HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 22, 19720-19726, May 31, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/22/19720    most recent M111901200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Richard, E. Articles by Hershfield, M. S. Search for Related Content PubMed PubMed Citation Articles by Richard, E. Articles by Hershfield, M. S. Social Bookmarking                      What's this?

Clustered Charged Amino Acids of Human Adenosine Deaminase Comprise a Functional Epitope for Binding the Adenosine Deaminase Complexing Protein CD26/Dipeptidyl Peptidase IV^*

Eva Richard^§, S. Munir Alam, Francisco X. Arredondo-Vega, Dhavalkumar D. Patel^¶, and Michael S. Hershfield^**

From the Departments of  Medicine, ^¶ Immunology, and  Biochemistry, Duke University Medical Center, Durham, North Carolina 27710

Received for publication, December 13, 2001, and in revised form, January 29, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Human adenosine deaminase (ADA) occurs as a 41-kDa soluble monomer in all cells. On epithelia and lymphoid cells of humans, but not mice, ADA also occurs bound to the membrane glycoprotein CD26/dipeptidyl peptidase IV. This "ecto-ADA" has been postulated to regulate extracellular Ado levels, and also the function of CD26 as a co-stimulator of activated T cells. The CD26-binding site of human ADA has been localized by homolog scanning to the peripheral 2-helix (amino acids 126-143). Among the 5 non-conserved residues within this segment, Arg-142 in human and Gln-142 in mouse ADA largely determined the capacity for stable binding to CD26 (Richard, E., Arredondo-Vega, F. X., Santisteban, I., Kelly, S. J., Patel, D. D., and Hershfield, M. S. (2000) J. Exp. Med. 192, 1223-1235). We have now mutagenized conserved 2-helix residues in human and mouse ADA and used surface plasmon resonance to evaluate binding kinetics to immobilized rabbit CD26. In addition to Arg-142, we found that Glu-139 and Asp-143 of human ADA are also important for CD26 binding. Mutating these residues to alanine increased dissociation rates 6-11-fold and the apparent dissociation constant K_D for wild type human ADA from 17 to 112-160 nM, changing binding free energy by 1.1-1.3 kcal/mol. This cluster of 3 charged residues appears to be a "functional epitope" that accounts for about half of the difference between human and mouse ADA in free energy of binding to CD26.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Adenosine deaminase (ADA)¹ catalyzes the hydrolytic deamination of adenosine (Ado) and 2'-deoxyadenosine (dAdo). The mouse enzyme structure consists of an 8-stranded -barrel surrounded by 8 -helixes, with the deep active site pocket and an essential Zn²⁺ ion located at the carboxyl-terminal end of the -barrel (1). In humans, ADA occurs as a soluble 41-kDa monomer in all tissues. An "ecto" form of ADA also occurs in some tissues of humans and cattle, but not mice, which consists of up to two ADA monomers bound to a homodimeric "ADA complexing protein" with subunit of 110 kDa (2-5). The latter has been identified as CD26/dipeptidyl peptidase IV (DPPIV), or simply CD26, a multifunctional membrane glycoprotein expressed on epithelia of liver, gut, kidney, and exocrine glands as well as on thymocytes and activated T lymphocytes (6, 7). CD26 has 6 NH₂-terminal cytoplasmic and 22 transmembrane amino acids; the 738-residue extracellular portion has a proximal glycosylated segment, a cysteine-rich domain, and a COOH-terminal region that contains the DPPIV active site. Several mostly hydrophobic amino acids located in the cysteine-rich domain of human CD26 have been implicated in ADA binding (8, 9). No structural information is available for CD26 from any source.

The DPPIV ectopeptidase of CD26 cleaves substrates with Pro or Ala at the penultimate position. Studies of rat strains and knockout mice lacking DPPIV suggest that it functions in the hydrolysis/absorption of prolyl dipeptides and in regulating the biologic activity of some hormones and chemokines (10-12). CD26 has been shown to act as a "co-stimulator" of antigen receptor-mediated activation of thymocytes and T lymphocytes (13-15). There is conflicting evidence about whether DPPIV activity influences the co-stimulatory function of CD26 (16, 17). The possible role of "ecto-ADA" bound to CD26 in the development and maintenance of immune function has also been a subject of interest.

