FEBS Letters 585 (2011) 3409–3414

journal homepage: www.FEBSLetters.org

Role of a propeller loop in the quaternary structure and enzymatic activity of prolyl dipeptidases DPP-IV and DPP9

Hung-Kuan Tang a,1, Ku-Chuan Chen a,1, Gan-Guang Liou b, Shu-Chun Cheng c, Chia-Hui Chien a, ⇑ ⇑ Hsiang-Yun Tang a, Li-Hao Huang a, Hui-Ping Chang d, Chi-Yuan Chou c, , Xin Chen a, a Division of Biotechnology and Pharmaceutical Research, National Health Research Institutes, Miaoli 350, Taiwan, ROC b Division of Molecular and Genomic Medicine, National Health Research Institutes, Miaoli 350, Taiwan, ROC c Department of Life Sciences and Institute of Genome Sciences, National Yang-Ming University, Taipei 112, Taiwan, ROC d Department of Applied Chemistry, National Chia Yi University, Chia Yi, Taiwan, ROC article info abstract

Article history: The dipeptidyl peptidase (DPP) family members, including DPP-IV, DPP8, DPP9 and others, cleave Received 3 September 2011 the peptide bond after the penultimate proline residue and are drug target rich. The dimerization Accepted 3 October 2011 of DPP-IV is required for its activity. A propeller loop located at the dimer interface is highly con- Available online 10 October 2011 served within the family. Here we carried out site-directed mutagenesis on the loop of DPPIV and identified several residues important for dimer formation and enzymatic activity. Interestingly, Edited by Miguel De la Rosa the corresponding residues on DPP9 have a different impact whereby the mutations decrease activ- ity without changing dimerization. Thus the propeller loop seems to play a varying role in different Keywords: DPPs. Dipeptidyl peptidase-IV Dipeptidyl peptidase 9 Propeller loop Structured summary of interactions: Dimerization DPP-IV and DPP-IV physically interact by comigration in gel electrophoresis (View interaction: 1, 2, 3, 4) Analytical ultracentrifugation DPP9 and DPP9 bind by circular dichroism (View interaction) DPP-IV and DPP-IV bind by circular dichroism (View interaction: 1, 2, 3, 4, 5) DPP-IV and DPP-IV bind by cosedimentation in solution (View interaction: 1, 2, 3, 4, 5) ADA binds to DPP-IV by surface plasmon resonance (View interaction: 1, 2, 3, 4, 5, 6) DPP9 and DPP9 bind by cosedimentation in solution (View interaction)

Ó 2011 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. Introduction diabetes [2]. It is involved in the degradation of two insulin-sensing hormones, glucagon-like peptide-1 and glucose-dependent insuli- The prolyl-cleaving dipeptidase family are serine proteases notropic polypeptide [3,4]. Chemical inhibitors of DPP-IV are effec- including dipeptidyl peptidase IV (DPP-IV), fibroblast activation tive at prolonging the half-life of these hormones and can be used in protein (FAP), DPP8, and DPP9 [1]. They preferentially cleave the the treatment of diabetes [5]. In addition, DPP-IV also plays an peptide bond after the penultimate proline residue. This unique important role in the regulation of adenosine signaling and in activity distinguishes this family from most cellular enzymes, and potentiating T-cell proliferation through interacting with adenosine makes them important to various biological functions in vivo [1]. deaminase (ADA) [6]. Determining the structure and mechanism of The most well-studied member of the family is DPP-IV (EC DPP-IV will help elucidate not only its function, but those of other 3.4.14.5), which is a validated drug target for human type II prolyl cleaving enzymes, whose functions are largely unknown. DPP-IV is a membrane protein with a short cytoplasmic tail and a

