A CD26-Controlled Cell Surface Cascade for Regulation of T Cell Motility and Chemokine Signals¹

The chemokines are a superfamily of small peptide molecules (8–20 kDa) produced by various cell types including leukocytes, epithelial cells, perivascular myofibroblasts (1), and tumor cells (2). Chemokines regulate the migration of inflammatory cells, stem cells, and cancer cells (3, 4, 5) to specific destinations within the organism (6) and determine the extravasation as well as the migration within target organs (7). In addition, chemokines induce release of proteinases from leukocytes, control angiogenesis (8), and regulate T cell differentiation and function as scavenger molecules (9). The chemokine superfamily includes >50 ligands and 20 receptors and is currently classified into 4 groups based on the amino acid motif that occurs within the first 2 N-terminal cysteines (10, 11). Chemokines bind to G-protein-coupled surface receptors on target cells (12), and almost all chemokine receptors recognize more than one chemokine (12, 13). Chemokines have been shown to be important orchestrators of the capacity of lymphocytes to extravasate and migrate that makes it possible for these cells to reposition themselves between a free-floating vascular state and active migration in tissues (14, 15). Although chemokines thus play a critical role for lymphocyte adhesion and migration, and hence for immune surveillance against pathogens, altered self-components, and cancer throughout the organism, the mechanisms via which chemokines regulate adhesion and migration are only partly known (16, 17, 18, 19, 20). Understanding of this regulation therefore requires deeper mechanistic insights and perhaps even identification of additional regulatory factors and pathways. This lack of understanding applies particularly to the formation of active cell edges formed by adhering and migrating cells that probably senses the environment and is critical for recognition, a key feature of the adaptive immune system.

Thrombospondin-1 (TSP-1)³ is a 420-kDa glycoprotein composed of three identical disulfide-linked polypeptide chains that display binding sites for various cell surface receptors including glycosaminoglycans; the integrins α_vβ₃, α₃β₁, α₄β₁, α₅β₁ and α₆β₁; calreticulin (CRT); CD91; CD36; and CD47 (21, 22, 23, 24, 25, 26, 27, 28, 29). T lymphocytes express endogenous TSP-1 associated with CD91 and CRT that seem to function as a cell surface ligand for regulation of T cell adhesion and migration (30, 31, 32). In vivo findings further suggest that TSP-1 has a role in immune regulation. Thus, TSP-1-deficient mice show inflammatory disturbances in lungs and other organs and TSP-1 or TSP-2 as well as CD47-deficient mice show defective delayed-type hypersensitivity with prolonged inflammation (33, 34, 35). Deficiency of TSP-1 also reduces Th17 differentiation and attenuates experimental autoimmune encephalomyelitis (36). The starting point for the present investigation was a working hypothesis, based on preliminary evidence that chemokines may influence T cell TSP-1, that the chemokine receptor system may cross-talk with the cell surface receptors that bind TSP-1. We further hypothesized that such a cross-talk might be important for regulation of T cell functions such as environmental recognition by active cell edges, adhesion, and migration, and perhaps contribute to the extensive lymphocyte infiltration seen in autoimmune and allergic inflammatory diseases. It is interesting in this context that the chemokine CXCL12, which is expressed in many tissues (37), binds to the chemokine receptor CXCR4, which is associated with the membrane-bound ectopeptidase dipeptidyl peptidase IV/CD26 (CD26), a cell surface protease with a widespread organ distribution, composed of two identical 110-kDa subunits, that inactivates chemokines and other peptides by selective removal of the N-terminal proline or alanine in the second position (38, 39, 40). CD26 affects T cell activation, proliferation, and cytokine production (41, 42, 43, 44), and CD26 inhibitors introduced into animal models have demonstrated a role for CD26 in the development of autoimmune diseases (45). However, the severity of autoimmune diseases including rheumatoid arthritis and experimental autoimmune encephalomyelitis is increased in CD26-deficient mice indicating that CD26 rather exerts a protective effect against disease development (46, 47). Our results show that CXCL12 and CCL5, which both are processed by CD26, are potent inducers of cell surface expression of TSP-1 and that TSP-1 induces expression of CD91 and CRT and stimulates cytoplasmic spreading and migration. Small interfering RNA (siRNA) silencing of CD26, inhibitors of CD26 as well as Ab-induced modulation of CD26 also stimulate this chemokine-induced molecular cascade, thus implicating CD26 as a suppressor factor for T cell formation of cytoplasmic projections and migration.

