Reduced virulence of the MARTX toxin increases the persistence of outbreak-associated Vibrio vulnificus in host reservoirs

Opportunistic bacteria strategically dampen their virulence to allow them to survive and propagate in hosts. However, the molecular mechanisms underlying virulence control are not clearly understood. Here, we found that the opportunistic pathogen Vibrio vulnificus biotype 3, which caused an outbreak of severe wound and intestinal infections associated with farmed tilapia, secretes significantly less virulent multifunctional autoprocessing repeats-in-toxin (MARTX) toxin, which is the most critical virulence factor in other clinical Vibrio strains. The biotype 3 MARTX toxin contains a cysteine protease domain (CPD) evolutionarily retaining a unique autocleavage site and a distinct β-flap region. CPD autoproteolytic activity is attenuated following its autocleavage because of the β-flap region. This β-flap blocks the active site, disabling further autoproteolytic processing and release of the modularly structured effector domains within the toxin. Expression of this altered CPD consequently results in attenuated release of effectors by the toxin and significantly reduces the virulence of V. vulnificus biotype 3 in cells and in mice. Bioinformatic analysis revealed that this virulence mechanism is shared in all biotype 3 strains. Thus, these data provide new insights into the mechanisms by which opportunistic bacteria persist in an environmental reservoir, prolonging the potential to cause outbreaks.

Opportunistic bacteria strategically dampen their virulence to allow them to survive and propagate in hosts. However, the molecular mechanisms underlying virulence control are not clearly understood. Here, we found that the opportunistic pathogen Vibrio vulnificus biotype 3, which caused an outbreak of severe wound and intestinal infections associated with farmed tilapia, secretes significantly less virulent multifunctional autoprocessing repeats-in-toxin (MARTX) toxin, which is the most critical virulence factor in other clinical Vibrio strains. The biotype 3 MARTX toxin contains a cysteine protease domain (CPD) evolutionarily retaining a unique autocleavage site and a distinct β-flap region. CPD autoproteolytic activity is attenuated following its autocleavage because of the β-flap region. This β-flap blocks the active site, disabling further autoproteolytic processing and release of the modularly structured effector domains within the toxin. Expression of this altered CPD consequently results in attenuated release of effectors by the toxin and significantly reduces the virulence of V. vulnificus biotype 3 in cells and in mice. Bioinformatic analysis revealed that this virulence mechanism is shared in all biotype 3 strains. Thus, these data provide new insights into the mechanisms by which opportunistic bacteria persist in an environmental reservoir, prolonging the potential to cause outbreaks.
Unlike obligate pathogens, opportunistic pathogens evolve to increase their fitness in environmental hosts by controlling their virulence, which may facilitate host-to-host spreading (1,2). Vibrio vulnificus is a life-threatening opportunistic pathogen that is becoming an increasing threat to human health worldwide because of global climate change (3,4). V. vulnificus-associated infections mostly occur through consumption of contaminated seafood and can result in primary septicemia with a high fatality rate (50%) in some severe cases (5,6). Exposure of open wounds to infected sea water or seafood products, which can cause wound infections and secondary septicemia, is also associated with a substantial mortality rate (25%) (7). Strains of V. vulnificus are classified into three biotypes based on their biochemical characteristics and phylogeny: biotype 1 strains such as MO6-24/O and CMCP6 cause the majority of human infections responsible for the entire spectrum of illness, including primary septicemia; biotype 2 strains are primarily eel pathogens; and biotype 3 strains, such as strain BAA87, cause wound infections and bacteremia and possess hybrid biochemical properties of both biotypes 1 and 2 (3,4). Biotype 3 strains have been exclusively isolated from outbreaks of severe wound infections and septicemia cases associated with a tilapia farm in Israel (8). Comparative genomic analysis revealed that biotype 3 is a distinct clone descended from the parental environmental population that acquired pathogenic potential by horizontal gene transfer from other Vibrio strains (9).
Meanwhile, the biotype 3 strains are less pathogenic than the biotype 1 strains (10). Among virulence factors produced by V. vulnificus strains, multifunctional autoprocessing repeats-in-toxin (MARTX) toxins are the primary exotoxins that regulate host inflammatory responses and immune defense (11) and facilitate colonization in the intestine and dissemination to distal organs (12)(13)(14)(15). The biotype 3 MARTX toxin harbors a distinct effector content derived from a putative progenitor MARTX toxin of the biotype 1 strain that contributes to the significantly reduced virulence of the strain (10). This study suggests that rather than showing increased potency, outbreak strains may retain decreased potency of key virulence factors as a strategy to enter the human food chain by persisting longer in natural hosts (10). However, the mechanisms by which the distinct effector content of the biotype 3 MARTX toxin leads to reduced pathogenicity are unclear.
MARTX toxins secreted by many bacterial pathogens contain disease-related, modularly structured effector domains that are released via processing events upon entry into host cells (16)(17)(18). The diversity of MARTX toxin effector domains correlates with distinct cytopathicities or cytotoxicities and with the overall toxicity of MARTX toxin-expressing strains (19,20). The cysteine protease domain (CPD) located at the end of the effector domain regions of all MARTX toxins directs the proteolytic processing of effector modules after it undergoes activation and autocleavage by binding to cytosolic inositol hexakisphosphate (InsP 6 ) (17,18). The makes caterpillars floppy-like (MCF) effector domain, which is present in about 30% of MARTX toxins, leads to effector module processing after binding to ADP-ribosylation factor (ARF) (16).
Our previous study revealed that effector domains within the biotype 3 MARTX toxin are unusually processed by its internal CPD and domain X effector (DmX, a homolog of MCF) (16). The CPD of biotype 3 MARTX toxin lacks the proteolytic function to process its associated effector domains, even though it has autocleavage activity in the presence of InsP 6 . In addition, only DmX is released from the toxin by the allosteric activator ARF after autocleavage and detachment from the autocleaved CPD, suggesting that these distinctive processing pathways may be correlated with the toxin potency and therefore the reduced virulence of this strain.
These previous data led us to hypothesize that the opportunistic biotype 3 strain has evolved to promote fitness in the host (e.g., tilapia) by relieving MARTX toxin-mediated virulence. In this study, we found that the biotype 3 MARTX toxin CPD evolved to contain an atypical N-terminal autocleavage site and a distinct C-terminal β-flap region that attenuates the autoproteolytic activity of the CPD required for the processing of associated effector domains, which leads to significantly reduced virulence in cells and mice. Bioinformatics analysis suggests that the V. vulnificus biotype 3 strains evolved to retain the atypical CPD and thus have lower MARTX toxindriven pathogenicity than biotype 1 and 2 strains, which maintain the functional MARTX toxin and exhibit potent virulence.

