Vibrio vulnificus Secretes an Insulin-degrading Enzyme That Promotes Bacterial Proliferation in Vivo*

Background: Vibrio vulnficus produces SidC, an extracellular insulin-degrading enzyme. Results: SidC causes degradation of insulin, leading to proliferation of the pathogen, and sidC was expressed in low-glucose conditions. Conclusion: Degradation of insulin by SidC correlated with the proliferation of the pathogen. Significance: V. vulnificus manipulates host endocrine signals through SidC, making the host environment more favorable for its own proliferation. We describe a novel insulin-degrading enzyme, SidC, that contributes to the proliferation of the human bacterial pathogen Vibrio vulnificus in a mouse model. SidC is phylogenetically distinct from other known insulin-degrading enzymes and is expressed and secreted specifically during host infection. Purified SidC causes a significant decrease in serum insulin levels and an increase in blood glucose levels in mice. A comparison of mice infected with wild type V. vulnificus or an isogenic sidC-deletion strain showed that wild type bacteria proliferated to higher levels. Additionally, hyperglycemia leads to increased proliferation of V. vulnificus in diabetic mice. Consistent with these observations, the sid operon was up-regulated in response to low glucose levels through binding of the cAMP-receptor protein (CRP) complex to a region upstream of the operon. We conclude that glucose levels are important for the survival of V. vulnificus in the host, and that this pathogen uses SidC to actively manipulate host endocrine signals, making the host environment more favorable for bacterial survival and growth.

Pathogens have evolved elaborate ways to survive and thrive within their hosts to improve their own chance for survival. Elucidation of the mechanisms behind such strategies can be an important step in defining targets for the development of novel antibacterial drugs. Little is known about the ways in which pathogens are able to utilize host catabolic substrates, and although there is a plethora of information about how pathogens modify host proteins, to our knowledge, little is known about bacterial modification of endocrine signals, including insulin.
Vibrio vulnificus causes infection when contaminated seafood is consumed or when it invades an open wound, and this infection can often lead to fatal septicemia in people who are immunocompromised or have an underlying condition such as liver disease, alcoholism, or diabetes mellitus (1,2). Several virulence factors, including hemolysin, elastolytic protease, Rtx toxin, and siderophores, have been identified in V. vulnificus (1)(2)(3). However, it is clear that the disease elicited by this pathogen requires the complex interaction of numerous factors, including many yet to be identified.
The existence of an insulin-degrading enzyme (IDE) 2 in human that plays a role in insulin turnover in tissues was suggested as early as the 1940s (4). However, attempts to identify and characterize IDEs were not successful because of its low concentration and poor stability (5). In addition to insulin, glucagon, ␤-endorphin, insulin-like growth factors I and II, and amyloid-␤ have also been reported as substrates for IDEs (5,6). Recently, IDE has attracted attention for its role in type 2 diabetes and Alzheimer disease, but additional studies are needed (6). A homologous bacterial protein was discovered in Escherichia coli, and the enzyme, called pitrilysin (or protease III), has a specificity for small substrates and is able to degrade both insulin and peptides smaller than 7 kDa (7). Pitrilysin is localized within the periplasmic space (8). This enzyme, like other IDEs, has an unusual metal-binding motif (HXXEH) characteristic of the inverzincin family of zinc metalloproteases (9,10). This motif is an inversion of the more common zinc-binding motif (HEXXH) of other metalloproteases (11). Biochemical properties such as substrate specificity, enzyme kinetics, and metal stoichiometry have been studied (12); however, no obvious phenotypic deficiencies were observed in pitrilysin-deficient E. coli mutants (13), and not much is known about the physiological functions of this enzyme in bacteria. IDE homologs are highly conserved between bacteria and mammals and, in bacteria, they are localized to the cytosol, the plasma membrane, or the periplasm (5). A previous study described a cytoplasmic insulin-degrading enzyme (vIDE) in V. vulnificus (14) and the gene encoding this enzyme was shown to be regulated by CRP-cAMP and by the sugar phosphotransferase system (15). However, a functional role for this enzyme remains to be elucidated.
Here we describe a novel secreted IDE from V. vulnificus named SidC that contributes to proliferation and survivability of the pathogen in a host through proteolytic degradation of the host endocrine signal insulin.

Experimental Procedures
Strains, Plasmids, and Culture Conditions-The bacterial strains and plasmids used in this study are listed in Table 1. E. coli strains were cultured in Luria-Bertani (LB) broth supplemented with appropriate antibiotics at 37°C. V. vulnificus strains were cultured in LB broth or thiosulfate citrate bile salt sucrose (TCBS) agar supplemented with appropriate antibiotics at 30 or 37°C.
