One site mutation disrupts dimer formation in human DPP-IV proteins.

DPP-IV is a prolyl dipeptidase, cleaving the peptide bond after the penultimate proline residue. It is an important drug target for the treatment of type II diabetes. DPP-IV is active as a dimer, and monomeric DPP-IV has been speculated to be inactive. In this study, we have identified the C-terminal loop of DPP-IV, highly conserved among prolyl dipeptidases, as essential for dimer formation and optimal catalysis. The conserved residue His750 on the loop contributes significantly for dimer stability. We have determined the quaternary structures of the wild type, H750A, and H750E mutant enzymes by several independent methods including chemical cross-linking, gel electrophoresis, size exclusion chromatography, and analytical ultracentrifugation. Wild-type DPP-IV exists as dimers both in the intact cell and in vitro after purification from human semen or insect cells. The H750A mutation results in a mixture of DPP-IV dimer and monomer. H750A dimer has the same kinetic constants as those of the wild type, whereas the H750A monomer has a 60-fold decrease in kcat. Replacement of His750 with a negatively charged Glu (H750E) results in nearly exclusive monomers with a 300-fold decrease in catalytic activity. Interestingly, there is no dynamic equilibrium between the dimer and the monomer for all forms of DPP-IVs studied here. This is the first study of the function of the C-terminal loop as well as monomeric mutant DPP-IVs with respect to their enzymatic activities. The study has important implications for the discovery of drugs targeted to the dimer interface.

Dipeptidyl peptidase IV (DPP-IV, 1 also known as CD26) (EC 3.4.14.5) is a well documented drug target for the treatment of type II diabetes (1). It is a serine protease involved in the in vivo degradation of two insulin-sensing hormones, glucagonlike peptide-1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP) (2,3). Either inhibiting the enzymatic activity of DPP-IV in various animal models or knocking out DPP-IV in mice and rats prolongs the half-lives of these two insulinsensing hormones, increases insulin secretion and improves glucose tolerance (4 -11). Hence inhibition of DPP-IV may be effective in the treatment of type II diabetes. Understanding the catalytic mechanism of DPP-IV is thus essential to discovering inhibitors for the treatment of the disease.
DPP-IV belongs to the prolyl oligopeptidase (POP) family, a subfamily of serine proteases (12,13). This class of prolyl peptidases includes DPP-IV, prolyl oligopeptidase (POP), DPP-II, DPP8, DPP9, and fibroblast activation protein (FAP) (12,13). Unlike classic serine proteases, the POP family of enzymes is highly selective toward peptides that have a proline residue at the penultimate position (18). The x-ray structures of DPP-IV and POP have shed light on the catalytic mechanisms, which differ significantly from those of the classic serine proteases, such as trypsin and subtilisin (12, 14 -18). DPP-IV consists of two domains, the ␣/␤ hydrolase domain and the ␤-propeller domain, with the active site inbetween (14 -17). The substrate specificity of DPP-IV is dictated by a prolinebinding pocket and a Glu 205 -Glu 206 motif at the active site (14 -17). Only small size peptides are hydrolyzed by this class of enzymes because of the unique propeller structure and/or side opening substrates used to access the active site (14 -17). Among the shared properties, the most obvious difference between POP and DPP-IV is the relationship of catalytic activity with respect to its quaternary structure. POP exists in solution as a monomer and is active in such a form (15). In contrast, DPP-IV is active only as a dimer or oligomer, and monomeric DPP-IV is speculated to be inactive, even though DPP-IV monomer has never been isolated and demonstrated to be inactive (14,19).
Based on the crystal structures of DPP-IV, two loops are located in the dimer interface and were proposed to be involved in dimer interaction, the C-terminal loop at the ␣/␤ hydrolase domain and the propeller loop extended from strand 2 of the fourth blade in the ␤-propeller domain (14,16,17) (Fig. 1A). The C-terminal loop of DPP-IV consists of the last 50 amino acid residues with two ␣-helices (amino acids 713-725 and 745-763) and one ␤-sheet (amino acids 726 -744) interacting with the same region from the other monomer across a 2-fold axis (Fig. 1B). Both hydrophobic and hydrophilic interactions have been proposed to be responsible for dimer formation (12)(13)(14)(15). This loop is highly conserved among the family of DPP-IVcontaining prolyl dipeptidases (Fig. 2), though the functional importance of the loop has not been addressed experimentally.