The biotype 3 MARTX toxin CPD is autocleaved atypically
Our previous study showed that the internal CPD (CPD BAA87 ) of MARTX toxin secreted by the V. vulnificus biotype 3 clinical strain BAA87 (MARTX BAA87 ) does not process its associated effector domains, even though CPD BAA87 itself is autocleaved in the presence of InsP 6 (16). A multiple sequence alignment of CPDs within MARTX toxins expressed by different Vibrio species, including V. vulnificus and Vibrio cholerae, revealed that CPD BAA87 has a distinct unique sequence (A 4090 -W 4091 -T 4092 ) instead of the conventional cleavage sequence, which is known as an X 1 -L-X 2 motif (where X 1 and X 2 are small amino acids such as Ala and Ser) (21) (Fig. 1A). The other distinctive feature of CPD BAA87 is the Cterminal region corresponding to the β-flap in the CPD (CPD cholerae ) of the V. cholerae MARTX toxin, which plays a critical role in InsP 6 binding and enzymatic function (17). The amino acid sequence of this region in CPD BAA87 is significantly different from that of other CPDs (Fig. 1A).
To identify the autocleavage site of CPD BAA87 , we carried out an in vitro autocleavage assay using N-terminally extended CPD BAA87 in the presence of InsP 6 and analyzed the product by Edman sequencing. As expected, CPD BAA87 was not processed at the unique sequence AWT. Rather, it was autocleaved between Leu4067 and Glu4068 (Fig. 1B), which is located 24 amino acids upstream of X 1 -L-X 2 , the conventional autocleavage site of other CPDs (Fig. 1A). Based on these results, we hypothesized that CPD BAA87 may have evolved to function differently from other CPDs to control the virulence of the biotype 3 strain via the MARTX toxin.
To test this hypothesis, we mutated the unique sequence AWT to ALA and analyzed the autocleavage activity. The mutant CPD (CPD AWT/ALA ) was preferentially and efficiently (comparable activity with CPD MO6-24/O of the MO6-24/O strain) autocleaved at the ALA site rather than at its authentic cleavage site, VLE (Fig. 1C). The cleavage sequence, ALA, was confirmed by Edman sequencing. The autocleavage activity at the VLE site was much weaker in CPD BAA87 than in CPD MO6-24/O (Fig. 1C). Noticeably, the autocleavage motif of CPD BAA87 contains a bulky side chain residue (glutamate), whereas the conventional cleavage motif has small amino acid residues at positions X 1 and X 2 (21), suggesting that this difference may cause the reduced autocleavage activity. Substitution of the VLE motif with ALA increased the autocleavage activity to a lever higher than that of WT CPD BAA87 (Fig. 1D). It should be noted that migration of the mutant CPD by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) was faster than that of WT CPD BAA87 . Collectively, these data suggest that CPD BAA87 is an altered CPD that evolved to moderate the virulence of the MARTX toxin.
The structure of CPD BAA87 reveals nonfunctionality for processing of associated effectors To uncover the molecular basis of the reduced function of CPD BAA87 with respect to both its autocleavage and the processing of associated effectors, we initially tried to crystallize the autocleaved CPD BAA87 , covering Glu4068-Asp4300, but could ultimately only crystallize CPD by deletion of its N-terminal region (Ala4090-Asp4300), suggesting that the N-terminal Glu4068-Asp4089 region may be flexible and not crystallizable. Its structure was determined at a resolution of 2.2 Å (Table S1). It should be noted that the expression construct of CPD BAA87 for crystallization included an N-terminal hexahistidine-tag, a tobacco etch virus protease recognition site, and restriction enzyme cloning site-derived residues (AM), followed by the unique sequence A 4090 -W 4091 -T 4092 (Fig. S1A). When incubated with InsP 6 , the CPD was autocleaved at the AMA 4090 site because of its high similarity to the conventional CPD autocleavage sequence (ALA). Purified CPD starts at Ala4090 (Fig. S1B), indicating that the determined structure is a representative autocleaved conformation of CPD BAA87 lacking the flexible N-terminal Glu4068-Asp4089 fragment.
The overall structure of the CPD BAA87 is similar to that of CPD cholerae . Its catalytic domain (Asn4099-Ser4262), which shows the most similarity with CPD cholerae , contains a sevenstranded β-sheet with five central parallel strands and two antiparallel capping strands flanked by three helices (Fig. 2, A and B). CPD BAA87 also contains a β-flap region (Ala4265-Lys4288) that has an entirely different conformation from the corresponding region of CPD cholerae (Fig. 2, A and C and Fig. S2). Although InsP 6 binds to CPD BAA87 covering residues 4056 to 4300 and residues 4090 to 4300 with binding affinity constants (K d ) of 3.95 and 2.43 μM, respectively (Fig. S3), an electron density map of the molecule could not be found, and thus, the structure model does not include InsP 6 (see below).
CPD cholerae forms a positively charged pocket for InsP 6 binding that is required for active site opening (17). The structural integrity of the β-flap in CPD cholerae is essential for forming both the pocket and the substrate-binding cleft adjacent to the dyad active residues. The residues Arg3610, Lys3611, and Lys3623 on the β-flap play critical roles in maintaining the integrity of the β-flap (Fig. 2D, upper panels) (17). Although all residues involved in formation of the InsP 6binding pocket in CPD cholerae are strictly conserved in CPD BAA87 , including Arg3610, Lys3611, and Lys3623, the structure of CPD BAA87 reveals a dissimilar conformation of the pocket (Fig. 2D, lower panels). In the structure of CPD cholerae , the region Gly3567-Phe3579, which includes the active residue Cys3568, forms a flexible long loop and exposes the active site to solvent (Fig. 2B). CPD BAA87 shows a different structural conformation in that the C terminal half of the corresponding region Asp4237-Phe4243 (Asp3573-Phe3579 in CPD cholerae ) is incorporated into the third helix of CPD BAA87 , resulting in a longer helix (Fig. 2B). Furthermore, the β-hairpin in the β-flap region in CPD BAA87 is rotated about 105 clockwise relative to the corresponding β-hairpin in CPD cholerae (Fig. 2C). These structural changes in CPD BAA87 are facilitated by several hydrogen bonds between His4183 and Ser4233, Gln4240 and Ser4264, Asp4237 and Ser4291, and Tyr4277 and Asn4290, as well as successive layers of hydrophobic interactions that allow the N-terminal Trp4091 to cover the crater-like hydrophobic surface (Fig. 2E). These events consequently change the locations of Arg4274, Lys4275, and Lys4287, which correspond to the residues Arg3610, Lys3611, and Lys3623 on the β-flap in CPD cholerae , moving them further from the InsP 6 -binding site (Fig. 2D, lower panels) and force the active residue Cys4232 into an improper position away from the other active residue, His4183 (Fig. 2C).
Overall, these data suggest that CPD BAA87 that is autocleaved at the site V 4066 -L 4067 -E 4068 constitutes an atypically structured CPD because of significant conformational changes in the N-terminal flexible region spanning Glu4068-Ala4090, the Trp4091 that caps the C-terminal hydrophobic region following the β-hairpin, the conserved catalytic core domain, and the C-terminal β-flap region, which forms a shield together with Trp4091 to block the active site (Fig. 3A). Consequently, unlike autocleaved CPD cholerae (Fig. 3B), the autocleaved CPD BAA87 becomes nonfunctional with respect to Evolutionary retarded MARTX toxin processing  Evolutionary retarded MARTX toxin processing the processing of linker regions between associated effector domains within the MARTX toxin.
We believe that the unavailability of an electron density map of InsP 6 may be a consequence of the significantly changed conformation of the β-flap region after autocleavage of CPD BAA87 (Fig. 3, A and B), which alters the locations of residues critical for InsP 6 binding (Arg4274, Lys4275, and Lys4287) (Fig. 2D). Mutation of Lys3611 or Lys3623 in CPD cholerae , which correspond to Lys4275 and Lys4287 in CPD BAA87 , abolished InsP 6 binding (17).