Abbreviations: DPP-IV, dipeptidyl peptidase IV; DPP, dipeptidyl peptidase; POP, transmembrane domain (residues 1–28). The crystal structure of the prolyl oligopeptidase; H-Gly-Pro-pNA, HGly-Pro-p-nitroanilide; DTT, dithiothreitol; ectodomain (residues 29–766) has revealed that it is homodimeric AUC, analytical ultracentrifugation; CD, circular dichroism; FAP, fibroblast activa- [7] (Fig. 1A). It contains an a/b-hydrolase domain and a b-propeller tion protein; ADA, adenosine deaminase; SPR, surface plasmon resonance. ⇑ domain, with the active site located between them. The dimeric Corresponding authors. Addresses: 155 Li-Nong St., Sec. 2, Taipei 112, Taiwan, interface included the C-terminal loop from the hydrolase domain ROC. Fax: +886 2 28202449 (C.-Y. Chou). 35 Keyan Rd., Zhunan town, Miaoli County 350, Taiwan, ROC. Fax: +886 37 586456 (X. Chen). and the b-propeller loop from the propeller domain. Previously, E-mail addresses: [email protected] (C.-Y. Chou), [email protected] (X. Chen). we have shown that a single mutation within the C-terminal loop 1 These authors contributed equally to this work. disrupts DPP-IV dimerization with a concomitant loss of enzymatic

0014-5793/$36.00 Ó 2011 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2011.10.009 3410 H.-K. Tang et al. / FEBS Letters 585 (2011) 3409–3414

Fig. 1. Schematic representation of the residues in the propeller loop of DPP-IV. (A) DPP-IV structure (1N1 M) with the propeller loop highlighted in yellow. Residues Y248, Y238 and Y256 are shown as green sticks. The catalytic triad S630-H740-D708 is represented in red and the D-D motif as orange. (B) Sequence alignment of the propeller loops from DPPs. GenBank accession numbers: NP_001926 (DPP-IV), Q12884 (FAP), NP_570629.2 (DPP6), AAG29766 (DPP8), NP_631898.3 (DPP9) and P42658 (DPP10) (C–E) Detailed interactions of Y248, Y238 and Y256 of DPP-IV with neighboring residues. The residues on the same subunit are shown in silver, and those on the opposite subunit in pink. The figures were drawn with PyMOL [28]. activity [8,9]. We also have observed that no equilibrium existing the propeller loop of DPP-IV to these areas using site-directed between the dimer and monomer states of DPP-IV, indicating that mutagenesis. Moreover, to understand whether the propeller loop the monomer conformation is different enough to prevent direct also affects other DPPs, we assessed the effect of similar mutations association with another monomer [8,9]. Until now, the contribu- targeting the putative propeller loop of DPP9. tions of the propeller loop to DPP-IV are unknown. Several other members of the DPP family have been identified with a various cellular functions. DPP8 and DPP9 are also dimeric 2. Materials and methods with DPP activity [10–13]. Unlike DPP-IV, they are found in the cytosol. DPP9 is the major DPP in the cytoplasm and responsible 2.1. Materials for at least 80% of the cleavage activity measured using dipeptide substrates [12]. Recently it has been found to be important to The human liver cDNA library and viral vector were obtained immune function [12]. Another member, FAP, is highly homolo- from Clontech (Mountain View, CA, USA) and fetal bovine serum, gous with DPP-IV. It has been implicated as a potential anti-cancer lipofectamine, and insect culture media from Invitrogen (Carlsbad, target [14]. Studying the structure and activity of these enzymes is CA, USA). The Western detection was obtained from Perkin- important to our understanding of their function in vivo. Elmer (Waltham, Massachusetts, USA) and bovine ADA from Roche The propeller loop is highly conserved among DPP-IVs from dif- (Branchburg, NJ, USA). Size-exclusion column, CM5 chip, and the ferent species (Supplemental Fig. 1A) and moderately conserved nickel affinity column were produced by GE healthcare (Uppsala, among the DPP family as a whole (Fig. 1B). It is an anti-parallel Sweden). Amicon filter with 10 kDa-cut was bought from Millipore two-stranded b-sheet, extending from blade 4 in the propeller do- (Billerica, MA, USA). Ala-Pro-pNA was purchased from Bachem main and interacting with the same loop of another subunit (Torrance, CA, USA). (Fig. 1A) [7]. DPP-IV has the longest propeller loop (residues 234– 260), while the loops from others are two to three residues shorter. 2.2. Plasmid construction, insect and mammalian cell culture Intriguingly, based on the structures, the propeller loops from the DPP family are generally longer than those from other prolyl- To construct the plasmids with mutations for expression in cleaving endopeptidases such as the prolyl oligopeptidases (POPs) insect cells, the baculovirus expression plasmid pBac-CD5-DPPIV [15], which are 12 to 15 amino acids long (Supplemental Fig. 1B). (residues 39–766) [8] and pBacMTeGFP-StrepDPP9 (StrepTagII in In this context, DPP-IV is only active as a dimer, while the POPs are N-terminus) [11] were used. Site-directed mutagenesis was active as a monomer. Whether the length of the propeller loop plays performed using quick-change mutagenesis [8,9]. The primers a part in the dimerization of DPP family remains unclear. used are listed in Supplemental Table 1. The nucleotide sequences Despite speculation [7], the function of the DPP-IV propeller were confirmed by autosequencing analysis. Sf9 and Hi5 cells were loop in terms of quaternary structure and enzymatic activity has grown in TNM-FH medium supplemented with 10% fetal bovine not been studied as yet. Here we investigated the contribution of serum and EXPRESS FIVEÒ SFM medium, respectively, at 27 °C H.-K. Tang et al. / FEBS Letters 585 (2011) 3409–3414 3411