Expression and Purification of SidC and CRP-A DNA fragment encoding 938 amino acids of SidC without an aminoterminal signal peptide region was amplified by PCR using the primers sid-OEF and sid-OEB (Table 2) and the product was subcloned into the pASK-IBA3 vector (IBA BioTAGnology, Göttingen, Germany). The resulting vector pASK-SidC was transformed into E. coli BL21(DE3). Expression of SidC with a Strep-tag at the carboxyl terminus was induced with 0.2 g/ml of anhydrotetracycline. After centrifugation, the bacterial pellets were suspended in buffer W (100 mM Tris-Cl, 150 mM NaCl), sonicated, and centrifuged at 7,000 rpm for 10 min. The resulting supernatant was passed through Strep-Tactin affinity resin (IBA BioTAGnology, Göttingen, Germany) and bound protein was eluted using buffer E (100 mM Tris-Cl, 150 mM NaCl, 2.5 mM desthiobiotin) according to the manufacturer's protocol. CRP was purified using a similar strategy. A DNA fragment encoding 310 amino acids of CRP was amplified by PCR using the primers crp-OEF and crp-OEB ( Table 2). The amplified PCR product was cloned into pASK-IBA3. The resulting plasmid pASK-CRP, which encodes CRP fused to a Strep-tag at the C terminus, was expressed in E. coli BL21(DE3) as described above. The purified SidC and CRP proteins were dialyzed and concentrated with buffer S (50 mM Tris-Cl (pH 8.0), 100 mM NaCl, 2 mM dithiothreitol, 10% glycerol) and with buffer WG (buffer W with 30% glycerol) using an Ultracell-30K centricon (Millipore, MA), respectively. To construct a V. vulnificus strain overexpressing SidC, a 3,076-bp DNA fragment of sidC was amplified by PCR using primers sid-comp-F1 and sid-comp-R1 (Table 2) and the resulting product was cloned behind the lac promoter in the vector pRK415 using HindIII and XbaI sites to generated pRK-sidC. The resulting construct was introduced into V. vulnificus strains.
Fractionation of Cellular Compartments and Subcellular Localization of SidC and Western Blot Analysis-Overnight cultures of the V. vulnificus strains were subcultured into fresh brain heart infusion broth (Difco, BD Biosciences, NJ) and cells and supernatants were separated by centrifugation when the A 600 value of the culture reached ϳ2.2. The supernatant was filtered by passing the culture through a 0.22-m syringe filter (Millipore, MA), and was concentrated using Ultracell-30K centricon (Millipore, MA). Fractionation of different cell compartments of V. vulnificus was carried out as described previously (16). After obtaining each fraction, the same volume of each subcellular fraction was separated by SDS-PAGE and transferred to a Hybond-P membrane (GE Healthcare). The membrane was incubated with polyclonal rat antiserum against purified SidC (1:1,000 dilution in blocking solution), and subsequently incubated with goat anti-rat immunoglobulin-G-AP (1:2000) (Santa Cruz Biotechnology, Santa Cruz, CA). SidC was visualized using the 5-bromo-4-chloro-3-indolyl phosphate/ nitro blue tetrazolium color development substrate (Promega, WI). Alkaline phosphatase activity was measured quantitatively as described previously (16).
Assessing the Degradation of Insulin and Other Substrates and Determination of Cleavage Sites of Insulin Digested by SidC-Purified SidC (0.05-1 M) was incubated with 20 g of recombinant human insulin or other substrates (glucagon, TGF-␣, IGF1, and IGF2, Sigma) in reaction buffer (50 mM Tris-Cl, pH 7.5) for 2 h at 37 ºC and then separated by 20% SDS-PAGE. To test the effectiveness of protease inhibitors, either EDTA or TPEN at a final concentration of 5 mM was added. The reaction was terminated by boiling the mixture in 1ϫ SDS- Construction of a sid-lux Fusion and an In-Frame Deletion of sidC-The upstream region (Ϫ257 to ϩ43, with respect to the translational start site) of the sid operon was amplified by PCR using primers sid-pro-F1 and sid-pro-B1 (Table 2), and cloned into the pGEM-T easy vector (Promega). The resulting plasmid was digested with KpnI and BamHI and cloned into pHK0011 (Table 1) to generate pPsid-lux, in which the promoter region of the sid operon is fused to luxAB. To construct an in-frame sidC deletion mutant, a 771-bp DNA fragment comprising the upstream region, and a 773-bp DNA fragment comprising the downstream region of the sidC were amplified using primers sid-KO-F1 and sid-KO-B1 and primers sid-KO-F2 and sid-KO-B2 (Table 2), respectively. Each fragment was digested with the restriction enzyme KpnI, and ligated with pGEM-T Easy vector. The resulting plasmids were digested by ApaI and SacI, and cloned into the pDM4 vector (Table 1) to generate pDM4-sidKO, which was then introduced into V. vulnificus MO6 -24/O by conjugation. Double crossover selection to construct a deletion of sidC in the chromosome was performed as described previously (17). The resulting strain was named ⌬sidC.