In humans, the highest levels of ADA occur in lymphoid cells, and inherited ADA deficiency causes profound lymphocyte depletion, resulting in the fatal disorder severe combined immunodeficiency (SCID) (18, 19). This condition has usually been attributed to toxic intracellular effects of Ado and dAdo. It has also been postulated that in the absence of ADA, elevated extracellular Ado might impair lymphocyte differentiation and function by causing aberrant signal transduction via G-protein-coupled Ado receptors (reviewed in Ref. 19). It has been suggested that ADA bound to CD26 controls levels of extracellular Ado and thus modulates Ado receptor signaling, presumed to be involved in T cell activation (7, 15, 20, 21). Alternatively, it has been suggested that ADA binding to CD26, unrelated to ADA catalytic activity, may be necessary for the co-stimulatory function of CD26 (21, 22).

Speculation that defective ADA-CD26 binding contributes to the pathogenesis of ADA-deficient SCID in humans prompted us to characterize the CD26-binding site of human ADA and to examine the effects of patient-derived ADA mutations on CD26 binding (23). For these studies, we exploited the observation that mouse ADA, although 83% identical in sequence to the human enzyme, is unable to form a stable complex with CD26 (5, 24, 25). By using a panel of recombinant human-mouse ADA hybrids, we localized the CD26-binding site of human ADA to the carboxyl-terminal end of an 18-amino acid segment, residues 126-143 (23). In mouse ADA this segment forms the peripheral 2-helix, located on the face opposite to the entrance to the active site pocket (1). Arg-142 in human ADA and Gln-142 in mouse ADA largely determined the capacity to bind CD26. ADA from a healthy adult whose only expressed ADA allele carries the R142Q mutation (26) showed markedly reduced binding to CD26 (23). These findings suggested that, as in the mouse, ADA binding to CD26 is not essential for normal immune function.

Our previous "homolog scanning" strategy focused on the 5 of 18 residues in the 2-helix at which the human and mouse ADA sequences have diverged. The methods used (gel filtration, nondenaturing PAGE, and immunoassay of cell-associated ADA) could detect only stable ADA-CD26 complexes. In the present study we have used alanine scanning mutagenesis (27) to evaluate the role in binding of conserved amino acids found in both human and mouse ADA, and we have used the more sensitive surface plasmon resonance (SPR) technique to evaluate kinetic and equilibrium binding constants.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Protein Reagents-- CD26/DPPIV was purified from rabbit kidneys, and DPPIV activity was assayed, as described (23). The two purified preparations used had DPPIV-specific activity of 7,667 or 7,883 µmol/min/mg and were essentially free of contaminating ADA activity. Purified recombinant human CD4 protein was obtained from Progenics Pharmaceuticals, Inc. (Tarrytown, NY). The 1C5 mouse mAb to purified human T lymphoblast ADA was prepared in this laboratory²; its binding specificity has been reported (23). Goat antibody to human ADA was a gift of Dr. Dan Wiginton, University of Cincinnati.

Construction and Expression of ADA Mutants-- Wild type human and mouse ADA cDNAs and several mutant ADA cDNAs involving amino acids 126-143 (Fig. 1) were available from a previous study (23). Alanine substitutions were introduced into the wild type human ADA cDNA by PCR mutagenesis essentially as described (28, 29). All final ADA cDNA PCR products were cloned into pBluescript II KS and fully sequenced using the ABI 377 PRISM DNA Sequencing Instrument (Applied Biosystems, Foster City, CA). [³⁵S]Methionine-labeled ADA was generated in vitro from ADA cDNAs in pBluescript, using the TNT Coupled Wheat Germ Extract System (Promega, Madison, WI). Translation products were analyzed by SDS-PAGE and fluorography, and they were also electrophoresed on cellulose acetate and stained for ADA activity in situ as described previously (28).

ADA cDNAs were ligated into the NcoI site of pZ plasmid (derived from pZC11 from which wild type human ADA cDNA had been excised (30)). pZ/ADA plasmids were used to transform Escherichia coli SØ3834, which has a deletion of the bacterial ADA gene (30). The conditions used for expression are as reported previously (23, 29), except that lysates of pZ/ADA cDNA-transformed E. coli SØ3834 from 100-ml cultures were dialyzed at 4 °C against 2 mM KCl, 150 mM NaCl, 1 mM KH₂PO₄, 8 mM Na₂HPO₄, pH 7.4 (PBS). ADA activity in these extracts was assayed, and ADA protein was analyzed by Western blotting using goat anti-human ADA antiserum and also the 1C5 mouse mAb to human ADA, as described previously (23).