Evolutionary insights into inhibition of MARTX toxin activation by CPD BAA87
The autocleavage activity of CPD MO6-24/O was much greater than that of CPD BAA87 (Fig. 1C). To elucidate the role of the βflap in the regulation of the enzymatic function of CPD BAA87 , we generated a chimeric CPD BAA87 (CPD BAA87 /β-flap MO6-24/O ) in which the β-flap is substituted with the β-flap of CPD MO6-24/O (β-flap MO6-24/O ). The chimeric CPD BAA87 /βflap MO6-24/O showed very low autocleavage activity (Fig. 4A), suggesting that the parental β-flap region is necessary for autocleavage in conjunction with the extensional N-terminal region of CPD BAA87 . Mutation of the unique sequence AWT to ALA in CPD BAA87 /β-flap MO6-24/O (CPD BAA87/ALA /βflap MO6-24/O ) to mimic CPD MO6-24/O resulted in comparable autocleavage activity to that of CPD MO6-24/O (Fig. 4B). Circular dichroism (CD) spectroscopy analysis revealed that there are no significant structural differences between CPD variants and WT CPD BAA87 (Fig. S4) We further evaluated the roles of the extended N-terminal region containing the unique autocleavage site and the β-flap region in CPD BAA87 in MARTX toxin activation. Consistent Evolutionary retarded MARTX toxin processing with the previous results, CPD BAA87 was incapable of processing the effector domains within MARTX BAA87 even though it was autocleaved in the presence of InsP 6 (Fig. 4C). The mutant CPD BAA87/ALA showing similar autocleavage activity as CPD MO6-24/O (Fig. 1C) also showed no effector processing activity (Fig. 4C). However, CPD BAA87 /β-flap MO6-24/O was processed between the Rho GTPase-inactivation domain (RID) and the alpha/beta hydrolase domain (ABH), although its autocleavage activity was reduced, as shown in Figure 4A, suggesting that this chimeric CPD forms an open conformation of the active site. The mutant CPD BAA87/ALA /β-flap MO6-24/O showed similar processing activity as CPD BAA87 /β-flap MO6-24/O (Fig. 4C). Consistent with these results, the affinities of the chimeric proteins for InsP 6 were lower than that of the WT (Fig. S5). These data suggest that the extended N-terminal region, including the distinct autocleavage site VLE and the β-flap in CPD BAA87 , has evolved to have moderate autocleavage activity and reduced effector domain processing, which may significantly reduce the MARTX toxin-mediated virulence of the biotype 3 strain.