Human plasma fibronectin and rat tendon collagen type I were purified and prepared as described elsewhere (48, 49). Poly-l-lysine (molecular mass, 5300) was purchased from Miles-Yeda. IL-2 and IL-4 were from Genzyme Diagnostics. Brefeldin A was from Sigma-Aldrich, and ICAM-1 was from R&D Systems. Human recombinant CD26 was obtained from Alexis Biochemicals. Anti-fibronectin (clone IST1, IgG1) was obtained from Sera-Lab. Anti-CD3 (clone SK7, IgG1) and anti-CD4 (clone SK3, IgG1) were obtained from BD Biosciences. Dynabeads were from Dynal. Mouse IgG and anti-CD8 (C8/144B, IgG1) were from Dako. Goat anti-mouse Ab was from Dakopatts. Anti-TSP-1 clone MBC 200.1 (also called TSP-Ab-9, IgG1), clone A6.1 (also called TSP-Ab-4, IgG1) and clone C 6.7 (also called TSP-Ab-3, IgG1), were from NEO-MARKERS. Anti-CD91 (clone A2MRα2, IgG1) was from Santa Cruz Biotechnology. Anti-CD26 (clone TA5.9) has been described previously (50). Anti-CD26 (clone BA5) was obtained from Dako. Anti-CRT (clone FMC75, IgG1) was from Biosite. Anti-CD47 (clone B6H12.2, IgG1) was from NEO-MARKERS. Biotinylated peroxidase and avidin were from Vector Laboratories. The peptides KRFYVVMWKK (4N1K) and KVFRWKYVMK (scrambled (Sc) 4N1K) were synthesized with Tri pep (Novum Research Park). Protease inhibitors used were GM6001, a metalloprotease inhibitor (Chemicon), Ile-Glu-Thr-Asp aldehyde, an inhibitor of granzyme B, Z-ValAlaAsp-fluoromethyl ketone, a caspase inhibitor (Calbiochem), and diprotin A and KR 62436, two inhibitors of CD26 (Sigma-Aldrich). The CD26 inhibitor vildagliptin was custom synthesized by GLSynthesis. Serum-free AIM-V medium was obtained from Invitrogen, and 10× RPMI 1640 was from Life Technologies.

Human peripheral T cells were purified as previously described and stimulated with anti-CD3 in the presence of IL-2 and IL-4 for 3–5 days before the experiments (51). The birch (Bet v 1)-specific T cell clone AF 24 was kindly provided by Dr. J. van Neerven (ALK). This cell line was regularly stimulated with anti-CD3 and cultured in the presence of IL-2 and IL-4 for 5–12 days before the experiments. All cells were cultured in RPMI 1640 supplemented with 2 mM l-glutamine, 0.16% sodium bicarbonate, 10,000 U/ml benzylpenicillin, 100,000 μg/ml streptomycin, and 10% FCS or in serum-free AIM-V medium. Peptides, brefeldin A, and mAbs when applied to study the influence on Ag expression or adhesion were added immediately before the experiments. To induce down-modulation of CD26, the lymphocytes were incubated with anti-CD26 (TA5.9) for 4 h (50).

The expression of CD26 was suppressed using the human T cell Nucleofector kit (Lonza) and a Nucleofector device (Amaxa Biosystems) as previously described (52). CD26-siRNA (human) and control-siRNA were obtained from Santa Cruz Biotechnology.