The concentration of ADA proteins in E. coli SØ3834 lysates is based on assay of ADA activity and a specific activity of 436.1 µmol/min/mg determined for human ADA from human T lymphoblasts, which was purified to homogeneity by ion exchange and adenosine-Sepharose affinity chromatography.² Our previous studies indicate that recombinant human and mouse ADA have the same catalytic activity per molecule (23).

Analysis of ADA-CD26 Binding-- Binding of [³⁵S]Met-labeled ADA in vitro translation products to rabbit CD26/DPPIV (2 nmol/min DPPIV activity) during a 2-h incubation at 37 °C was analyzed by non-denaturing PAGE as reported previously (23). Binding of recombinant ADA in lysates of E. coli SØ3834 to the CD26-expressing, ADA-deficient, HTLV-1-transformed AlNe human T lymphoblastoid cell line (LCL) was assessed by flow cytometry with the 1C5 mAb to human ADA as reported previously (23).

SPR analysis of ADA binding to purified rabbit kidney CD26/DPPIV was performed with a BIAcore 3000 instrument (BIAcore, Uppsala, Sweden). For these studies, rabbit kidney CD26 and recombinant human CD4 glycoprotein (a nonspecific control) were coupled to the CM-dextran surface of separate flow cells of the same Research Grade CM5 Sensor Chip, using amine coupling reagents provided by the manufacturer. PBS was used as running buffer at a flow rate of 15 µl/min. Immediately prior to immobilization, rabbit CD26/DPPIV (140 µg/ml in PBS) was diluted 2-fold with 20 mM sodium acetate, pH 3.7, and CD4 (1 mg/ml) was diluted with 3 volumes of 10 mM sodium acetate, pH 4.0. Then 20 µl of the CD26 and 30 µl of CD4 solutions were injected at a flow rate of 5 µl/min to immobilize ~4000-6000 response units (RU) of these proteins, after which residual coupling sites were blocked with ethanolamine. A third flow cell of the CM5 Chip underwent activation and blocking, but without protein, and served as a blank surface for subtracting bulk, nonspecific signal.

Interaction kinetics were determined at 25 °C by injecting 45 µl of lysates of E. coli SØ3834-expressing wild type or mutant ADA proteins at a flow rate of 15 µl/min. The dissociation phase was examined for 5 min. At the end of each cycle, bound ADA was removed, and the chip was regenerated by injecting 5 µl of 10 mM glycine HCl, pH 3, at 100 µl/min. (CD26 was sensitive to the pH of the regeneration buffer. Buffers with pH >3 did not remove bound ADA, and pH <3 destabilized the CD26 surface. Under the conditions used, CD26 immobilized on the flow cell was stable for 24-28 cycles.) ADA-containing lysates were diluted 2-, 4-, and 8-fold with PBS containing BSA (1 mg/ml), and 4 concentrations of each ADA protein were examined for each binding analysis. Typically, 150-600 RU of ADA protein were bound to the immobilized CD26/DPPIV. Specific signal due to ADA binding was derived by subtracting the RU obtained with the CD4 or no-protein blank flow cells from the response obtained with the CD26 flow cell. Due to the number of ADA mutants, experiments were performed with several CM5 chips. Results were confirmed by testing several different preparations of individual recombinant ADA proteins; in all experiments wild type human ADA was also run as a reference. The kinetic parameters of the binding reactions were determined using BIAevaluation 3.1 software, using the 1:1 interaction model.

For determining the equilibrium dissociation (K_eq) constant for wild type human ADA, CD26 was immobilized at 1200 RU. ADA at concentrations of 0.35, 0.14, 0.07, 0.035, and 0.0175 µM was injected at 45 µl/min. K_eq was calculated from the slope of a Scatchard plot of the data corrected for nonspecific binding.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Mutagenesis and Expression of ADA in E. coli SØ3834-- We showed previously that replacing the entire 2-helix (residues 126-143) of human ADA with the corresponding murine segment abolished stable binding of human ADA to purified rabbit kidney CD26, as assessed by gel filtration fast protein liquid chromatography (23). No human-to-mouse replacements outside this segment diminished CD26 binding. Among the single human-to-mouse substitutions at the 5 non-conserved positions within the 2-helix, replacing Arg-142 of human ADA with mouse Gln caused the largest reduction in stable binding to rabbit CD26 (23). We have now evaluated the effects of these mutations, as well as mutations of conserved residues within the 2-helix, on CD26 binding using SPR.