The structural modulation of CPD BAA87 reduces virulence in vivo
Full activation of the MARTX toxin effector domains depends on internal proteases such as CPD and MCF that cleave the toxin, releasing the effector domains, which dysregulate host cell functions (16,17). To assess whether the processing activity of CPD BAA87 directly influences the virulence of the MARTX toxin, we utilized a variant of the V. vulnificus MO6-24/O strain in which exotoxins including hemolysin, elastase, secretory protease, and secretory phospholipase A 2 were deleted to evaluate MARTX toxin-specific effects (16).  (Fig. 5A).
We first evaluated the virulence of the engineered strains in infected HeLa cells by measuring cell rounding, which is caused by inhibition of Rho GTPase via the effector RID within the MARTX toxin (22,23). Most cells infected with the parental strain became rounded by 120 min, whereas cell rounding was significantly reduced in cells infected with the MO6-24/O/MARTX BAA87 strain (Fig. 5, B and C). The strain containing the CPD MO6-24/O -mimicking enzyme showed comparable virulence to the parental strain (Fig. 5, B and C). These results further suggest that the structural modulation of CPD BAA87 is linked to the significantly reduced virulence of the biotype 3 strain through inhibition of MARTX toxin processing.
We further assessed the significance of MARTX toxin processing by CPD BAA87 during the pathogenesis of V. vulnificus in vivo. Mice challenged with the MO6-24/O/MARTX BAA87 strain showed 50% mortality at 48 h postinfection (hpi), whereas the parental strain resulted in 100% mortality within 15 hpi (Fig. 5D). Strikingly, no mice challenged with the strain containing the CPD MO6-24/O -mimicking enzyme survived at 15 hpi, which is comparable with the mortality of the parental strain, further demonstrating that the structural modulation of CPD BAA87 is the determinant of the virulence moderation of V. vulnificus biotype 3 strains.