[8,9]. Baculoviruses containing the DPP-IV and DPP9 mutants were kinetic parameters were obtained by fitting the results to the generated as described previously [8,9,13]. Michaelis–Menten equation as described previously [9]. The plasmid pIRES-CD5-A39-DPPIV that expresses secreted DPP- IV and its mutants were constructed. The plasmids were transfected 2.4. Non-reducing SDS–PAGE and Western blot analysis into HEK293Tcells using Lipofectamine. The concentrated media containing 5 lg DPP-IV were separated by non-reducing SDS–PAGE The separation of the dimers and monomers present in the cul- at 4 °C. ture media of HEK293T was carried out as described previously [17].

2.5. Analytical ultracentrifugation (AUC) 2.3. Expression, purification and determination of the kinetic constants of the DPP-IV and DPP9 AUC of the DPP proteins was performed as described previously [8,9,13], at concentrations of 0.1–0.5 mg/ml. Briefly, sedimentation His-tagged DPP-IV was expressed and purified from baculovi- velocity experiments were performed using a double-sector rus-infected insect cells as described in our previous study [8]. centerpiece at 20 °C with a rotor speed of 42,000 rpm. OD280 was StrepTagII-tagged DPP9 was expressed and purified from baculovi- monitored in a continuous mode every 10 min for 3 h. The multiple rus-infected Sf9 cells as described in our previous study [11]. Tag- scans were then fitted to a continuous size distribution model ging the amino-terminus of DPP-IV did not affect the quaternary using SEDFIT [18] (Fig. 2A), which calculated the physical parame- structure or its enzymatic activity [16]. Protein purity was deter- ters of each molecular species. The distributions were analyzed at a mined by SDS–PAGE (Supplemental Fig. 2). Size-exclusion chroma- confidence level of P = 0.95 by maximal entropy regularization and tography was utilized to separate the monomers of DPP-IV from at a resolution N of 200 for molar masses between 0 and 500 kDa. the dimers for the activity measurement [8,9]. The kinetic constants were determined using the chromogenic 2.6. Circular dichroism (CD) and melting temperature determination substrate Ala-Pro-pNA [8,9,13]. The reaction was performed in PBS (pH 7.5) and monitored at OD405 by detecting the released CD was carried out on a Jasco J-815 spectropolarimeter (Tokyo, pNA. The enzyme concentrations used were 10–20 nM for wild- Japan) as described previously [19]. The ellipticity at 222 nm was type DPP-IV, the dimeric mutants, and DPP9 proteins. The enzyme recorded at varying temperature ranging from 30 °Cto85°C. The concentrations used were 200–500 nM for the monomeric DPP-IV results were fitted to the two-state unfolding model [20] to calcu- mutants, and 100 nM for the DPP9 mutants. The steady state late the melting temperature (Tm) of the DPPs.