The human ADA mutants studied include alanine substitutions for Gln-135, Gln-138, Glu-139, Gly-140, Glu-141, Arg-142, and Asp-143 (Fig. 1). In the mouse ADA structure, residues at all of these positions (except Gly-140) have surface-exposed side chains (1). We also substituted Lys and Glu for Arg-142 to evaluate the importance of charge. When expressed in E. coli SØ3834 under standardized conditions, all of these mutants had substantial ADA activity, ranging from 35 to 96% that obtained with wild type human cDNA (Table I). Western blots of equal amounts of ADA activity indicated that the ADA mutants all have essentially the same activity per molecule as wild type human ADA (Fig. 2). Thus, none of these mutations appears to cause a major conformational change.

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Alignment of amino acids 126-143 (2-helix) of human and mouse ADA and the mutants studied.

View larger version (46K): [in this window] [in a new window]   Fig. 2.   Western blot of recombinant human ADA mutants with mAb 1C5 to human ADA. All lanes in both A and B were loaded with the same amount of ADA activity (42 nmol/min). A, lane 1, wild type human ADA; lane 2, E128D; lane 3, A131D; lane 4, G134N; lane 5, Q135A; lane 6, Q138A; lane 7, E139A; lane 8, G140A; lane 9, E141A; lane 10, R142Q; and lane 11, D143A. B, lane 1, wild type human ADA; lane 2, R142A; lane 3, R142K; lane 4, R142E; lane 5, R142Q/D143A; lane 6, E128D/R142Q; lane 7, E128D/R142Q/D143A; lane 8, E128D/A131D/R142Q/D143A; and lane 9, H1-125/M126-143/H144-363.

In preliminary studies, we explored the ability to express and purify human ADA as a fusion protein with glutathione S-transferase and as a histidine-tagged derivative. In neither case could catalytically active wild type ADA protein be obtained in good yield. Because the SPR method has been used for studying complex mixtures (31-33), we performed our binding studies with crude lysates of ADA-expressing E. coli SØ3834 cells. Appropriate controls reported in our earlier studies (23, 29) and described below establish the validity of the results obtained.

Binding of Recombinant ADAs to Rabbit CD26-- We used a Biacore 3000 instrument to assess the kinetics of ADA/CD26 binding by SPR measurements. In these studies, we monitored simultaneously the binding of ADA in lysates of E. coli SØ3834 to immobilized rabbit CD26, human CD4 glycoprotein (nonspecific control), and to a control flow cell with no immobilized protein (Fig. 3A). None of the ADA-containing lysates studied showed significant binding to CD4 or to the no-protein control flow cell, and lysate prepared from untransformed E. coli SØ3834 did not show binding to CD26 (not shown), indicating that interaction of ADA with CD26 was specific.

View larger version (27K): [in this window] [in a new window]   Fig. 3.   BIAcore analysis of the interaction of wild type (Wt) human and mouse ADA with immobilized rabbit CD26. A, human ADA (1.3 µM) was pumped over flow cells of a single CM5 chip on which the following had been immobilized: 1, rabbit CD26 (circles and dashed line); 2, human CD4 glycoprotein (triangles); 3, no-protein control (no symbol); 4, specific signal for binding of wild type human ADA to CD26, derived by subtracting the RU obtained with the CD4 flow cell from response obtained with the CD26 flow cell. B, kinetic analysis of binding of human ADA to CD26 (concentrations of ADA are indicated). Data shown are from one of four experiments. C, Scatchard analysis of equilibrium binding data for binding of human ADA to rabbit CD26 (see "Experimental Procedures" for details). The data were obtained in a separate experiment from that shown in B. D, kinetic analysis of binding of mouse ADA to CD26 (concentrations of ADA injected are indicated). Data shown are from one of three experiments. RU, response units; C, concentration, µM; Keq, equilibrium binding constant.

