Crystal structures reveal a thiol protease-like catalytic triad in the C-terminal region of Pasteurella multocida toxin

Abstract

Pasteurella multocida toxin (PMT), one of the virulence factors produced by the bacteria, exerts its toxicity by up-regulating various signaling cascades downstream of the heterotrimeric GTPases Gq and G12/13 in an unknown fashion. Here, we present the crystal structure of the C-terminal region (residues 575–1,285) of PMT, which carries an intracellularly active moiety. The overall structure of C-terminal region of PMT displays a Trojan horse-like shape, composed of three domains with a “feet”-,“body”-, and “head”-type arrangement, which were designated C1, C2, and C3 from the N to the C terminus, respectively. The C1 domain, showing marked similarity in steric structure to the N-terminal domain of Clostridium difficile toxin B, was found to lead the toxin molecule to the plasma membrane. The C3 domain possesses the Cys–His–Asp catalytic triad that is organized only when the Cys is released from a disulfide bond. The steric alignment of the triad corresponded well to that of papain or other enzymes carrying Cys–His–Asp. PMT toxicities on target cells were completely abrogated when one of the amino acids constituting the triad was mutated. Our results indicate that PMT is an enzyme toxin carrying the cysteine protease-like catalytic triad dependent on the redox state and functions on the cytoplasmic face of the plasma membrane of target cells.

Pasteurella multocida is a pathogenic bacterium causing a wide variety of mammalian and avian diseases, including swine atrophic rhinitis, fowl cholera, and hemorrhagic septicemia of cattle and buffaloes. Zoonotic infections in humans often occur because of bites, scratches, or being licked by domestic pet animals. Serotypes A and D of the organism produce a protein toxin [Pasteurella multocida toxin (PMT)], which draws attention both as a virulence factor responsible for the pathogenesis of swine atrophic rhinitis and as a previously unrecognized substance with unique biochemical activities. Swine atrophic rhinitis is a respiratory disease that causes a characteristic turbinate atrophy with severe bone loss. PMT is considered to impair turbinate osteogenesis mainly by inhibiting osteoblastic differentiation, which results in the bone loss (1, 2). Currently, PMT is an important component of the vaccine against this disease. In the field of basic cellular biology, the toxin is known to act on various types of cultured cells as a highly potent mitogen, even causing anchorage-independent growth in some fibroblastic cells (3, 4). To date, PMT has been found to activate phospholipase Cβ (PLCβ), Rho, and MAP kinases, such as ERK and JNK, and tyrosine phosphorylation of focal adhesion kinase and paxillin in target cells (5–10). These alterations in biosignaling events are considered to originate from the activation of two distinct pathways dependent on the heterotrimeric GTPases Gq and G12/13 (10–13). However, the molecular mechanism of the action of PMT remains unknown. The toxin is at least believed to bind a putative receptor on target cells, to be internalized by endocytosis, and to enter the cytoplasm from the endosome (7, 14), implying that the toxin possesses an enzymatic region to modify as-yet-unknown intracellular target molecules. PMT consists of a single polypeptide chain of 1,285 aa. Several lines of evidence indicate that the N-terminal region of the toxin binds to target cells and that the C-terminal region carries the intracellularly active moiety, the nature of which remains unknown (15–18). The N-terminal region is partly homologous to Escherichia coli cytotoxic necrotizing factors, CNF1 and CNF2 (19, 20). In contrast, in the C-terminal region of PMT, there is neither a known catalytic motif nor a region homologous with any other toxin or enzyme, and therefore no structural information has been available for elucidation of the intracellular action of the toxin.

In this study, we determined the crystal structure of the C-terminal region of PMT (C-PMT) (residues 569–1,285). C-PMT was found to consist of three distinct domains. The most N-terminal domain contributes to the membrane localization of C-PMT. The most C-terminal domain organizes a cysteine protease-like catalytic triad that depends on its redox state. Our results indicate that PMT is an enzyme with the Cys–His–Asp triad, which functions on the cytoplasmic face of the plasma membrane.