Y248A 0.6 D A 0.5

0.4

0.3 E Y248T

0.2

0.1

Absorbance at 280 nm 0.0 F Y248F

6.26.46.66.87.0 c (M) Radius (cm) B wt DPP-IV G Y238A

c (M) C DPP-IV-del H Y256A

Molar mass (kDa)

Molar mass (kDa)

Fig. 2. AUC of DPP-IVs and its mutants. (A) A typical trace of absorbance at 280 nm of wild-type DPP-IV during an experiment. Symbols represent experimental data, and lines are the results fitted to the Lamm equation using SEDFIT [18]. (B–H) the molar mass distribution of wild-type DPP-IV, DPP-IV-del, Y248A, Y248T, Y248F, Y238A, and Y256A mutants from the best-fit analysis. The protein concentrations were about 0.1–0.5 mg/ml. The residual bitmaps of raw data and the best-fit results are shown in the insets. 3412 H.-K. Tang et al. / FEBS Letters 585 (2011) 3409–3414

2.7. Surface plasma resonance (SPR) According to this analysis, we found that the hydroxyl group of Y248 may interact with the carbonyl oxygen of Y256 and the main SPR analysis of the binding of DPP-IV and its mutants to bovine chain amide nitrogen of K258 from the other subunit (Fig. 1C). More- ADA was performed using a BIAcore-2000 (BIAcore, Uppsala, over, the hydrophobic phenyl group of Y248 is less than 4 Å away Sweden) [9]. Briefly, ADA was immobilized on a CM5 sensor chip from other residues on the opposite subunit, including L235, P257, using amine-coupling reagents. After immobilization, 4500–8800 K258 and Y256 (Fig. 1C). These interactions by Y248 are completely response units were detected. The interaction kinetics was deter- conserved in DPPs including FAP and DPP6 as examined by dimplot. mined by injecting 90 ll of DPP-IV proteins at a flow rate of To investigate the significance, three mutations, Y248A, Y248F, and 30 ll/min for 3 min and then PBS for 5 min to achieve dissociation. Y248T, were generated. As shown by AUC and kinetic assay, Y248F The protein concentrations of DPP-IV and its mutants ranged from remained dimeric and had a similar activity to wild-type DPP-IV 94 nM to 1.5 lM. The evaluation software provided by the manu- (Fig. 2F, Table 1). In contrast, Y248A and Y248T were almost com- facturer was used to analyze and calculate kon and koff using a pletely dissociated into monomers (Fig. 2D and E). The activity of 1:1 interaction model. the monomeric Y248T was decreased 1300-fold with a reducedkcat and an unchanged Km, while no activity was detected when Y248A was examined (Table 1). In addition, we did not detect any activity 3. Results using other dipeptidase, endopeptidase or tri-peptidase substrates as detailed above. 3.1. A shortened propeller loop results in monomeric DPP-IV with a complete loss of activity 3.3. Disruption of intra-molecular interactions results in monomeric DPP-IVs To determine the importance of the propeller loop, a deletion of residues 236–253 within the loop was carried out by replacing Other than intermolecular interaction, the propeller loop also these residues with three Gly residues to create DPP-IV-del. has several potential intramolecular interactions with the residues Measured by AUC, wild-type DPP-IV was dimeric (Fig. 2B), with a from the same subunit; these involve Y238 and Y256 (Fig. 1D and molecular mass of 187 kDa, a Stokes’ radius of 5.1 nm, and a sedi- E). The two residues are highly conserved among DPP enzymes mentation coefficient of 8.7 S (Supplemental Table 2). (Supplemental Fig. 1A). To investigate their importance, Y238 Interestingly, the shortening of the propeller loop resulted in and Y256 were individually mutated to Ala. By AUC analysis, both the formation of exclusively monomer (Fig. 2C), with a molecular mutations resulted in a mixture of dimers and monomers (Fig. 2 mass of 98 kDa, a Stokes’ radius of 4.4 nm, and a sedimentation G and H and Supplemental Table 2). The dimers and monomers coefficient of 5.2 S. The f/fo ratios of the proteins were 1.3 and did not equilibrate with each other and could be isolated by gel- 1.4, respectively, indicating that both proteins were globular. No filtration chromatography. The enzymatic activities of the dimeric enzymatic activity was associated with the monomeric DPP-IV mutants were similar to that of wild-type DPP-IV, while no activity using Ala-Pro-pNA as the substrate (Table 1). We also investigated was detected with either monomer (Table 1). whether the deletion mutant had different substrate specificity. Next, to determine the quaternary structures of the DPP-IVs None of the dipeptide substrates from the X-Pro or Ala-X libraries generated in mammalian cells, we transfected the expression plas- [13], various endopeptidase substrates, such as Z-Gly-Pro-pNA, mids into HEK293T cells. Culture media were collected and the Z-Ala-Ala-pNA and Z-Ala-Pro-pNA, or various tripeptidase proteins were separated by non-reducing SDS–PAGE. Similar to substrates, such as Ala-Phe-Pro-pNA and Ala-Ala-Pro-pNA, were AUC, the DPP-IVs mutants, including Y238A, Y248A, Y248T, and cleavable by DPP-IV-del. Y256A, were all monomeric (Fig. 3, lanes 2, 3, 5 and 6). This indi- cates that what we observed is a direct consequence of the muta- 3.2. Propeller loop interaction is essential for the formation of dimeric tion and not an in vitro purification artifact. The reason for the low DPP-IVs expression level of Y256A was not clear (lane 6). However, differ- ent to the measurement by AUC, Y248F was a mixture of dimer The propeller loop of DPP-IV is involved in both inter- and intra- and monomers, with a ratio of 2 to 1 (lane 4). This suggests that molecular interaction, as analyzed by dimplot for protein–protein the hydroxyl group of Y248 may contribute to the stability of the interaction (http://www.biochem.ucl.ac.uk/bsm/ligplot/ligplot.html). protein. The evidence for this is the lower Tm of Y248F (see below).