In this article, the quaternary structures and catalytic activities were studied and compared among endogenous DPP-IV from human semen, recombinant wild-type and mutant DPP-IVs expressed in baculoviral-infected insect cells. The role of the highly conserved C-terminal loop for dimerization was investigated. For the first time, we have isolated and characterized the biochemical properties of monomeric mutant DPP-IV proteins altered at residue His 750 of the loop, which is highly conserved among DPP-IV-containing prolyl dipeptidases (Fig. 2).

EXPERIMENTAL PROCEDURES
Materials-The enzyme substrate H-Gly-Pro-pNA and dipeptide Gly-Pro were purchased from Bachem. Fetal bovine serum was from Hyclone. Lipofectin and the insect culture media, Grace and Express Five media, were from Invitrogen. Human liver cDNA library and linear viral vector were from Clontech. The ECL Western detection kit was from PerkinElmer Life Sciences. Q Sepharose TM High Performance, CNBr-activated Sepharose 4B and Superdex 200 prepacked-columns were from Amersham Biosciences. The chemical cross-linker dithiobis-succinimidyl propionate (DTSP) was from Pierce. Bovine adenosine deaminase (ADA) was from Roche Applied Science.
Construction of the Secreted DPP-IV Expression Plasmid-The baculovirus expression plasmid pBac8-CD5 was constructed with the secretion tag CD5. Vector pBac-PAK8 (Clontech) was modified by inserting an MT-EGFP cassette at EcoRV site to facilitate the selection of the virus expressing EGFP (20). CD5 coding sequence was amplified by PCR from human Jurkat cell cDNA with the following primers: 5Ј-CGGGATCCATGCCCATGGGGTCTCT-3Ј and 5Ј-CCGCTCGAGC-CGAGGCAGGAAGC-3Ј. The CD5 cDNA fragment was released by digestion with BamH1 and XhoI before ligation into pBac-PAK8, resulting in pBac8-CD5. The expression plasmid of DPP-IV with the secretion tag CD5 is constructed as follows. The human cDNA fragment of DPP-IV containing amino acids 39 -766 was amplified by PCR from a human liver cDNA library with the primers 5Ј-CCGCTC-GAGAAAAACTTACACTCTA-3Ј and 5Ј-GCGTCGACCTAAGGTAAA-GAGAAACATTG-3Ј, and cloned into pCR®-Blunt II-Topo vector (Invitrogen). The DPP-IV cDNA was then released by digestion with XhoI and EcoRI before ligation into the vector pBac8-CD5. Site-directed mutagenesis of DPP-IV was carried out using Pfu Turbo DNA Polymerase (Stratagene). The primers used for generating H750A and H750E mutants are 5Ј-AGCACACCAAGAAATATATACCCAC-3Ј and 5Ј-GTGGGTATATATTTCTTGGTGTGCT-3Ј for H750E, and 5Ј-AGCA-CACCAAGCTATATATACCCAC-3Ј and 5Ј-GTGGGTATATATAGCTT-GGTGTGCT-3Ј for H750A, respectively. All DPP-IV cDNA fragments cloned were sequenced to verify that they contain no additional mutations other than those desired.
Insect Cell Culture, DNA Transfection, Virus Selection, and Amplification-Sf9 cells were grown in Grace medium supplemented with 10% fetal bovine serum at 27°C. The transfection of DNA to Sf9 cells, and the selection and amplification of the recombinant virus were carried out as described (21). For expression and purification purposes, Hi5 cells, instead of Sf9 cells, were used. Hi5 cells were infected at a multiplicity of infection of 1.0 TCID 50 unit/cell (TCID 50 is 50% tissue culture infectious dose), determined to be the optimal condition for protein expression as described (21), and the cells were harvested at 72-h post-transfection.
Purification of DPP-IV Proteins from Hi5 Insect Cells and Human Semen-The purification of wild-type recombinant DPP-IV was carried out as described (14). ADA affinity columns were prepared as described (22). For the purification of both H750A and H750E mutant proteins, only the ADA column was used with the omission of Triton X-100 in both the washing and elution solutions. Human semen DPP-IV protein was purified from healthy Asian male donors as described (22). The elution buffer for protein bound on an ADA column did not contain Triton X-100. Freezing at Ϫ80°C does not change either the quaternary structure (determined by AUC) or the enzymatic activities of DPP-IV proteins described in this study. The purity of the protein was determined by SDS-PAGE, and proteins were visualized with Coomassie Blue. The amount of protein was determined by the method of Bradford using bovine serum albumin as the standard.
