Adhesion to Vascular Cell Adhesion Molecule 1 and Fibronectin COMPARISON OF a4P1 (VLA-4) AND a4@p, ON THE HUMAN B CELL LINE JY*

Most mononuclear leukocytes and cell lines express the integrin a4D1 (VLA-4) heterodimer. In this study we have used Northern blotting and immunoprecipi- tation experiments to demonstrate that a B lymphoblastoid cell line (JY) expressed the integrin f17 subunit in association with a4. These a4B7-positive JY cells bound poorly or not at all to VLA-4 ligands (soluble form of vascular cell adhesion molecule 1 (sVCAM-1) and the CS1 region of fibronectin). In contrast, a positive variant of JY cells (selected to express a mix- ture of a4b1 and a“&) bound avidly to VLA-4 ligands, and this binding was completely inhibitable by anti-a4 and anti-dl monoclonal antibodies. Thus, expression appears to be a critically important component of VLA-4-mediated binding to its ligands. After either JY or JY-& cells were stimulated for 15 min with the phorbol ester 12-O-tetradecanoylphorbol-13-acetate, the ma- jority of adhesion to VCAM or fibronectin remained a4- and bI-dependent, but a low amount of adhesion to sVCAM- 1 or fibronectin became a4-dependent, B1-in-dependent, thus suggesting a role for a4&. In summary, we have found (i) that a4@7 makes little or no contribution

in germinal centers (12,13) and adhesion of hematopoietic progenitor cells to VCAM-1 on bone marrow stroma (14,15). Also, binding of melanoma cell VLA-4 to VCAM-1 on activated endothelium is hypothesized to contribute to metastasis (16). Besides binding to VCAM-1, VLA-4 also mediates adhesion to a domain (called CSl) within the Hep-2 region of fibronectin (17)(18)(19). This interaction may have important functional relevance during maturation of bone marrow progenitor cells (20), during localization of sensitized T lymphocytes to antigenic sites (21), and during embryogenesis (22). In other experiments, VLA-4 has been found to trigger homotypic aggregation (23,24), by a mechanism that probably does not involve binding to .
Like other integrins, VLA-4 is composed of an a subunit (a4) and a / 3 subunit (8'). However, preliminary evidence has indicated that human a4 might sometimes associate with a B subunit other than B1. For example, a human B cell line (called JY) expresses a high level of a4, but little or no detectable p1 subunit (5), and freshly isolated human B lymphocytes contain approximately 2-fold more a4 than p1 (26).
It is assumed that this previously observed "excess" a4 is associated with another /3 subunit, other than pl , because integrin heterodimer formation has been established as being an obligatory prerequisite for cell surface expression (27)(28)(29)(30). On some mouse lymphoid cells a4 has been shown to associate with at least one other 8 subunit (called pp) which may be involved in lymphocyte homing (31, 32). A possible human homologue of the murine @p is p7 (33,34), but it has not yet been determined whether human a4 might associate with the p7 subunit.
During studies of the functions of VLA-4 (a4@' heterodimer), antibodies directed toward the a4 subunit have often been utilized, and thus difficulties in interpretation could arise if a4 associates with multiple / 3 subunits. In this regard, it has not yet been determined whether another a4 heterodimer (ie. not containing &) might mediate adhesion functions similar to those mediated by VLA-4 (a4&). Furthermore, while the a4 subunit of VLA-4 has clearly been shown to be required for cell binding to VCAM-1 and fibronectin, direct evidence for involvement has been lacking.
In this paper we have made use of &-negative J Y cells and J Y cells selected for p1 expression cells) to study the contribution of to VLA-4 functions. Also, we have identified B7 as the subunit associated with a4 on &-negative JY cells, and compared its functional contribution with that of 8'.