Table 1 3.4. Thermodenaturation and the interaction of the DPP-IV mutants a Kinetic constants of the wild-type and mutant DPP-IVs and DPP9s. with ADA

À1 À1 À1 Quaternary structure kcat (s ) Km (lM) kcat/Km (s lM ) To determine whether the mutations introduced in this study DPP-IV Wild type dimer 26.0 ± 3.6 11 ± 2 2.4 ± 0.5 affect protein thermostability, the Tm’s were measured by CD. For DPP-IV-del monomer ND ND ND Y248A monomer ND ND ND Y248T monomer 0.02 ± 0.002 12 ± 2 0.002 ± 0.0004 Y248F dimer 20.4 ± 3.4 10 ± 2 2.0 ± 0.5 Y238A dimer 18.3 ± 1.2 8 ± 1 2.3 ± 0.3 Y238A Monomer ND ND ND Y256A dimer 22.1 ± 1.1 15 ± 1 1.5 ± 0.1 Y256A monomer ND ND ND DPP9 Wild type dimer 33.0 ± 3.0 99 ± 7 0.33 ± 0.03 DPP9-del dimer 6.4 ± 0.6 320 ± 320 0.02 ± 0.02 Y334A dimer 1.8 ± 0.5 4128 ± 899 0.0004 ± 0.0001 Fig. 3. DPP-IVs expressed in HEK293T cells. Plasmids encoding DPP-IV and mutants ND: enzymatic activity non-detectable. were transfected in HEK293T cells and separated by electrophoresis on non- a The data were calculated from three independent measurements using differ- reducing SDS–PAGE followed by Western blotting. Lanes 1–7, wild-type DPP-IV, ent batches of enzymes. The substrate used was H-Ala-Pro-pNA. Y238A, Y248A, Y248F, Y248T, Y256A and F713A. Lane 8 represents the control pIRES plasmid. H.-K. Tang et al. / FEBS Letters 585 (2011) 3409–3414 3413

Table 2 Thermal denaturation temperatures of DPP-IVs and DPP9s.a

Quaternary structure Tm (°C) DPP-IV Wild type dimer 70.3 DPP-IV-del monomer 57.3 Y248A monomer 59.5 Y248T monomer 57.9 Y248F dimer 60.7 Y238A dimer 63.9 Y238A monomer 60.5 Y256A dimer 60.3 Y256A monomer 61.7 DPP9 Wild type DPP9 58.9 DPP9-del dimer 52.9 Y334A dimer 60.9

a For each protein, the Tm measurement was carried out twice using two different batches of enzymes and the difference was less than 1 °C. One set of the data is presented.

the wild-type DPP-IV, the Tm was around 70 °C, while those for monomeric DPP-IVs were between 57 and 60 °C(Table 2). Interest- ingly, the Tm’s for the dimeric mutants were between 60 and 63 °C, indicating that, even though the mutations may not disrupt the dimerization, protein thermostability is still decreased. Interaction of DPP-IV with ADA is through the propeller domain of DPP-IV, opposite the propeller loop [21]. We used SPR to deter- mine whether the interaction with ADA was changed significantly Fig. 4. AUC of DPP9 and its mutants. From top to bottom: the molar mass in the monomeric DPP-IVs. Wild-type DPP-IV interacted with ADA distribution of wild-type DPP9, DPP9-del and Y334A from the best-fit AUC analysis. with Kd of 21 nM, while the mutants that interacted with ADA had The protein concentrations were 0.2–0.5 mg/ml. The residual bitmaps are shown in a higher Kd by 4-fold or less (Table 3). The results suggest that there the insets. is a limited conformational change in the propeller domain upon monomerization. indicating that the putative propeller loop contributes to protein 3.5. Similar mutations in DPP9 have different effects to those in DPP-IV thermostability.

