A region of the integrin VLA alpha 4 subunit involved in homotypic cell aggregation and in fibronectin but not vascular cell adhesion molecule-1 binding.

The VLA-4 (α4β1) integrin is involved in the adhesion of cells to fibronectin and vascular cell adhesion molecule-1 (VCAM-1). In order to study α4 structure-function relationships, we have expressed mutated α4 subunit by transfection into VLA-4-negative K562 cells. Substitutions at α4 residues Arg89-Asp90, which show the highest surface probability indexes inside the N-terminal α4/80 fragment, resulted in a reduction in the reactivity of all anti-α4 epitope A monoclonal antibodies (mAbs) tested, compared with the reactivity with anti-α4 epitopes B1, B2, and C mAb, both by transfectant flow cytometry, and by immunoprecipitation and SDS-polyacrylamide gel electrophoresis analysis of transfectant surface-iodinated proteins. In contrast, substitutions at nearby residues, Gln101, Pro102, and Ile108 did not affect the reactivity of any anti-α4 mAb representing the known α4 epitopes. Homotypic cell aggregation triggered by anti-α4 epitope A mAb was prevented in the transfectants expressing mutated α4 Arg89-Asp90Asp residues, while cell aggregation was fully achieved with either anti-α4 epitope B2 or anti-β1 mAb. Mutations at α4 residues Gln101, Pro102, and Ile108 did not affect the homotypic cell aggregation of the transfectants expressing these mutations. In addition, the adhesion of mutant Arg89-Asp90 α4 transfectants to the connecting segment-1-containing fibronectin-40 (FN-40) fragment of fibronectin was diminished compared to wild type α4 transfectants, as well as to other mutant α4 transfectants. This adhesion to FN-40 was restored when the activating anti-β1 TS2/16 mAb was present in the adhesion assays. In contrast, adhesion to VCAM-1 was not affected by mutations at Arg89-Asp90, nor at Gln101, Pro102, and Ile108 α4 residues. Altogether, these results indicate that α4 residues Arg89 and Asp90 are included in a region involved in homotypic cell aggregation, as well as in adhesion to FN-40, but not to VCAM-1.

In addition to most types of leukocytes, VLA-4 is also expressed on various nonhematopoietic tumor cells such as melanomas (20) and during muscle differentiation at the stage of myotubes (21). The ␣4 subunit is expressed on nonlymphoid tissues in developing mouse embryo (22,23), and it has been reported that the absence of a functioning ␣4 gene results in defects in placental and cardiac development, leading to embryonic lethality (24). ␣4␤7 integrins are expressed on most lymph node T and B cells (25), on subsets of CD4 ϩ memory T cells (18), and on lymphocytes present in rheumatoid synovium (26). The ␣4 subunit can be expressed at the cell surface as an uncleaved 150-kDa form, or as proteolytically cleaved fragments of 80 and 70 kDa, designated previously as ␣ 4/150 and ␣ 4/80,70 , respectively (27).
Most data accumulated so far on the in vivo role of the ␣4 integrins have come from studies using anti-␣4 mAb in animal models. Lung antigen challenge (28 -30), experimental allergic encephalomyelitis (31,32), ulcerative colitis (33), contact hypersensitivity (34,35), and diabetes (36 -38), are among the processes where ␣4 integrins play a significant role. The inhibitory effect of the anti-␣4 mAb in these processes comes from the blockade of VLA-4/ligand interactions, resulting in an inhibition of leukocyte recruitment.
Functional epitope mapping of the ␣4 subunit with a wide panel of anti-␣4 mAb revealed the existence of three topographically distinct epitopes (39). Epitope A anti-␣4 mAb were able to induce homotypic aggregation, blocked partially adhesion to the CS-1-containing FN-40 fibronectin fragment and did not inhibit adhesion to VCAM-1. Epitope B mAb were subdivided into B1 and B2, both blocking adhesion to FN-40 and VCAM-1, with the difference that B2 mAb also triggered homotypic cell aggregation, while B1 mAb did not (39). The only effect so far described for epitope C mAb is the blocking of homotypic aggregation (39).
A precise localization of the ␣4 residues involved in VLA-4mediated functions will contribute to the understanding of the interactions between VLA-4 and its ligands and could help in the designing of compounds aimed at blocking this interaction during unwanted inflammatory processes. In the present work we have analyzed the effect of amino acid substitutions at selected positions in the amino-terminal end of the ␣4 integrin subunit, in the adhesion of K562 cells expressing transfected mutant ␣4 to fibronectin and VCAM-1, as well as in homotypic cell aggregation.
Cell Surface Iodination and Immunoprecipitation-K562 cells were surface-labeled with Na[ 125 I] (Amersham Corp., UK) and solubilized as previously reported (27). For immunoprecipitation, the supernatants were precleared with protein A-Sepharose beads, followed by incubation at 4°C with antibodies. The immunocomplexes were harvested by incubation with protein A-Sepharose beads, boiled, and analyzed by SDS-PAGE using 7% polyacrylamide gels and nonreducing conditions.
Cell Adhesion and Aggregation Assays-For cell adhesion, transfectants or K562 cells were labeled in complete medium with the fluorescent dye BCECF-AM (Molecular Probes, The Netherlands), and added in RPMI medium containing 0.4% bovine serum albumin to 96-well dishes (Costar) (5 ϫ 10 4 cells/well) previously coated with FN-40 or sVCAM-1. After incubation for 20 min at 37°C, unbound cells were removed by three washes with RPMI medium, and adhered cells were quantified using a fluorescence analyzer (CytoFluor 2300, Millipore Co.). Homotypic cell aggregation assays were performed essentially as previously described (50). Briefly, 10 5 cells in complete medium were incubated with 1/20 final dilution of culture supernatant mAb, and the degree of cell aggregation was measured at 3, 7, and 20 h in a semiquantitative manner using the method described by Rothlein and Springer (51).