Discussion
Opportunistic pathogens evolutionarily alter their virulence for adaptation to the environment and to facilitate infection of susceptible hosts (24). In this study, we found that the opportunistic pathogen V. vulnificus biotype 3, which caused an outbreak in humans associated with tilapia farming, may have evolved a form of an MARTX toxin with attenuated virulence. This may have allowed it to persist in the host for a longer time. MARTX toxins are expressed across multiple bacterial species and genera and are considered the primary virulence factors of V. vulnificus (13,25). Once translocated into host cells, MARTX toxins undergo a proteolytic processing event via their internal CPD that releases functionally discrete effector domains into the host cell to dysregulate cellular substrates (16)(17)(18).
Genome-wide single nucleotide polymorphism genotyping revealed that the V. vulnificus biotype 3 genome is based on the core genome of a biotype 1 strain belonging to the clade B group (26). Evolutionary analysis of the rtxA1 genes of various V. vulnificus strains suggested that the biotype 3 MARTX toxin may have been generated from the MARTX toxin of a biotype 1 clade B strain via a recombination event (10,27). The biotype 3 MARTX toxin contains five effector domains, including domain of unknown function (DUF1), RID, ABH, ExoY, and DmX (a homolog of MCF), whereas a representative biotype 1 MARTX toxin consists of DUF1, RID, ABH, MCF, and Ras/Rap1-specific endopeptidase domain (RRSP). It has been proposed that modification of the effector domain content in the biotype 3 MARTX toxin results in altered toxin potency and contributes to the emergence of V. vulnificus strains with outbreak potential (10). That study demonstrated that restoring the biotype 3 toxin to the biotype 1 progenitor toxin by replacing the ExoY and DmX domains with MCF and RRSP significantly increased virulence in mice. However, that study replaced the effector domains ExoY and DmX as well as the CPD with MCF, RRSP, and CPD from biotype 1 strain LOS6966.
Sequence alignment revealed that the CPD of the biotype 1 strain LOS6966 is a conventional CPD, enabling it to process associated effector domains (Fig. S6). We previously reported that the MCF homolog DmX in the biotype 3 MARTX toxin is autocleaved via allosteric activation by host ARF (16). Another study also showed that DmX is N terminally autocleaved by the interaction with ARF (28). These data, together with the present results, led us speculate that upon entry into host cells, the biotype 3 MARTX toxin undergoes autocleavage of the internal CPD by binding Evolutionary retarded MARTX toxin processing to InsP 6 in the cytoplasm (Fig. 6A, left panel), and then, DmX is cleaved from the ExoY by activation of ARF (Fig. 6A,  middle panel). Consequently, DmX is released from the toxin, but other effectors are not, because it becomes a nonfunctional protease after autocleavage of the CPD (Fig. 6A, right panel). This atypical MARTX toxin processing event may be related to the significantly attenuated virulence of the biotype 3 strain, rather than the modification of effector content. Substitution of the CPD with a conventional CPD would result in release of DUF1, RID, ABH, ExoY, and DmX from the toxin, significantly increasing the virulence of the pathogenic strain. It should be mentioned that a conventional CPD-like CPD MO6-24/O does not cleave between DUF1 and RID (16). Note that the free forms of RID, ABH, ExoY, and DmX target and dysregulate Rho-family GTPases (22,29), autophagy and endosome pathways (30), the cAMP signaling pathway (31), and the Golgi, as well as an unknown substrate(s) (16, 28), respectively.