Functional Comparison of a4p1 and a4P7
tide (RREYSRFEKEQQQLNWKQDS) derived from the cytoplasmic domain of @7 (33,34). The FN-120 chymotryptic fragment of fibronectin was purchased from Telios Pharmaceuticals (La Jolla, CA), and the FN-40 fragment was prepared as previously described (45). Briefly, fibronectin was isolated from human plasma using gelatin-Sepharose and then cleaved with chymotrypsin. The resulting 40-kDa fragment was purified using heparin-Sepharose, followed by DEAE chromatography (45). The 25-amino acid CS1 peptide (DELPQLVTLPHPNLHGP-EILDVPST) derived from fibronectin and recombinant soluble VCAM-1 (sVCAM-1) were gifts from Dr. R. Lobb (Biogen Inc., Cambridge, MA). The latter was prepared by genetic alteration of the native molecule (the form with seven Ig domains) such that the transmembrane and cytoplasmic regions are deleted (46). Establishment and Analysis of JY-@, Cells-The B lymphoblastoid cell line JY was previously described as lacking @,, but expressing a4 (5). Attempts were made to transfect JY cells with p1 cDNA in the pFneo expression vector, and select for @,-positive JY cells. However, in multiple experiments, the transfection procedure itself resulted in selection for P1 expression, regardless of the presence or absence of (3, cDNA. This was possibly due to a selective advantage conferred by endogenous P1 on cells grown for extended periods at low viability. Thus, a few subcloned JY cell populations were found to express variable levels of cell surface pl, as assessed by flow cytometry. The @,-positive cells were further enriched by magnetic bead selection, using the anti-@, mAb A-1A5. When grown in normal conditions (RPMI 1640 supplemented with 10% fetal bovine serum) the elevated expression of O1 cells), or lack of p1 expression (in JY cells), was each remarkably stable despite continuous culture for nearly 6 months.
Cell Adhesion Assays-Cell attachment to FN-40, FN-120, CS1 peptide, and sVCAM-1, was carried out essentially as previously described (3), except that bound cells were detected by fluorescence. Briefly, protein ligands were coated onto 96-well microtiter plates (Flow Labs), and then 0.1% bovine serum albumin was added to block nonspecific adhesion. Cells were labeled by incubation with the fluorescent dye BCECF (Molecular Probes, Eugene, OR) for 30 min and then washed once in PBS containing 0.5 mM EDTA, once in 0.1% bovine serum albumin in RPMI, and finally resuspended in the same solution. The labeled cells (5 X 10') were then added to each well of a ligand-coated microtiter plate and incubated for 25 min at 37 "C followed by three washes with 0.1% bovine serum albumin in RPMI. For inhibition studies, monoclonal or polyclonal antibodies were present during the 25-min adhesion period. After washing, cells remaining attached to the plate were analyzed using a Fluorescence Concentration Analyzer machine (IDEXX Co., Portland, ME). After subtraction of background cell binding (assessed using bovine serum albumin-coated wells), values for cells bound/mm2 were calculated from the equation,Cells bound/mm2 = BF/TF X 50,000 cells added per we11/38-mm2 area per well, where B F = bound fluorescence after washing and TF = total initial fluorescence. For example, if 100% binding were observed, then 50,000/38 = 1316 cells bound/mm2. We ascertained that fluorescence was linearly proportional to cell number [from 50,000 cells (1316 cells/mm2) down to 150 cells (4 cells/mm2)]. Background binding was typically less than 10% of the total, and assays are reported as the mean of triplicate determinations. Standard deviations typically ranged from 20 to 50 cells/mm2. Also, for every assay, a microscope was used to frequently monitor cell attachment by visual inspection.
Radiolabeling, Immunoprecipitation, and Northern Blot Analysis-Cells were surface-labeled with lZ5I using lactoperoxidase and lysed in the presence of 0.5% Nonidet P-40. Then immunoprecipitation and sodium dodecyl sulfate-polyacrylamide gel electrophoresis analyses (on 7% polyacrylamide gels) were carried out as previously described (47). For Northern blot analysis, total RNA was prepared as previously described (48), separated by electrophoresis on formaldehyde-agarose gels, transferred to Zetabind nylon membrane (Cuno of p7 cDNA (33) that had been labeled with 32P by random priming.