Results

Intracellular Activities of C-PMT.

Previous studies indicated that the intracellular toxic activity of PMT was attributable to its C terminus (15, 17). To confirm this finding, we introduced the recombinant C-PMT (residues 569–1,285) directly into Swiss 3T3 cells with the aid of HVJ liposomes and examined the cells for elevated levels of DNA synthesis and PLC activity, which have usually been determined for the quantitative measurement of PMT toxicity (2, 4, 9). C-PMT and the holotoxin stimulated inositol phosphate accumulation, indicating PLC activation, and DNA synthesis [supporting information (SI) Fig. 6] when introduced with the HVJ liposomes. In contrast, when extracellularly added, C-PMT was inactive, whereas the holotoxin stimulated both PLC activity and DNA synthesis. These results indicate that C-PMT is equipped to exert intracellular toxicity, but not to bind to and enter cells.

Three Distinct Domains of C-PMT.

The recombinant C-PMT was crystallized as described in ref. 21, and the crystal structure was solved by single isomorphous replacement techniques with a mercury derivative and subsequently refined at a resolution of 1.9 Å (SI Table 1). The N-terminal 29 residues, including Met–Gly residues, a hexa-histidine tag peptide, a 15-residue spacer peptide (21), and the first 6 residues (569–574) of C-PMT were not visible in the defined electron density map and therefore were omitted from the model (Fig. 1). The size of the C-PMT molecule was approximately 87 × 84 × 30 Å. The overall structure of C-PMT displays a Trojan horse-like shape, composed of three domains arranged in “feet,” “body,” and “head,” which were designated domains C1 (residues 575–719), C2 (residues 720–1,104), and C3 (residues 1,105–1,285), respectively (Fig. 1). The “feet” C1 domain is composed of seven helices (H1–H7). A DALI search through the database of protein three-dimensional structures (22) revealed significant structural homology between the N-terminal four helices of the C1 domain and the N-terminal domain of Clostridium difficile toxin B [Protein Data Bank (PDB) ID code 2BVL] (Z = 11.1, rms deviation = 2.6 Å) (Fig. 2A). In addition, when the region of the first four helices (residues 590–670) was subjected to a sequence homology search by PSI-BLAST, the corresponding region of toxin B was proposed as a sequence homologue with 21% identity and 41% similarity. C. difficile toxin B and a related lethal toxin from Clostridium sordellii were reported to associate with anionic lipids through their N-terminal domains. This feature provides the toxins with the lipid membrane environment in which their enzymatic reaction against geranyl-geranylated Rho GTPases is facilitated (23, 24). Therefore, we examined whether the C1 domain, like the N-terminal domains of the clostridial toxins, has the ability to localize to the lipid membrane (Fig. 2B). C-PMT expressed in 293T cells showed plasma membrane localization, whereas C-PMT lacking the first four helices of the C1 domain [ΔC1(4H), residues 672–1,285] did not. Similarly, EGFP tagged with the four helices of the C1 domain [C1(4H), residues 569–671] was found in the cell membrane region, whereas native EGFP was distributed in the nucleus and cytoplasm. These results indicate that the C1 domain possesses a signal that leads the toxin to the cell membrane.

Fig. 1.

Overall structure of C-PMT. (A) Ribbon diagram of C-PMT. Helices and β-strands are drawn as coils and flat arrows, respectively. C-PMT has a Trojan horse-like shape and is composed of three different domains, C1 (blue), C2 (green), and C3 (magenta), which correspond to the feet, body, and head, respectively. The two subdomains of the C2 domain are defined by green and light green. The red “N” and “C” indicate the N and C terminus, respectively. The image was prepared with the program PYMOL. (B) Amino acid sequence and secondary structure elements of C-PMT. The C1, C2, and C3 domains are represented by the same colors as shown in A. Lines below the sequence indicate the secondary structure, with boxes for helices (H) and arrows for β-strands (β).

