstrated that a human elastase IIIA (an elastase very similar to human elastase IIIB) cDNA probe hybridizes to RNA prepared form porcine pancreas, it is possible that this elastase type is not found in commercial porcine pancreatic elastase preparations. This possibility may explain the lower specific activity of the porcine pancreatic elastase preparations.
Human pancreatic elastase IIIB is similar (22) to another human elastase, cholesterol-binding pancreatic protease (34) (also called human pancreatic elastase I (35)). The preferred sites for cholesterol-binding pancreatic protease cleavage have FIG. 6. Crude mucus from mouse small intestine and isolated mouse elastase alter the mobility of the A 2 peptide in activated samples of Stx2d on SDS-PAGE. Stx2d (5.6 g) was incubated for 2 h at 37°C with trypsin (1.25 g in all samples) plus buffer (Control), plus mucus (Mucus, 200 g), or plus Rotofor peak fractions that contained mouse elastase (Activator, 3 g, as estimated by densitometry). This incubation followed a preincubation at 37°C for 30 min in the absence or presence of elastatinal (ϩElastatinal, 40 g). The samples were concentrated by microspin concentration and resolved on a 4 -12% NuPAGE Bis-Tris gel (Novex) and then electrotransferred to nitrocellulose. The Western immunoblot was developed with affinity-purified anti-Stx2d A 2 antiserum incubated overnight at 1:1000 followed by a conjugated secondary antibody and chemoluminesence-based detection. The arrows at the left indicate the mobility of the activated (bottom arrow) versus the nonactivated (top arrow) Stx2d A 2 peptides. Note that elastatinal alone did not alter the mobility of the Stx2d A 2 subunit (data not shown). The cytotoxicity-enhancing activity for each sample was determined by testing a portion of each sample on Vero cells. The fold activation was determined by dividing the CD 50 of each sample by the CD 50 for the buffer-treated Stx2d. Fold activation results less than or equal to 1 were considered 0-fold activation. been determined to be the peptide bonds at the carboxyl ends of Ala, Val, Leu, Ser, His, and Thr (34). This observation is relevant to our studies because one of the two differences between the A 2 subunit of activatable Stx2d2 and nonactivatable Stx2c is a Ser at position 291 in Stx2d2 in the place of a Phe in Stx2c (5). This Ser in Stx2d2 lies on the N-terminal side of a Leu found in both activatable and nonactivatable variants of Stx. We speculate that perhaps this leucine-serine bond is the site of the elastase cleavage that results in the activation of Stx2d2. Consistent with this speculation, the Stx2d A 2 peptide demonstrates a slight increase in electrophoretic mobility following elastase treatment (Fig. 6). Elastase treatment of Stx2d also increased the isoelectric point of the A 2 peptide, a finding that is consistent with the removal of a negatively charged amino acid like the glutamate at the C terminus of the Stx2d A 2 (5).
Identification of the precise site at which the Stx2d A 2 is clipped by elastase and how this effect serves to increase the cytotoxicity of Stx2d is the subject of a separate ongoing study. 2 Elastase cleaved Stx2d at two sites: between the A 1 and A 2 peptides (Fig. 5) and within the A 2 peptide (Figs. 6 and 7). We were unable to determine whether the nicking of Stx2d at the first site (either by elastase or another protease) was required prior to enhancement of cytotoxicity by elastase. We were also unable to define the precise location at which the Stx2d A was clipped to the A 1 and A 2 peptides by elastase, which left open the possibility that this site is different from that of the trypsin cleavage site. That such a putative alternate nicking site might directly cause the activated phenotype of elastase-treated Stx2d seems unlikely for the following reasons. First, mouse crude small intestinal mucus nicks both the Stx2d and the Stx2 A subunits to the A 1 and A 2 peptides but only activates Stx2d (5), even though the amino acid sequence between the Cys residues (amino acids 241-260 of Stx2d, the probable elastase clipping region) are identical for Stx2d and Stx2. Second, the second nicking event that occurred in the A 2 of activable Stx2d after elastase treatment (Figs. 6 and 7) that was not seen in similarly treated Stx2 (data not shown) correlated with activation (Figs. 6 and 7). Third, Stx2d was nicked (Fig. 5) to the A 1 subunit and A 2 peptide by concentrations of elastase that are too low to significantly enhance Stx2d cytotoxicity (Fig. 4B). Fourth, although soybean trypsin inhibitor blocked enhancement of Stx2d cytotoxicity by both crude mucus and isolated elastase (Table II), this protease inhibitor did not inhibit nicking by elastase (Fig. 5). This latter observation suggests that the elastase may act at amino acids residues at two different locations in Stx2d: one at the putative nick site between the two Cys residues (2) and the other at the putative activation site near the C terminus of Stx2d (Figs. 6 and 7). The amino acid residue at the nick site may be susceptible to elastase even in the presence of soybean trypsin inhibitor, whereas the amino acid residue at the activation site may not.
In this study we showed that the cytotoxic activity of Stx2d is increased by various elastases of mouse and porcine origin. These proteolytic enzymes are found in the intestinal mucus content of many species including humans (21,22,35,36), where they would presumably be available for the activation of Stx2d produced by infecting E. coli. We have shown that strains of E. coli that express activable Stx2d are exquisitely virulent in a streptomycin-treated mouse oral challenge model (8), and we now hypothesize that in that model mouse pancreatic elastase can activate Stx2d in vivo and increase the virulence of the infecting strain (5,6). Elastatinal is nontoxic to mice (25), and it may be possible to decrease the virulence of strains of STEC that express Stx2d through the use of orally administered elastatinal. FIG. 7. Isolated mouse elastase treatment alters the isoelectric point of the Stx2d A 2 peptide. Stx2d (5 g) was pretreated with 1.25 g of trypsin for 15 min at 37°C and then incubated for 2 h at 37°C with 200 l of HIC eluant (containing Stx2d activating activity) or buffer alone. The samples were concentrated by microspin concentration and separated by two-dimensional electrophoresis. In the first dimension, the proteins were separated based on isoelectric point in a pH range 3/10 and 4/6 ampholyte mixture. In the second dimension, proteins were separated on 15% continuous SDS-PAGE. The separated proteins were electrotransferred to nitrocellulose, and the Western immunoblot was developed with affinity-purified anti-Stx2d A 2 antiserum and anti-Stx2 antiserum incubated in combination overnight at 1:1000 followed by a conjugated secondary antibody and chemoluminesence based detection. A, buffer-treated Stx2d. B, HIC eluant-treated Stx2d. C, same as B superimposed over A. The single arrow at the left indicates the electrophoretic mobility of the Stx2d A 1 peptide in the second dimension, whereas the double arrow indicates the electrophoretic mobility of the Stx2d A 2 peptide in the second dimension. The buffer-treated Stx2d A 2 peptide focused 91 mm from the cathode end of the isoelectric focusing gel at a pH of ϳ4.78, whereas the HIC eluant-treated Stx2d A 2 peptide focused 82 mm from the cathode end of the isoelectric focusing gel at a pH of ϳ5. 15. The numbers at the left of A and the corresponding marks on each panel indicate the position of molecular mass markers in kilodaltons for the second dimension. The cytotoxicity of the sample in B was enhanced 17-fold for Vero cells versus the cytotoxicity of the sample in A.