Structural and Kinetic Analysis of the Substrate Specificity of Human Fibroblast Activation Protein α

Fibroblast activation protein α (FAPα) is highly expressed in epithelial cancers and has been implicated in extracellular matrix remodeling, tumor growth, and metastasis. We present the first high resolution structure for the apoenzyme as well as kinetic data toward small dipeptide substrates. FAPα exhibits a dipeptidyl peptidase IV (DPPIV)-like fold, featuring an α/β-hydrolase domain and an eight-bladed β-propeller domain. Known DPPIV dipeptides are cleaved by FAPα with an ∼100-fold decrease in catalytic efficiency compared with DPPIV. Moreover, FAPα, but not DPPIV, possesses endopeptidase activity toward N-terminal benzyloxycarbonyl (Z)-blocked peptides. Comparison of the crystal structures of FAPα and DPPIV revealed one major difference in the vicinity of the Glu motif (Glu203-Glu204 for FAPα; Glu205-Glu206 for DPPIV) within the active site of the enzyme. Ala657 in FAPα, instead of Asp663 as in DP-PIV, reduces the acidity in this pocket, and this change could explain the lower affinity for N-terminal amines by FAPα. This hypothesis was tested by kinetic analysis of the mutant FAPα/A657D, which shows on average an ∼60-fold increase in the catalytic efficiency, as measured by kcat/Km, for the cleavage of dipeptide substrates. Furthermore, the catalytic efficiency of the mutant is reduced by ∼350-fold for cleavage of Z-Gly-Pro-7-amino-4-methylcoumarin. Our data provide a clear understanding of the molecular determinants responsible for the substrate specificity and endopeptidase activity of FAPα. Fibroblast activation protein α (FAPα) is highly expressed in epithelial cancers and has been implicated in extracellular matrix remodeling, tumor growth, and metastasis. We present the first high resolution structure for the apoenzyme as well as kinetic data toward small dipeptide substrates. FAPα exhibits a dipeptidyl peptidase IV (DPPIV)-like fold, featuring an α/β-hydrolase domain and an eight-bladed β-propeller domain. Known DPPIV dipeptides are cleaved by FAPα with an ∼100-fold decrease in catalytic efficiency compared with DPPIV. Moreover, FAPα, but not DPPIV, possesses endopeptidase activity toward N-terminal benzyloxycarbonyl (Z)-blocked peptides. Comparison of the crystal structures of FAPα and DPPIV revealed one major difference in the vicinity of the Glu motif (Glu203-Glu204 for FAPα; Glu205-Glu206 for DPPIV) within the active site of the enzyme. Ala657 in FAPα, instead of Asp663 as in DP-PIV, reduces the acidity in this pocket, and this change could explain the lower affinity for N-terminal amines by FAPα. This hypothesis was tested by kinetic analysis of the mutant FAPα/A657D, which shows on average an ∼60-fold increase in the catalytic efficiency, as measured by kcat/Km, for the cleavage of dipeptide substrates. Furthermore, the catalytic efficiency of the mutant is reduced by ∼350-fold for cleavage of Z-Gly-Pro-7-amino-4-methylcoumarin. Our data provide a clear understanding of the molecular determinants responsible for the substrate specificity and endopeptidase activity of FAPα. FAPα 1The abbreviations used are: FAPα, fibroblast activation protein α; DPPIV, dipeptidyl peptidase IV; Z, benzyloxycarbonyl; AFC, 7-amino-4-trifluoromethylcoumarin; AMC, 7-amino-4-methylcoumarin; CHES, 2-(cyclohexylamino)ethanesulfonic acid. 1The abbreviations used are: FAPα, fibroblast activation protein α; DPPIV, dipeptidyl peptidase IV; Z, benzyloxycarbonyl; AFC, 7-amino-4-trifluoromethylcoumarin; AMC, 7-amino-4-methylcoumarin; CHES, 2-(cyclohexylamino)ethanesulfonic acid. is expressed on reactive stromal fibroblasts in over 90% of common human epithelial cancers (1Garin-Chesa P. Old L.J. Rettig W.J. Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 7235-7239Crossref PubMed Scopus (482) Google Scholar), in granulation tissue of healing wounds, and in bone and soft tissue sarcomas (2Rettig W.J. Garin-Chesa P. Beresford H.R. Oettgen H.F. Melamed M.R. Old L.J. Proc. Natl. Acad. Sci. U. S. A. 1988; 85: 3110-3114Crossref PubMed Scopus (185) Google Scholar). Expression of FAPα is not detected in fibroblasts of benign epithelial tumors or normal adult tissues.FAPα is a type II transmembrane serine protease and has both in vitro dipeptidyl peptidase activity capable of cleaving N-terminal dipeptides from polypeptides (3Niedermeyer J. Enenkel B. Park J.E. Lenter M. Rettig W.J. Damm K. Schnapp A. Eur. J. Biochem. 1998; 254: 650-654Crossref PubMed Scopus (58) Google Scholar) and collagenolytic activity capable of degrading gelatin and type I collagen (4Levy M.T. McCaughan G.W. Abbott C.A. Park J.E. Cunningham A.M. Muller E. Rettig W.J. Gorrell M.D. Hepatology. 1999; 29: 1768-1778Crossref PubMed Scopus (238) Google Scholar, 5Park J.E. Lenter M.C. Zimmermann R.N. Garin-Chesa P. Old L.J. Rettig W.J. J. Biol. Chem. 1999; 274: 36505-36512Abstract Full Text Full Text PDF PubMed Scopus (419) Google Scholar). Both functions utilize a common active site in FAPα (5Park J.E. Lenter M.C. Zimmermann R.N. Garin-Chesa P. Old L.J. Rettig W.J. J. Biol. Chem. 1999; 274: 36505-36512Abstract Full Text Full Text PDF PubMed Scopus (419) Google Scholar). It has been suggested that FAPα promotes tumor growth and proliferation (4Levy M.T. McCaughan G.W. Abbott C.A. Park J.E. Cunningham A.M. Muller E. Rettig W.J. Gorrell M.D. Hepatology. 1999; 29: 1768-1778Crossref PubMed Scopus (238) Google Scholar, 5Park J.E. Lenter M.C. Zimmermann R.N. Garin-Chesa P. Old L.J. Rettig W.J. J. Biol. Chem. 1999; 274: 36505-36512Abstract Full Text Full Text PDF PubMed Scopus (419) Google Scholar, 6Huber M.A. Kraut N. Park J.E. Schubert R.D. Rettig W.J. Peter R.U. Garin-Chesa P. J. Invest. Dermatol. 2003; 120: 182-188Abstract Full Text Full Text PDF PubMed Scopus (122) Google Scholar, 7Cheng J.D. Dunbrack Jr., R.L. Valianou M. Rogatko A. Alpaugh R.K. Weiner L.M. Cancer Res. 2002; 62: 4767-4772PubMed Google Scholar). In contrast, other studies have suggested that FAPα has tumor suppressor activity (8Rettig W.J. Garin-Chesa P. Healey J.H. Su S.L. Ozer H.L. Schwab M. Albino A.P. Old L.J. Cancer Res. 1994; 53: 385-392Google Scholar, 9Tsujimoto H. Nishizuka S. Redpath J.L. Stanbridge E.J. Mol. Carcinogen. 1999; 26: 298-304Crossref PubMed Scopus (25) Google Scholar, 10Ramirez-Montagut T. Blachere N.E. Sviderskaya E.V. Bennett D.C. Rettig W.J. Garin-Chesa P. Houghton A.N. Oncogene. 2004; 23: 5435-5446Crossref PubMed Scopus (86) Google Scholar) and showed that this activity is independent of its enzymatic activity (10Ramirez-Montagut T. Blachere N.E. Sviderskaya E.V. Bennett D.C. Rettig W.J. Garin-Chesa P. Houghton A.N. Oncogene. 2004; 23: 5435-5446Crossref PubMed Scopus (86) Google Scholar). The role of FAPα in tumor growth and invasion and the exact molecular mechanisms the enzyme utilizes remains unknown.The closest homolog of FAPα is dipeptidyl peptidase IV (DP-PIV). In contrast to FAPα, DPPIV is ubiquitously expressed and plays a role in various processes such as T cell co-stimulation, chemokine biology, glucose metabolism, and tumorigenesis (11De Meester I. Korom S. Van Damme J. Scharpe S. Immunol. Today. 1999; 20: 367-375Abstract Full Text Full Text PDF PubMed Scopus (404) Google Scholar, 12Drucker D.J. Expert Opin. Investig. Drugs. 