A sensorgram obtained with wild type human ADA over the range 0.15-1.3 µM (Fig. 3B) demonstrates high affinity binding to rabbit CD26, characterized by a rapid association rate (k_on) of 2.2 ± 0.2 × 10⁴ M¹ s¹, and a slower dissociation rate (k_off) of 3.8 ± 0.4 × 10⁴ s¹. The apparent equilibrium dissociation constant (K_D, based on k_off/k_on) was 17 ± 3 nM (mean ± S.D. of 3 separate experiments) (Table II). Very similar rates were obtained in experiments performed with two different preparations of purified rabbit CD26 and with human ADA from several different E. coli SØ3834 lysates. The dissociation constant determined under equilibrium conditions (K_eq) was 20 nM (Fig. 3C). These values for K_eq and apparent K_D determined by SPR are similar to previous estimates of 6-11 nM determined for the binding of ¹²⁵I-labeled human and bovine ADA to glutaraldehyde-fixed membranes prepared from rabbit kidney cortex (5), and 12 nM for the binding of ¹²⁵I-labeled bovine ADA to human CD26-expressing transformed murine hybridoma cells (34).

By comparison with human ADA, wild type mouse ADA showed low affinity binding to rabbit CD26, characterized by almost 5-fold slower association and 50-fold faster dissociation rates, resulting in an apparent K_D of 5397 nM (Fig. 3D and Table II). The difference in free energy of binding (G) for wild type human and mouse ADA was 3.4 kcal/mol (Fig. 4). Replacing amino acids 126-143 of human ADA with the murine 2-helix segment (construct H1-125/M126-143/H144-363 in Tables I and II) caused a large increase in k_off (1.3 × 10² s^-1), resulting in an apparent K_D of 320 nM, almost 20-fold higher than for wild type human ADA but 10-fold lower than for mouse ADA (Table II). The G (relative to wild type human ADA) associated with this substitution was 1.8 kcal/mol (Fig. 4) or about half of the total difference in binding energy for wild type mouse and human ADAs.

View larger version (28K): [in this window] [in a new window]   Fig. 4.   Contributions of residues to net binding energy. The difference in free energy of binding to rabbit CD26 between wild type human ADA and the indicated human ADA mutants or wild type mouse ADA (G) are shown as solid bars. The G between wild type mouse ADA and the indicated mouse ADA mutants is shown as open bars. Hum (M126-143) is a derivative of human ADA in which amino acids 126-143 have been replaced with this segment from mouse ADA (referred to as H1-125/M126-143/H144-363 in Tables I and II). Mus (H126-143) is a derivative of mouse ADA in which amino acids 126-143 have been replaced with this segment from human ADA (referred to as M1-125/H126-143/M144-352 in Tables I and II). All G values are derived from G values shown in Table II. Positive G values indicate that a mutation decreased affinity for CD26, whereas negative G values indicate that affinity was increased.

The effects of point mutations within the 126-143 segment on binding to rabbit CD26 are illustrated in Fig. 5 and summarized in Table II. The naturally occurring human-to-mouse replacement R142Q had a minor effect on k_on but increased k_off by 6-fold (Fig. 5, A and E), resulting in an apparent K_D of 113 nM. The R142A, R142K, and R142E mutations also showed markedly reduced binding to CD26, due mainly to 9-16-fold increased dissociation rates, resulting in apparent K_D values of 160, 176, and 192 nM, respectively (Table II). The G values associated with these mutations were 1.3-1.4 kcal/mol (Fig. 4).

View larger version (32K): [in this window] [in a new window]   Fig. 5.   BIAcore sensorgrams showing the interactions of selected human ADA mutants with immobilized rabbit CD26. A-D, kinetic analysis of binding (performed as described in legend for Fig. 3B) with the following mutants: A, R142Q; B, D143A; C, E139A; and D, Q135A. E, comparison of representative kinetic traces for wild type human ADA (1.3 µM), wild type mouse ADA (1.5 µM), and the indicated human ADA mutants: Q135A (1.6 µM), E139A (1.1 µM), R142Q (0.8 µM), and D143A (1.1 µM). Data shown are from one of two experiments with Q135A and three experiments each for E139A, R142Q, and D143A. All analyses were performed with the same biosensor chip, except for wild type mouse ADA.