Evaluation of the Effect of Purified SidC on the Concentration of Insulin Secreted by INS-1 Cells-The amount of insulin secreted by INS-1 cells was measured. Cells were preincubated in KRBB buffer (25 mM HEPES, pH 7.4, 115 mM NaCl, 5 mM KCl, 2.5 mM CaCl 2 , 1 mM MgCl 2 , 25 mM NaHCO 3 ) containing 0.5% BSA for 3 h and stimulated with either 3 (LG) or 15 mM glucose (HG), with or without SidC, for an additional hour as described (18).
Cytotoxicity Assays-Cytotoxicity assays were performed using the CytoTox96 Non-radioactive Cytotoxicity assay kit (Promega). Briefly, HepG2 was seeded at 2 ϫ 10 5 cells per well into 24-well culture plates and grown overnight in DMEM at 37°C in the presence 5% CO 2 . Zero, 1, and 2 M SidC were added to HepG2, incubated for 1-2 h, then cytotoxicity was determined by measuring the level of released lactic dehydrogenase according to the manufacturer's protocol.

Assessment of Insulin Activity, Glucose Level, and Proliferation of V. vulnificus in the Blood of Infected Mice-C57BLKS
and db/db mice were purchased from DBL (Korea), and ICR (CD-1) mice were purchased from Orient Bio Inc. (Korea). All mice used to assess the effects of SidC were fasted for 16 h before the experiment and only water was provided. For the intraperitoneal glucose tolerance test (IPGTT), db/db mice were injected intraperitoneally with glucose (1 g/kg of body weight) along with 100 l of either the concentration buffer alone or 1.6 M SidC per mouse. Blood was drawn from the tail vein at the designated times and was used to detect glucose and insulin levels as described previously (18). To determine the effect on insulin and glucose levels in the host after infection of vibrio strains, wild type V. vulnificus MO6 -24/O, ⌬sidC, MO6 -24/O (pRK-sidC), and ⌬crr (pRK-sidC) cells were cultured overnight in LB medium and subcultured into fresh LB

Cloning of sidC and crp sid-OEF
Nucleotides modified for the generation of restriction sites or for site-directed mutagenesis are underlined.

An Insulin-degrading Enzyme Facilitates Host Infection
broth. When the culture reached a cell density (A 600 ) of ϳ0.7, the cells were harvested, washed twice, and then diluted to the same cell density using phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , and 2 mM KH 2 PO 4 , pH 7.4). V. vulnificus cells (about 10 7 colony forming units) were subcutaneously injected into mice. At 2-h time points after injection of V. vulnificus strains, the mice were administered filter-sterilized glucose (1.5 g/kg body weight) orally. Blood samples were collected from the heart at time points after the V. vulnificus injection. The blood glucose level was measured immediately using Accu-check Performa (Roche Diagnostics). To determine the level of proliferation of the strains within the host, the number of viable cells present in the blood was measured as follows. A blood sample was diluted in PBS and the diluted samples were then spread onto LBS agar (LB agar with 2% NaCl) in duplicate. The rest of the blood sample was incubated at 4 ºC for coagulation and centrifuged for serum separation. Insulin levels were determined in serum using an Ultrasensitive EIA kit (Alpco Diagnostics, NH) as per the manufacturer's instructions. The results were analyzed using GraphPad Prism 5.
Cloning of crp-The 886-bp DNA fragment comprising the promoter region and the coding region of the crp was amplified by PCR using the primers crp-F and crp-B ( Table 2). The resulting product was cloned into pBBR1-MCS2 (Table 1) to construct pBBR1-crp. Plasmids pBBR1-MCS2 and pBBR1-crp were conjugated into wild type V. vulnificus MO6 -24/O and the ⌬crp strain (19).
Luciferase Assay-Overnight cultures of V. vulnificus strains were inoculated into appropriate medium. Samples were diluted 125-fold with PBS. After adding 0.006% (v/v) n-decylaldehyde, luminescence was measured using a luminometer (Lumat LB 9507, Berthold Technologies, Bad Wildbad, Germany). Specific transcriptional levels were expressed as light units normalized to cell density (relative light units), as described previously (20).
