Thy-1 (CD90) Is an Interacting Partner for CD97 on Activated Endothelial Cells

Leukocyte extravasation into perivascular tissue plays a key role in inflammatory diseases. This recruitment requires leukocyte interaction with vascular endothelium and consists of multiple consecutive steps, including the capture of circulating leukocytes, subsequent leukocyte rolling, arrest, firm adhesion, and ensuing diapedesis. The cascade occurs by sequential activation-dependent interactions between endothelial cell (EC) adhesion molecules and their specific ligands on leukocytes.

Thy-1 (CD90), a highly glycosylated GPI-anchored cell surface protein with a molecular mass ∼35 kDa, is a receptor on EC, belonging to the Ig superfamily, and is involved in arrest and firm adhesion of leukocytes to the endothelium (1, 2). In humans, Thy-1 expression is restricted to activated EC, fibroblasts, neuronal cells, and a subset of peripheral CD34⁺ stem cells (3). Adhesion of neutrophils to activated Thy-1⁺ EC, mediated by Thy-1/Mac-1 (CD11b/CD18) interaction, is one attachment mechanism facilitating their subsequent migration into lesions of psoriatic skin (4–6). However, blocking of CD11b/CD18 did not result in complete inhibition of Thy-1–mediated adhesion of myeloid cells to activated EC (6). This suggests the presence of an additional interacting partner of Thy-1 involved in Thy-1–mediated adhesion of myeloid cells to activated EC.

CD97, a member of the epidermal growth factor (EGF)–seven-span transmembrane (TM7) subfamily of adhesion (class B2) G protein-coupled receptors (7), shows a hematopoietic expression profile that merits its consideration as a potential ligand for Thy-1 on activated EC. CD97 is a cell surface receptor present in peripheral neutrophils, monocytes, and activated lymphocytes (8). CD97 is expressed as a heterodimer of a noncovalently bound extracellular α-chain, represented by tandemly arranged EGF domains and a stalk, and a β-chain, composed of the TM7 and a short intracellular portion. Both chains result from intracellular autocatalytic cleavage (9). The CD97 α-chain has been thought to be shed from the membrane of CD97-expressing cells. It is very likely identical to soluble CD97 (sCD97), detected in synovial fluid of rheumatoid arthritis patients (10). As the result of alternative splicing in humans, three isoforms exist: CD97(EGF1,2,5), CD97(EGF1,2,3,5), and CD97(EGF1–5). CD97 shows remarkable homology to the EGF-TM7 receptor EMR2 (CD312) (11), especially within the EGF-like domains. Although both receptors are present at high levels in immune cells, the overall expression pattern, ligand binding, and, thus, function are dissimilar (12).

Signal transduction through CD97 and EMR2 is still unknown. Because truncation of the TM7 region disrupted CD97-increased single random cell migration (13), and binding of a specific Ab to EMR2 regulated human neutrophil function (14), signaling through EGF-TM7 receptors seems very likely.

In this study, we identified Thy-1 as a potential new ligand of CD97 and demonstrated that PMNC interact specifically, via CD97, with Thy-1 expressed in activated EC, thus mediating leukocytic adhesion.

The CD97^EGF mAb (clone BL-Ac/F2 (8) detects a glycosylation-dependent epitope within the first two EGF-like domains of CD97 and EMR2 (15). The CD97^stalk mAb (clone CLB-CD97/3) (16) binds to the stalk region of the CD97 α-chain. The rabbit polyclonal CD97 Ab was purchased from Sigma-Aldrich Chemie (Munich, Germany). The mAb 1B5 binds to the fourth EGF-like domain present only in the largest isoform of CD97 and EMR2 (17). The Thy-1 mAb clone AS02 does not block Thy-1–dependent cell adhesion, whereas clone BC9 does (3, 18). The CD55 mAb (clone BRIC 216), binding to the short consensus repeat 3 domain of CD55, was purchased from the International Blood Reference Group laboratory (Bristol, U.K.). The CLB-CD97L/1 mAb, binding to the short consensus repeat 1 domain of CD55 (19), and the EMR2 mAb (clone 1A2) (11) were kind gifts of J. Hamann (University of Amsterdam, Amsterdam, The Netherlands). Both CD55 mAb inhibit binding of erythrocytes to COS-7 cells transfected with CD97(1,2,5) cDNA (19). The mAb to αvβ3 integrin (clone LM609) and the polyclonal Ab to α5β1 integrin were purchased from Millipore (Schwalbach, Germany). MAb directed to ICAM-1 (CD54; clone R6.5.D6), CD11b (clone X-5), and CD18 (clone IB4) were purchased from the American Type Culture Collection (LGC Standards, Wesel, Germany), BMA Biomedicals (Augst, Switzerland), or Calbiochem (Darmstadt, Germany).