2003; 12: 87-100Crossref PubMed Scopus (231) Google Scholar). Interestingly, an active DPPIV/FAPα heterodimer has been identified on the surface of migratory connective tissue cells and contributes to cell migration necessary for connective tissue repair (13Scanlan M.J. Raj B.K. Calvo B. Garin-Chesa P. Sanz-Moncasi M.P. Healey J.H. Old L.J. Rettig W.J. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 5657-5661Crossref PubMed Scopus (432) Google Scholar, 14Ghersi G. Dong H. Goldstein L.A. Yeh Y. Hakkinen L. Larjava H.S. Chen W.T. J. Biol. Chem. 2002; 277: 29231-29241Abstract Full Text Full Text PDF PubMed Scopus (139) Google Scholar).To investigate the molecular determinants that are responsible for the substrate specificity of FAPα, we solved the crystal structure of the extracellular domain and measured steady state catalytic parameters for various dipeptide substrates. We also used site-directed mutagenesis to confirm the importance of active site residue Ala657 in catalysis.MATERIALS AND METHODSEnzyme Expression and Purification—The cDNA encoding human FAPα was isolated by PCR from IMAGE clone 4731931 (ATCC 6670182). The extracellular domain of FAPα (residues 27–760) was cloned into a modified pFastBacHTb vector (Invitrogen). The final construct contains the coding sequence for a baculovirus gp67 signal peptide followed by residues 27–760 of FAPα and a His6 tag. The theoretical molecular mass for the final expressed protein is 90 kDa. Human DPPIV was cloned and expressed as described previously (15Aertgeerts K. Ye S. Tennant M.G. Kraus M.L. Rogers J. Sang B.C. Skene R.J. Webb D.R. Prasad G.S. Protein Sci. 2004; 13: 412-421Crossref PubMed Scopus (151) Google Scholar). PCR-based site-directed mutagenesis was performed to generate the single amino acid substitutions FAPα/A657D and DPPIV/D663A using the constructs of the respective wild-type forms. Expression and purification of all enzymes was performed following the protocol as described previously (15Aertgeerts K. Ye S. Tennant M.G. Kraus M.L. Rogers J. Sang B.C. Skene R.J. Webb D.R. Prasad G.S. Protein Sci. 2004; 13: 412-421Crossref PubMed Scopus (151) Google Scholar).Crystallization and Structure Determination—Diffraction quality crystals were obtained at 4 °C using Syrrx's automated Nanovolume Crystallisation® technology (16Hosfield D. Palan J. Hilgers M. Scheibe D. McRee D.E. Stevens R.C. J. Struct. Biol. 2003; 142: 207-217Crossref PubMed Scopus (74) Google Scholar). The reservoir solution was 0.05 m CHES, pH 9.5, 22% polyethylene glycol 6000, and 1.26 m LiCl. Crystals appeared in about 5 days and grew to a maximum size within 7–10 days. For x-ray data collection, crystals were flash-frozen at 100 K using 25% v/v ethylene glycol as a cryoprotectant. Data were collected at beamline 5.0.3 (17Snell G. Cork C. Nordmeyer R. Cornell E. Meigs G. Yegian D. Jaklevic J. Jin J. Stevens R.C. Earnest T. Structure. 2004; 12: 537-546Abstract Full Text Full Text PDF PubMed Scopus (99) Google Scholar) of the Advanced Light Source (ALS) and integrated and scaled using HKL2000 (18Otwinowski Z. Minor W. Methods Enzymol. 1997; 276: 307-326Crossref Scopus (38368) Google Scholar). Initial phases were obtained by molecular replacement using the program MOLREP (19Vagin A. Teplyakov A. J. Appl. Crystallogr. 