Fig. 2.

Structure and membrane-targeting feature of the C1 domain. (A) Comparison of the C1 domain (blue) with the N-terminal domain of C. difficile toxin B (orange). A stereoview of wire ribbon models is shown. (B) Localization of ectopically expressed C-PMTs. C-PMT, ΔC1(4H), C1(4H)–EGFP, and EGFP were expressed in 293T cells and were located by fluorescence microscopy (green). The nuclei and the plasma membrane region were visualized by staining cells with DAPI (blue) and anti-human CD46 antibody (red), respectively. The images are presented at the same magnification. (Scale bar: 10 μm.)

The “body” C2 domain is the largest of the C-PMT domains, and is composed of 18 helices and nine β-strands (Fig. 1). This domain is separable into two subdomains with typical α/β-structures, which are generally considered characteristic of nucleotide-binding proteins. The first subdomain, which corresponds to the hind feet of the Trojan horse, consists of five helices and three β-strands. The long helix (H14) connects the “hind feet” subdomain to the second subdomain, which has 11 helices and six β-strands. A homology search with DALI proposed folylpolyglutamate synthetase (PDB ID code 1FGS) (Z = 5.3) and cdc14bs phosphatase (PDB ID code 1OHC) (Z = 5.1) as structural homologues of the second subdomain. Both enzymes interact with the phosphate groups of substrates or cosubstrates, which may provide a clue as to the function of the C2 domain as discussed below. The long helix H14 hydrophobically interacts with H7 of the C1 domain and with H13 and H22 of the C2 domain. These hydrophobic interactions between helices appear to determine the central major frame of the C-PMT molecule.

The “head” C3 domain, which is connected to the C2 domain through the V-shaped long loop (1,087–1,104), shows a typical α/β-protein fold, with seven β-strands (β10–β16) and eight helices (H26–H33) (Fig. 1B and see SI Fig. 7). This domain is composed of two subdomains; the first is constructed with H26–H31 and β13–β15, and the second is constructed with H32 and β11, β12, and β16. These two subdomains provide the cleft space, which is surrounded in particular by H29, H30, H32, β12, β14, and the loop between H28 and H29 (referred to as “loop_H28-H29” hereafter) (Fig. 3A and see SI Figs. 7 and 8A).

Fig. 3.

The amino acid residues in the cleft space of the C3 domain. (A) A magnified view of the C3 domain. The amino acid residues that were mutated in this study are shown as stick models. Residues of those mutations that resulted in an abrogation of the toxic activity are labeled in black, whereas residues of those mutations that had no effect are labeled in gray. Hx, helices. (B) Effects of amino acid substitutions examined by an SRE-SEAP reporter assay. The quiescent Swiss 3T3/SRE-SEAP cells were incubated with PMT and PMT mutants, as described in Materials and Methods. (Left) The PMT mutants that were judged to be inactive. (Right) Those PMT mutants that were judged to be active. Each point represents the mean ± SD of triplicate samples. The representative results from three independent experiments are shown. Statistical significance was determined by ANOVA (∗, P < 0.01 compared with the cells treated with wild-type PMT).

Cleft Space of the C3 Domain as an Active Center.