Patients aged ≥18 y (n = 15; 11 males; mean age, 49.7 ± 20 y) with plaque-type psoriasis that had been refractory to topical treatment with external glucocorticoids or vitamin D3 analogs within the last 4 wk were included in the study. Age- and sex-matched normal subjects (n = 12; 10 males; mean age, 47.7 ± 14 y) were used as controls. The study was approved by the University of Leipzig Ethics committee. All persons gave their written consent prior to their enrollment into the study. The study was conducted in accordance with the guidelines of the World Medical Association’s Declaration of Helsinki.

Cryostat sections of patients with psoriasis were stained for double immunofluorescence and analyzed by laser-scanning microscopy or for simple immunohistology, as described (20).

Human dermal microvascular EC (HDMEC) and HUVEC were prepared, as described (21). HDMEC were cultured in EGM-MV media (Promocell, Heidelberg, Germany). Only preparations with >95% CD31⁺ EC were used (21). For induction of Thy-1, EC in the first or second passage were stimulated with 10 ng/ml PMA (Invitrogen, Karlsruhe, Germany) for 24 h. PMNC of normal subjects and psoriatic patients were isolated, as described (5). The purity was >95% CD15⁺ cells. PMNC were labeled with 0.1 μM carboxyfluorescein diacetate succinimidyl ester (CFDA; Invitrogen) for 15 min on ice. In some experiments, CFDA-labeled PMNC were activated with 10 ng/ml TNF-α (Immunotools, Friesoythe, Germany) for 30 min.

CD97(EGF1,2,5) and CD97(EGF1–5) cDNA cloned into the pcDNA3.1 Zeo(+) vector (22) were used as basic templates for the generation of new constructs. pEGFP-N1 (BD Biosciences, Heidelberg, Germany) generates a fusion protein consisting of CD97 with enhanced GFP, allowing the direct monitoring of transfected cells without further labeling.

Wild-type HT1080, WiDr, and CHO cells (American Type Culture Collection) were stably transfected with 1 μg construct DNA by electroporation (290 V, 1500 μF; Multiporator; Eppendorf, Hamburg, Germany). A total of 7.5 μg/ml geneticin (Invitrogen) was added after 24 h to select stable clones. From each transfection, 3 clones of >10 were randomly selected by high CD97 mean fluorescence intensity (MFI) in flow cytometry. Control cells contained either the empty plasmid (empty) or the inverted construct (mock). CHO clones stably expressing human Thy-1 were generated as described (18).

Constructs encoding sCD97(EGF1,2,5), sCD97(EGF1,2,5) δ stalk (without the stalk), sCD97(EGF1–5), soluble EMR2 (sEMR2)(EGF1,2,5), sEMR2(EGF1,2,3,5), and sEMR2(EGF1–5) fused to a murine Fc-tag (mFc) were used (17, 23). mFc-sCD97 and mFc-control protein were biotinylated with the RTS AviTag E. coli Biotinylation Kit from Roche Diagnostics (Mannheim, Germany).

