Cell surface adenosine deaminase binds and stimulates plasminogen activation on 1-LN human prostate cancer cells.

Adenosine deaminase (ADA) is expressed intracellularly by all cells, but in some tissues, it is also associated with the cell surface multifunctional glycoprotein CD26/dipeptidyl peptidase IV. By modulating extracellular adenosine, this "ecto-ADA" may regulate adenosine receptor signaling implicated in various cellular functions. CD26 is expressed on the surface of human prostate cancer 1-LN cells acting as a receptor for plasminogen (Pg). Since ADA and Pg bind to CD26 at distinct but nearby sites, we investigated a possible interaction between these two proteins on the surface of 1-LN cells. Human ADA binds to CD26 on the surface 1-LN cells and immobilized CD26 isolated from the same cells with similar affinity. In both cases, ADA binding is diminished by mutation of ADA residues known to interact with CD26. ADA was also found to bind Pg 2 in the absence of CD26 via the Pg kringle 4 (K4) domain. In the presence of 1-LN cells or immobilized CD26, exogenous ADA enhances conversion of Pg 2 to plasmin by 1-LN endogenous urinary plasminogen activator (u-PA), as well as by added tissue Pg Activator (t-PA), suggesting that ADA and Pg bind simultaneously to CD26 in a ternary complex that stimulates the Pg activation by its physiologic activators. Consistent with this, in melanoma A375 cells that bind Pg, but do not express CD26, the rate of Pg activation was not affected by ADA. Thus, ADA may be a factor regulating events in prostate cancer cells that occur when Pg binds to the cell surface and is activated.

Adenosine deaminase (ADA), 1 which converts adenosine and 2Ј-deoxyadenosine to inosine and 2Ј-deoxyinosine, has a wide phylogenetic distribution from bacteria to humans (1). Humans with inherited ADA deficiency have a profound lymphopenia causing severe combined immunodeficiency disease (2). Although mainly cytosolic, an "ecto" form of ADA also exists on activated human T lymphocytes and epithelial cells of many organs because of the association of ADA with a multifunctional cell membrane glycoprotein, CD26/dipeptidylpeptidase IV (3)(4)(5)(6)(7). A high affinity between ADA and CD26 also prevails in some other mammalian species such as rabbits and cattle, but not in others, such as mice and rats (8,9).
It has been suggested that CD26-associated ecto-ADA regulates the level of extracellular adenosine and thereby controls signal transduction via a family of adenosine receptors (10,11). CD26 is a co-stimulator of T lymphocyte activation (12), and in vitro studies suggest that ADA binding modulates this costimulatory function (10,11,13). It has been proposed that a lack of "ecto-ADA" contributes to severe combined immunodeficiency in patients with inherited ADA deficiency (10,11,13). Human ADA binds to CD26 via a carboxyl-terminal amino acid segment of the peripheral ␣2 helix, comprising residues 128 -143 (14,15). Binding of recombinant human ADA to CD26 was greatly reduced by mutating Arg142 to Gln, the amino acid occupying this position in mouse ADA, which does not bind CD26 (14). ADA from a healthy adult who expressed only the enzyme with a R142Q mutation (16) showed markedly reduced binding to CD26, suggesting that, as in mice, binding of ADA to CD26 is not essential for immune function in humans (14).
CD26 expressed at high levels on 1-LN human prostate tumor cells has been shown to act as a receptor for plasminogen type 2 (Pg 2) (17). In addition to promoting Pg 2 activation by urinary-type Pg activator (u-PA), CD26-associated Pg 2 also initiates a signal transduction cascade, which regulates expression of matrix metalloproteinase-9 (MMP-9) by these cells (17). Pg 2 binds via its ␣2,3-linked sialic acid residues of the Thr 345 O-linked carbohydrate chain to a region of CD26 located between amino acids 313 and 319 (18). This epitope responsible for binding to Pg 2 is conserved in human and mouse CD26 (19,20). ADA binding occurs at a region between amino acids 340 and 345 of human CD26 (21). The Pg and ADA binding regions are located in close proximity in the crystal structure of CD26 (22).