1997; 30: 1022-1025Crossref Scopus (4121) Google Scholar). The coordinates of human DPPIV (Protein Data Bank entry 1R9M) were used as a search model (15Aertgeerts K. Ye S. Tennant M.G. Kraus M.L. Rogers J. Sang B.C. Skene R.J. Webb D.R. Prasad G.S. Protein Sci. 2004; 13: 412-421Crossref PubMed Scopus (151) Google Scholar). The highest solution from the translation function was subjected to rigid body refinement followed by maximum likelihood refinement as implemented in REFMAC (20Murshudov G.N. Vagin A.A. Dodson E.J. Acta Crystallogr. Sect. D. 1997; 53: 240-255Crossref PubMed Scopus (13781) Google Scholar). Rigid body refinement and torsional dynamics refinement were followed by multiple rounds of manual building with Xfit (21McRee D.E. J. Struct. Biol. 1999; 125: 156-165Crossref PubMed Scopus (2019) Google Scholar). All stages of model refinement were carried out with bulk solvent correction and anisotropic scaling.Determination of Catalytic Activity—Kinetic parameters of DPPIV and FAPα for four different substrates were determined at room temperature using varying concentrations of the substrate in 100 μl of 20 mm Tris, pH 7.4, 20 mm KCl, 0.1 mg/ml bovine serum albumin, and 1% Me2SO in 96-well half-area plates. Initial rates were monitored at Ex 400 nm and Em 505 nm for 7-amino-4-trifluoromethycoumarin (AFC) substrate and at Ex 375 nm and Em 460 nm for 7-amino-4-methylcoumarin (AMC) substrate using Molecular Devices SpectraMax Gemini. Assays were performed in duplicate for each substrate. MDL data analysis toolbox was used for analysis of Michaelis-Menten kinetic parameters.RESULTS AND DISCUSSIONOverall Structure of FAPα—The asymmetric unit is composed of a homodimer with approximate dimensions of 60 × 80 × 125 Å (Fig. 1). The final R and Rfree values of the refined model are 0.22 and 0.28, respectively. Additional information on the data collection and refinement statistics is provided in the supplemental material. Each subunit contains two topologically distinct domains: the β-propeller (residues 54–492) and the α/β-hydrolase domain (residues 27–53 and 493–760). Eight blades form the β-propeller domain with each blade comprising a three- or four-stranded antiparallel β-sheet. The organization of the blades in the propeller domain results in the formation of a central pore that has a length of ∼27 Å and a diameter of ∼14 Å.The FAPα catalytic triad is composed of residues Ser624, Asp702, and His734 and is located at the interface of the β-propeller and the α/β-hydrolase domain (Fig. 1). The active site is accessible in two ways: through a central hole formed by the eight blades of the β-propeller domain or through a cavity formed between the β-propeller and the hydrolase domain. The side opening has a diameter of ∼24 Å in contrast to the narrower β-propeller opening (∼14 Å).FAPα has five potential N-linked glycosylation sites on asparagine residues 49, 92, 227, 314, and 679. Four are located in the β-propeller domain, and one is located in the hydrolase domain. We observed electron density for sugar residues for four of the five sites for subunit A and for all five sites of subunit B. The exact contribution of glycosylation to folding, dimerization, and catalysis is unclear.Dimerization of FAPα is required for its catalytic function (5Park J.E. Lenter M.C. Zimmermann R.N. Garin-Chesa P. Old L.J. Rettig W.J. J. Biol. Chem. 1999; 274: 36505-36512Abstract Full Text Full Text PDF PubMed Scopus (419) Google Scholar). The homodimer interface buries a total solvent-accessible area of ∼1400 Å2. The monomers are related by a 2-fold dyad axis. The dimer interface involves β8 of the peptidase central β-sheet, the last two α-helices (αE and αF), the loop between β6 and αD, and the anti-parallel β-sheet composed of β-strands β1* and β2* (Fig. 1).FAPα and DPPIV have a sequence identity of 50% and root mean square difference values for the overlay of 713 residues were 1.01, 1.55, and 1.75 Å for the C-α, main chain, and all heavy atoms, respectively.FAPα Dipeptidyl Peptidase Activity—Under physiological conditions, FAPα is involved in extracellular matrix degradation, but its physiological substrates remain unidentified. On the other hand, DPPIV is an exopeptidase implicated in the degradation of many