Electrophoresis Mobility Shift Assay-A 395-bp DNA fragment of the upstream region of the sid operon (Ϫ247 to ϩ148, with respect to the translation start site) was amplified by PCR using primers sid-F and 32 P-labeled sid-B (Table 2). For the gel shift assay, 5 ng of the labeled probe was incubated with increasing amounts of purified CRP protein (0 to 400 nM) in a 20-l reaction in the binding buffer containing 10 mM Tris-Cl (pH 8.0), 75 mM NaCl, 1 mM dithiothreitol (DTT), and 1 mM cyclic AMP (cAMP) for 30 min at 37°C. The binding mixtures were resolved on a 5% neutral polyacrylamide gel after addition of 5% glycerol. Gels were exposed to a BAS-MP 2040s IP plate (Fujifilm, Tokyo, Japan) and scanned by BAS-1500 (Fujifilm, Tokyo, Japan).
Site-directed Mutagenesis of the CRP Binding Site in the Upstream Region of the sid Operon-For site-directed mutagenesis of the CRP binding site, the 266-bp DNA fragment (Ϫ344 to Ϫ78, with respect to the translation start site) and 343-bp DNA fragment (Ϫ97 to ϩ246, with respect to the translation start site) of the sid upstream region were amplified by PCR using primers sid-mtF1 and sid-mtB1 and primers sid-mtF2 and sid-mtB2 ( Table 2). The resulting products were cloned into the pGEM-T easy vector (Promega) to generate pGEM-sidA-crpmt using the In-fusion HD cloning kit (Clontech Laboratories, Takara Bio Inc., Shiga, Japan). The region upstream to sidA (Ϫ257 to ϩ43, with respect to the translation start site) containing the mutagenized CRP binding site was amplified by PCR from pGEM-sidA-crpmt using primers sid-pro-F1 and sid-pro-B1, and then cloned into pHK0011 as described above. The resulting plasmid was named pPsid mt -lux, in which the region upstream to the sid operon with a mutation at the CRPbinding site is transcriptionally fused to luxAB, and was conjugated into V. vulnificus strains.
Semi-quantitative Measurement of Levels of Cytokines in Blood of Mice Infected by V. vulnificus-Serum was obtained from ICR mice infected with wild type V. vulnificus MO6 -24/O and the ⌬sidC strain. Levels of cytokines were semi-quantitatively determined using a mouse cytokine antibody array C3 kit (RayBiotech, Inc.) following the manufacturer's instructions. The results were analyzed using a GelQuant Pro v13 (DNR/Bio-Imaging Systems Ltd., Israel).

Results
SidC Is a Virulence Factor Regulating V. vulnificus Survival within the Host-To adapt and survive within the host, V. vulnificus generates factors to manipulate the physiological status of the host, making the host environment favorable for its own survival. We took two approaches to identify novel virulence factors in V. vulnificus: transcriptomic comparisons of V. vulnificus isolates with different pathogenic potencies (21) and in vivo expression technology (22,23) to identify genes that were expressed at higher levels in a mouse model than in vitro (24). From these two independent methods, a putative insulin-degrading enzyme with the conventional N-terminal secretory signal peptide was identified (annotated as: VVMO6_00510 for strain MO6 -24/O (accession number YP_004187735)). Analysis of the deduced amino acid sequence of this gene using the Blastx database in NCBI showed that there is a conventional signal leader sequence at the amino terminus. This protein also contains the inverted zinc-binding motif that is typical to IDEs (data not shown). Because both of these methods identified a specific protease with putative insulin-degrading activity, we hypothesized that this enzyme would be expressed preferentially within the host and therefore, may be important for survival and pathogenicity of V. vulnificus. VVMO6_00510 is encoded by the third gene in an operon that also includes genes encoding VVMO6_00508 (a putative NADPH:quinone reductase) and VVMO6_00509 (a hypothetical protein). We named the three genes in this operon sidABC for secreted-insulin degrading enzyme. The sidA and sidB genes have no detectable effect on the expression or function of SidC (data not shown).