Anti-CD3-activated cells were biotinylated in solution, 3 × 10⁶/well, and after adhesion to tissue culture dishes (Falcon 35 3001), 5 × 10⁶/well, coated overnight at 4°C with fibronectin (20 μg/ml) using reagents for biotinylation from the Cellular Labeling and Immunoprecipitation Kit (Roche). The reaction was stopped and cells were washed and lysed with 1 ml of lysis buffer on ice for 30 min, followed by centrifugation at 12,000 × g for 10 min at 4°C. The cells were allowed to adhere for 30 min. Where stated, cells were treated with 4N1K at 50 μΜ or scrambled 4Ν1Κ (Sc4N1K) at 50 μΜ or inhibitor of CD26, diprotin A, at 50 μM for 30 min before biotinylation. Adherent cells were washed and biotinylated while remaining attached (0.5 ml of biotinylation buffer/well). Adherent cells were scraped off the plate, washed, lysed, and centrifuged as above. For preclearing, each tube with lysate was incubated with protein G agarose at 4°C overnight. Agarose beads were discarded and lysates mixed with 1 μg of TSP-1 Ab or CD91 Ab and incubated at 4°C overnight. Protein G agarose was added to each tube followed by incubation at 4°C overnight. Beads were washed, and 20 μl of reducing sample buffer were added followed by boiling for 4 min. Proteins were separated on 6% SDS-PAGE gels.

Glass chambers were coated with ICAM-1 (2 μg/ml) or fibronectin (10 μg/ml) overnight at 4°C and washed. Anti-CD3 activated cells in serum-free AIM-V medium (15 μl of 1 × 10⁶ cells/ml per position) were allowed to adhere to the coated surface for 30 min at 37°C followed by fixation with 2% paraformaldehyde at 4°C for 20 min. For detection of CD26, the cells were subsequently permeabilized using 0.1% saponin. Mouse anti-human Abs to TSP-1, CD91, or CD26 were added, followed by detection of the primary Ab with the biotin-avidin-HRP-based Vectastain ABC Kit (Vector Laboratories) and a peroxidase substrate (3-amino-9-ethylcarbazole). Pictures were taken of each slide using the program NIKON ACT-1 and a NIKON Eclipse E1000M microscope at ×600. Mean staining intensity (arbitrary units) was determined using the Image J picture analysis program.

Collagen type 1 was diluted in serum-free RPMI 1640 and H₂O (8:1:1), applied in plastic petri dishes, 1 ml/dish (30 mm; BD Biosciences), and allowed to polymerize at room temperature. The chemokines were dissolved in the gel, whereas Abs and peptides were present with the cells in migration experiments. A total of 1.0 × 10⁶ cells in AIM-V medium was added to each well and allowed to migrate for different times. The cells were fixed in 2.5% glutaraldehyde and washed twice with PBS. Cell morphology and cell migration were routinely, unless otherwise stated, evaluated in nine fixed positions in each well and at 50-μm intervals throughout the gel by the use of an inverted microscope (Nikon Eclipse TE300) and a digital depth meter (Heidenheim ND221). The results are given as maximal infiltration depth or mean number of infiltrating cells/field (×20 objective) per infiltration depth (50 μm for the first two layers immediately beneath the gel surface and 100 μm for other layers further down). The infiltrating cells were identified in situ in the collagen gels using immunocytochemistry after fixation in paraformaldehyde.

To study cell adhesion, plastic petri dishes (90 mm; Heger) were coated with ICAM-1 (2 μg/ml) or fibronectin (10 μg/ml) and extensively washed before use. The cells (10,000/position) in AIM-V medium were incubated on the substrates in a humidified CO₂ incubator at 37°C for 15 or 30 min. Cells were fixed in 2.4% cold glutaraldehyde for 10 min, and unbound cells were removed by gentle aspiration. The number of adherent cells per microscopic field (×20 objective) was counted. Chemokines, diprotin A, and peptides were present with the cells in adhesion experiments. A total of 1.0 × 10⁶ cells in AIM-V medium were added and allowed to migrate for different times. The cells were fixed in 2.5% glutaraldehyde, except for immunocytochemistry, for which 2% paraformaldehyde was used, and washed twice with PBS. Cell adhesion was routinely, unless otherwise stated, evaluated in five fixed positions.