chemokines, neuropeptides, and peptide hormones, and its kinetic properties toward various peptide substrates have been studied extensively (24Lambeir A.M. Proost P. Durinx C. Bal G. Senten K. Augustyns K. Scharpe S. Van Damme J. De Meester I. J. Biol. Chem. 2001; 276: 29839-29845Abstract Full Text Full Text PDF PubMed Scopus (231) Google Scholar, 25Lambeir A.M. Durinx C. Proost P. Van Damme J. Scharpe S. De Meester I. FEBS Lett. 2001; 507: 327-330Crossref PubMed Scopus (97) Google Scholar, 26Lambeir A.M. Proost P. Scharpe S. De Meester I. Biochem. Pharmacol. 2002; 64: 1753-1756Crossref PubMed Scopus (31) Google Scholar, 27Zhu L. Tamvakopoulos C. Xie D. Dragovic J. Shen X. Fenyk-Melody J.E. Schmidt K. Bagchi A. Griffin P.R. Thornberry N.A. Sinha Roy R. J. Biol. Chem. 2003; 278: 22418-22423Abstract Full Text Full Text PDF PubMed Scopus (204) Google Scholar). A series of fluorogenic substrates that contain a free N terminus and a Pro present in the penultimate position was used to compare the specific exopeptidase activity of FAPα with DPPIV. The kcat/Km values for cleavage of Gly-Pro (GP)-AFC, Ala-Pro (AP)-AFC, and GP-AMC by FAPα are much lower than 105 m-1 s-1 and are about 100 times lower compared with DPPIV.To investigate the endopeptidase activity of both enzymes, we tested their catalytic activity toward the N-terminal blocked dipeptide substrate Z-GP-AMC. This substrate is hydrolyzed by FAPα with a kcat/Km of 5.3 × 104 m-1 s-1 but is resistant to cleavage by DPPIV (Table I).Table IKinetic constants for the hydrolysis of chromogenic and fluorogenic dipeptide substrates by the wild-type form and point mutants of FAPα and DPPIV The data represent the mean of two determinants, which differed by less than 15%. ND, non-determined.FAPα wtDPPIV wtFAPα/A647DDPPIV/D663AAP-AFCkcat (s-1)ND242054Km (μm)>2001312120kcat/Km (m-1 s-1)2.1×1041.8×1061.7×1064.5×105GP-AFCkcat (s-1)ND4222NDKm (μm)>2008495>200kcat/Km (m-1 s-1)4.3×1035.0×1052.3×1051.1×105GP-AMCkcat (s-1)ND251463Km (μm)>100050100540kcat/Km (m-1 s-1)3.0×1035.0×1051.4×1051.2×105Z-GP-AMCkcat (s-1)NDNDNDNDKm (μm)NDNDNDNDkcat/Km (m-1 s-1)5.3×1049.01.5×1021.6×103 Open table in a new tab Active Site and Substrate Binding Sites of FAPα—Structural comparison of the active sites of FAPα and DPPIV revealed similar S2-S2′ specificity pockets. The S1′ (numbered according to Berger and Schechter (28Berger A. Schechter I. Philos. Trans. R. Soc. Lond. B Biol. Sci. 1970; 257: 249-264Crossref PubMed Scopus (375) Google Scholar)) subsite in FAPα is flat and could accommodate most amino acids. The S2′ active site pocket is lined by Trp623 and Tyr745. These residues would be expected to interact with large aliphatic side chains. The S1 specificity pocket in FAPα is a well defined hydrophobic pocket lined by Tyr625, Val650, Trp653, Tyr656, Tyr660, and Val705. This site optimally accommodates a proline residue. Large hydrophobic and aromatic residues can be modeled in the hydrophobic S2 pocket, defined by residues Arg123, Phe350, Phe351, Tyr541, Pro544, Tyr625, and Tyr660.The charged N-terminal end of substrate peptides is recognized by two glutamates (Glu203 and Glu204). Both residues are located on an α-helix (residues 198–204) that protrudes from the β-propeller domain into the active site (Figs. 1 and 2). Glu204 has a single acid oxygen atom solvent-accessible and available for binding. The second acid oxygen atom is effectively buried in a hydrophobic pocket defined by Trp199, Val200, and Ala657 and probably forms a hydrogen bond, at least transiently, with the hydroxyl hydrogen of Tyr656. The importance of the Glu motif (Glu205-Glu206) in DPPIV catalysis has been confirmed by single point mutants that abolish the enzyme