Evolutionary retarded MARTX toxin processing
Intriguingly, sequence analysis of CPDs from V. vulnificus MARTX toxins revealed that the functionality of the CPD correlates with the biotype. This indicates that most MARTX toxins of biotype 1 and 2 strains preserve conventionally functional CPDs for both self-cleavage and associated effector domain processing, whereas all biotype 3 strains retain attenuated CPD autoproteolytic activity (Fig. 6B).
Strong evidence suggests that climate change is influencing outbreaks and changing the epidemiology of Vibrio infections on a worldwide scale and will increase the spread of aquatic Vibrio pathogens, with detrimental effects on human and animal health (32)(33)(34). Although the V. vulnificus biotype 3 strain was originally isolated from outbreaks associated with tilapia farming in Israel (8), a biotype 3-associated case with Figure 6. The functionality of CPDs correlates with the evolutionary classification of V. vulnificus biotypes. A, proposed model for MARTX BAA87 processing. Upon entry into host cells, the CPD of MARTX BAA87 is allosterically activated by binding to InsP 6 and autocleaved (left panel). DmX released from the autocleaved CPD is then allosterically activated via the interaction with ARF and autocleaved (middle panel). The released DmX complexed with ARF disrupts the Golgi structure, and DmX may induce cell shrinking via modification of an unknown target(s) (16,28). B, phylogenetic tree of CPDs from different V. vulnificus strains. Amino acid sequence analysis revealed that CPDs can be classified into three groups based on their sequences and predicted functional discrepancies, shown as arcs outside of the tree. The biotypes of V. vulnificus strains are indicated with different symbols: biotype 1, red circle; biotype 2, blue circle; biotype 3, green circle; unidentified, magenta diamond. ARF, ADP-ribosylation factor; CPD, cysteine protease domain; DmX, domain X effector; DUF1, domain of unknown function; ExoY, ExoY-like adenylate cyclase domain; MARTX, multifunctional autoprocessing repeats-in-toxin; RID, Rho GTPase-inactivation domain.
Evolutionary retarded MARTX toxin processing primary septicemia was also reported in Japan (35), indicating that the habitat of the V. vulnificus biotype 3 has extended to East Asia and might spread worldwide in the future. Our study providing new insights into the mechanisms by which opportunistic bacteria make evolutionary trade-offs to increase their fitness in the environmental reservoir should expand our understanding of the regulation of virulence across a wide range of emerging infectious diseases.

Experimental procedures
Bacterial strains, plasmids, and culture media The bacterial strains and plasmids used in this study are listed in Table S2. Escherichia coli and V. vulnificus strains were grown in Luria-Bertani medium at 37 C and in Luria-Bertani supplemented with 2.0% (w/v) NaCl at 30 C, respectively, with appropriate antibiotics.  Table S3. The resulting PCR products were treated with restriction enzymes corresponding to sites present in the primers and ligated into linearized pET21d, pProEX, or pPosKJ vectors for protein expression. To construct the pHis-parallel1 vector expressing chimeric CPD BAA87 (residues 4056-4300) in which the β-flap (residues 4265-4291 of CPD BAA87 ) was substituted with that of CPD MO6-24/O (residues 3760-3786), DNA fragments encoding the flanking region of β-flap BAA87 (11,521-12,792 bp and 12,874-13,591 bp of rtxA1 from V. vulnificus strain BAA87) and β-flap MO6-24/O (11,358 bp of rtxA1 from V. vulnificus strain MO6-24/O) were amplified, respectively, and simultaneously inserted into pHis-parallel1 using the NcoI/XhoI restriction sites via a one-step sequence-and ligation-independent cloning method (36). The DNA fragment encoding chimeric CPD BAA87 was amplified and ligated with NdeI/XhoI-digested pPosKJ to generate pPosKJ_CPD BAA87 /βflap MO6-24/O . In addition, pPosKJ_CPD BAA87/ALA and pPosKJ_CPD BAA87/ALA /β-flap MO6-24/O expressing CPD BAA87/ ALA or CPD BAA87/ALA /β-flap MO6-24/O , respectively, were produced by site-directed mutagenesis using pPosKJ_CPD  or pPosKJ_CPD BAA87 /β-flap MO6-24/O , respectively, as a template. Each plasmid was transformed into the E. coli NiCo21 (DE3) strain (New England Biolabs), and transformed cells were grown to an absorbance at 600 nm (A 600 ) of 0.5 to 0.6. Expression of recombinant proteins was induced by adding 0.5 mM isopropyl-β-D-thiogalactopyranoside (LPS Solution). Transformed cells were further incubated at 18 C for 18 h and harvested by centrifugation at 4000g for 5 min. The harvested cells were resuspended in buffer A [300 mM NaCl, 50 mM Tris-HCl (pH 7.5), 5% glycerol, and 1 mM DTT] supplemented with 10 mM imidazole and lysed with a high-pressure homogenizer (Nano DeBEE, B.E.E. International). The lysates were centrifuged at 28,000g at 4 C for 1 h. The supernatants were loaded onto a column with nickel-nitrilotriacetic acid resin (Qiagen). The column was washed with 1 l of buffer A supplemented with 20 mM imidazole. Then, the hexa-histidine (His 6 )-tagged or His 6 -VHb-tagged recombinant proteins were eluted with buffer A supplemented with 250 mM imidazole. If necessary, the His 6 tag or VHb tag was removed by treatment with recombinant tobacco etch virus protease during dialysis against buffer B [150 mM NaCl, 50 mM Tris-HCl (pH 7.5), 5% glycerol, and 1 mM DTT] at 4 C for 10 h. Nonspecific or undigested proteins were eliminated by loading onto nickelnitrilotriacetic acid resin and further purified by sizeexclusion chromatography using a HiLoad 16/60 Superdex 75 column (GE Healthcare Life Science) preequilibrated with buffer B. The purified proteins were concentrated using Amicon Ultra-15 100 K or 10 K columns (EMD Millipore), flash-frozen in liquid nitrogen (LN 2 ), and kept at −80 C until use.

