N-linked glycosylation of dipeptidyl peptidase IV (CD26): effects on enzyme activity, homodimer formation, and adenosine deaminase binding

Abstract

The type II transmembrane serine protease dipeptidyl peptidase IV (DPPIV), also known as CD26 or adenosine deaminase binding protein, is a major regulator of various physiological processes, including immune, inflammatory, nervous, and endocrine functions. It has been generally accepted that glycosylation of DPPIV and of other transmembrane dipeptidyl peptidases is a prerequisite for enzyme activity and correct protein folding. Crystallographic studies on DPPIV reveal clear N-linked glycosylation of nine Asn residues in DPPIV. However, the importance of each glycosylation site on physiologically relevant reactions such as dipeptide cleavage, dimer formation, and adenosine deaminase (ADA) binding remains obscure. Individual Asn-->Ala point mutants were introduced at the nine glycosylation sites in the extracellular domain of DPPIV (residues 39-766). Crystallographic and biochemical data demonstrate that N-linked glycosylation of DPPIV does not contribute significantly to its peptidase activity. The kinetic parameters of dipeptidyl peptidase cleavage of wild-type DPPIV and the N-glycosylation site mutants were determined by using Ala-Pro-AFC and Gly-Pro-pNA as substrates and varied by <50%. DPPIV is active as a homodimer. Size-exclusion chromatographic analysis showed that the glycosylation site mutants do not affect dimerization. ADA binds to the highly glycosylated beta-propeller domain of DPPIV, but the impact of glycosylation on binding had not previously been determined. Our studies indicate that glycosylation of DPPIV is not required for ADA binding. Taken together, these data indicate that in contrast to the generally accepted view, glycosylation of DPPIV is not a prerequisite for catalysis, dimerization, or ADA binding.

Figure 1.

(A) Sequence alignment of human, rat, mouse, bovine, and pig DPPIV. The potential glycosylation sites predicted by using the Prosite pattern N-{P}-[ST]-{P} are indicated in bold and underlined; an asterisk indicates the catalytic triad residues: Ser 630, Asp 708, and His 740. (B) Ribbon diagram of the three-dimensional structure of DPPIV. The glycosylation sites are labeled, and the sugar molecules that were modeled into the electron density maps are shown in green as ball-and-stick representations. Seven of the nine glycosylation sites were found in the β-propeller domain of the molecule (blue), and two glycosylation sites were observed in the α/β-hydrolase domain (red) of the molecule. Disufide bonds are shown in yellow. Residues (L294, L340, V341, A342, and R343) that have been reported to play a role in ADA binding (Abbott et al. 1999) are labeled in red and shown in gold as ball-and-stick representations. Residues of the catalytic triad are shown in blue as ball-and-stick representations. (C) Three-dimensional structure of DPPIV with space-filled representation of β-propeller domain. Glycosylation sites N219 and N85 are labeled and are located at the entrance of the central pore of the β-propeller domain. The central pore (approximate diameter of 15 Å) creates an entrance to the active-site of the molecule.

Figure 1.

(A) Sequence alignment of human, rat, mouse, bovine, and pig DPPIV. The potential glycosylation sites predicted by using the Prosite pattern N-{P}-[ST]-{P} are indicated in bold and underlined; an asterisk indicates the catalytic triad residues: Ser 630, Asp 708, and His 740. (B) Ribbon diagram of the three-dimensional structure of DPPIV. The glycosylation sites are labeled, and the sugar molecules that were modeled into the electron density maps are shown in green as ball-and-stick representations. Seven of the nine glycosylation sites were found in the β-propeller domain of the molecule (blue), and two glycosylation sites were observed in the α/β-hydrolase domain (red) of the molecule. Disufide bonds are shown in yellow. Residues (L294, L340, V341, A342, and R343) that have been reported to play a role in ADA binding (Abbott et al. 1999) are labeled in red and shown in gold as ball-and-stick representations. Residues of the catalytic triad are shown in blue as ball-and-stick representations. (C) Three-dimensional structure of DPPIV with space-filled representation of β-propeller domain. Glycosylation sites N219 and N85 are labeled and are located at the entrance of the central pore of the β-propeller domain. The central pore (approximate diameter of 15 Å) creates an entrance to the active-site of the molecule.