aminopeptidase activity (29Abbott C.A. McCaughan G.W. Gorrell M.D. FEBS Lett. 1999; 458: 278-284Crossref PubMed Scopus (99) Google Scholar).Fig. 2Superposition of FAPα with DPPIV showing a detailed view of the residues around the Glu motif. The interactions in DPPIV between active site residues and the N terminus of the hexapeptide (NPY6, YPSKPD, blue), present in the crystal structure of DPPIV (yellow), are shown as dashed lines (15Aertgeerts K. Ye S. Tennant M.G. Kraus M.L. Rogers J. Sang B.C. Skene R.J. Webb D.R. Prasad G.S. Protein Sci. 2004; 13: 412-421Crossref PubMed Scopus (151) Google Scholar). Residues in FAPα are orange. Amino acid residues are labeled in italic and bold for DPPIV and FAPα, respectively. The figure was made using the program Xfit (21McRee D.E. J. Struct. Biol. 1999; 125: 156-165Crossref PubMed Scopus (2019) Google Scholar).View Large Image Figure ViewerDownload Hi-res image Download (PPT)Detailed comparison between DPPIV and FAPα revealed a more pronounced negatively charged pocket in DPPIV due to the presence of Asp663 (corresponding residue in FAPα: Ala657), which is in direct contact with Glu206 (Figs. 2 and 3). Asp663 in DPPIV pushes Glu206 further into the active site and exposes both acid oxygen atoms to the S2 pocket. This suggests that all three acidic residues may contribute to the recognition and anchoring of the free N terminus of peptide substrates (Figs. 2 and 3) in DPPIV.Fig. 3Surface representation of the active site of FAPα (A) and DPPIV (B). The hexapeptide, NPY6 (YPSKPD), is represented as a stick model and is present in the crystal structure of DPPIV (15Aertgeerts K. Ye S. Tennant M.G. Kraus M.L. Rogers J. Sang B.C. Skene R.J. Webb D.R. Prasad G.S. Protein Sci. 2004; 13: 412-421Crossref PubMed Scopus (151) Google Scholar) and modeled in the structure of FAPα. The enzymes are represented as solid surfaces, colors represent positive and negative electrostatic potentials (blue, electropositive; white, neutral; red, electronegative). The relative difference in electrostatic effect around the Glu motif is represented in the figures. Residues of the decapeptide are numbered according to Berger and Schechter (28Berger A. Schechter I. Philos. Trans. R. Soc. Lond. B Biol. Sci. 1970; 257: 249-264Crossref PubMed Scopus (375) Google Scholar). The figures were made with MOE (Chemical Computing Group, 1010 Sherbrooke St. West, Suite 910, Montreal, Quebec H3A 2R, Canada).View Large Image Figure ViewerDownload Hi-res image Download (PPT)Molecular Determinants for the Endopeptidase Activity of FAPα—To analyze the importance of Ala657 in FAPα and the corresponding residue Asp663 in DPPIV for the recognition of the N-terminal NH3+ group of substrate peptides, the mutant proteins, FAPα/A657D and DPPIV/D663A, were made and analyzed kinetically (Table I). The kcat/Km values for the cleavage by FAPα/A657D of GP-AFC, AP-AFC, and GP-AMC are on average increased by ∼60-fold compared with the wild-type form of FAPα. The measured kinetic values (kcat, Km, and kcat/Km) are similar to the values obtained for wild-type DPPIV and confirm the importance of Ala657 in substrate recognition and enzyme catalysis. DPPIV/D663A shows an ∼4-fold decrease in the catalytic efficiency for N-terminal dipeptide cleavage relative to the wild-type form of the enzyme. Replacement of Asp663 by an alanine turns DPPIV into a more efficient endopeptidase as demonstrated by the ability of this mutant to cleave Z-Gly-Pro-AMC. On the other hand, mutation of Ala657 to an Asp in FAPα reduces the kcat/Km for cleavage of Z-Gly-Pro-AMC by ∼350-fold.The structural and kinetic data described here provide a better understanding of the molecular mechanisms that determine