To gain insight into the role of SidC in the host and its role in the survival of V. vulnificus, we compared the survival of wild type V. vulnificus (MO6 -24/O), a sidC knock-out strain (⌬sidC), and wild type harboring the sidC gene cloned in a multicopy vector (Fig. 1A). Compared with wild type, ⌬sidC can barely survive within the host, whereas the strain complemented with sidC in high copy can survive at least 2-fold more than wild type cells. We also compared the survival of wild type and ⌬sidC inside wild type mice and mice suffering type 2 diabetes (db/db mice, C57BLKS-m Lepr db ) (25) (Fig. 1B). The db/db mice are characterized by extreme physiological changes, hyperinsulinemia and hyperglycemia, compared with wild type mice. Infection with wild type V. vulnificus resulted in significantly lower insulin levels in both mouse strains. Glucose levels in db/db mice are, as expected, significantly higher than in wild type mice. However, there were no significant differences in glucose levels between mice infected by wild type V. vulnificus or ⌬sidC. There were at least 4-fold more viable wild type bacterial cells in the db/db mice compared with those in non-diabetic mice (Fig. 1B), suggesting that there were favorable factors for V. vulnificus growth and survival in the diabetic mice. Growth of ⌬sidC mutants in the diabetic mice, however, was greatly impaired (Fig. 1B). These results imply that there may be factors that V. vulnificus must overcome via SidC despite the existence of favorable factors in a diabetic host. In addition to decreased survival within the host, ⌬sidC bacteria were also less pathogenic than wild type (Fig. 2). Wild type V. vulnificus killed 50% of infected mice in less than 6 h, whereas a ⌬sidC took 10 h to reach this level of killing (Fig. 2). These results suggest that SidC may play an important role not only in survival within the host, but also in pathogenicity, and that high levels of blood glucose could provide a condition favorable for V. vulnficus proliferation in vivo.
SidC Is a Secreted Insulin-degrading Enzyme-SidC is a member of the inverzincin family of zinc metalloproteases (9,11), but is only distantly related to other IDEs (Fig. 3, A and B). SidC is distinct from other IDEs in that it has a conventional secretory sequence (26) at the N terminus that is absent in the IDE of E. coli (7) or in the cytosolic IDE of V. vulnificus (14) (Fig.  3C). Mammalian enzymes such as human IDE also can be secreted extracellularly (6,27). However, those IDEs are secreted in a non-conventional way mediated by the SlyX motif (EKPPHY) located at the C terminus (28). We found that SidC of V. vulnificus was detected in supernatant, periplasmic cell fractions, and cytoplasmic fractions from bacterial culture suggesting that SidC may have an extracellular role (Fig. 3D).
These results led us to test whether SidC has insulin-degrading activity, as does human IDE. V. vulnificus SidC protein that was overexpressed and purified from E. coli showed insulindegrading activity in a concentration-dependent manner (Fig.  4A). Like other zinc metalloproteases, zinc is required for SidC activity because TPEN, a zinc-specific chelator, and EDTA abolished the insulin-degrading activity of SidC (Fig. 4A). Cleavage sites on insulin digested by SidC were determined by analyzing the enzymatic products using LC-MS/MS (Fig. 4B). SidC cleaves more sites than human IDE (hIDE) or rat IDE (rIDE) (29, 30). Among 14 cleavage sites, four of them (Leu 13 -Tyr 14 of chain A, and His 10 -Leu 11 , Ala 14 -Leu 15 , and Phe 25 -Tyr 26 of chain B) were identical with those of both hIDE and rIDE. One cleavage site (Tyr 16 -Leu 17 of chain B) and four cleavage sites (Ser 12 -Leu 13 of chain A, Arg 22 -Gly 23 , Gly 23 -Phe 24 , and Phe 24 -Phe 25 of chain B) were identical with those of hIDE or rIDE, respectively (29,30). Five cleavage sites were unique to SidC.  (VVMO6_00860). B, the neighbor-joining phylogenetic tree of insulin-degrading enzymes. Phylogenetic analysis was conducted using MEGA version 4 (48). The distance scale is shown under the phylogenetic tree. C, sequence of the amino terminus of SidC. Charged amino acid residues and hydrophobic residues characteristic of a conventional leader sequence are indicated by asterisks and in boldface, respectively. D, the protein encoded by SidC is exported outside of the cell. Subcellular localization of SidC was determined by Western blot hybridization using polyclonal rat antiserum against purified SidC. 1, Wild type V. vulnificus MO6/24-O; 2, ⌬sidC; 3, MO6 (pRK-sidC). Alkaline phosphatase activity was measured quantitatively as described previously (16). Alkaline phosphatase and ␤-galactosidase activity values are the average from three independent experiments, and standard deviations are indicated.