Staining intensity in immunocytochemistry experiments is presented as mean arbitrary units ± SEM, and the Mann-Whitney U test was used to evaluate differences between groups. For analysis of adhesion and spreading, mean values ± SD are shown. For determination of differences between inhibitor-treated lymphocytes and control lymphocytes, the paired Student t test was used. A two-tailed value of <0.05 was considered statistically significant.

To study the possible influence of chemokines on the cell surface expression of TSP-1, we exposed anti-CD3-activated lymphocytes (generally >92% CD3 positive cells) in suspension and on a fibronectin substrate to CXCL12 and CCL5, two chemokines processed by CD26 (53, 54). The cells were cultured in serum-free medium to exclude interference of exogenous TSP-1. Anti-CD3-activated cells were used because we intended to examine cell migration and formation of pseudopodia that develop poorly in lymphocytes fresh from the blood (55). Both chemokines were potent inducers of cell surface expression of TSP-1, CD91, and CRT as determined by SDS-PAGE of immunoprecipitated biotinylated cell surface components from cells in suspension (Fig. 1,A) and quantitative immunocytochemistry of adherent cells fixed in situ to prevent Ab-induced perturbations of Ag expression and turnover of Ags during the detection procedure (Fig. 1, C–G). In contrast, CXCL12 and CCL5 (not shown) did not affect the cell surface expression of CD3 and CD4 (Fig. 1,H). Brefeldin A inhibited the cell surface expression of TSP-1 induced by CXCL12 (Fig. 1,E), indicating that the chemokines induced TSP-1 transport to the cell surface from an intracellular source. The effect of CXCL12 on the expression of TSP-1 was concentration dependent with a peak at 50–200 ng/ml (Fig. 1,D). Immunoprecipitation of surface-biotinylated cells demonstrated intact 175-kDa TSP-1 as well as 130- and 115-kDa fragments that also could be identified as TSP-1 using Western immunoblotting (Fig. 1,B). It is also evident from Fig. 1,B that CXCL12 increased the amount of intact 175-kDa TSP-1 in comparison with control cells. In contrast, these showed a prominent 115-kDa TSP-1 band that was weak in cells cultured in the presence of CXCL12. In addition, there were multiple minor bands immunoprecipitated with anti-TSP-1 Abs as revealed by the gradient gel in Fig. 1,A. It is possible that some of the multiple bands seen on the gel in Fig. 1,A represent components that associate with TSP-1, such as CRT or CD47. CXCL12 also markedly increased the cell surface expression of TSP-1 in a T cell clone as revealed by SDS-PAGE gels of immunoprecipitated biotinylated cell surface components (results not shown). In contrast to CXCL12 and CCL5, CXCL8, which binds to and induces functional responses in activated T cells (56) but is not processed by CD26 (57), did not affect TSP-1 expression (Fig. 1,C). The results in Fig. 1 indicate that CD26-processed chemokines induce cell surface expression of TSP-1 through a brefeldin A-sensitive transport mechanism and also expression of CD91 and its coreceptor CRT, which has been shown to coprecipitate with cell surface TSP-1 in T cells (32).

TSP-1 and a peptide mimetic of a sequence in TSP-1, 4N1K, were found to be inducers of cell surface expression of CD91 whereas they did not affect CD3 expression (Fig. 2). In contrast, a scrambled control peptide, Sc4N1K, did not induce CD91 expression. This suggests that the chemokine-induced cell surface expression of CD91 demonstrated in Fig. 1 is dependent on TSP-1.