In this study, we investigated binding of ADA to 1-LN cells and to isolated, immobilized CD26 purified from 1-LN cell membranes, in the absence or presence of Pg 2. ADA binding to human melanoma cells, which do not express CD26 (23), was also examined. The specificity of binding to CD26 was determined in experiments with recombinant wild type human ADA and a mutant ADA with greatly reduced affinity for CD26. Our findings indicate that activation of Pg 2 by its physiologic activators u-PA and tissue-type Pg activator (t-PA) are stimulated by ADA when both Pg 2 and ADA are bound to either 1-LN cells or immobilized CD26. However, the rate of Pg 2 activation was not enhanced when ADA bound to melanoma cells, which lack the CD26 receptor. ADA activity is not affected by Pg 2. These results are the first evidence demonstrat-ing a connection between the Pg activation system and the protein complex CD26-ADA.

EXPERIMENTAL PROCEDURES
Materials-Culture media were purchased from Invitrogen. 125 I-labeled Bolton-Hunter reagent for protein iodination was obtained from PerkinElmer Life Sciences. All other reagents were of the highest grade available.
Proteins-ADA was purified from human monocytic leukemia U937 cells grown in 10 liters of RPMI 1640 culture medium containing 10% fetal bovine serum. The protein was purified to homogeneity using a combination of ion-exchange and affinity chromatographies as described by Aran et al. (24).
Human Pg was purified and separated into its two classes of isoforms, Pg 1 and Pg 2, as described previously (25,26). CD 26 from 1-LN cell membranes was purified as described previously (17). Pg kringles 1-3 (K1-3) and kringle 4 (K4) were prepared by limited proteolysis of Pg with porcine pancreatic elastase and purified as described by Sottrup-Jensen et al. (27). Pg kringle 5 (K5) was obtained by digestion of mini-Pg with pepsin and purified as described by Cao et al. (28). Iodination of ADA was carried out using 125 I-labeled Bolton-Hunter reagent according to the manufacturer's protocol. The specific activity varied from 500 to 700 cpm/ng.
Construction and Expression of ADA Mutants-Wild type human and mouse ADA cDNAs and the hybrid human ADA cDNA in which the amino acid sequence 126 -143 was substituted by the corresponding segment from mouse cDNA (H1-125/M126 -143/H144 -363) were constructed and expressed as described previously (14,29). To prepare ADA for CD26 binding studies, the cells from 100-ml overnight cultures were sonicated in 3 ml of lysis buffer (29). ADA enzymatic activity and concentration in these lysates was determined as described previously (14,29).
Antibodies-Antibodies to CD26 were raised in rabbits according to standard protocols, using purified CD26 (17) as the antigen, by Research Genetics (Huntsville, AL). The IgG fraction specific against CD26 was purified on a column containing CD26 covalently attached to Sepharose 4B. The mouse monoclonal antibody 1C5 was prepared against ADA purified from human T cell leukemia cells as described previously (14).
Cell Cultures-The human prostate tumor cell line 1-LN, a generous gift of Dr. Philip Walther of the Department of Surgery, Duke University Medical Center, Durham, NC, was grown in RPMI 1640 medium supplemented with 10% (v/v) fetal bovine serum, 100 units/ml penicillin G, and 100 g/ml streptomycin. The human melanoma A375 cell line was obtained from the American Type Culture Collection (Manassas, VA) and cultured in Dulbecco's Modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum, 2 mM L-glutamine, 10 g/ml penicillin G, and 100 g/ml streptomycin.
Ligand Binding Analysis-Cells were grown in tissue culture plates until the monolayers were confluent. The cells were washed in Hanks' balanced salt solution. All binding assays were performed at 4°C in RPM1 1640 containing 2% bovine serum albumin (BSA). Increasing concentrations of 125 I-labeled ADA were incubated with cells for 60 min in 96-well strip plates. Free ligand was separated from bound ligand by aspirating the incubation mixture and washing the cell monolayers rapidly three times with RPMI 1640 containing 2% BSA. Wells were stripped from the plates, and radioactivity was determined in an Amersham Biosciences LKB Biotechnology® 1272-␥ counter. Molecules of ligand bound were calculated after subtraction of nonspecific binding measured in the presence of 100 M unlabeled ADA. Estimates of dissociation constants (K d ) and maximal binding of ADA (B max ) were determined by fitting data directly to a nonlinear regression program using the statistical program SYStat® for Windows 98.