Polyacrylamide Gel Electrophoresis and Western Blot Analysis-Purified proteins were run on a 4 -20% gradient native polyacrylamide gel with the gel running system from Amersham Biosciences. SDS-PAGE and Western blot analysis were conducted as described (23). Rabbit anti-DPP-IV antibody was generated in house using purified semen DPP-IV as the antigen.
Kinetic Constant Determinations-To measure the kinetic parameters, the chromogenic substrate H-Gly-Pro-pNA was utilized to initiate the reaction, which was monitored at OD 405 nm as a function of time (21). The enzyme concentrations used in the reaction were 10 nM for wild-type and H750A proteins, and 100 nM for the H750E protein, respectively. The initial rate was measured with less than 10% substrate depletion for the first 10 to 300 s. The steady state parameters, k cat and K m , were determined from initial velocity measurements at 0.5 to 5 K m of the substrate concentrations. Lineweaver Burk plots were analyzed using non-linear regression of the Michaelis-Menten equation. Correlation coefficients greater than 0.99 were obtained.
Chemical Cross-linking in Intact Cells and Size Exclusion Chromatography-Chemical cross-linking in intact cells was conducted as described (24). Size exclusion chromatography was conducted at 4°C. Purified proteins (0.5 ml at a concentration of 5 M) were applied to a Superdex 200 10/30 column (10 ϫ 300 -310 mm) pre-equilibrated with PBS. The sample was eluted with the same buffer at 0.3 ml/min and 0.25 ml fractions were collected. The Superdex 200 10/30 column was calibrated with the Stokes Radii of ferritin (6.1 nm), catalase (5.22 nm), aldolase (4.81 nm), albumin (3.55 nm), ovalbumin (3.05 nm), and chymotrypsinogen A (2.09 nm) from Amersham Biosciences.
Analytical Ultracentrifugation-DPP-IV proteins at concentrations of around 0.1 to 0.2 mg/ml (1.2-2.3 M) were used for AUC analysis with either PBS, high salt (100 mM Tris-HCl, 50 mM NaCl, 0.5 M Na 2 SO 4 , pH 7.5) or low salt (100 mM Tris-HCl, 50 mM NaCl, pH 7.5) buffers as indicated. Buffer was changed using an Amicon device and DPP-IV proteins were allowed to equilibrate for at least 4 h or longer as indicated in the text at 25°C after buffer changes. The sedimentation coefficients (S) of the enzyme were estimated by a Beckman-Coulter XL-A analytical ultracentrifuge with an An60Ti rotor as described (25). Sedimentation velocity analysis was performed at 40,000 rpm at 25°C with standard double sector aluminum centerpieces. The UV absorption of the cells was scanned every 5 min for 4 h. Sedimentation equilibrium was performed at 20°C with six-channel open centerpieces and then centrifuged at 12,000 rpm for 12 h. The data from both sedimentation velocity and sedimentation experiments were analyzed with the SedFit version 8.7 program to obtain molecular weights and sedimentation coefficients (25). Sednterp version 1.07 program is used to obtain solvent density, viscosity, Stokes' radius (R s ) and anhydrous frictional ratio (f/f o ).
Dilution Experiment-Enzyme concentrations ranging from 200 to 1.6 nM were used in the dilution experiments. The experiments were carried out with consecutive 2-fold dilutions in PBS containing 0.1% bovine serum albumin and 1 mM DTT. The solution after dilution was incubated at 25°C for 16 h to ensure attainment of dimer-monomer equilibrium. The reaction was initiated by adding the substrate H-Gly-Pro-pNA at a final concentration of 10 M for both wild-type DPP-IV and H750A proteins. The initial rate of the reaction was recorded and converted to specific activity.