Consistent with our earlier results (23), 3 of the other 4 human-to-mouse replacements within the 126-143 segment (E128D, A131D, G134N) showed small effects (1.4-2.5-fold on K_D; G 0.5 kcal/mol) on CD26 binding (Table II and Fig. 4). However, D143A had an apparent K_D of 112 nM when measured by SPR, due mainly to an increase in k_off (Fig. 5, B and E). Thus, R142Q and D143A reduced the affinity of human ADA for rabbit CD26 by the same magnitude (about 7-fold). The reason a larger effect was not observed with D143A in our previous studies (23), which analyzed binding by fast protein liquid chromatography, is unclear. However, the effect observed with D143A using SPR was reproducible and consistent with an inability to detect stable binding of the D143A protein to rabbit CD26 by non-denaturing PAGE (data not presented). In other experiments, only those human-to-mouse constructs that altered Arg-142 and Asp-143 diminished CD26 binding significantly. Those that altered both residues (R142Q/D143A, E128D/R142Q/D143A, and E128D/A131D/R142Q/D143A), like the construct that contained the entire mouse 126-143 segment, markedly increased k_off to 1.2-1.6 × 10² s¹, resulting in apparent K_D values of 260-360 nM and Gs of 1.7-1.8 kcal/mol (Fig. 4).

Effects Due to Alanine Mutagenesis of Conserved Amino Acids-- The on-rate for the E139A mutant was similar to that of wild type human ADA, but the complex with CD26 dissociated about 11-fold faster (Fig. 5, C and E), resulting in an apparent K_D of 160 nM (Table II). The G of 1.3 kcal/mol due to E139A was similar to G values of 1.3 and 1.1 kcal/mol for R142A and D143A, respectively (Fig. 4). The other Ala substitutions (Q135A, Q138A, G140A, and E141A) had small effects on CD26 binding (1.7-2.2-fold on apparent K_D; G 0.5 kcal/mol) (Table II, Fig. 4, and Fig. 5, D and E). In other studies not shown, the ³⁵S-labeled in vitro translation products of the Q135A, Q138A, G140A, and E141A mutants showed normal binding to rabbit CD26 as assessed by nondenaturing PAGE. The E139A, R142A, and D143A, as well as the R142K and R142E translation products all showed no detectable binding by this method, as we previously found with the R142Q mutant (23).

Introducing the Q142R mouse-to-human mutation in mouse ADA slightly increased affinity for CD26 (apparent K_D values of 3460 versus 5397 nM for wild type mouse ADA). Making the double mutant Q142R/A143D or introducing the entire human 126-143 segment into mouse ADA decreased k_off by 5-fold, resulting in about a 10-fold decrease in apparent K_D to 661 and 591 nM, respectively (Table II). The G values for these substitutions, relative to wild type mouse ADA, were 1.2 and 1.3 kcal/mol (Fig. 4).

Binding of Recombinant ADAs to Human T Cell-associated CD26-- To examine binding to human CD26, we used the AlNe T LCL derived from an immunodeficient patient homozygous for a splice site mutation in ADA intron 10, which results in a highly unstable mutant ADA protein (35). AlNe cells express high levels of CD26, and in the absence of exogenous ADA they do not bind anti-ADA mAb 1C5, as determined by flow cytometry (Fig. 6, A and B) (23). mAb 1C5 recognizes an epitope near the COOH terminus of human ADA, which is absent in mouse ADA, and it reacts approximately as well with all of the ADA mutants studied as with the wild type human ADA (Fig. 2). Exogenous ADA activity bound to AlNe cells has been found to partition between monomeric 41-kDa ADA and a larger species that co-elutes with DPPIV activity (23).

View larger version (37K): [in this window] [in a new window]   Fig. 6.   Flow cytometric analysis of binding of mAb 1C5 to AlNe human T LCL induced by exogenous ADA. A and B, the AlNe cells in A had been incubated with lysate of untransformed E. coli SØ3834, and those in B with lysate of SØ3834 expressing wild type (wt) human ADA (400 nmol/min per ml of medium). Subsequent binding of mAb 1C5 is depicted by filled histograms. Also shown in B is the binding of control IgG (open histogram). C and D, binding of 1C5 mAb induced by wild type human ADA or the indicated human ADA mutants. In these experiments AlNe cells were incubated with 0.3-400 nmol/min of ADA activity per ml of medium. After washing, cell surface-associated ADA was determined by reactivity with mAb 1C5 using flow cytometry. Results are from one of two experiments.