The results in Fig. 1 suggested that chemokines metabolized by CD26, such as CXCL12 and CCL5 (53, 54, 58), that cleave N-terminal dipeptides from various chemokines regulate the cell surface expression of CD91 and TSP-1 through CD26. Therefore, we examined the possible influence of the CD26 inhibitors diprotin A, KR 62436, and vildagliptin on the cell surface expression of TSP-1 in anti-CD3-activated lymphocytes using quantitative immunocytochemistry and SDS-PAGE of immunoprecipitated biotinylated cell surface proteins (Fig. 3). Diprotin A, KR 62436, and vildagliptin increased the cell surface expression of TSP-1 and CD91 (Fig. 3, A and B). The fact that vildagliptin has a high specificity for CD26 (59, 60) indicated that CD26 counteracts the expression of these components. In contrast, an inhibitor of granzyme B (Ile-Glu-Thr-Asp aldehyde), another protease expressed by T cells, did not increase TSP-1 and CD91 expression. A CD26-modulating mAb to the ADA-binding site in CD26 (50) also enhanced the cell surface expression of TSP-1, further indicating that CD26 inhibits TSP-1 expression. The enhancing effects of chemokines and CD26 inhibitors were similar in magnitude and not additive indicating that they affect the same pathway (Fig. 3,C). Diprotin A increased the expression of intact 175-kDa TSP-1 and a high molecular mass protein identified as CD91 (not shown) as well as several bands of lower molecular mass, immunoprecipitated with anti-TSP-1 Abs, (130, 115, 70, and 55 kDa; Fig. 3,D) similar to CXCL12 and CLL5 as demonstrated in Fig. 1. The 130- and 115-kDa bands were identified as TSP-1 using Western immunoblotting with an anti-TSP-1 Ab (TSP-4; Fig. 1,B). The other bands may represent degradation products of TSP-1 or coprecipitated components, such as CRT (70 kDa) or CD47 (50 kDa). Thus, inhibition of CD26 and chemokines processed by CD26 had virtually the same effect on the expression of TSP-1 and CD91. This indicates that CD26 suppresses the cell surface expression of TSP-1 and CD91 and that chemokines interacting with CD26 abrogate this suppression. GM6001 decreased the intensity of intact TSP-1, enhanced the 115-kDa band, and induced a 70-kDa band (Fig. 3 D). This points to the possibility that a metalloprotease also influences TSP-1 expression, and further suggestive evidence for involvement of several proteases is provided by the loss of cell surface TSP-1 when GM6001 was combined with diprotin A.

CD26 may inhibit the expression of TSP-1 and CD91 if capable of binding to TSP-1 and inhibit its stimulatory effect leading to CD91 expression. To study the possible influence of CD26 on TSP-1 the two molecules were mixed at increasing enzyme: substrate (w/w) ratios and incubated at 37°C for 15 h. CD26, at a relatively low concentration, prevented reduction of TSP-1 as showed by the persistence of a 420-kDa band. In addition, CD26 induced a 320-kDa band, most probably derived from the 420-kDa band, 140-, 90-, and 70-kDa bands (Fig. 4). Moreover, a 110-kDa band appeared, that did not seem to be CD26, although it had the same molecular mass, because it was not present in the control lane with CD26 alone. The results in Fig. 4 suggest that CD26 can associate with TSP-1, but further studies are needed to elucidate whether CD26 can cleave TSP-1.

In the light of the previous findings that endogenous TSP-1 participates in the regulation of T cell migration and adhesion (30, 32) together with the present evidence that chemokines and CD26 affect TSP-1 expression (Figs. 1 and 3), it seemed reasonable to assume that chemokines as well as CD26 may influence T cell adhesion and migration through TSP-1. To investigate this, we examined the influence of chemokines and diprotin A on adhesion of lymphocytes to ICAM-1 coated on a plastic surface. CXCL12 and diprotin A stimulated polarized cytoplasmic spreading with formation of pseudopodia in T cells adhering to these substrata whereas control cells exhibited a relatively uniform circumferential spreading along the cell periphery (Fig. 5). A peptide mimetic of a sequence in TSP-1 (4N1K) prevented spontaneous spreading (Fig. 5) as well as the potentiation of spreading induced by diprotin A and CXCL12 (not shown), probably through interference with the association of endogenous TSP-1, whereas a scrambled control peptide (Sc4N1K) had a negligible effect on spreading.