In PMNC-to-cell–adhesion assays, EC or Thy-1– or mock-transfected CHO cells were cultured in a 96-well plate. Confluent EC were stimulated for 24 h to induce stronger Thy-1 expression. A total of 2 × 10⁴ CFDA-labeled unstimulated or TNF-α–activated PMNC were added to the wells. After incubation for 45 min at 37°C, unbound PMNC were removed by washing with PBS. Adherent PMNC were lysed by the addition of 50 μl 10% SDS to the wells. Fluorescence was quantified using a fluorimetric plate reader (SpectraMax; Molecular Devices, Sunnyvale, CA). Fluorescence and adherent cell number correlated in a linear fashion (data not shown).

In cell line to cell-adhesion assays, 2 × 10⁵ HT1080, WiDr, or CHO cells stably overexpressing CD97 as an enhanced GFP-fusion protein were added to a well of an eight-chamber slide precultured with Thy-1– or mock-transfected CHO cells to confluence. After incubation for 30 min at 37°C, unbound cells were removed by several washes with PBS. Adherent cells were fixed with ice-cold methanol for 10 min at −20°C, rinsed in PBS, and mounted. The number of adherent cells was counted in 20 fields at 40× magnification under a fluorescence microscope.

In blocking experiments, cells were preincubated with a specific or control IgG mAb (10 μg/ml) for 30 min at 37°C in the case of endothelial or CHO monolayers or at 4°C in the case of PMNC. EC were preincubated with Ab directed to Thy-1, CD55, αvβ3 or α5β1 integrin, or ICAM-1. PMNC were preincubated with mAb to CD18 or CD97.

In cell-to-protein adhesion assays, 96-well MaxiSorb plates (Fisher Scientific, Schwerte, Germany) were coated with 500 ng human Fc (hFc)–Thy-1 or hFc-control protein/well in TBS, 2 mM CaCl₂, 0.1 mM MgCl₂ overnight. Plates were washed with TBS and blocked with TBS/1% BSA for 1 h at room temperature. A total of 5 × 10⁵ CFDA-labeled PMNC, preincubated with IgG ctr-mAb or mAb to CD97, were added to the coated wells. Fluorescence was measured as described above. In other experiments, cells of three CHO clones, differing in the expression level of CD97, were added to the coated wells.

In protein-binding assays, 1 μg purified sCD97(EGF1,2,5), sCD97(EGF1,2,5) δ stalk, sCD97(EGF1–5), sEMR2(EGF1,2,5), sEMR2(1,2,3,5), or sEMR2(EGF1–5) was added to immobilized hFc–Thy-1 or hFc-control protein-coated wells and incubated for 2 h at room temperature. Plates were washed with TBS, 0.1% BSA, 0.05% Tween 20, 2 mM CaCl₂, and 0.1 mM MgCl₂. CD97 binding to immobilized Thy-1 was performed in the absence or presence of 10 μg/ml mAb to Thy-1 or CD97, respectively. Bound proteins were detected by addition of the biotinylated CD97^EGF mAb and HRP-labeled streptavidin (DakoCytomation, Hamburg, Germany), followed by tetramethylbenzidine substrate (Fisher Scientific). Color reaction was measured at 450 nm. To evaluate whether binding depends on the presence of Ca²⁺, binding was performed with buffers containing Ca²⁺ or 1 mM EDTA without Ca²⁺.

In the sCD97-to-cell–binding assay, 1 × 10⁵ Thy-1–transfected or control CHO cells and unstimulated or activated HUVEC were incubated with 0.1 μg sCD97(EGF1,2,5), sCD97(EGF1–5), or mFc-control protein for 60 min at room temperature, followed by PE-labeled goat anti-mouse (Dako) to detect the Fc-tag in flow cytometry. Alternatively, biotinylated sCD97 could also be detected by PE-streptavidin (Dako).

To block binding of sCD97 to EC, cells were preincubated with 10 μg/ml the respective Ab for 1 h at 37°C prior to sCD97 addition. Nonbound protein was washed away several times with PBS. EC were analyzed by flow cytometry.

Statistical analysis was performed using the Student t test or Mann–Whitney U test; p values < 0.05 were regarded as significant.