Human DPP-IV Protein Is a Dimer in Intact Cells and in
Vitro-From the crystal structures, the human recombinant DPP-IV was shown to be a homodimer whereas DPP-IV purified from porcine kidney is a homotetramer (14,16,17,26). In addition, previous studies showed that purified DPP-IV proteins from various sources migrated at sizes corresponding to either dimer or tetramer/oligomer according to gel filtration experiments (19,(27)(28)(29)(30). To determine the physiologically relevant oligomerization state of DPP-IV, we performed chemical cross-linking in the DPP-IV-containing Caco-2 cells. The chemical cross-linker used was DTSP, a primary amine-specific cross-linker with moderate chain length. As shown in Fig. 3A, DPP-IV could form a dimer (240 kDa) in intact cells, twice the size of the monomer (ϳ120 kDa) (Fig. 3A, lane 2). The crosslinker DTSP is specific, because the addition of the DTT abolishes dimer formation (Fig. 3A, lane 1). The formation of dimer is DTSP-dependent since in the absence of DTSP, no dimer formation was observed (Fig. 3A, lanes 3 and 4).
Next we determined whether endogenous DPP-IV purified from human semen (sDPP-IV) forms dimers in vitro. The purified protein was quite pure as demonstrated by SDS-PAGE (Fig. 3B, lane 1). By measuring its kinetic constants (k cat and K m values), we confirmed that purified sDPP-IV was as active as reported previously (Table I) (31). On a native gel, sDPP-IV runs predominantly as a dimer of about 200 kDa with the presence of minor but higher molecular mass species (Fig. 3C,  lane 1). It elutes at a position corresponding to a 400 kDa protein with a Stokes' radius of 5.9 nm, determined by gel filtration chromatography ( Fig. 4A and Table II). Cross-linking of the purified protein in vitro showed that the protein is dimeric with a mass of 250 kDa (data not shown). We then used AUC to determine the hydrodynamic properties of sDPP-IV. As shown in Fig. 5A, sDPP-IV is undoubtedly homodimeric with a sedimentation coefficient of 9.1 S (Table II) and a molecular mass of 225 kDa. Notably, there is only a single peak corresponding to the dimer in the AUC experiment, suggesting that the dimer is the predominant form under the conditions tested. For sDPP-IV, the value of the anhydrous frictional ratio f/f o is 1.4, indicating that the protein is non-spherical. Therefore, gel filtration does not provide an accurate measurement of sDPP-IV's quaternary structure and molecular weight, because of the protein's non-globular shape. The aberrant mobility in gel fil-tration was also observed in previous studies with DPP-IV proteins purified from either human fibroblast cells or urine (29,32).
Properties of the Baculoviral-expressed DPP-IV Proteins-Determination of the residues important for dimer formation necessitated the generation of mutant DPP-IVs. We chose baculoviral-infected insect cells to express both wild-type and mutant DPP-IV proteins for the in vitro biochemical studies. Whether the recombinant DPP-IV (rDPP-IV) is also dimeric in solution and has comparable biochemical properties, despite the difference in glycosylation between the endogenous sDPP-IV and rDPP-IV, were investigated. rDPP-IV from baculoviral-infected insect cells was purified and found to be as active as endogenous sDPP-IV, based on determination of k cat and K m values (Fig. 3B, lane 2 and Table  I). The purified rDPP-IV runs at the position of around 250 kDa in both native gel and gel filtration chromatography (Stokes' radius of 5.6 nm) (Fig. 3C, lane 2, Fig. 4B, and Table II). Determined by velocity AUC experiments, the molecular mass of rDPP-IV is 183 kDa with a sedimentation coefficient value of 8.4 S (Fig. 5B and Table II). The value of the anhydrous frictional ratio (f/f o ) is 1.4, the same as that of the sDPP-IV, indicating that rDPP-IV is also a non-spherical dimer. Therefore, as demonstrated here, rDPP-IV expressed from the baculoviral-infected insect cells is suitable for studying the quaternary structure and enzymatic properties of DPP-IV.
The size difference between sDPP-IV and rDPP-IV in SDS-PAGE, native gel, gel filtration and sedimentation experiments, reflects the difference in the extent and nature of the glycosylation and the non-spherical nature of the dimeric proteins. This is also consistent with the difference observed in Stokes' radii between these two wild-type proteins in AUC (Table II).