Next, to show insulin-degrading activity within a cellular system, we employed INS-1 pancreatic beta cells, which can secrete insulin in response to high glucose. Insulin levels were reduced in the presence of SidC in a concentration-dependent manner (Fig. 4C). To determine whether the lower insulin levels were due to a cytotoxic effect of the protein, we determined the effects of SidC on the human liver carcinoma cell line HepG2 (31). As shown in Fig. 5A, SidC does not have a significant cytotoxic effect on the HepG2 cell line. We also examined the cytotoxic effect of SidC in INS-1 cells and determined that it is not cytotoxic up to a concentration of 2 M of SidC (Fig. 5B).
SidC Reduces Insulin Levels but Increases Glucose Levels in Vivo-SidC is a unique IDE because it can be secreted via the N-terminal signal peptide (Fig. 3, C and D). Therefore, we hypothesized that after infecting a host, V. vulnificus secretes SidC to degrade insulin in the bloodstream, thereby increasing blood glucose levels, improving survivability and enhancing pathogenicity. To test this hypothesis, we injected mice with purified SidC and measured both insulin and glucose levels in the bloodstream using the IPGTT. Interestingly, glucose clearance was significantly reduced in SidC-injected mice compared with the control throughout the experiment, showing that SidC promoted glucose intolerance in mice (Fig. 6A). Because the treatment of SidC reduced glucose-elevated insulin levels in cultured beta cells, we suspected that impaired glucose tolerance might be attributed to the reduction in plasma insulin levels by SidC. Indeed, whereas the intraperitoneal injection of glucose promoted higher insulin concentration in the plasma of control mice. SidC-injected mice showed reduced plasma insulin levels even in the presence of a supra-physiological concentration of glucose (Fig. 6B). This result led us to speculate that after infecting a host, V. vulnificus may actively degrade insulin by secreting SidC to affect the host environment.
SidC Facilitates Survival of V. vulnificus in Mice by Modulating Host Insulin Levels-The effects of injection of purified SidC into mice should likely be different from an active infection with V. vulnificus, which will secrete SidC during proliferation and consume host nutrients, such as glucose, for survival. To determine the effects of SidC produced by V. vulnificus within the host, we infected mice with both wild type and ⌬sidC bacteria and monitored both insulin and glucose levels as well as viable bacterial cell counts over time (Fig. 7). Four hours after  infection, insulin levels in the blood of mice infected with wild type V. vulnificus were significantly lower than in mice infected with the ⌬sidC mutant (Fig. 7A) confirming that secreted SidC degrades insulin in vivo. We also examined glucose levels and observed that 4 h after infection, glucose levels in the blood of mice infected with wild type V. vulnificus were significantly lower than in mice infected with the ⌬sidC mutant (Fig. 7B). Importantly, viable cell counts of wild type V. vulnificus were significantly higher than those of the ⌬sidC mutant at 4 h postinfection and thereafter (Fig. 7C). Taken together, these results demonstrate that SidC is important for V. vulnificus survival and proliferation within the host by degrading insulin.
We expected that, due to a lower level of insulin, the level of glucose in blood from mice infected by wild type V. vulnificus would be higher than that in mice infected by ⌬sidC. But our results did not match this prediction (Fig. 7B). One possible explanation for the low level of glucose in the blood from mice infected by wild type V. vulnificus is that SidC produced by infecting V. vulnificus also degrades glucagon, resulting in a decreased blood glucose level. We actually observed that glucagon can serve as a substrate for SidC (data not shown). Another possibility is that glucose was being consumed rapidly by the pathogen. To test these possibilities, we employed the V. vulnificus strain ⌬crr that has a null mutation in crr (Table 1), which encodes a crucial factor for glucose uptake; thus ⌬crr cannot utilize glucose as a carbon source (32). We compared insulin levels, glucose levels, and viable cell counts in blood samples of mice infected by wild type MO6 -24/O, ⌬sidC, ⌬crr(pRK-sidC), or MO6 -24/O(pRK-sidC). We assessed the effects of SidC produced by the pathogen employing the oral glucose tolerance test (OGTT) to minimize the influence of SidC on glucagon (33).
As expected, levels of insulin in blood samples from mice infected with wild type MO6 -24/O, ⌬crr(pRK-sidC), or MO6 -24/O(pRK-sidC) were significantly lower than levels from mice infected with ⌬sidC (Fig. 8A). The viable cell counts of wild type MO6 -24/O and MO6 -24/O(pRK-sidC) were significantly higher than for ⌬sidC (Fig. 8B). However, ⌬crr(pRK-sidC) showed barely detectable levels of viable cells compared with other strains, putatively because this strain cannot utilize glucose due to its inability to transport glucose. Consistent with this, glucose levels in the blood of mice infected by this strain remains high, whereas glucose levels in mice infected with wild type and MO6 -24/O(pRK-sidC) are lower (Fig. 8C). These data suggest that the elevated levels of glucose caused by SidC degradation of insulin are subsequently lowered by infecting the mice with V. vulnificus, which consumes the glucose to support growth. Taken together, these results show that SidC facilitates growth of V. vulnificus in the host by reducing insulin levels, and that degradation of insulin was correlated with increased proliferation of the infecting bacteria.