Adhesion of PMNC to activated Thy-1⁺ EC, mediated by the interaction of Thy-1 and Mac-1, facilitates their migration into lesions of psoriatic skin (4–6). The presence of an additional interacting partner for Thy-1 in PMNC has been suggested, because blocking of Mac-1 did not result in complete inhibition of Thy-1–mediated adhesion of PMNC (6). In this study, we examined whether interaction of Thy-1 and CD97 is involved in the adhesion of PMNC to activated EC.

Infiltrating leukocytes located in epidermal microabscesses and in dermal inflammatory infiltrates of psoriatic skin lesions were CD97⁺. Microvascular EC adjacent to these leukocytes expressed Thy-1 (Fig. 1A).

According to Wetzel et al. (6), we confirmed that psoriatic PMNC adhered more to activated EC compared with PMNC of normal subjects (Fig. 1B). However, expression of the Thy-1 counterreceptor Mac-1 was not different on psoriatic and healthy PMNC (Fig. 1C). In contrast, psoriatic PMNC showed higher expression of CD97 than did normal PMNC, whereas expression of EMR2, the close subfamily member of CD97, was comparable between psoriatic and healthy PMNC (Fig. 1C).

Stimulation of healthy PMNC with TNF-α, resulting in an enhanced adhesion of these PMNC to Thy-1–transfected cells, thus imitating characteristics of psoriatic PMNC, increased the expression of CD97 1.5-fold. The expression of EMR2 and CD11b was unchanged (Fig. 1D). Psoriatic PMNC showed no additional increase in CD97 levels after TNF-α treatment (data not shown). Accordingly, TNF-α–activated PMNC adhered more strongly to stimulated micro- and macrovascular EC compared with control PMNC (Fig. 1E). In summary, our data suggested the possible involvement of CD97 in adhesion of PMNC to activated EC.

PMNC showed stronger adhesion to activated, compared with unstimulated, micro- and macrovascular EC, indicating the upregulation of an endothelial receptor that mediates PMNC adhesion (Fig. 2). Unstimulated EC slightly expressed Thy-1. Activation increased the percentage of Thy-1⁺ HUVEC (Fig. 2A) and HDMEC (Fig. 2B), as well as the expression level of Thy-1 in these cells.

Because adhesion of PMNC to activated Thy-1⁺ HDMEC and to Thy-1–transfected cells can be blocked only partially with Ab to CD18 (6), we suggested additional interaction partners for endothelial Thy-1 on PMNC. To examine whether CD97 is involved in binding of PMNC to Thy-1⁺ cells, the adhesion of PMNC to HUVEC and HDMEC was determined upon blocking by specific mAb.

First, PMNC were pretreated with different mAb to CD97 (Fig. 3A, 3C). As a positive control, PMNC were preincubated with mAb to CD18, which indeed partially blocked the binding of PMNC to HUVEC (Fig. 3A) and HDMEC (Fig. 3C). Interestingly, both CD97 mAb directed to the stalk (CD97^stalk) or to the EGF-like domains (CD97^EGF) blocked PMNC binding. The CD97^stalk mAb showed a stronger blocking compared with the CD97^EGF mAb in both EC. In a parallel assay, adhesion of CD55⁺ erythrocytes to CD97-transfected HT1080 cells, performed as described by Hamann et al. (24), was strongly inhibited by the CD97^EGF mAb, thus demonstrating the blocking function of this mAb (data not shown). Consequently, our data indicated that the stalk of CD97 is involved in the interaction of CD97 with Thy-1. The combination of mAb to CD18 and CD97^stalk caused a further significant decrease in adhesion of PMNC compared with blocking with either mAb alone (Fig. 3C). This indicated that both molecules mediate the adhesion of PMNC to activated EC.