H750 Is Important for Dimer Formation and Stability-One important interaction between two monomers of DPP-IV is provided by the C-terminal loop located at the ␣/␤ hydrolase domain (14,16,17) (Fig. 1, A and B). Based on the sequence alignment of the prolyl dipeptidases presented in Fig. 2, the highly conserved C-terminal loop might play an important role in the dimerization of DPP-IV and other prolyl dipeptidases. There are several residues conserved in this region from amino acids 713-766 (Fig. 2). We chose to mutate His 750 based on the following criteria. His 750 is completely conserved and intimately involved in the dimer interface based on our examination of the crystal structure. Compared with other conserved residues in the region, it is farthest from the triad residues Asp 708 and His 740 , thus it is least likely to disturb the position of the catalytic triad if mutated. His 750 was mutated to Glu (H750E), to introduce an opposite charge, or Ala (H750A) to simply remove the bulky imidazole ring. Both mutant proteins were purified and found to run at positions around 250 and 140 kDa in the native gel (Fig. 3, B and C, lanes 3 and 4). k cat of H750A is about 3-fold smaller than that of the wild-type rDPP-IV whereas K m is not changed (Table I). Interestingly, k cat /K m of H750E is about 300 times smaller than that of the wild-type rDPP-IV, with a 10-fold increase for K m and a 30-fold decrease for k cat (Table I).
To determine if the difference in the activities of the proteins might be correlated with their quaternary structures, we investigated whether the proteins remained dimeric. First, gel filtration experiments were used to study the quaternary structures of H750A and H750E. As shown in Fig. 4C, H750A is a mixture of two peaks, with the Stokes' radius and predicted mass of 5.7 nm and 328 kDa, and 4.6 nm and 144 kDa, respectively. Interestingly, there is only single peak for H750E, with Stokes' radius of 4.6 nm and predicted mass of 144 kDa (Fig.  4D). The Stokes' radii determined from gel filtration experiments are summarized in Table II. Velocity AUC was thus used to analyze the quaternary structures of mutant proteins. As shown in Fig. 5, C and D, as well as Table II, H750E consists only of monomer while H750A is a mixture of dimer and monomer, under the same condition as that used for the wild-type DPP-IV. The sedimentation coefficient for H750E is 5.5 S with a molecular mass of 96 kDa, whereas H750A consists of both the 5.5 S monomer and 8.5 S dimer with molecular masses of 99 and 188 kDa, respectively (Table II). These results indicate that the H750A mutation is not enough to transform all dimer into monomer in PBS buffer, which represents physiological conditions. However, the H750E mutation is sufficient to disrupt DPP-IV completely to the monomer, based on AUC analysis.
Because H750A is a mixture, the kinetic constants of 31 s Ϫ1 and 77 M for k cat and K m values, respectively, are derived from both monomer and dimer forms (Table I). Thus we used gel  filtration experiment to separate these two forms of H750A before subjecting them separately to sedimentation equilibrium analysis and the measurement of the enzymatic activities. As shown in Fig. 6, interestingly, both dimer and monomer maintained their subunit compositions without converting into monomer or dimer, respectively, after incubation at room temperature for up to 48 h. This indicates that a dynamic equilibrium between dimer and monomer of H750A is extremely slow or non-existent.
The kinetic constants were measured for the monomer and the dimer of H750A after separation by gel filtration. As shown in Table I, dimeric H750A has activity similar to that of the wild-type rDPP-IV indicating that, in the absence of change in quaternary structure, the mutation did not perturb the enzymatic activity. However, monomeric H750A has a 60-fold decrease in the k cat but a similar K m value. Therefore, the quaternary structure of enzymes correlates with the enzymatic activities since both monomeric H750A and H750E have much lower catalytic activities compared with those of the dimeric rDPP-IV or H750A.
Since the sedimentation velocity depends on both size and shape of the protein, we carried out a sedimentation equilib-  rium analysis to determine unambiguously the molecular masses for all forms of DPP-IVs studied here. Summarized in Table II, for sDPP-IV, rDPP-IV, H750A dimer, H750A monomer and H750E, the molecular masses determined are 225, 187, 186, 98, and 98 kDa, respectively. The values obtained are comparable with those from sedimentation velocity experiments, consistent with no dynamic equilibration. All the mutant proteins analyzed have the f/f o values of 1.3-1.4, indicating that they are all non-spherical in shape.