Expression of SidC Is Repressed by Glucose via CRP-cAMP-To determine how sidC is regulated, we sought to determine the growth conditions where sidC is best transcribed. Among various bacterial growth media, sidC was expressed more quickly and at higher levels during growth in brain heart infusion medium, which is an undefined medium made from animal tissue, compared with LB-rich medium or AB minimal medium (3). Also, sidC was expressed more abundantly at 37 ºC compared with 28 ºC (Fig. 9A), which is consistent with our previous in vivo expression technology screening results, which showed that sidC is expressed at higher levels in vivo compared with ex vivo (24). Numerous genes associated with sugar metabolism are regulated in response to glucose levels through cAMP and the activator protein CRP (34), and it has been shown that expression of the V. vulnificus cytoplasmic IDE, insulysin, is regulated in this way (15). We predicted that sidC expression may be regulated in a similar manner. In fact, expression of the sid operon generally increased as cells entered stationary phase, and subsequent addition of glucose (but not sucrose) led to a dramatic reduction in expression levels (Fig. 9B). In a crp-null mutant, expression levels were significantly lower (Fig. 9C), suggesting a role for CRP-mediated regulation. We located a consensus CRP binding site (35) in the region upstream of the sid operon (Fig. 9D) and gel mobility shift assays showed that the CRP-cAMP complex bound to this region of the DNA (Fig.  9E). Furthermore, site-directed mutagenesis of the putative CRP-binding site (Fig. 9D) abolished this binding (Fig. 9E). Using luxAB reporter fusions, we then showed that both the presence of CRP and the presence of an intact CRP-binding site are required for expression of the sid genes (Fig. 9F). We conclude that sidC is up-regulated under glucose-limiting conditions, and that this regulation is mediated by the CRP-cAMP complex.

Discussion
Many pathogenic bacteria express enzymes that modify host proteins during infection. Some of these enzymes catalyze the addition of chemical groups onto host factors and others modify amino acids (36). Proteolysis of host factors is also known to

An Insulin-degrading Enzyme Facilitates Host Infection
occur, as exemplified by the cleavage of adhesin by a Bacteroides fragilis enterotoxin (37) and the cleavage of host factors by botulinum toxin (38). Through these post-translational modifications, an infecting pathogen can disrupt host signal transduction pathways, important cellular defense mechanisms, or normal cellular processes to facilitate its own survival and propagation. However, specific hydrolysis of human endocrine signals has not yet been documented. In this study, we identified and characterized SidC, a novel bacterial insulin-degrading enzyme that enhances the pathogenicity of V. vulnificus. SidC is the first example of an extracellular enzyme secreted by a pathogenic bacterial species that affects the physiology of the host through the proteolytic degradation of a host endocrine signal, insulin.
It is challenging for pathogens to acquire nutrients and energy during infection of a host. Some viruses stimulate an increase in host intracellular glucose levels to increase the energy available for replication by affecting signal transduction pathways associated with sugar metabolism (39,40). Intracellular Salmonellae have been shown to relay on host intracellular glucose (41). Intracellular Brucella abortus also needs glucose for chronic infection (42). However, little is known about how extracellular bacterial pathogens access host metabolites (43). We observed that diabetic mice supported the growth of infecting V. vulnificus to higher levels than wild type mice (Fig. 1B), suggesting that higher blood glucose levels are advantageous to V. vulnificus. The opportunistic pathogen Saccharomyces cerevisiae uses glucose as a primary carbon source in vivo, and the ability to sense and respond to glucose is important for its survival (44). These results suggest that host blood glucose could be an important source of carbon and energy for other extracellular pathogens, including V. vulnificus.