To analyze the influence of CD26 and chemokines on lymphocyte migration anti-CD3-activated cells were allowed to migrate into three-dimensional collagen gels, a well established test system for analysis of lymphocyte migration (30, 61, 62, 63, 64), in the presence of CXCL12, the CD26 inhibitors vildagliptin, diprotin A, and KR 62436. In addition, the influence of modulation of CD26 by Ab to CD26 (TA5.9) on T cell migration was examined. Migration was determined by counting the number of cells at different depths of the collagen. It can be seen in Fig. 6,A that diprotin A and vildagliptin stimulated T cell migration into collagen. The CD26-specific Ab (TA5.9) also enhanced T cell migration significantly, whereas an Ab to CD8 did not affect migration (Fig. 6,A). A mAb to CD47, B6H12, previously found to block responses to CD47 ligands (25, 65), inhibited diprotin A-induced as well as CXCL12-induced stimulation of T cell migration (Fig. 6, B and C). In contrast, Abs to CD4 and CD8 did not affect T cell migration. This indicates that CXCL12 induced T cell migration through TSP-1 and CD47. The stimulatory effects of CXCL12 and diprotin A were not additive, indicating that they affect the same pathway (Fig. 6,D). Given the possibility that the inhibitors of CD26 may affect T cell migration through granzyme B we examined the influence of CD26 inhibitors as well as a granzyme B inhibitor on T cell migration (Fig. 6 E). CD26 inhibitors were potent inducers of migration, whereas the granzyme B inhibitor as well as an inhibitor of caspases Z-ValAlaAsp-fluoromethyl ketone did not influence migration. The granzyme B inhibitor enhanced elongation of the cells (not shown) but did not stimulate migration. These results strongly indicate that CD26 counteracts T cell migration. The fact that CXCL12 and CD26 inhibitors had indistinguishable effects on migration and adhesion-dependent cytoplasmic spreading respectively indicate that they stimulate the same pathway. This indicates that the stimulation of migration and polarity by CXCL12 reflects abrogation of a suppressive effect of CD26.

To further investigate the influence of CD26 on TSP-1 and CD91 expression and motility in T cells, a siRNA-mediated gene silencing of CD26 was performed (Fig. 7,A). CD26 siRNA but not control siRNA suppressed the cell surface expression of CD26 and increased TSP-1 and CD91 expression significantly but did not affect the expression of CD4 (Fig. 7,A). CD26 siRNA markedly increased the tendency to cytoplasmic spreading and development of locomotor morphology (Fig. 7,B). Furthermore, CD26 siRNA also markedly increased T cell migration into a collagen gel, whereas control siRNA did not increase T cell migration (Fig. 7 C). This strengthens the conclusion that CD26 is a major inhibitor of the cell surface expression of TSP-1 and of T cell motility.

The present results unveil a previously unknown CD26-controlled intraplasma membrane pathway for T cell regulation of adhesion, particularly cytoplasmic spreading, and migration connecting chemokines/chemokine receptors with TSP-1, CD91, including coreceptor CRT, and CD47 (31, 32, 66). CD26 thus down-regulates TSP-1 expression, cell polarization, and migration at the plasma membrane level, where environmental interactions take place. CXCL12 and CCL5 seem to abrogate this CD26-dependent negative regulation and hence are potent inducers of a molecular cascade characterized by up-regulated cell surface expression of TSP-1 that stimulates expression of CD91. This indicates that the CD26-controlled chemokine-activated TSP-1-CD91 cascade is essential for the motile capacity that enables T cells to carry out immune surveillance throughout the organism. The involvement of CD47, a TSP-1 receptor for induction of growth, apoptosis, and differentiation of lymphocytes as well as for the regulation of T cell inflammation (34, 67, 68, 69, 70), suggests that this cascade may also have bearing on other aspects of immune regulation besides T cell adhesion and migration. Furthermore, the fact that chemokines as well as TSP-1, CD47, CD91, and CD26 are present also in nonlymphoid cells means that interactions in cis between these components within the same plasma membrane may have implications also for other cell types.