Dilution Experiment-Because the monomer and dimer of H750A did not equilibrate under the conditions tested, we wanted to know whether dilution of the enzyme to very low concentrations would facilitate the dissociation of the dimer into monomer. The dilution method has been used to study the dimer-monomer equilibrium of several herpes viral proteases with K d values in the nanomolar range (33)(34)(35). If the monomeric DPP-IV has a very low activity as observed for monomeric H750A and H750E, dilution of the protease to a concentration near or below the K d value might result in the formation of low activity monomeric DPP-IV. As a result, the specific activity measured will be decreased (33)(34)(35). We did not observe decreased specific activity for either sDPP-IV or rDPP-IV, even at a concentration as low as 1.6 nM (Fig. 7, A and B). Similarly, the specific activity of H750A protein was also constant over the range of 200 -1.6 nM (Fig. 7C). These results along with sedimentation equilibrium experiments (Fig. 6) are consistent with the lack of a dynamic equilibrium between monomeric and dimeric H750A and the small K d of wild-type DPP-IV dimer.
Effects of Salts, Active Site Inhibitor, and Dipeptide Product on DPP-IV Structure-A high concentration of anti-chaotropic salts, such as sulfate, phosphate, and citrate, enhances the stability of dimer for several herpes virus proteases (34, 36 -38). High salt buffer (0.5 M sodium sulfate) has been used to probe the dimer-monomer equilibrium for dimeric HCMV protease (36). We determined whether the same high salt (0.5 M sodium sulfate) had any effect on the quaternary structure of DPP-IVs studied in this article. As shown in Fig. 8 and Table II, high salt did not change the composition of the subunits for any DPP-IV, since the molecular masses and Stokes' radii measured by sedimentation velocity still correspond to dimer, mixture of dimer-monomer and monomer for rDPP-IV, H750A and H750E, respectively, similar to results obtained in PBS buffer. However, they all show significant global conformational changes as indicated by dramatic shifts in sedimentation coefficients ( Fig. 8 and Table II). The S values for the dimeric forms of rDPP-IV and H750A change from 8.4 to 5.0 S, and the monomeric H750A and H750E from 5.5 to 3.3 S (Fig. 8 and Table II). Interestingly, the kinetic constants of these DPP-IVs in high salt are comparable to the corresponding ones in PBS buffer with a slight increment in the K m for rDPP-IV and H750A (Table I). This result suggests that the interaction between the monomers of DPP-IV is quite different from that in HCMV protease. The subunit composition and activity of DPP-IVs were also studied in low salt buffer. We did not find any differences in either AUC analysis or catalytic activities for rDPP-IV, H750A, or H750E protein, as compared with those in PBS buffer (data not shown).
To determine whether substrate could induce dimerization of DPP-IV, we performed AUC for H750A in the presence of the proline-mimetic inhibitor and Gly-Pro dipeptide product, because the substrate Gly-Pro-pNA is cleaved by the enzyme. The proline-mimetic inhibitor, 1-(2-amino-2-cyclohexyl-acetyl)-2cyano-(S)-pyrrolidine, targets to the active site and has an IC 50 value of less than 50 nM (6). The monomeric forms of H750A and H750E do not shift to dimer in the presence of either dipeptide product or the inhibitor in either PBS or high salt buffer (data not shown).

DISCUSSION
Dimerization is an important way to regulate the activities of many proteins, such as herpes viral and retroviral proteases, SARS 3C protease, caspase 9, and STATs (39 -43). In this article, we have studied the catalytic activity of DPP-IV with respect to its quaternary structure. Despite a much lower extent of glycosylation, human DPP-IV expressed in insect cells has similar biochemical properties, catalytic activities, and dimer structure, compared with those of the endogenous human semen DPP-IV. Using cross-linking and analytical ultracentrifuation (AUC), we showed that DPP-IV is dimeric both in vivo and in vitro.