DNA cloning and protein purification
Crystallization, X-ray diffraction, and structure determination The DNA fragment encoding CPD (residues 4090-4300) was cloned into pHis-parallel1 using NcoI/XhoI restriction sites, and recombinant proteins were purified as described above. Specifically, the partially purified CPDs were incubated with a five-fold excess of InsP 6 at 37 C for 1 h in cleavage assay buffer [60 mM NaCl, 20 mM Tris-HCl (pH 7.5), and 250 mM sucrose] (17). The activated CPDs were further purified by size-exclusion chromatography using a HiLoad 16/60 Superdex 75 column (GE Healthcare Life Science). Finally, purified CPDs were concentrated to 21 mg/ml using an Amicon Ultra-15 10 K column (EMD Millipore) in buffer B, and after the addition of 2 mM InsP 6 , crystals were grown in 96-well MRC crystallization plates (Swissci) via the sittingdrop vapor-diffusion method. Crystals were produced at 20 C after 3 days in reservoir solution containing 1.0 M potassium sodium tartrate and 0.1 M MES:NaOH (pH 6.0). For Xray diffraction experiments, the CPD crystals were cryoprotected using the reservoir solution supplemented with 23% glycerol and stored in LN 2 until diffraction. Then, X-ray diffraction data were collected at 2.2 Å resolution from the CPD crystals at beamline 5C at the Pohang Accelerator Laboratory. X-ray diffraction data were processed and scaled using the HKL2000 software package (37). The initial structure of CPD BAA87_4090-4300 was solved by molecular replacement in Phaser-MR from PHENIX using the V. cholerae CPD structure (PDB code, 3EEB) as a template model (17,38). Model building was carried out with the COOT program, and refinement, including the translation-liberation-screw procedure, was implemented using phenix.refine from PHENIX (38,39). The overall statistics of data collection and structure refinement are summarized in Table S1.

In vitro processing assay and Edman sequencing
To investigate the autocleavage activity of CPDs (CPD BAA87 , CPD BAA87/ALA , CPD BAA87 /β-flap MO6-24/O , and CPD BAA87/ALA /β-Evolutionary retarded MARTX toxin processing flap MO6-24/O ), the indicated CPDs (5 μM) were incubated at 37 C for 1 h or the indicated durations in the absence or presence of InsP 6 (5 μM) in cleavage assay buffer. The reaction was stopped by addition of SDS-PAGE sample buffer and boiling at 98 C for 5 min. Proteins were loaded, separated on 15% polyacrylamide gels, and visualized by staining with Coomassie brilliant blue dye. The band intensities of cleaved and uncleaved CPDs from three independent experiments were quantified using the ImageJ program of the National Institutes of Health (40) and plotted using GraphPad Prism (version 6). For identification of the autocleavage site of CPD BAA87 , N-terminally His 6 -tagged CPD BAA87 (residues 4090-4300) or His 6 -VHb-fused CPD BAA87 (residues 4056-4300) was incubated with InsP 6 at 37 C for 1 h in cleavage assay buffer. The cleaved CPD BAA87 was further purified by size-exclusion chromatography on a Superdex 75 10/300 GL column (GE Healthcare Life Science) preequilibrated with buffer B. Then, the cleaved CPD BAA87 was concentrated using an Amicon Ultra-15 10 K filter (EMD Millipore), and N-terminal amino acid sequences were analyzed using a PPSQ-51A protein sequencer (Shimadzu). In addition, following autocleavage, the N-terminal amino acid sequences of CPD AWT/ALA and CPD VLE/ ALA were confirmed by Edman sequencing.
To analyze the effector domain-processing activity of the CPDs, the purified effector domains of MARTX BAA87 toxin (residues 1959-4060, 0.05 mg/ml) were incubated with the indicated CPDs (CPD BAA87 , CPD BAA87/ALA , CPD BAA87 /βflap MO6-24/O , and CPD BAA87/ALA /β-flap MO6-24/O ; 0.05 mg/ml) in the absence or presence of InsP 6 (1 mM). The reactions were separated on 10% polyacrylamide gels and visualized by staining with Coomassie brilliant blue dye. Additionally, the proteins were transferred to Immobilon-P polyvinylidene fluoride membranes (0.45 μm pore size; EMD Millipore) before staining, and the N-terminal amino acid sequences of the processed effector domains were analyzed.