the substrate specificity of FAPα. Comparison of our results with the reported structural and kinetic data on the related enzyme DPPIV reveals a reduced acidity in the active site of FAPα due to the presence of Ala657 in FAPα instead of Asp663 as in DPPIV. As a consequence, the affinity for N-terminal peptides is reduced, and increased endopeptidase activity is observed. FAPα 1The abbreviations used are: FAPα, fibroblast activation protein α; DPPIV, dipeptidyl peptidase IV; Z, benzyloxycarbonyl; AFC, 7-amino-4-trifluoromethylcoumarin; AMC, 7-amino-4-methylcoumarin; CHES, 2-(cyclohexylamino)ethanesulfonic acid. 1The abbreviations used are: FAPα, fibroblast activation protein α; DPPIV, dipeptidyl peptidase IV; Z, benzyloxycarbonyl; AFC, 7-amino-4-trifluoromethylcoumarin; AMC, 7-amino-4-methylcoumarin; CHES, 2-(cyclohexylamino)ethanesulfonic acid. is expressed on reactive stromal fibroblasts in over 90% of common human epithelial cancers (1Garin-Chesa P. Old L.J. Rettig W.J. Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 7235-7239Crossref PubMed Scopus (482) Google Scholar), in granulation tissue of healing wounds, and in bone and soft tissue sarcomas (2Rettig W.J. Garin-Chesa P. Beresford H.R. Oettgen H.F. Melamed M.R. Old L.J. Proc. Natl. Acad. Sci. U. S. A. 1988; 85: 3110-3114Crossref PubMed Scopus (185) Google Scholar). Expression of FAPα is not detected in fibroblasts of benign epithelial tumors or normal adult tissues. FAPα is a type II transmembrane serine protease and has both in vitro dipeptidyl peptidase activity capable of cleaving N-terminal dipeptides from polypeptides (3Niedermeyer J. Enenkel B. Park J.E. Lenter M. Rettig W.J. Damm K. Schnapp A. Eur. J. Biochem. 1998; 254: 650-654Crossref PubMed Scopus (58) Google Scholar) and collagenolytic activity capable of degrading gelatin and type I collagen (4Levy M.T. McCaughan G.W. Abbott C.A. Park J.E. Cunningham A.M. Muller E. Rettig W.J. Gorrell M.D. Hepatology. 1999; 29: 1768-1778Crossref PubMed Scopus (238) Google Scholar, 5Park J.E. Lenter M.C. Zimmermann R.N. Garin-Chesa P. Old L.J. Rettig W.J. J. Biol. Chem. 1999; 274: 36505-36512Abstract Full Text Full Text PDF PubMed Scopus (419) Google Scholar). Both functions utilize a common active site in FAPα (5Park J.E. Lenter M.C. Zimmermann R.N. Garin-Chesa P. Old L.J. Rettig W.J. J. Biol. Chem. 1999; 274: 36505-36512Abstract Full Text Full Text PDF PubMed Scopus (419) Google Scholar). It has been suggested that FAPα promotes tumor growth and proliferation (4Levy M.T. McCaughan G.W. Abbott C.A. Park J.E. Cunningham A.M. Muller E. Rettig W.J. Gorrell M.D. Hepatology. 1999; 29: 1768-1778Crossref PubMed Scopus (238) Google Scholar, 5Park J.E. Lenter M.C. Zimmermann R.N. Garin-Chesa P. Old L.J. Rettig W.J. J. Biol. Chem. 1999; 274: 36505-36512Abstract Full Text Full Text PDF PubMed Scopus (419) Google Scholar, 6Huber M.A. Kraut N. Park J.E. Schubert R.D. Rettig W.J. Peter R.U. Garin-Chesa P. J. Invest. Dermatol. 2003; 120: 182-188Abstract Full Text Full Text PDF PubMed Scopus (122) Google Scholar, 7Cheng J.D. Dunbrack Jr., R.L. Valianou M. Rogatko A. Alpaugh R.K. Weiner L.M. Cancer Res. 2002; 62: 4767-4772PubMed Google Scholar). In contrast, other studies have suggested that FAPα has tumor suppressor activity (8Rettig W.J. Garin-Chesa P. Healey J.H. Su S.L. Ozer H.L. Schwab M. Albino A.P. Old L.J. Cancer Res. 1994; 53: 385-392Google Scholar, 9Tsujimoto H. Nishizuka S. Redpath J.L. Stanbridge E.J. Mol. Carcinogen. 1999; 26: 298-304Crossref PubMed Scopus (25) Google Scholar, 10Ramirez-Montagut T. Blachere N.E. Sviderskaya E.V. Bennett D.C. Rettig W.J. Garin-Chesa P. Houghton A.N. Oncogene. 