Incubating AlNe cells with 0.3-400 nmol/min (0.01-19.5 µM) of recombinant Q135A, Q138A, G140A, and E141A human ADA mutants induced 1C5 mAb binding in a manner similar to that induced by wild type human ADA, indicating essentially normal binding to CD26 (Fig. 6C). 1C5 binding was detected at all concentrations of these proteins tested. Induction of 1C5 binding was reduced with the E139A and D143A mutants and could not be detected at lower concentrations of these proteins (Fig. 6C). At the highest concentration tested, 400 nmol/min (19.5 µM), wild type ADA induced 20-fold more 1C5 binding than did E139A and 3.5-fold more than D143A. Very similar results were found in a second experiment (not shown). The effect of D143A (Fig. 6D) was similar to that obtained previously with the R142Q mutant (23). We also compared the binding of R142A, R142K, and R142E mutants to AlNe cells. 1C5 binding was not detected with these mutants at 0.3-42 nmol/min of ADA activity. Wild type human ADA induced 65-fold more 1C5 binding than R142A, 40-fold more than R142K, and 100-fold more than R142E mutant at 400 nmol/min ADA activity (Fig. 6D).

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Protein-protein interfaces generally occupy a large area on the surfaces of the interacting partners and have large patches of hydrophobic residues (36-38). However, studies involving alanine scanning mutagenesis have indicated that a subset of residues within interfaces makes a disproportionate contribution to binding energy (39-41). "Functional epitopes" defined in this manner correspond well to conserved amino acids found by computer analysis to occur in spatially similar environments in the crystal structures of complexes involving members of protein "interface families" (42). Functional epitopes often involve charged or polar amino acids that cluster into "hot spots" near the center of the binding site, where they are protected from interaction with bulk solvent by an "O-ring" consisting of surrounding hydrophobic residues (41, 42).

The binding of human ADA to CD26 may fit this pattern. Thus, using human-mouse hybrids (homolog scanning) in our previous study (23) and alanine mutagenesis in the present work, we have shown that the ability of human ADA to form a stable complex with CD26 is due largely to 3 charged amino acids, Glu-139, Arg-142, and Asp-143. These same 3 residues are conserved in bovine ADA (23), which also binds CD26 (for reference, Fig. 7 shows the positions of Glu-139, Gln-142, and Ala-143 at the carboxyl-terminal end of the 2-helix of mouse ADA). SPR analysis of binding after mutating residues 139, 142, and 143 of human ADA to alanine shows that their side chains each contribute 1.1-1.3 kcal/mol to binding of CD26. The E139A and R142A mutations had equivalent effects on binding to both rabbit CD26 and to human T LCL-associated CD26; the D143A mutation had a somewhat weaker effect.

View larger version (62K): [in this window] [in a new window]   Fig. 7.   Location of the Glu-139, Gln-142, and Ala-143 residues in the mouse ADA crystal structure. The RasMol model uses coordinates reported by Wilson et al. (1). Residues Glu-139, Gln-142, and Ala-143 are identified; the rest of the 2-helix (residues 126-138 and 140, 141) is shown as a gold space-filling structure. For orientation, the carboxyl-terminal helix (residues 337-351) is shown in light blue. The bound inhibitor (green) and zinc ion (yellow) are shown as space-filling structures at the active site (based on Fig. 3, see Ref. 23).

The naturally occurring R142Q mutation, as well as the site-directed Ala, Glu, and Lys mutations of Arg-142 all reduced the affinity of human ADA to CD26, associated with 7-11-fold increased apparent K_D values and with G values of 1.1-1.4 kcal/mol. Thus, despite similar conformational restrictions and configurational entropy of the side chains of Glu, Gln, Lys, and Arg (43), these results indicate that the positively charged guanidinium group of Arg-142 is critical for ADA binding to CD26. The magnitude of the G values for the mutants studied suggests that Arg-142 may be involved in a critical hydrogen bond (44).

The number of human ADA residues that contact CD26 (the "structural epitope") is probably much larger than the functional epitope we have defined. Nevertheless, the sum of the G values for the E139A, R142A, and D143A mutants, 3.7 kcal/mol, is essentially identical to the 3.4 kcal/mol difference binding energy between human and mouse ADA. Replacing only the 18-residue 2-helix of human ADA with that of mouse ADA, which changes 5 residues but does not alter Glu-139, was associated with a G of 1.8 kcal/mol, the same as for the R142Q/D143A double mutant of human ADA. Thus, almost half of the difference in binding energy between human and mouse ADA may be due to other differences in their structures. Some residues of mouse ADA outside the 2-helix may have an inhibitory effect on the interaction with rabbit CD26. Consistent with this, introducing the human 2-helix into mouse ADA or making the Q142R/A143D mouse-to-human double mutant changed the binding free energy of mouse ADA only by about 1.3 kcal/mol.