Inc., Meriden, CT), and then RNA was probed with a 2.3-kb fragment

RESULTS
Selection of JY Cells Expressing the P1 Subunit-As demonstrated by flow cytometry, the B lymphoblastoid cell line JY expressed substantial levels of cell surface a4 but approximately 17-fold less 8, (Fig. 1). This result suggested that at least 94% of the a4 was associated with a different B subunit. Flow cytometry was carried out using a FACScan instrument (Becton Dickinson, Mountainview, CA) as previously described (3). From these J Y cells, a variant population of cells was selected with PI expression elevated over 10-fold as demonstrated by flow cytometry (Fig. 1). At the same time, a4 expression was not increased in the JY-Pl cells, and in fact was diminished slightly compared to the level in J Y cells. Expression of a5 was also observed in the JY-P, cells, although it was undetectable in J Y cells. Expression of a', a*, a3, and a6 was not observed in either the J Y or JY-P, cells (not shown). From mean fluorescence intensity values, it was estimated that approximately 50% of the newly expressed PI in JY-PI cells was associated with a6 , leaving the other 50% to be associated with cy4. Because only part of the total a4 (-42 MFI units) appeared to be associated with P1 , it is assumed that the remaining a4 (-24 MFI units) may be a associated with a different / 3 subunit.

Adhesion Properties of Unstimulated JY and JY-PI Cells-
To analyze the functional consequences of PI expression in J Y cells, adhesion assays were carried out as shown in Fig. 2. To determine the involvement of specific integrins in adhesion to each ligand, antibody inhibition experiments were of a4p1 and a4p7 performed. As shown in Fig. 3A, attachment of JY-@, cells to FN-40, CS1 peptide, and sVCAM-1 was uninhibited in the presence of either control mAb, or the a n t i d MAb PlD6, but adhesion was almost completely inhibited by the anti-a4 MAb HP2/1. In contrast, adhesion to the FN-120 fragment of fibronectin was substantially inhibited by the a n t i d mAb, but only minimally inhibited by the anti-a4 mAb. Adhesion to all of these substrates was highly inhibited by the anti-@, mAb BlE11. In another experiment (Fig. 3B), attachment to sVCAM-1 by JY-P1 cells was almost completely abolished by two different anti-& antibodies (mAb 13, 4B4), but not by nonblocking anti-& antibodies (A-1A5, TS2/16; not shown) or by negative control antibodies (P3). Together the results in Fig. 3 are consistent with VLA-5-mediated binding to FN-120, and VLA-4-mediated binding to FN-40, the CS1 peptide, and to sVCAM-1 for JY-P, cells. In contrast, attachment of JY cells to the various ligands was too low (<lo cells/mm2 in this experiment) to allow accurate analysis of antibody inhibition.
Adhesion Properties of TPA-stimulated J Y and JY-P1 Cells-Whereas unstimulated JY cells showed minimal adhesion to sVCAM-1 or to fibronectin, stimulation with TPA (50 nM for 15 min) caused increased adhesion to both fibronectin and sVCAM-1, up to a level comparable to that seen for unstimulated JY-@, cells (Fig. 4, A and B). Adhesion to fibronectin and sVCAM-1 by JY-P1 cells was also up-regulated by TPA stimulation (Fig. 4). In a control experiment, neither stimulated nor unstimulated JY or JY-@, cells showed any adhesion to collagen (coated from 0.075 up to 5 pg/ml, not shown). Because TPA caused up-regulation of JY adhesion functions, it then became feasible to carry out monoclonal antibody inhibition experiments (Fig. 5). As shown, adhesion of
JY cells to either sVCAM-1 or to fibronectin was completely inhibited by an anti-a4 antibody. Adhesion was also inhibited to a large extent by an anti-P1 antibody, but a small amount of &-independent adhesion to both sVCAM and fibronectin appeared to remain. Lack of complete inhibition was not due to insufficient anti-@, antibody, because (i) the antibody was titered to give maximal inhibition, and (ii) complete inhibition was obtained in a positive control experiment using RD cells (Fig. 5). As with the JY cells, the JY-@, cells also showed a small amount of a4-dependent, @,-independent adhesion to sVCAM-1. These results (Fig. 5) contrasted with the results in Fig. 3B, which showed complete inhibition of unstimulated JY-P1 adhesion by anti-@, antibodies.