We showed that the C-terminal loop of DPP-IV is essential for dimer formation and optimal catalytic efficiency. As the dimer interface formed by the C-terminal loop is 2-fold symmetric (Fig. 1, A and B), a single mutation is therefore functionally equivalent to double mutations in this dimeric enzyme. This is the first study where monomeric DPP-IVs, H750A, and H750E, were generated, purified to homogeneity, and studied. Detailed kinetic analysis showed that monomeric H750A has a 60-fold drop of the k cat with no change in the K m , whereas both k cat and K m of H750E are remarkably changed, with a more severe effect on k cat (30-fold reduction) than K m value (10-fold increment). The result is particularly interesting since it reveals that the monomers of DPP-IV are not void of activity as previously speculated. Instead, much lower activities compared with the dimeric DPP-IV are associated with the monomeric DPP-IVs. The difference in the K m between these two monomeric DPP-IV mutant proteins, H750A and H750E, might be caused by a charge effect, affecting the conformation of the active site and/or the binding of the substrate. The data also suggest that the structure of DPP-IV is sensitive to packing interactions around His 750 . His 750 is located in the vicinity of several bulky hydrophobic residues, such as Val 726 , Val 728 , and Phe 730 , with the exception of the charged residue Asp 729 . The carbonyl of Val 728 is within hydrogen bonding distance of the imidazole ring of His 750 , as marked on Fig. 1C. The drastic effect of H750E on disrupting the dimeric DPP-IV to monomer might be caused by charge repulsion generated between Glu 750 (H750E) of one monomer and Asp 729 of the other (Fig. 1C). On the other hand, generation of the monomeric H750A suggests that the interaction mediated by the imidazole ring with the neighboring residues are crucial for dimer stability, further stressing the critical role of this residue for the C-terminal loop.
The interaction between the C-terminal loops of DPP-IV is most likely to hold the catalytic triad and the active site in an optimal position for catalysis. The formation of the monomer upon losing the dimer interaction might result in the disorientation of the loop. Since two of the three triad residues (Asp 708 and His 740 ) are located on the C-terminal loop and close to the actual dimerization interface (17), the optimal alignment of the triad needed for catalysis, the conformation of the substrate binding pocket or/and the position of an oxyanion hole might be affected upon monomer formation. The studies on dimeric HCMV and HIV proteases have revealed that upon the introduction of the deletion/mutation at the dimer interface, the active site configuration is changed and a loop involved in oxyanion hole stabilization is distorted (39,44). The failure of the DPP-IV propeller loop to hold the dimer together upon the introduction of the mutation on the C-terminal loop emphasizes the importance of the C-terminal loop in dimer formation and maintenance. In the DPP-IV-containing prolyl dipeptidase family (Fig. 2), it is not clear whether other members adopt similar quaternary structures similar to that of DPP-IV. Because the C-terminal loop of DPP-IV is very highly conserved (Fig. 2), it is likely that this loop is a general dimerization motif used by other members of the prolyl dipeptidases.
We have identified the His 750 residue, completely conserved as well among different prolyl dipeptidases, as essential for dimer formation. In addition, we found that high salt induces significant global conformational changes without affecting the subunit composition and the catalytic activities of the DPP-IVs. Therefore, salt has much less effect and is not capable of disrupting the dimer of DPP-IV to monomer or promoting dimer stability. This is contrary to HCMV protease, whose dimer is stabilized by high salt with a concomitant increase in catalytic activities (34,36,39,45). Based on our data (Figs. [5][6][7][8], the interaction between the monomers in DPP-IV is much stronger than that of the HCMV protease, supported by the lack of salt-induced effect for DPP-IV. One of the most surprising findings of this study is that there is no dynamic equilibration between the dimer and monomer of either wild-type or mutant DPP-IVs in vitro (Figs. 6 and 7). The formation of dimeric H750A by the insect cells may be assisted and promoted in vivo by chaperone proteins in the endoplasmic reticulum or by a local high concentration of the proteins during synthesis. Once dimeric H750A is formed in vivo, it does not dissociate into monomer again in vitro (Figs. 6 and 7). This indicates that there are additional interactions present in the dimer interface to compensate for the loss of the interaction by the imidazole ring of H750. The dilution experiments are consistent with the AUC experiments, indicating that there is no change of the dimer-monomer composition. This might explain the fact that up to the present time, there is no report of the isolation of the monomeric form of wild-type DPP-IV. In addition, the presence of the dipeptide product or the inhibitor failed to promote the dimerization, demonstrated in this study. These data suggested that there is not sufficient activation energy to shift either monomer to dimer or dimer to the monomer form.
To address the drug resistance commonly observed with current active site inhibitors, there is an alternative approach directed toward finding novel drugs targeting protein interface to inhibit protease activity (46,47). In fact, several drugs targeting the viral protease interface are in different phases of clinical trials (48). In this dimerization context, it is possible that we might identify a possible "binding site" at the dimer interface resulting in disrupting the active dimer to inhibit DPP-IV activity. The study presented here elucidates the reaction mechanism of DPP-IV and may facilitate the anti-diabetic drug discovery.