Isothermal titration calorimetry analysis
To measure the binding affinity between CPD BAA87 and InsP 6 , isothermal titration calorimetry measurements were carried out using a VP-isothermal titration calorimetry Microcalorimeter (Microcal). Following the purification of enzymatically inactive CPD C/S (residues 4056-4300; C4232S) and CPD C/S (residues 4090-4300; C4232S) as described above, the proteins were dialyzed against buffer B for 12 h at 4 C. Proteins (80 μM) were degassed for 20 min using a ThermoVac (Microcal), then placed in the reaction cell, and InsP 6 (1 mM) dissolved in the same dialyzed buffer was inserted into the syringe for titration. The reaction heat data from InsP 6 in the syringe into the buffer in the cell was subtracted from that of InsP 6 in the syringe into the protein in the cell. All data were processed and fitted using the Origin program (version 7) supplied with the instrument. The chimeric proteins, CPD BAA87 /β-flap MO6-24/O C/S(4056-4300) and CPD BAA87 /β-flap MO6-24/O C/S(4090-4300), were prepared in the same manner as described above. Following dialysis, 0.75 mM InsP 6 in a syringe was titrated into 32 μM of each chimeric protein in the reaction cell. The data were processed using the Origin program (version 7).

Quantification of round HeLa cells and mouse survival experiments
To analyze CPD-mediated cytotoxicity by V. vulnificus variants, HeLa cells (5 × 10 5 /well) grown in 6-well plates (Ther-moFisher Scientific) were treated with PBS (mock) or V. vulnificus (the parental strain, the MO6-24/O/MARTX BAA87 strain, MO6-24/O/MARTX BAA87 /CPD/β-flap MO6-24/O, or the MO6-24/O/MARTX BAA87 /CPD MO6-24/O-mimic strain) at an multiplicity of infection of 0.1. Images of three random fields from each well were obtained at the indicated time-points using the Floid Cell Imaging Station (ThermoFisher Scientific), and rounded HeLa cells were manually counted using the cell counter plug-in of the NIH ImageJ program. For mouse survival tests, 5-week-old female ICR (CrljOri:CD1) mice were purchased from Orientbio and adapted to the laboratory environment for 24 h. Parental, MO6-24/O/MARTX BAA87 , or MO6-24/ O/MARTX BAA87 /CPD MO6-24/O-mimic strains of V. vulnificus were grown to an A 600 of 0.5 in Luria-Bertani supplemented with 2.0% (w/v) NaCl and harvested by centrifugation at 1000g for 3 min. The bacterial cells were washed once and diluted to 5 × 10 7 colony forming units/ml using PBS. Subsequently, mice (n = 10 per each group) were subcutaneously administered 50 μl of the bacterial suspension on the dorsal side under light isoflurane anesthesia (10, 16) and were monitored every 2 h for 48 hpi. The Institutional Animal Care and Use committee of the Korea Research Institute of Bioscience and Biotechnology approved all mouse experiment protocols (approval no. KRIBB-ACE-18186).

Analysis of CPDs and construction of the phylogenetic tree
Amino acid sequences corresponding to the CPDs of registered MARTX toxins from different V. vulnificus strains were downloaded from the National Center for Biotechnology Information database (https://www.ncbi.nlm.nih.gov/protein) Evolutionary retarded MARTX toxin processing and aligned using ClustalW. The sequences were analyzed by the maximum likelihood method, and the phylogenetic tree was constructed using Molecular Evolutionary Genetics Analysis X software (41). The CPDs were classified by comparing residues of the putative autocleavage sites and βflap regions with those of CPD BAA87 and CPD MO6-24/O .

Circular dichroism spectroscopy
CD spectroscopy was carried out to analyze and compare the structural conformations of CPD BAA87 and its variants. WT CPD BAA87 and its variants (0.1 mg/ml) in 150 mM NaCl, 50 mM Tris-HCl (pH 7.5), and 5% glycerol were subjected to CD spectroscopy at 20 C using 1-mm path length quartz cuvettes and a Jasco J-815 CD spectrophotometer (Jasco). CD spectra were acquired over the wavelength range 200 to 260 nm and were converted into mean residue ellipticity (MRE, degree cm 2 dmol −1 ).

Data availability
The structure presented in this paper has been deposited in the Protein Data Bank, www.wwpdb.org (PDB ID code 7D5Y).