Biochemical Analysis of Integrins in J Y and JY-0, Cells-To identify the endogenous @ subunit associated with a4 on unselected JY cells, immunoprecipitation experiments were carried out. Using an anti-a4 mAb (Fig. 6, lane 2), a trace of the typical 150-kDa a4 band was obtained, in addition to substantial amounts of the 80 kd and 70-kDa cleavage products often derived from a4 (49). Also, an associated p subunit of approximately 100 kDa was coprecipitated with a4 (lane 2). This p subunit probably was identified as p7 because polyclonal anti-P7 antiserum detected an apparent a4P7 complex (lane 4 ) , almost indistinguishable from that seen in lane 2 using the anti-a4 antibody. In contrast, p1 was not detected using the anti-& antibody A-1A5 (lane 3 ) , or using two different anti& polyclonal antibodies (not shown).
From JY-pl cells, the anti-& antibody yielded a prominent pattern of a4 and p1 bands (Fig. 6, lane 7) similar to those obtained using anti-a4 (lane 6). Consistent with the flow cytometry data (Fig. l), p1 expression appeared substantially elevated in the JY-pl cells (compare lanes 7 and 3 ) , while a4 expression was not increased (compare lanes 6 and 2). Also from the JY-p, cells, an a4p7 heterodimer was detected at somewhat lower amounts (lane 8) than obtained from the JY cells (lune 4 ) , consistent with a portion of a4 being associated with p1 instead of p7. No proteins were detected in the negative control lanes (lanes 1, 5 ) , and in other experiments, the a5 subunit was readily coprecipitated in association with p1 from JY-pl cells, but not from JY cells (not shown). In agreement with the results shown in Fig. 6, Northern blot analysis confirmed that p7 mRNA was selectively present in both JY and JY-pl cells (Fig. 7) and in the positive control IEL cells, but not in a variety of other cell lines (lymphoid, myeloid, carcinoma, sarcoma, and melanoma).

Contribution of the p1 Subunit to V U -4 Function-Pre-
vious studies had directly demonstrated that the integrin a4 subunit was critically involved in adhesion to both fibronectin and to VCAM-1, and indirect evidence was also provided for Dl involvement (3). Here we provide additional evidence that the integrin p1 subunit is critically involved in these functions. Whereas a4-positive JY cells with little detectable p1 expression showed little or no adhesion to VLA-4 ligands (VCAM-1 and fibronectin), B1-positive JY cells showed a high degree of adhesion to the same ligands.
Because replacement of p7 with p1 resulted in a marked alteration of a4 heterodimer function, a role for the p1 subunit in determining ligand binding avidity and/or specificity is suggested. In agreement with this, the ligand binding specificity of other integrin heterodimers also varies depending on the identity of the associated p subunits. For example, a6p1 -28s -18s FIG. 7. Northern blot analysis of JY and JY-P1 cells compared to other cell types. Total RNA was prepared (5 pgllane) from intraepithelial lymphocytes (IEL) that had been stimulated for 2-3 weeks with TGFp, and 30 pg each was analyzed from the T lymphoblastoid cell line Jurkat, J Y and JY-pl cells, the carcinoma cell line CCL-228, the rhabdomyosarcoma cell line RD, and the melanoma cell line LOX. For the erythroleukemic cell line K-562,50 pg/ml was analyzed. Ethidium bromide staining of 28 S and 18 S RNA was carried out to verify that similar levels of RNA were loaded in each sample (except for the IEL cells). RNA was probed with a 2.3-kb fragment of 81 cDNA (33) that had been labeled with '*P by random priming. binds to laminin, whereas a6P4 does not (50).