We assumed that the C3 domain was the biologically active center of PMT for the following reasons: (i) a PMT mutant that was C-terminally truncated by 155 aa no longer showed intracellular activity (15), and (ii) the amino acid substitution of Cys¹¹⁶⁵, His¹²⁰⁵, or His¹²²³ in the C3 domain eliminated the intracellular activity of PMT (16, 18). Thus, we examined the structure of the C3 domain more closely. At the cleft space of the C3 domain, the loop_H28-H29, which is surrounded by water molecules and a phosphate ion probably derived from the precipitant, thus, implying solvent-accessibility, is bound up by a disulfide bond between Cys¹¹⁵⁹ and Cys¹¹⁶⁵ and faces His¹²⁰⁵ (Fig. 3A). His¹²²³, the substitution of which also eliminates the toxic activity (16), is in the vicinity of this region, although its side chain has the opposite orientation to the cleft space. His¹²⁰⁵ appeared to form a hydrogen bond with Asp¹²²⁰ at a distance of 2.64 Å. His¹²⁰⁵ and Asp¹²²⁰ were individually mutated to leucine and alanine, respectively, and the toxic activity of each mutant was assessed by the secreted embryonic alkaline phosphatase (SEAP) assay, which is based on the fact that PMT stimulates the serum response element (SRE)-driven transcriptional activity in a Rho-dependent manner (12). Both mutations resulted in abrogation of the toxic activity (Fig. 3B). In many hydrolytic enzymes, a His–Asp pair composes a catalytic triad with a nucleophilic center residue, such as serine or cysteine. At the cleft space of the C3 domain, Ser¹¹⁶⁹ and Ser¹²²² neighbor the His¹²⁰⁵–Asp¹²²⁰ interface (Fig. 3A), implying that one of these serine residues might take part in the triad. However, substitutions of these residues with alanine did not reduce the toxic activity (Fig. 3B), which indicates that the serine residues are not involved in the catalytic action of the toxin.

Catalytic Triad in the C3 Domain.

We next examined the possibility that the His–Asp pair forms a triad with cysteine. In the loop_H28-H29, Cys¹¹⁵⁹ and Cys¹¹⁶⁵ form a disulfide bond (Fig. 3A). The substitution of Cys¹¹⁶⁵ with serine, glycine, or arginine abolishes the PMT activities (15, 17, 18), which was confirmed in this study (Fig. 3B). It was also reported that substitution of Cys¹¹⁵⁹ with glycine resulted in a significant reduction in the toxic activity (18). However, we found that PMT was fully active when Cys¹¹⁵⁹ was replaced by serine, which is structurally more similar to cysteine than glycine (Fig. 3B), indicating that the loss of activity on the mutation of Cys¹¹⁶⁵ is not due to a collapse of the overall structure caused by the breaking of the disulfide bond, but that the side chain group of Cys¹¹⁶⁵ might be involved in the biological activities of PMT. This finding prompted us to dissect the structure of PMT mutants in which Cys¹¹⁵⁹ and Cys¹¹⁶⁵ are independently replaced by serine (designated C-PMT_1159S and C-PMT_1165S, respectively) so that the disulfide bond does not form. Each mutant protein was crystallized and its structure was solved, at a resolution of 2.6 Å for C-PMT_1159S and 2.4 Å for C-PMT_1165S. C-PMT_1159S was crystallized in a dimeric form, in which the C3 domain differs from that of C-PMT in its position relative to the C1 and C2 domains (see SI Materials and Methods and Table 1). In overall structure, the C1 and C2 domains in C-PMT_1159S correspond well to those of C-PMT. However, marked structural changes were found in the loop_H28-H29 compared with that of C-PMT (Fig. 4 A–C); Ser¹¹⁶⁴ is shifted by 3.07 Å, and this acts to extend the H29 helix, which displaces Cys¹¹⁶⁵ toward the putative catalytic pocket. Furthermore, Cys¹¹⁶⁵ then forms a thiolate imidazolium ion pair with His¹²⁰⁵, and the triad Cys¹¹⁶⁵–His¹²⁰⁵–Asp¹²²⁰ emerges (Fig. 4B). C-PMT_1159S maintains the cleft space, in which a thiol group of Cys¹¹⁶⁵ is exposed on the molecular surface (SI Fig. 8B). Interestingly, in C-PMT, the corresponding position is occupied by water molecules, implying the susceptibility of this region to environmental alterations. The crystal structure of C-PMT_1165S also shows a similar folding to C-PMT_1159S in the loop_H28-H29. The side chain of Ser¹¹⁶⁵ corresponding to Cys¹¹⁶⁵ of C-PMT and C-PMT_1159S is also turned out toward His¹²⁰⁵, and a pseudo triad, Ser–His–Asp, is organized (Fig. 4C).