The structure of DPP9 has not been solved. Based on computer 4. Discussion modeling and , it is speculated that DPP9 has a similar domain structure to DPP-IV [22]. The proposed propeller In this study, we showed that the propeller loop is critical for loops of DPP9 from various species are also very conserved (Sup- the dimerization and consequently the enzymatic activity of plemental Fig. 1C). Based on this, we generated mutations includ- DPP-IV. Interestingly, similar mutations in DPP9 have a different ing a deletion mutation by replacing residues 317–334 with three impact and decrease the enzymatic activity without changing its Glys (DPP9-del) and Y334A. Y334 corresponds to Y256 of DPP-IV dimerization. Thus, the propeller loop seems to play different role By AUC, both DPP9-del and Y334A were dimers (Fig. 4) and they among the DPPs. We also identified three residues, Y238, Y248, and both had residual enzymatic activity with a significantly increased Y256 in the propeller loop of DPP-IV that are important to the pro- K and decreased k (Table 1). Overall, the k /K values were m cat cat m tein dimerization. These residues are highly conserved among 1/16.5 and 1/825 of the wild-type one. Intriguingly, different from DPPs. Based on protein-ligand analysis, the hydrogen bonding DPP-IV, the propeller loop of DPP9 affects enzymatic activity dras- and hydrophobic interactions between Tyrs and their neighboring tically without affecting the dimerization. A thermostability analy- residues are also present in FAP and DPP6 [23,24]. Thus, our studies sis of DPP9 was also carried out to evaluate the mutational effects can help an understanding of the biochemical properties of this (Table 2). Wild-type DPP9 and the Y334A mutant had a T around m family, which are potentially drug target-rich. 60 °C, respectively, while the DPP9-del mutant had a T of 53 °C, m DPP9 is the most abundant enzyme contributing to the prolyl dipeptidase activity in the cytoplasm and important to immuno- Table 3 logical functioning of cells [12]. We showed that dimerization of Kinetics and binding affinities of the monomeric DPP-IVs for ADA by SPR. DPP9 is dramatically different from that of DPP-IV. Either deletion

À1 À1 À1 a or mutation within the propeller loop fails to change dimerization Quaternary structure kon (M s ) koff (s ) Kd (nM) even though activity is significantly decreased. In our previous Wild type DPP-IV 4.2 ± 1.1 Â 103 8.7 ± 0.6 Â 10À5 21 ± 5.7 DPP-IV-del monomer 4.0 ± 1.4 Â 103 3.3 ± 1.2 Â 10À4 83 ± 41.9 studies, we have found that a single mutation in the C-terminal Y248A monomer 5.4 ± 0.5 Â 103 2.8 ± 0.4 Â 10À4 52 ± 8.9 loop of DPP9 renders the protein enzymatically inactive while Y248T monomer 12.8 ± 2.4 Â 103 2.8 ± 1.3 Â 10À4 22 ± 11 the protein remained dimeric [11]. Therefore, our studies suggest 3 À4 Y238A monomer 6.8 ± 0.8 Â 10 5.4 ± 0.7 Â 10 79 ± 13.8 that the DPP8/9 and DPP-IV interfaces interact differently, even 3 À4 Y256A monomer 5.7 ± 0.3 Â 10 3.3 ± 0.2 Â 10 58 ± 4.7 though their C-terminal loop and b-propeller loop are highly a Kd was calculated as koff/kon, where kon and kon are the association and disso- conserved [8]. The underlying mechanism remains to be under- ciation rate constants, respectively. stood and awaits a high resolution structure for DPP9 in the future. 3414 H.-K. Tang et al. / FEBS Letters 585 (2011) 3409–3414