Second, EC were pretreated with mAb directed to EC-specific Ags (Fig. 3B, 3D). As expected, adherence of PMNC to HUVEC was decreased by the Thy-1–blocking mAb (Fig. 3B). Thus, blocking of HDMEC (Fig. 3D) was expanded to ICAM-1, a known ligand of Mac-1 (CD11b/CD18), as well as to the other potential interacting partners of CD97 on EC, such as CD55 and αvβ3 and α5β1 integrin. Strongest blocking was demonstrated with the Thy-1–blocking and the ICAM-1 mAb. Using two CD55 mAb, we demonstrated slight blocking with the mAb BRIC 216. In a preassay, we confirmed that both CD55 mAb used, BRIC 216 and CLB-CD97/L1, nearly completely blocked adhesion of erythrocytes to CD97-transfected HT1080 cells (data not shown). Adhesion of PMNC to activated HDMEC was also blocked by αvβ3 and α5β1 integrin Ab, but not as strongly as by the Thy-1 blocking mAb. The combination of the Thy-1–blocking mAb and the ICAM-1 mAb caused an additive effect, whereas the combination of the Thy-1–blocking mAb and the CD55 mAb (BRIC 216) did not.

The slight adhesion of PMNC to unstimulated EC could not be inhibited by mAb directed to CD18 or CD97 on PMNC or by mAb to the various Ags in EC (Fig. 3A–D).

To clarify whether CD97 mediates the increased binding of activated PMNC to Thy-1⁺ EC, we performed blocking experiments using TNF-α–stimulated PMNC (Fig. 3E). After preincubation of activated PMNC with the CD18, CD97^EGF, or CD97^stalk mAb, adhesion of activated PMNC was blocked significantly. Blocking of CD97 by the CD97^stalk mAb decreased adhesion of activated PMNC to the level of adhesion of unstimulated PMNC to activated EC.

Next, we examined whether CD97-overexpressing cells adhered to Thy-1⁺ CHO cells more strongly compared with the corresponding wild-type or CD97 mock-transfected cells. CHO, WiDr, and HT1080 clones stably overexpressing CD97 were generated to rule out that the measured effects depend on the parental cell line transfected. Wild-type WiDr and HT1080 cells showed basal expression of CD97. However, the selected CD97-overexpressing clones showed higher levels of CD97 compared with wild-type or CD97 mock cells (Fig. 4A).

Irrespective of the cell line, CD97-overexpressing clones adhered strongly to Thy-1–transfected CHO cells (Fig. 4B). The number of adherent cells was comparably low in all control combinations: adhesion of CD97 mock-transfected clones to vector-transfected CHO cells, CD97-overexpressing clones to vector-transfected CHO cells, and CD97 mock-transfected CHO or WiDr clones to Thy-1–transfected CHO cells. A greater number of wild-type HT1080 cells (data not shown) or CD97 mock-transfected HT1080 cells adhered to Thy-1–transfected CHO cells, because HT1080 wild-type cells express CD97 at a significant level. However, binding of CD97-overexpressing HT1080 clones to Thy-1–transfected CHO cells was significantly greater compared with binding of HT1080 mock-transfected cells (Fig. 4B).

Next, we examined whether PMNC and CD97-overexpressing CHO clones bind specifically to immobilized Thy-1 protein in a cell-to-protein adhesion assay. PMNC showed strong adhesion to hFc–Thy-1–coated wells but only weak binding to an irrelevant immobilized hFc-control protein (Fig. 5A, 5B). This adhesion could be partially inhibited by mAb to CD97. The CD97^stalk mAb showed the strongest inhibition.

To verify whether the number of adherent cells depends on the expression level of CD97, we examined three CHO clones with different CD97 expression levels (Fig. 5C). Clearly, the number of adherent CD97-overexpressing CHO cells correlated with the MFI of the different clones (Fig. 5D).

To further examine the interaction between CD97 and Thy-1, we analyzed binding of the purified proteins hFc–Thy-1 and sCD97(EGF1,2,5) or sCD97(EGF1–5) in a cell-free in vitro ligand-binding assay. Control binding to an irrelevant hFc-control protein was used to exclude unspecific binding of the Fc-tag. Both sCD97 isoforms bound to immobilized Thy-1 but not to the hFc-control protein (Fig. 6A). Binding was concentration dependent and saturable.