Figure 1.

(A) Sequence alignment of human, rat, mouse, bovine, and pig DPPIV. The potential glycosylation sites predicted by using the Prosite pattern N-{P}-[ST]-{P} are indicated in bold and underlined; an asterisk indicates the catalytic triad residues: Ser 630, Asp 708, and His 740. (B) Ribbon diagram of the three-dimensional structure of DPPIV. The glycosylation sites are labeled, and the sugar molecules that were modeled into the electron density maps are shown in green as ball-and-stick representations. Seven of the nine glycosylation sites were found in the β-propeller domain of the molecule (blue), and two glycosylation sites were observed in the α/β-hydrolase domain (red) of the molecule. Disufide bonds are shown in yellow. Residues (L294, L340, V341, A342, and R343) that have been reported to play a role in ADA binding (Abbott et al. 1999) are labeled in red and shown in gold as ball-and-stick representations. Residues of the catalytic triad are shown in blue as ball-and-stick representations. (C) Three-dimensional structure of DPPIV with space-filled representation of β-propeller domain. Glycosylation sites N219 and N85 are labeled and are located at the entrance of the central pore of the β-propeller domain. The central pore (approximate diameter of 15 Å) creates an entrance to the active-site of the molecule.

Figure 2.

SDS-PAGE of purified wild-type DPPIV and DPPIV glycosylation mutants. (A) Mutants expressed in Sf9 insect cells. (B) Mutants expressed in Hi5 insect cells. Approximately 1000 ng purified material was loaded on a 4% to 20% gradient gel for each sample. (Lane 1) Wild-type DPPIV; (lane 2) N85A; (lane 3) N92A; (lane 4) N150A; (lane 5) N219A; (lane 6) N229A; (lane 7) N281A; (lane 8) N321A; (lane 9) N520A; (lane 10) N685A.

Figure 3.

(A) BIAcore analysis: sensorgram showing binding of ADA to DPPIV. DPPIV was immobilized on a CM5 Biacore sensor chip, and ADA was injected at various concentrations (indicated on figure) over the DPPIV surface. Specific signal for binding of ADA to DPPIV was measured by subtracting response units (RU) obtained from no-protein control. Data shown are from one of four experiments. (B) Titration of binding of wild-type DPPIV to ADA by HP-SEC. Purified samples of wild-type DPPIV and ADA were incubated for 2 h at 37°C using DPPIV/ADA ratios of 2 : 1 (dashed line), 1 : 1 (dotted line), and 1 : 2 (dashed/dotted line). ADA binding was confirmed by a shift in the retention time from 8.94 (DPPIV only, full line) to 8.51 (DPPIV/ADA complex, dotted and dashed/dotted lines). An intermediate complex with a retention time of 8.67 was observed when a DPPIV/ADA ratio of 2 : 1 was used (dashed line).

Figure 3.

(A) BIAcore analysis: sensorgram showing binding of ADA to DPPIV. DPPIV was immobilized on a CM5 Biacore sensor chip, and ADA was injected at various concentrations (indicated on figure) over the DPPIV surface. Specific signal for binding of ADA to DPPIV was measured by subtracting response units (RU) obtained from no-protein control. Data shown are from one of four experiments. (B) Titration of binding of wild-type DPPIV to ADA by HP-SEC. Purified samples of wild-type DPPIV and ADA were incubated for 2 h at 37°C using DPPIV/ADA ratios of 2 : 1 (dashed line), 1 : 1 (dotted line), and 1 : 2 (dashed/dotted line). ADA binding was confirmed by a shift in the retention time from 8.94 (DPPIV only, full line) to 8.51 (DPPIV/ADA complex, dotted and dashed/dotted lines). An intermediate complex with a retention time of 8.67 was observed when a DPPIV/ADA ratio of 2 : 1 was used (dashed line).