We have attempted to crystallize monomeric DPP-IVs unsuc- [3] Conarello, S.L. et al. (2003) Mice lacking dipeptidyl peptidase IV are protected cessfully. Cryo-EM study was therefore carried out on Y248A against obesity and insulin resistance. Proc. Natl. Acad. Sci. USA 100, 6825– 6830. (Supplemental Fig. 3). The result confirmed that the monomeric [4] Marguet, D. et al. (2000) Enhanced insulin secretion and improved glucose DPP-IV largely maintains its conformation except for the propeller tolerance in mice lacking CD26. Proc. Natl. Acad. Sci. USA 97, 6874–6879. loop region. Missing electron density was found in the propeller [5] Deacon, C.F., Hughes, T.E. and Holst, J.J. (1998) Dipeptidyl peptidase IV inhibition potentiates the insulinotropic effect of glucagon-like peptide 1 in loop, which is likely to be due to its dynamic movement. As the the anesthetized pig. Diabetes 47, 764–769. result, the conformational changes in the loop may distort the [6] Dong, R.P. and Morimoto, C. (1996) Role of CD26 for CD4 memory substrate binding pocket and/or the catalytic triad, which could function and activation. Hum. Cell 9, 153–162. [7] Rasmussen, H.B., Branner, S., Wiberg, F.C. and Wagtmann, N. (2003) Crystal then lead to very reduced enzymatic activity. Indeed, our kinetic structure of human dipeptidyl peptidase IV/CD26 in complex with a substrate assays suggested that the monomeric mutants have either a very analog. Nat. Struct. Biol. 10, 19–25. low or undetectable enzymatic activity. In addition, the affinity [8] Chien, C.H., Huang, L.H., Chou, C.Y., Chen, Y.S., Han, Y.S., Chang, G.G., Liang, P.H. and Chen, X. (2004) One site mutation disrupts dimer formation in human between the ADA and monomeric DPP-IV does not change signifi- DPP-IV proteins. J. Biol. Chem. 279, 52338–52345. cantly, confirming that the mutations did not influence overall [9] Chien, C.H., Tsai, C.H., Lin, C.H., Chou, C.Y. and Chen, X. (2006) Identification of folding. Some mutations do not affect the DPP-IV dimerization hydrophobic residues critical for DPP-IV dimerization. Biochemistry 45, 7006– but do lower the T , suggesting that the residues may contribute 7012. m [10] Bjelke, J.R., Christensen, J., Nielsen, P.F., Branner, S., Kanstrup, A.B., Wagtmann, the protein stability. N. and Rasmussen, H.B. (2006) Dipeptidyl peptidase 8 and 9 specificity and Our study seems to also reveal differences in the propeller loops molecular characterization compared to dipeptidyl peptidase IV. Biochem. J. among two types of prolyl cleaving enzymes, the DPPs and POPs. 396, 391–399. [11] Tang, H.K., Tang, H.Y., Hsu, S.C., Chu, Y.R., Chien, C.H., Shu, C.H. and Chen, X. POPs and DPPs share a similar two domain structure with the (2009) Biochemical properties and expression profile of human prolyl active site located between the two domains [1]. POP is a dipeptidase DPP9. Arch. Biochem. Biophys. 485, 120–127. prolyl-cleaving endopeptidase and its monomer is active [15].Itis [12] Geiss-Friedlander, R., Parmentier, N., Moller, U., Urlaub, H., Van den Eynde, B.J. and Melchior, F. (2009) The cytoplasmic peptidase DPP9 is rate-limiting for a potential drug target for neurodegenerative disease [25]. Three degradation of proline-containing peptides. J Biol. Chem. 284, 27211–27219. structures are available including porcine, Myxococcus xanthus, [13] Lee, H.J., Chen, Y.S., Chou, C.Y., Chien, C.H., Lin, C.H., Chang, G.G. and Chen, X. and Sphingomonas capsulata POPs [26,27]. Although the propeller (2006) Investigation of the Dimer Interface and Substrate Specificity of Prolyl Dipeptidase DPP8. J. Biol. Chem. 281, 38653–38662. loop of POPs also extends from the propeller domain, they are much [14] Santos, A.M., Jung, J., Aziz, N., Kissil, J.L. and Pure, E. (2009) Targeting fibroblast shorter than those of the DPPs (Supplemental Fig. 1B). The propeller activation protein inhibits tumor stromagenesis and growth in mice. J. Clin. loop of POP interacts with the hydrolase domain. However, muta- Invest. 