The blocking, but not the nonblocking, Thy-1 mAb inhibited binding up to 20% (Fig. 6B). Both the CD97^EGF and CD97^stalk mAb also disturbed Thy-1 binding. The CD97^stalk mAb showed the strongest inhibition of Thy-1 binding, suggesting the involvement of the stalk of CD97 in Thy-1–CD97 interaction.

To clarify this point in more detail, we compared binding of sCD97(EGF1,2,5) and sCD97(EGF1,2,5) δ stalk to hFc–Thy-1. sCD97 without the stalk did not bind hFc–Thy-1 (Fig. 6C), indicating that the stalk is essential for CD97 binding to Thy-1. Furthermore, we examined whether the three isoforms of sEMR2 bind to hFc–Thy-1. EMR2 shows 97% amino acid sequence identity to CD97 within the EGF-like domains but only 46% amino acid sequence identity in the stalk region. None of the sEMR2 isoforms bound hFc–Thy-1 (Fig. 6C). Furthermore, withdrawal of Ca²⁺ did not disturb Thy-1–CD97 interaction (Fig. 6D). Because binding to the EGF-like domains depends on Ca²⁺, this result also suggested the involvement of the CD97 stalk in binding Thy-1.

Finally, we examined binding of sCD97 to Thy-1⁺ cells. Binding of sCD97(EGF1,2,5) and sCD97(EGF1–5), but not mFc-control protein, to EC and to Thy-1–transfected CHO cells was detected via the mFc-tag or the biotin-tag in flow cytometry (Fig. 7A). Unstimulated EC, slightly expressing Thy-1, showed weak binding of both sCD97(EGF1,2,5) and sCD97(EGF1–5). Activated EC showed an increase in sCD97 binding compared with unstimulated EC.

Both sCD97 isoforms also bound to Thy-1–transfected CHO cells (Fig. 7B, 7C). However, we observed significant binding of sCD97(EGF1–5), but not sCD97(EGF1,2,5), to vector-transfected CHO cells. Because wild-type CHO cells express chondroitin sulfate B, a potential ligand of the longest, but not the shortest and middle, CD97 isoform (17), it is very likely that sCD97(EGF1–5) bound to chondroitin sulfate B. We tested this hypothesis by blocking the interaction site for chondroitin sulfate B on sCD97(EGF1–5). Pretreatment of sCD97(EGF1–5) with either chondroitin sulfate B or the mAb 1B5 abolished the binding to vector-transfected CHO cells, whereas binding to Thy-1–transfected cells was diminished but still detectable (Fig. 7C). Binding of sCD97(EGF1–5) to Thy-1⁺ EC could be partially inhibited by the blocking mAb to Thy-1 and by the CD55 mAb BRIC 216 (Fig. 7D).

The adhesive interaction of PMNC with activated EC is an essential step in the process of PMNC accumulation at sites of inflammation in vivo. In this study, we demonstrated that CD97 is involved in the adhesion of PMNC to activated EC. Binding of CD97 and Thy-1 was shown at various levels. On the one hand, PMNC, CD97-overexpressing cells, and sCD97 bound to activated EC, Thy-1–overexpressing cells, as well as immobilized Thy-1 protein. On the other hand, adherence and binding could be blocked partially and specifically with the corresponding Ab.

CD97 belongs to the receptors that are able to interact with ligands different in structure and expression profiles. The long extracellular α-chain, probably identical to naturally occurring sCD97, provides many possibilities for binding different cellular and extracellular matrix ligands.