Solid-phase Radioligand Binding Studies-To study specific binding of ADA to immobilized CD26 purified from 1-LN cells, 96-well strip plates were coated with CD26 (1 g/ml in 0.1 M sodium carbonate, pH 9.6; 200 l/well; 37°C; 2 h). After coating, plates were washed with 200 l of 10 mM sodium phosphate/100 mM NaCl, pH 7.4, containing 0.05% Tween 80 (PBS-Tween) to remove unbound protein. Nonspecific sites were blocked by incubation with PBS-Tween containing 2% (w/v) BSA at room temperature for 1 h. Plates were rinsed twice with 200 l of PBS-Tween, air-dried, and stored at 4°C. For assays, increasing concentrations of 125 I-labeled ADA, with or without a 50-fold excess of unlabeled ligand, were added to triplicate wells and incubated at 22°C for 1 h. Following incubation, the supernatants were removed, and the plates were rinsed three times with 200 l of PBS-Tween. Wells were stripped from the plates, and radioactivity was measured in an Amer-sham Biosciences LKB Biotechnology® 1272-␥ counter. Specific binding was calculated by subtraction of nonspecific binding measured in the presence of unlabeled ligand. A similar technique was used to immobilize Pg on 96-well culture plates. Conditions for binding of 125 I-labeled ADA to immobilized Pg and measurements of radioactivity were similar to those described above.
Cellular Enzyme-linked Immunoabsorbent Assay (CELISA) to Measure Binding of Cloned Human ADA to Cell Surface CD26 -Binding of wild type and hybrid human ADA (H1-125/M126 -143/H144 -363) to 1-LN or A375 cells was evaluated with a CELISA assay (30). Cells were grown in 96-well plates until confluent. Cell monolayers were incubated in RPMI 1640 with increasing concentrations of wild type and hybrid human ADAs for 90 min at 4°C. To avoid nonspecific binding of primary antibody, the cell monolayers were then blocked with 2% BSA. Excess ADA and BSA were removed by rinsing the plates three times with 0.2 ml of methanol at Ϫ20°C for 20 min. At this time, cells were rinsed three times in PBS and then incubated with 0.2 ml of PBS containing 100 ng of anti-human ADA monoclonal antibody 1C5 at 22°C for 90 min or with a nonspecific mouse antibody/monoclonal antibody. Then, cells were rinsed three times with PBS and incubated with a goat alkaline phosphatase-conjugated anti-mouse IgG (0.2 ml of PBS containing 50 ng of secondary antibody) at 22°C for 90 min. Bound IgG was monitored by hydrolysis of the alkaline phosphatase substrate p-nitrophenyl phosphate at a wavelength of ⌬A 405 nm using an Anthos Labtec® kinetic plate reader. Bound ADA was expressed as ⌬A 405 nm /min.
Plasmin Generation on the Surface of 1-LN Cells or Immobilized CD26 -Plasmin generation by endogenous u-PA was measured on confluent 1-LN cell monolayers grown in 96-well plates. Cells were incu- Gel Electrophoresis-Electrophoresis was performed on polyacrylamide gels (1.2 mm thick; 14 cm ϫ 10 cm) containing 0.1% SDS. A discontinuous Laemmli buffer system was used (31). Visualization of the proteins was carried out by staining the gel with 0.25% Coomassie Brilliant Blue R-250 in 45% (v/v) methanol/10% (v/v) acetic acid. The dye-conjugated M r markers (Bio-Rad) used were: BSA (M r 86,000), carbonic anhydrase (M r 44,000), and soybean trypsin inhibitor (M r 33,000).

RESULTS
Binding of ADA to 1-LN Cells and Immobilized CD26 Isolated from 1-LN Cell Membranes-Analyses of the binding isotherm resulting from incubation of 1-LN cells with increasing concentrations of 125 I-labeled human ADA (Fig. 1A), purified from U937 cells (Fig. 1A, inset), suggest that ADA binds to these cells in a dose-dependent manner with high affinity (K d of 20 Ϯ 3.7 nM) and to a large number of sites (B max of 28 Ϯ 4.3 ϫ 10 5 sites/cell). Analyses of the binding isotherm of ADA to CD26 immobilized on cell culture plates (Fig. 1B) indicates a single class of binding sites (a K d of 18 Ϯ 5.4 nM and a B max of 0.8 nmol of bound ADA/nmol CD26). This K d value is within the same order of magnitude to previous estimates of K d determined for the binding of recombinant human ADA to purified rabbit CD26 (15).