Four residues located within an extracellular cysteine-rich domain of human CD26 (Leu-294, Leu-340, Val-341, Ala-342, and Arg-343) have been shown to be required for the binding of cell surface-associated human CD26 to bovine ADA (8, 9). These results, and the observation that CD26 dissociated from an ADA-Sepharose affinity column at very low ionic strength (45), led to the prediction (9) that nonconserved hydrophobic residues of human ADA might be involved in binding to CD26 (possibly Leu-346, Ala-350, and Gly-352 near the COOH terminus, which are, respectively, Arg, Glu, and Gln in mouse ADA). However, replacing all residues of human ADA beyond 247 with the COOH terminus of mouse ADA did not alter binding to CD26 (23).

It is possible that Glu-139, Arg-142, and Asp-143 of human ADA interact with charged residues of CD26. However, mutations that disrupt electrostatic interactions between proteins generally alter k_on but not k_off (46), whereas the Ala mutations of these residues in ADA all had much larger effects on k_off. Also, in studies not presented we observed <2-fold effects on the association rate when binding of wild type human ADA to CD26 was studied at 20-300 mM NaCl. Increasing ionic strength would be expected to cause a large decrease in association rate if binding were driven by electrostatic attraction (47, 48).

In summary, Glu-139, Arg-142, and Asp-143 clustered at the end of a peripheral helical segment of human ADA comprise a functional epitope for binding CD26. Rather than an electrostatic interaction, these charged amino acids may interact with a largely hydrophobic region of CD26, as has been observed in the complex of lysozyme with anti-lysozyme mAb HyHEL-63 (49). A full definition of the ADA-CD26-binding site must await determination of the structures of these proteins in their free and liganded states. However, it is conceivable that the largely nonpolar residues of human CD26 shown to be necessary for binding ADA may be part of a hydrophobic O-ring that serves to shield the polar binding residues of ADA from solvent (41).

	ACKNOWLEDGEMENTS

We thank Aron Boney and Casey Paleos for their excellent technical assistance.

	FOOTNOTES

^* This work was supported in part by Grants RO1 DK20902 (to M. S. H.) and R01 AI47604 (to D. D. P.) from the National Institutes of Health, by a grant from Enzon, Inc. (to M. S. H.), and by a Scientist Development grant from the American Heart Association (to S. M. A.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ Supported by Fellowship 98/9329 from Fondo de Investigación Sanitaria, Instituto de Salud Carlos III, Ministerio de Sanidad y Consumo, Spain.

^** To whom correspondence and reprint requests should be addressed: Box 3049, Duke University Medical Center, Durham, NC 27710. Tel.: 919-684-4184; Fax: 919-684-4168; E-mail: msh@biochem.duke.edu.

Published, JBC Papers in Press, March 18, 2002, DOI 10.1074/jbc.M111901200

² M. S. Hershfield, unpublished data.

	ABBREVIATIONS

The abbreviations used are: ADA, adenosine deaminase; DPPIV, dipeptidyl peptidase IV; SCID, severe combined immunodeficiency; LCL, lymphoblastoid cell line; PCR, polymerase chain reaction; SPR, surface plasmon resonance; RU, response units; mAb, monoclonal antibody.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				W. A. Weihofen, J. Liu, W. Reutter, W. Saenger, and H. Fan Crystal Structure of CD26/Dipeptidyl-peptidase IV in Complex with Adenosine Deaminase Reveals a Highly Amphiphilic Interface J. Biol. Chem., October 8, 2004; 279(41): 43330 - 43335. [Abstract] [Full Text] [PDF]

				M. Gonzalez-Gronow, M. S. Hershfield, F. X. Arredondo-Vega, and S. V. Pizzo Cell Surface Adenosine Deaminase Binds and Stimulates Plasminogen Activation on 1-LN Human Prostate Cancer Cells J. Biol. Chem., May 14, 2004; 279(20): 20993 - 20998. [Abstract] [Full Text] [PDF]

				K. Aertgeerts, S. Ye, L. Shi, S. G. Prasad, D. Witmer, E. Chi, B.-C. Sang, R. A. Wijnands, D. R. Webb, and R. V. Swanson N-linked glycosylation of dipeptidyl peptidase IV (CD26): Effects on enzyme activity, homodimer formation, and adenosine deaminase binding Protein Sci., January 1, 2004; 13(1): 145 - 154. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) All Versions of this Article: 277/22/19720    most recent M111901200v1 Alert me when this article is cited