The first identified ligand was CD55 (decay-accelerating factor) (19), which binds to the first two EGF domains of CD97 (24). CD55 is present in resting EC at a high level, whereas Thy-1 is only slightly expressed on unstimulated EC but is increased after activation. In our study, CD97⁺ cells, either PMNC or CD97-transfected cells, bound much more to activated, compared with unstimulated, EC, which did not indicate binding via CD55. Moreover, pretreatment of activated EC with Thy-1– or CD55-blocking mAb resulted in stronger inhibition of adherence by the Thy-1–blocking mAb. Additionally, in normal human skin, CD55 is present in vascular structures, but its expression is decreased in nonlesional psoriatic skin and virtually abolished in lesional psoriatic skin (25). Furthermore, CD97–CD55 interaction is not involved in human leukocyte adhesion to porcine EC in transplantation models (26). All of the data indicated that, although one of the examined CD55 mAb slightly blocked adhesion of PMNC on HDMEC, it is unlikely that CD97 in myeloid cells interacts with CD55 in EC in psoriatic lesions.

Subsequently, α5β1 and αvβ3 integrins were identified as interacting proteins for human sCD97(1,2,5) and sCD97(EGF1–5) on macrovascular EC (27). These sCD97 forms chemoattracted HUVEC in a migration, as well as a Matrigel-based invasion, assay. The Arg-Gly-Asp tripeptide present only in the stalk region of human, but not mouse, CD97 was partially involved in these effects (27). αvβ3 integrin is present in EC in psoriatic lesions (28). In our study, binding of CD97 to α5β1 and αvβ3 integrins may be partially involved in the adhesion of PMNC to activated microvascular EC, because blocking Ab to these integrins slightly inhibited the attachment of PMNC. We used the same Ab that inhibited binding of sCD97 to HUVEC, as shown recently (27). In contrast to our study, Wang et al. (27) used only sCD97 to evaluate binding to α5β1 and αvβ3 integrins in EC. We performed cell-to-cell adhesion assays to confirm binding of surface-associated CD97 to EC. Moreover, the investigators examined HUVEC for binding of sCD97, although microvascular EC are more relevant for migration of PMNC to inflammatory lesions.

Furthermore, the fourth EGF domain of CD97 and EMR2 and, thus, only the longest isoforms of these receptors, interacts with proteoglycans containing chondroitin sulfate B (17). Chondroitin sulfate B was identified as a ligand by the use of multivalent fluorescent beads containing the EGF-like domains of EMR2 in human tissue sections (17). The described binding pattern (i.e., that of chondroitin sulfate B) did not resemble the expression pattern of human Thy-1. Although our data showed binding of sCD97(EGF1–5) to chondroitin sulfate B in wild-type and vector-transfected CHO cells, binding to Thy-1–transfected CHO cells was much stronger and was not prevented by blocking the chondroitin sulfate B interaction site. This indicated a specific interaction between sCD97(EGF1–5) and Thy-1 and a minor role for binding to chondroitin sulfate B at activated EC.

The phenomenon that several ligands could bind to one receptor, and vice versa, is described as redundancy and also includes interactions in which both partners are cellular receptors. Because the ligands of CD97 differ in expression pattern and structure, the same is true for the known interacting proteins of Thy-1. Its interaction with αvβ3 in melanoma is one mechanism whereby these tumor cells adhere to activated endothelium (18). Furthermore, Thy-1 mediates adhesion of PMNC to EC by interaction with Mac-1 (CD11b/CD18) (5).

In this article, we described CD97 as a new interaction partner of Thy-1, mediating the adhesion of PMNC to activated EC, whereby activated, compared with unstimulated, PMNC adhered more strongly. After blocking with the CD97^stalk mAb, activated PMNC adhered only to the same extent as did unstimulated PMNC. In blocking experiments with combinations of mAb against CD97 and CD18, we achieved a significant decrease in adhesion compared with blocking with each mAb alone. However, blocking was not complete. As we showed in protein–protein binding assays, the available CD97 mAb did not completely block the binding site for Thy-1 on CD97. Moreover, in addition to CD97 and CD18, other ligands for Thy-1 on PMNC may exist.