Fig. 4.

Dynamic alterations in the loop_H28-H29 upon breakage of the disulfide bond. (A–C) The loop_H28-H29 and the proximal regions in C-PMT (A), C-PMT_1159S (B), and C-PMT_1165S (C) are shown. A σ_A-weighted omit electron density map (contoured at 3 σ) is superimposed on the stick model of key residues, Cys¹¹⁵⁹ (Ser¹¹⁵⁹ in B), Cys¹¹⁶⁵ (Ser¹¹⁶⁵ in C), His¹²⁰⁵, Asp¹²²⁰, and Gln¹²²⁵. (D) Comparison of the active sites among C-PMT_1159S (light green) and C-PMT_1165S (light blue) with papain (pink), AvrPphB (lime green), and N-acetyltransferase (orange). His¹²⁰⁵, Asp¹²²⁰, and Cys¹¹⁶⁵ of C-PMT_1159S were superimposed on the corresponding residues of the counterpart proteins by Lsqkab in the CCP4 program suite (30).

The homology search by secondary structure mapping (SSM) (www.ebi.ac.uk/msd-srv/ssm) for C-PMT_1165S showed the best match in the secondary structure surrounding the putative active pocket with Pseudomonas syringae AvrPphB (PDB ID code 1UKF) (25), a cysteine protease. In addition, the secondary structures of the C3 domain and AvrPphB have structural equivalents in the papain-like cysteine protease fold. The papain-like fold of the C3 domain includes most of the central antiparallel β-sheets (β11–β15) and H29. A search for structural homology by DALI also revealed that the putative catalytic pocket of the C3 domain bore similarities to Salmonella typhimurium arylamine N-acetyltransferase (PDB ID code 1E2T) (Z = 3.2) (26), which has a Cys–His–Asp catalytic triad. The catalytic triads of C-PMT_1159S and C-PMT_1165S resemble those seen in AvrPphB, N-acetyltransferase, and papain (PDB ID code 1POP) (Fig. 4D). In the superimposed structures, the catalytic cysteines (serine for C-PMT_1165S) and histidines align very well, as well as the third catalytic residues; aspartic acid in C-PMT_1159S, C-PMT_1165S, AvrPphB, and N-acetyltransferase and asparagine in papain. At the active site of papain, Gln¹⁹ functions to make the oxyanion hole together with the amino group of the essential Cys, which is known to stabilize the main-chain carbonyl group of the P1 residue of the substrate (27). Asn⁹³ of AvrPphB is considered to take part in the formation of the oxyanion hole (25). In C-PMT_1159S and C-PMT_1165S, Gln¹²²⁵ resides at the corresponding position; the deviation in distance between the amino groups of Cys and Gln was 0.12 Å, when compared between C-PMT_1159S and papain. Substitution of Gln¹²²⁵ with alanine or glutamic acid eliminated the toxic action of PMT (Fig. 3B). These results suggest that Gln¹²²⁵ participates in the catalytic action probably by forming part of the oxyanion hole.

Discussion

In this study, we determined the crystal structure of C-PMT, which is fully active when directly introduced into mammalian cells, and predicted a number of its structure–function relationships (Fig. 5). C-PMT was found to consist of three domains, C1, C2, and C3, each of which seems to make a distinct contribution to the intracellular toxic effects even though the role of the C2 domain remains to be established. The C1 domain was found to structurally resemble the phospholipid-binding domain of C. difficile toxin B and to actually mediate the plasma membrane-localization of C-PMT. The C2 domain is composed of two subdomains, each of which shows a typical α/β-fold. This domain, the largest, appears to stabilize the entire structure of C-PMT. The C3 domain possesses the latent cysteine protease-like catalytic triad: On the breaking of the Cys¹¹⁵⁹–Cys¹¹⁶⁵ disulfide bond, Cys¹¹⁶⁵ is displaced and participates in the triad with His¹²⁰⁵ and Asp¹²²⁰.