Expression and Function of a4P7-We have demonstrated that (i) the a4p7 heterodimer is endogenously expressed on unstimulated JY cells, and (ii) the a4p7 heterodimer is not an effective adhesion receptor for known VLA-4 ligands, despite being present at high levels. However, we also found that (iii) a low amount of a4-dependent, pl-independent adhesion was observed on phorbol ester-stimulated JY cells, thus suggesting a possible role for a4p7 in binding to VCAM-1 and to fibronectin. At present, we tentatively assume that all or most of the observed a4-dependent, P1-independent adhesion is mediated by a4p7. However, we must emphasize that direct proof for this point awaits the availability of anti-P7 antibodies.
Nonetheless, assuming that all a4 on JY cells is associated with either p1 or p7, the flow cytometry data indicates that there may be approximately 17-fold less expression of a4p1 compared to a4p7. Similarly, immunoprecipitation showed substantial a4p7 but no detectable a4p1. Nonetheless, after 15min TPA stimulation (a period too short to alter integrin expression), anti$, antibodies still caused 60-80% inhibition of JY adhesion to VLA-4 ligands. On unstimulated JY-pl cells, anti$, antibodies caused 100% inhibition of cell adhesion, despite comparable levels of p1 and p7 expression (as shown by both flow cytometry and immunoprecipitation). Thus if a4p7 is truly an adhesion receptor for fibronectin or VCAM-1, it is substantially less effective than a4p1. We hypothesize that other ligands for a4p7, perhaps with more avidity, remain to be discovered. It is rather intriguing that in the same JY cellular environment, the presumed a4p7 function appeared to require functional up-regulation, whereas a4p1 was already functional in the unstimulated cell. Although a precise biochemical mechanism for functional up-regulation is not yet known, recent work with the integrin p 2 subunit has suggested a role for the p chain cytoplasmic tail in determining functional upregulation in response to TPA (51). In this regard, it is notable that the cytoplasmic domain of human p7 (33,34) is only -38% similar to that of human p1 (52), thus allowing the hypothesis that these distinct domains could participate to different extents in the mechanism of functional up-regulation.
Our evidence for the presence of p7 on the surface of a B lymphoblastoid cell line (JY) is consistent with previous Northern blot experiments showing p7 expression on lymphoblastoid cells (33,34). The size of p7 (sometimes slightly smaller than Pl), its association with a4, and its expression on lymphoid cells are all properties resembling those of the murine pp subunit. If the human p7 subunit is indeed the homologue of the murine pp subunit, then a4D7 could possibly contribute to lymphocyte "homing" to Peyer's Patches, as shown for a 4 @ p (31,53). Other recent evidence regarding p7 (54,55) suggests that it may not only complex with a4, but also may be part of a novel integrin-like complex found on the surface of rat (56), mouse (55), and human (57-59) intraepithelial lymphocytes. Thus 87, like the integrin Plr &, and p3 subunits, defines a subfamily of integrin heterodimers, with each subfamily having multiple distinct a subunits associated with a common p subunit.
In conclusion, we have established that the p1 subunit is essentially involved in determining the specificity of VLA-4 binding to its ligands. Also, we found that the a4p7 heterodimer expressed on JY cells appears to be minimally active as a receptor for VLA-4 ligands, perhaps partially due to an absolute requirement for functional up-regulation before adhesion can be detected. This information will be of fundamental importance for sorting out the diverse roles of a4containing integrins on leukocytes and other cells.