Binding of Recombinant ADAs to Purified CD26 and to CD26-expressing versus CD26-nonexpressing Cells-We examined the capacity of recombinant ADA constructs expressed in lysates of Escherichia coli Sဧ3834 to compete for binding of purified, 125 I-labeled human ADA to human CD26 immobilized on culture plates. These experiments (Fig. 2) show that only wild type recombinant human ADA competed significantly for CD26 binding. Lysates expressing mouse ADA or the human ADA mutant H1-125/M126 -143/H144 -363, in which the mouse ␣2 helix is substituted for the human segment, showed no more evidence of binding to human CD26 than did a lysate of E. coli Sဧ3834 transformed with the carrier plasmid pZ alone, used as a negative control. These data confirm that the amino acid segment 126 -143 of human ADA comprises the functional epitope for binding to CD26.
The binding of recombinant ADAs to 1-LN cells was compared with their binding to human melanoma A375 cells, which do not express CD26, using a CELISA assay as described under "Experimental Procedures." As was found with purified CD26, wild type recombinant human ADA, but neither mouse ADA nor the H1-125/M126 -143/H144 -363 mutant, showed binding to 1-LN cells (Fig. 3A), indicating that ADA binding to 1-LN cells was exclusively via cell surface CD26. Consistent with this conclusion, and with previous studies of 125 I-labeled ADA binding to several other melanoma cell lines (23), neither wild type nor mutant recombinant human ADA constructs showed any specific binding to A375 cells (Fig. 3B).
Binding of ADA to Immobilized Human Pg 2-A close examination of the human and murine CD26 segment containing amino acids 310 -350 (19, 20) reveals a conserved Pg2 binding region (Fig. 4) (18). Confirming our previous finding (18), Fig. 5 shows that 125 I-labeled ADA binds to Pg 2 immobilized on cell culture plates in a dose-dependent, saturable manner. 125 Ilabeled ADA also binds to Pg 2 electroblotted to a nitrocellulose membrane after electrophoresis (Fig. 5, inset, lane 2). Analysis of the binding isotherm demonstrates that ADA binds to Pg 2 with high affinity (K d of ϳ0.28 M and B max of 0.93 nmol of bound ADA per nmol Pg).
Since both Pg 2 and ADA bind to CD26, and ADA binds directly to Pg 2, we investigated the capacity of Pg 2 to influence the binding of ADA to CD26 (Fig. 6). Increasing concentrations of Pg enhanced the binding of ADA to the complex CD26-Pg 2 (Fig. 6A). Furthermore, the binding of 125 I-labeled ADA to immobilized Pg 2 was not affected by either K1-3 or K5 domains of Pg, whereas the K4 domain was an effective inhibitor (Fig. 6B). These results suggest that the interaction of ADA and Pg 2 is mediated by K4.
Effect of the Interaction between Pg 2 and ADA on Pg Activation or ADA Enzymatic Activity-Pg 2 binds to CD26 on 1-LN cells, where it is rapidly activated by u-PA, the primary Pg activator at the surface of these cells (17). We studied the effect of increasing concentrations of ADA on Pg 2 activation by 1-LN cell monolayers at a single Pg 2 concentration (0.2 M) (Fig. 7A) and also on Pg 2 activation by t-PA on plates containing immobilized CD26 isolated from 1-LN cell membranes (Fig. 7B). In both cases, ADA stimulated Pg 2 activation in a dose-dependent manner, when either u-PA or t-PA was the activator. The enzymatic activity of ADA (0.1 M) was not affected by increasing concentrations of Pg 2 (values of ⌬A 265 nm /min of 0.15 and 0.14 in the absence or presence of 1 M Pg 2, respectively). These experiments suggest that ADA facilitates Pg 2 activation, possibly by inducing a conformational change in Pg 2.