We believe that SidC is directly secreted into the bloodstream of mice because we detected V. vulnificus in the blood of infected mice. However, the mechanism by which V. vulnificus gains access to the bloodstream has yet to be studied. SidC is expressed at higher levels when V. vulnificus infects a host, and was also detected at higher levels when cells were grown in an undefined, animal-tissue containing medium. This, together with the fact that low sugar levels stimulated sidC expression, led us to speculate that the role of this enzyme is to manipulate glucose levels in the host. Injection of purified SidC led to a reduction in insulin levels and an elevation of glucose levels in the blood of mice (Fig. 6). Blood insulin levels and glucose levels in mice infected with wild type V. vulnificus were both significantly lower than those in mice infected with a sidC-deletion strain (Fig. 7). It is possible that blood glucose levels increased as a result of SidC activity, and then decreased as the sugar was consumed by the pathogen. Alternatively, it is possible that the reduced blood glucose level may be due to decreased levels of glucagon caused by SidC-mediated degradation, because we observed that SidC can also use glucagon as a substrate (data not shown). A recent study indicated that the route of glucose administration determines the effect of IDE on glucose tolerance (33). The insulin level in IPGTT is significantly lower than that in OGTT. Therefore, during IPGTT, IDE degrades proportionally more glucagon and thereby results in lower glucose levels during IPGTT (33). Meanwhile, OGTT does not decrease the blood insulin level, providing a better condition to monitor the effect of IDE on insulin (33). To circumvent the potential effect of IDE on glucagon, we performed V. vulnificus infection experiments using mice on which glucose was orally administered ( Figs. 7 and 8).
We hypothesized that the lower glucose levels in mice infected by wild type V. vulnificus is due to the consumption of the sugar by the proliferating pathogen. To explore this, we employed the ⌬crr strain harboring pRK-sidC, a strain that cannot utilize glucose due to an inability to transport sugar. The reason for the introduction of an extra copy of sidC under the vector promoter is that sidC located on the chromosome cannot be expressed because the sugar group translocation system (34) regulates the expression of sidC (Fig. 9). Therefore, the chromosomal sidC under its own cognate promoter would not be expressed without Crr. Our assumption was that, if the reduced level of glucose was caused not by bacterial consumption but by some other factors (for instance, degradation of glucagon by SidC), the glucose levels in mice infected by ⌬crr (pRK-sidC) would also be as low as those infected by wild type. The results clearly showed that the blood glucose level in mice infected by ⌬crr (pRK-sidC) is even higher than in mice infected by wild type bacteria. It is noteworthy that the presence of V. vulnificus strains expressing SidC led to a reduction in blood insulin levels, and that the amount of blood glucose is inversely proportional to the level of proliferation of the infecting pathogen (Figs. 7 and 8).
It also is noteworthy that the proliferation of MO6 -24/ O(pRK-sidC) is higher than that of wild type, even if the consumption of blood glucose by these two strains are not significantly distinct (Fig. 8, B and C). SidC may have additional functions for the pathogenesis of V. vulnificus. Human IDE has multiple roles in the cell, including degradation of both small molecular weight proteins and ␤-amyloid (5,6,27), and therefore SidC may affect host physiology in multiple ways as well. Using a mouse cytokine antibody array, we eliminated the possibility that SidC degrades host cytokines (data not shown). Other possible roles for SidC-associated pathogenicity remain to be explored.
Regardless of the underlying mechanism for the pathogenicity of SidC, our results may explain why there is an increased risk of V. vulnificus infections in patients with diabetes mellitus (45), and may explain the higher risk of other bacterial infections as well (46). SidC homologs are present in other pathogenic Vibrio species, such as Vibrio cholerae and Vibrio parahemeolyticus (identity %/similarity %; 71/84 and 74/86, respectively) (data not shown). The SidC homolog in V. cholerae is up-regulated during the late stages of infection, as measured using a recombination-based in vivo expression technology (47), suggesting a role for SidC in the pathogenicity of this Blood samples were collected 4 h after infection, and tested for insulin levels (A), viable bacterial cell counts (B), and glucose levels (C) as described under "Experimental Procedures." Each symbol represents an individual mouse. Long bars indicate the mean, and upper and lower bars represent the mean Ϯ S.E. for each group. Statistical significance was determined using the Mann-Whitney test. species as well. More in-depth study of this novel pathogenic mechanism could lead to new options for both treating and preventing infections caused by pathogenic Vibrio species. Likewise, further understanding of other previously unrecognized catabolite-scavenging capabilities used by pathogenic microorganisms could address the increasing challenges in treating infectious diseases.
Author Contributions-K. K., I. H. K., and S. K. designed the study and wrote the paper. Y. W. and N. P. purified SidC. J. P. and J. L. performed the cytotoxicity tests on human cells. I. H. K., I. K., K. L., and A. K. performed mice experiments. I. H. K. performed experiments shown in Fig. 9. All authors reviewed the results and approved the final version of the manuscript.