119, 3613–3625. [15] Polgar, L. (2002) The prolyl oligopeptidase family. Cell. Mol. Life Sci. 59, 349– tions within the POP’s propeller loop do not affect either catalytic 362. activity or substrate specificity [27]. On comparison, mutations [16] Aertgeerts, K. et al. (2004) Crystal structure of human dipeptidyl peptidase IV within the propeller loop of DPP-IV and DPP9 affected enzymatic in complex with a decapeptide reveals details on substrate specificity and activity significantly, either through dimer maintenance, as is the tetrahedral intermediate formation. Protein Sci. 13, 412–421. [17] Chung, K.M. et al. (2010) The dimeric transmembrane domain of prolyl case with DPP-IV, or in a less obvious way as with DPP9. In sum- dipeptidase DPP-IV contributes to its quaternary structure and enzymatic mary, we have found that the propeller loop is essential for the activities. Protein Sci. 19, 1627–1638. enzymatic activity of DPPs, but that it has different structural roles [18] Schuck, P. (2000) Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and Lamm equation modeling. among the various DPPs. Biophys. J. 78, 1606–1619. [19] Chen, X., Cook, R.K. and Rubenstein, P.A. (1993) Yeast actin with a mutation in Acknowledgment the ‘‘hydrophobic plug’’ between subdomains 3 and 4 (L266D) displays a cold- sensitive polymerization defect. J. Cell. Biol. 123, 1185–1195. [20] Pace, C.N. (1990) Measuring and increasing protein stability. Trends We thank Dr. Wei-Hau Chang for the electron microscopy and Biotechnol. 8, 93–98. Gu-Gang Chang, Peter Rubenstein, Ann Marie Stanley and Wang [21] Weihofen, W.A., Liu, J., Reutter, W., Saenger, W. and Fan, H. (2004) Crystal structure of CD26/dipeptidyl-peptidase IV in complex with adenosine Chung for helpful suggestions. The study was financially supported deaminase reveals a highly amphiphilic interface. J. Biol. Chem. 279, 43330– by NRPGM (to X.C.) and an individual grant (to C.Y.C.) from Na- 43335. tional Science Council and National Health Research Institutes, Tai- [22] Rummey, C. and Metz, G. (2007) Homology models of dipeptidyl peptidases 8 and 9 with a focus on loop predictions near the active site. Proteins 66, 160– wan, ROC (to X.C.). 171. [23] Aertgeerts, K. et al. (2005) Structural and kinetic analysis of the substrate Appendix A. Supplementary data specificity of human fibroblast activation protein alpha. J. Biol. Chem. 280, 19441–19444. Supplementary data associated with this article can be found, in [24] Strop, P., Bankovich, A.J., Hansen, K.C., Garcia, K.C. and Brunger, A.T. (2004) Structure of a human A-type potassium channel interacting protein DPPX, a the online version, at doi:10.1016/j.febslet.2011.10.009. member of the dipeptidyl aminopeptidase family. J. Mol. Biol. 343, 1055– 1065. References [25] Szeltner, Z. and Polgar, L. (2008) Structure, function and biological relevance of prolyl oligopeptidase. Curr. Protein Pept. Sci. 9, 96–107. [1] Rosenblum, J.S. and Kozarich, J.W. (2003) Prolyl peptidases: a serine protease [26] Fulop, V., Bocskei, Z. and Polgar, L. (1998) Prolyl oligopeptidase: an unusual subfamily with high potential for drug discovery. Curr. Opin. Chem. Biol. 7, beta-propeller domain regulates proteolysis. Cell 94, 161–170. 496–504. [27] Shan, L., Mathews II and Khosla, C. (2005) Structural and mechanistic analysis [2] Bergman, A.J. et al. (2006) Pharmacokinetic and pharmacodynamic properties of two prolyl endopeptidases: role of interdomain dynamics in catalysis and of multiple oral doses of sitagliptin, a dipeptidyl peptidase-IV inhibitor: a specificity. Proc. Natl. Acad. Sci. USA 102, 3599–3604. double-blind, randomized, placebo-controlled study in healthy male [28] DeLano, W.L. (2002) The PyMOL Manual, DeLano Scientific, San Carlos, CA. volunteers. Clin. Ther. 28, 55–72.