We next assessed Pg 2 activation by t-PA in the presence of ADA and poly-D-lysine, which mimics a fibrin surface, a physiological substrate for Pg activation by t-PA. In this system, we studied the effect of Pg K1-3, K4, and the anti-fibrinolytic amino acid 6-AHA, which inhibits binding of Pg or t-PA to fibrin. The results (Table I) demonstrate that poly-D-lysine enhances the effects of ADA on Pg activation by t-PA. Furthermore, these experiments demonstrate that K4 effectively suppresses the ADA effect, suggesting again that K4 is the region in Pg responsible for the interaction with ADA. 6-AHA was an effective inhibitor of ADA-induced Pg activation. This raises the possibility that Pg and its activators must be closely localized by substrate molecules, such as fibrin or CD26, in order for ADA to enhance the rate of Pg activation. To test this hypothesis, we examined the effect of ADA on Pg activation by A375 cells, which do not express CD26 on their surface (Table II). In contrast to results with 1-LN cells, Pg activation was not substantially changed in the presence of two concentrations of ADA. Taken together, these results suggest that ADA-facili-tated Pg activation observed with 1-LN cells was mediated by a specific interaction of ADA, Pg, and CD26 on the cell surface. DISCUSSION In addition to suggestions that ADA binding may influence the function of CD26 in T lymphocytes, in vitro studies have suggested other possible roles for ADA, which may be independent of its enzymatic activity (1,32). For example, in cerebellar neurons, exogenous ADA increased the influx of 45 Ca 2ϩ and hydrolysis of polyphosphoinositides (33). Since the binding regions for Pg and ADA in CD26 are in close proximity (18,21,22), we investigated the possible influence of the interaction of these proteins on Pg 2 activation. Our findings suggest a novel role for the complex CD26-ADA, which may not require the catalytic activity of ADA.
We have shown that ADA binds to CD26 on 1-LN cells and that ADA binds to Pg 2, either alone or when Pg 2 is bound to immobilized CD26. The enzymatic activity of ADA was not affected by binding to Pg 2. ADA stimulated the activation of Pg by u-PA on the surface of 1-LN cells or by t-PA when Pg 2 was bound to immobilized CD26. ADA also stimulated Pg 2 activation by t-PA in the presence of poly-D-lysine, which mimics a fibrin surface, thereby suggesting that extracellular ADA may act as a pro-fibrinolytic factor. The interaction between ADA and Pg is mediated by K4. ADA appears to mimic the effect of tetranectin, a specific K4-binding protein occurring in plasma at ϳ10 mg/liter, which enhances Pg activation by t-PA in the presence of poly-D-lysine (34). This effect of tetranectin on Pg activation is inhibited by 6-AHA, suggesting a mechanism mediated by the L-lysine binding site on K4 (34). All these effects are similar to those described above for the interaction between ADA and Pg 2.
Our data suggest a mechanism that permits acceleration of the rate of Pg activation when both Pg and ADA are bound to CD26 in prostate carcinoma 1-LN cells, which is not available in A375 melanoma cells, which do not express CD26. However, Pg is found colocalized with tetranectin at the invasive front of malignant melanoma lesions, suggesting a coordinated role for tetranectin and Pg, mediated by the L-lysine binding site on K4 (35); this Pg 2 activation may facilitate the migration of melanoma tumor cells (35). Therefore, both the evolution of melanoma and prostate cancer may depend on the interaction of ligands specific for Pg K4 to regulate the rate of Pg activation at their invasive fronts.
In this context, the levels of CD26 and ADA have been evaluated in prostate cancer. The specific activity of ADA in cancerous human prostate has been reported to be similar or slightly less than in normal prostatic tissue (36,37). However, ADA levels in benign prostatic hypertrophy (BPH) tissue were significantly higher than in normal prostate (36). CD26 levels are significantly elevated in prostate cancers relative to BPH tissues, but CD26 levels were similar to those of the tumor in BPH tissue adjacent to the tumor (38). These studies, along with our observations, suggest that one of the local factors influencing growth of epithelial cells in prostate cancer may be high levels of the protease plasmin generated by BPH tissue in  the immediate growth environment of the tumor, resulting from high levels of CD26 and ADA, which facilitate Pg localization and activation on the surface of these cells.