It is well established that chemokines regulate T cell migration through G-protein-coupled receptors and that CD26 is involved in the truncation of CCL5, CCL4, and CXCL12, which is considered to inactivate or reduce their cell-stimulating capacity (53, 71, 72). However, the present results suggest that the regulatory influence of chemokines and CD26 is more complex and that chemokines stimulate T cells through abrogation of a CD26-dependent suppression. This abrogated suppression probably reflects a different handling of TSP-1 in comparison with that of cells not stimulated by chemokines. It is reasonable to assume that the CD26 control mechanism counteracts the capacity of TSP-1 to stimulate CD91 expression and as a consequence of this TSP-1 is not associated with and processed by CD91 and TSP-1 turnover maintained at a low level. CD26 may thus be postulated to have a dual role as a regulator of T cell functions being engaged either in suppression of the TSP-1-CD91 cascade or in the processing of chemokines stimulating the same cascade. Chemokines processed by CD26 may be presented to T cells associated with extracellular matrix components, on cell surfaces such as endothelial cells, or in solution, and may thus increase the locomotor capacity of T cells through TSP-1 and CD91 by acting as CD26 inhibitors.

Transient chemokine-induced inhibition of CD26 may up-regulate the motility of vascular lymphocytes as part of a commitment to extravasation. Under normal physiological conditions with respect to chemokine concentrations and endothelial presentation, CD26 probably controls the TSP-1-CD91 cascade effectively. However, inflammation-induced chemokine production provoked by microorganisms, autoantigens, or allergens may lead to persistent inhibition of CD26 and down-regulate its suppressive effect on this cascade. Thus, it is possible that persistent down-regulation of the suppressive CD26 effect on TSP-1 expression and T cell motility may account for a propensity of certain individuals to develop autoimmune and other types of T cell-dependent inflammation. Involvement in TSP-1 regulation may explain the protective effect of CD26 against autoimmunity, as demonstrated in knockout mice (46, 47). It is interesting in this context that recent studies of regulatory T cells, which are important for ensuring that the immune system does not attack self and does not overreact to external Ags, suggest that the ability of regulatory T cells to suppress bystander T cells depends on their capacity to produce TSP-1 (73). The present finding that CD26 exerts a negative control of T cell activity through TSP-1 taken together with the notion that CD26 suppresses autoimmune disease in in vivo models (46, 47) suggest that TSP-1, CD91, CD47, CD26, and chemokine/chemokine receptors are components in a multimolecular network for integrated positive and negative T cell regulation of immunity.

In conclusion, the present investigation provides evidence for the existence of a CD26-controlled chemokine-activated cell surface cascade for regulation of T lymphocyte functions through cis receptor communication, as outlined in Fig. 8. According to this model, CD26-processed chemokines inhibit the capacity of CD26 to suppress the formation of active cell edges and motility in T cells through TSP-1 and CD91. This CD26-mediated suppression seems to target a mechanism that probably plays a pivotal role for T cell recognition and interaction with the environment and that may be important for the pathogenesis of immunological diseases. An inherent poor CD26 expression or down-regulation of the CD26-dependent suppression owing to overproduction of chemokines may thus induce sustained chronic T cell stimulation that aggravates inflammatory processes. This implies that CD26 is protective against sustained T cell stimulation by Ags and chemokines. It is possible that this protective function is impaired secondary to overstimulation and sometimes even primarily defective in inflammatory disorders such as autoimmune diseases.

The authors have no financial conflict of interest.