Fig. 5.

Model of the domain architecture of membrane-associated C-PMT. A ribbon diagram of C-PMT_1159S is presented. The C1, C2, and C3 domains are in blue, green, and magenta, respectively. Note that the membrane-targeting region in the C1 domain (yellow oval) and the catalytic cleft in the C3 domain (red oval) reside on the same face of the molecule, which faces the cytoplasmic surface of the membrane.

Previous reports have shown that PMT binds to an unknown receptor on target cells, is internalized by endocytosis, and escapes from the acidic endosome into the cytoplasm. The N-terminal region of the toxin is believed to mediate these procedures (14, 17). We hypothesize about the subsequent action of PMT as follows: Although it is unknown whether the N-terminal region is clipped off when the toxin molecule enters the cytoplasm, the C-terminal region, at least, corresponding to C-PMT, is transferred to the vicinity of the plasma membrane with the aid of the C1 domain. During this step, the catalytic triad of the C3 domain must be completely organized through the breaking of the Cys¹¹⁵⁹–Cys¹¹⁶⁵ disulfide bond, which may occur spontaneously in the reducing environment of the cytoplasm or may be catalyzed by an enzymatic reducing system such as thioredoxin/thioredoxin reductase as reported for some bacterial toxins (28, 29). Eventually, the catalytic reaction should take place on the cytoplasmic face of the membrane, where a target molecule of PMT should reside (Fig. 5). The roles of the C2 domain in the toxic actions could not be clearly defined by analyzing its structure. The C2 domain has two typical α/β-folds, which are known to provide biochemically active sites in various enzymes. We speculate that the C2 domain helps to recognize and hold the target molecule or an additional cosubstrate. In addition, we do not exclude the possibility that PMT may be a bifunctional enzyme; one activity being ascribed to the C3 domain and the other to the C2 domain. This idea may explain the pleiotropic effects of PMT, which simultaneously stimulates distinct pathways downstream of Gq and G12/13 (12, 13).

Although the targets of PMT have yet to be identified, it is noteworthy that C-PMT was located in the vicinity of the plasma membrane. In the clostridial toxins, the first 18 residues composing the first short helix were reported to be responsible for binding to phospholipids (23, 24). The corresponding helix of C-PMT is situated on the same molecular face as the catalytic cleft (Fig. 5). Therefore, if its structure is unchanged, even after the toxin is associated with the membrane, the catalytic cleft should be closely directed to the cytoplasmic face of the membrane, which may help to narrow down the candidates for target molecules. The most probable targets are the heterotrimeric GTPases, Gq and G12/13, which are believed to be associated with the hepta-helical transmembrane receptors beneath the plasma membrane. Alternatively, another molecule involved in the GTPase-dependent pathways beneath the membrane could be a target. The hepta-helical transmembrane receptors, Gβγ-subunits, and other accessory factors, including RGS proteins, may be candidates. One or some of these molecules could be modified by the catalytic action of the triad in the C3 domain. In general, the Cys–His–Asp triad is known to basically perform acyl-hydrolysis (e.g., peptidase activity) or an acyl-transfer (e.g., transglutaminase activity) reaction. The toxin did not show any peptidase activity against the commercially available substrates that we examined (data not shown). What serves as a substrate for PMT and how the toxin modifies it are questions that remain to be answered.