The physiological relevance of Thy-1 in activated EC for the recruitment of inflammatory cells was demonstrated very recently in thioglycollate-induced peritonitis and acute and chronic lung inflammation using Thy-1–deficient mice (29). In this study, Thy-1 mediated the adhesion of granulocytes and monocytes to activated EC. This interaction plays a pivotal role in the control of the emigration of granulocytes and monocytes from blood into peripheral tissue during inflammation. Consequently, the altered number and composition of extravasated leukocytes affect the inflammatory tissue microenvironment, including the chemokine/cytokine and protease pattern (29).

The biological role of Thy-1, like that of CD97, is context dependent (30). Thy-1 distribution in mice differs from that in humans. Thy-1 is also expressed in rat activated EC (31, 32). But in contrast to humans, mouse Thy-1 is present on the surface of mouse thymocytes and peripheral T cells. The physiological ligand or interacting molecule for mouse Thy-1 in the lymphoid compartment has not been identified (33).

Our data raise a number of interesting questions that must be clarified in detail in future studies. First, which domain(s) of CD97 mediate binding of Thy-1? Our results indicated that the stalk region of CD97 is involved in this interaction, because stronger blocking of CD97 binding to Thy-1⁺ cells or to Thy-1 protein was seen with the CD97^stalk compared with the CD97^EGF mAb. sCD97(EGF1,2,5) without a stalk, as well as the different sEMR2 constructs, failed to bind Thy-1, although the EGF-like domains of both proteins are almost identical. Moreover, binding did not need calcium that is necessary for binding to the CD97 EGF-like domains. However, both CD97 mAb used prevented binding of Thy-1 protein only partially, indicating that they did not bind to the specific interaction site of CD97 and Thy-1.

Second, are there discrepancies between the binding affinities or properties of CD97 expressed at the cell surface and in soluble forms of CD97? In our adhesion assays using PMNC in activated EC, we observed a clear inhibiting effect of the Thy-1–blocking mAb but only a weak effect of the CD55 mAb. However, binding of sCD97 to activated EC was equally well blocked by the Thy-1–blocking and CD55 mAb. Overall, our data agree with those of Hamann et al. (24) and Kwakkenbos et al. (34): cell surface-associated CD97(EGF1–5) showed only a very weak binding to CD55, whereas binding of sCD97(EGF1–5) to HEK293 cells was clearly mediated by CD55, as shown by blocking of this interaction with a CD55-specific mAb.

Third, does CD97–Thy-1 binding play a significant role in vivo? The interaction of CD97 with Thy-1 in EC is probably restricted to sites of inflammation, because Thy-1 is expressed in EC at a significant level only in regions in which cell activation and inflammation occurred (35). Several murine experimental studies indicated the involvement of CD97 in PMNC accumulation at inflammatory sites. Targeting mouse CD97 by mAb inhibited the accumulation of neutrophils at sites of inflammation, thereby affecting antibacterial host defense (36) and inflammatory disorders (37). Otherwise, accumulation of PMNC at sites of inflammation was not affected in CD97-deficient mice (20, 38) that display no overt phenotype at steady-state, except for a mild granulocytosis, which increases under inflammatory conditions. Interestingly, application of CD97 mAb blocked neutrophil trafficking after thioglycollate-induced peritonitis in wild-type, but not in CD97 knockout, mice (20). Consequently, CD97 mAb induced an inhibitory effect that disturbed normal granulocyte trafficking, which was not perturbed in the absence of the molecule (20). Overall, comparison of the consequences of mAb treatment and gene targeting implied that CD97 mAb actively inhibited the innate response, presumably at the level of granulocyte or macrophage recruitment to sites of inflammation in mice.

In summary, we identified Thy-1 as a potential new ligand of CD97. PMNC interact specifically via CD97 with Thy-1 upregulated on activated EC. Thus, Thy-1–CD97 is probably involved, together with other receptor–ligand pairs, in mediating leukocyte adhesion at sites of inflammation.

The expression constructs for the various sCD97 and sEMR2 isoforms were kindly provided by M. Stacey (Sir William Dunn School of Pathology, Oxford, U.K., now University of Leeds, Leeds, U.K.).

The authors have no financial conflicts of interest.