2004; 23: 5435-5446Crossref PubMed Scopus (86) Google Scholar) and showed that this activity is independent of its enzymatic activity (10Ramirez-Montagut T. Blachere N.E. Sviderskaya E.V. Bennett D.C. Rettig W.J. Garin-Chesa P. Houghton A.N. Oncogene. 2004; 23: 5435-5446Crossref PubMed Scopus (86) Google Scholar). The role of FAPα in tumor growth and invasion and the exact molecular mechanisms the enzyme utilizes remains unknown. The closest homolog of FAPα is dipeptidyl peptidase IV (DP-PIV). In contrast to FAPα, DPPIV is ubiquitously expressed and plays a role in various processes such as T cell co-stimulation, chemokine biology, glucose metabolism, and tumorigenesis (11De Meester I. Korom S. Van Damme J. Scharpe S. Immunol. Today. 1999; 20: 367-375Abstract Full Text Full Text PDF PubMed Scopus (404) Google Scholar, 12Drucker D.J. Expert Opin. Investig. Drugs. 2003; 12: 87-100Crossref PubMed Scopus (231) Google Scholar). Interestingly, an active DPPIV/FAPα heterodimer has been identified on the surface of migratory connective tissue cells and contributes to cell migration necessary for connective tissue repair (13Scanlan M.J. Raj B.K. Calvo B. Garin-Chesa P. Sanz-Moncasi M.P. Healey J.H. Old L.J. Rettig W.J. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 5657-5661Crossref PubMed Scopus (432) Google Scholar, 14Ghersi G. Dong H. Goldstein L.A. Yeh Y. Hakkinen L. Larjava H.S. Chen W.T. J. Biol. Chem. 2002; 277: 29231-29241Abstract Full Text Full Text PDF PubMed Scopus (139) Google Scholar). To investigate the molecular determinants that are responsible for the substrate specificity of FAPα, we solved the crystal structure of the extracellular domain and measured steady state catalytic parameters for various dipeptide substrates. We also used site-directed mutagenesis to confirm the importance of active site residue Ala657 in catalysis. MATERIALS AND METHODSEnzyme Expression and Purification—The cDNA encoding human FAPα was isolated by PCR from IMAGE clone 4731931 (ATCC 6670182). The extracellular domain of FAPα (residues 27–760) was cloned into a modified pFastBacHTb vector (Invitrogen). The final construct contains the coding sequence for a baculovirus gp67 signal peptide followed by residues 27–760 of FAPα and a His6 tag. The theoretical molecular mass for the final expressed protein is 90 kDa. Human DPPIV was cloned and expressed as described previously (15Aertgeerts K. Ye S. Tennant M.G. Kraus M.L. Rogers J. Sang B.C. Skene R.J. Webb D.R. Prasad G.S. Protein Sci. 2004; 13: 412-421Crossref PubMed Scopus (151) Google Scholar). PCR-based site-directed mutagenesis was performed to generate the single amino acid substitutions FAPα/A657D and DPPIV/D663A using the constructs of the respective wild-type forms. Expression and purification of all enzymes was performed following the protocol as described previously (15Aertgeerts K. Ye S. Tennant M.G. Kraus M.L. Rogers J. Sang B.C. Skene R.J. Webb D.R. Prasad G.S. Protein Sci. 2004; 13: 412-421Crossref PubMed Scopus (151) Google Scholar).Crystallization and Structure Determination—Diffraction quality crystals were obtained at 4 °C using Syrrx's automated Nanovolume Crystallisation® technology (16Hosfield D. Palan J. Hilgers M. Scheibe D. McRee D.E. Stevens R.C. J. Struct. Biol. 2003; 142: 207-217Crossref PubMed Scopus (74) Google Scholar). The reservoir solution was 0.05 m CHES, pH 9.5, 22% polyethylene glycol 6000, and 1.26 m LiCl. Crystals appeared in about 5 days and grew to a maximum size within 7–10 days. For x-ray data collection, crystals were flash-frozen at 100 K using 25% v/v ethylene glycol as a cryoprotectant. Data were collected at beamline 5.0.3 (17Snell G. Cork C. Nordmeyer R. Cornell E. Meigs G. Yegian D. Jaklevic J. Jin J. Stevens R.C. Earn