The minimal region of PMT required for the intracellular action has been a subject of controversy. One research group reported that the region from residue 720 to the C terminus was intracellularly active (17), whereas another group reported that the shortest active fragment was from residue 581 to the C terminus and that a fragment from residue 701 was inactive (15). It was found in this study that C-PMT (residue 569 to the C terminus) is active and that the C1 domain (residues 575–719) probably plays a role in the intracellular action. Elimination of the first four helices of the C1 domain from C-PMT significantly reduced the toxic activity (S.K. and M. Miyazawa, unpublished data), indicating that the C1 domain may not be essential but may be supportive for the intracellular action. Thus, we consider that C-PMT represents the intracellularly active moiety of the toxin. This does not conflict with the fact that the membrane translocation domain is reportedly composed of regions spanning residues 402–457 (14). Furthermore, the receptor-binding domain could be located on the N-terminal side of the translocation domain. Resolution of the structure of the N-terminal remainder of PMT may provide insights into the overall organization of the functional domains.

Materials and Methods

Fluorescence Microscopy.

293T cells in 0.5 ml of DMEM and 10% FBS were seeded at 2 × 10⁵ cells per well into 24-well plates containing glass coverslips (Matsunami, Osaka, Japan). After incubation overnight, the cells were transfected with 1.0 μg of pcDNA3–C-PMT, pcDNA3–ΔC1(4H), pcDNA3–C1(4H)–EGFP, pcDNA3–EGFP, or pcDNA3 by using 2.0 μl of Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. After a 24-h transfection, the cells were fixed with 3.0% paraformaldehyde in PBS for 10 min, treated with 10% FCS in PBS for 10 min, and subjected to staining of the plasma membrane with polyclonal anti-human CD46 rabbit antibody (a gift from T. Seya, Hokkaido University Graduate School of Medicine, Sapporo, Japan) and Alexa 546-conjugated anti-rabbit IgG antibody (Invitrogen). After being permeabilized with 0.1% saponin in PBS for 30 min, the cells were sequentially stained by anti-PMT mouse monoclonal antibody 1B6, Alexa 488-conjugated anti-mouse IgG antibody (Invitrogen), and DAPI (Invitrogen). Polyclonal and monoclonal anti-PMT antibodies were raised by immunizing rabbits and mice, respectively, with PMT as an antigen. The immunolabeled cells and C1(4H)–EGFP and EGFP-expressing cells were subjected to microscopy with an epifluorescence microscope (BX50; Olympus, Tokyo, Japan).

SEAP Reporter Assay.

Swiss 3T3 cells integrated with the SRE-SEAP reporter gene were established by retroviral infection by using a pCXbsr retroviral vector bearing a blasticidin S resistance gene and designated Swiss 3T3/SRE-SEAP cells. Swiss 3T3/SRE-SEAP cells were seeded at 1 × 10⁴ cells per well into a 96-well microtiter plate and incubated at 37°C for 2 days to become quiescent. The quiescent cells were incubated with PMT or its mutants for 24 h. Fifty microliters of the culture supernatant was transferred into a well of the 96-well plate and heated at 65°C for 10 min to inactivate the endogenous alkaline phosphatases; SEAP is not inactivated by this procedure because of its thermostability. The samples were mixed with 150 μl of 50 mM Tris·HCl at pH 8.0, including 0.4% para-nitrophenyl phosphate and 10 mM l-homoarginine, and incubated at 37°C for 60 min. The amount of para-nitrophenol released was determined by measuring absorbance at 405 nm referenced at 620 nm with an ELISA microtiter plate reader (Multiskan, Titertek, Germany). The arbitrary units of the SEAP level were calculated as follows: SEAP unit = (Abs at 405 nm × 1000)/[Abs at 620 nm × volume (ml) × time (min)].

Supporting Materials and Methods.

Detailed information on the production of vectors and recombinant proteins, structural analyses of crystals, and the procedures used in the experiments yielding data presented in SI figures are described in SI Materials and Methods and Table 2.

Abbreviations

C1(4H)

the first four helices of the C1 domain

ΔC1(4H)

the C1 domain lacking the first four helices

C-PMT

C-terminal region of PMT

PLC

phopholipase C

PMT

Pasteurella multocida toxin

SEAP

secreted embryonic alkaline phosphatase

SRE

serum response element.