Expression of Mutated ␣4
Subunits-Analyses of hydrophilicity profiles according to Kyte and Doolitle (52) and surface probability according to Emini (see Kyte and Doolitle (52)), indicated that within the first 200 amino acids in the aminoterminal end of the integrin ␣4 subunit there are several regions ranging from 4 to 10 amino acids long showing high indexes of surface probability (Fig. 1A). Residues Arg 89 -Asp 90 are contained inside one of these regions and show the highest surface probability index of this amino-terminal end. Other amino acid stretches displaying high indexes of surface expo-sure include amino acids Gln 101 , Pro 102 , Asp 138 , and Leu 139 (Fig. 1A). Therefore, we performed site-directed mutagenesis at residues Arg 89 -Asp 90 , Gln 101 -Pro 102 , and Asp 138 -Leu 139 , according to the scheme shown on Fig. 1B. We also made single mutations at residues Ile 108 and Gly 130 , which exhibit low surface exposure indexes (Fig. 1, A and B). The mutated, as well as wild type, full-length ␣4 cDNAs in the expression vector pFNeo were transfected into VLA-4-negative K562 cells, and the resulting geneticin-resistant transfectants were designated as shown in Fig. 1B.
We next surface iodinated wild type and mutant ␣4 transfectants, and after solubilization the cell extracts were immunoprecipitated with anti-␣4 antibodies representing epitopes A, B1, and C, as well as with anti-␤1 antibodies, followed by SDS-PAGE. The 4M7, QP(101-102)HL, and I108M cells showed a characteristic pattern of ␣4 structural forms in ␣4-K562 transfectants (27). Thus, most of ␣4 is expressed at the cell surface as a cleaved ␣ 4/80,70 form, with little expression of the uncleaved ␣ 4/150 form (Fig. 3). No differences in terms of ␣4 structure and expression levels with the various anti-␣4 antibodies were found amongst these cells (Fig. 3). In contrast, the anti-␣4 epitope A HP1/1 and HP1/7 immunoprecipitates from the RD(89 -90)SA transfectants showed a dramatic reduction in the amount of cell surface ␣4 subunit compared with that found in HP2/1 and B-5G10 immunoprecipitates, which were identical to their counterparts in the other three transfectants (Fig. 3). Also, anti-␤1 immunoprecipitates from RD(89 -90)SA cells did not change with respect to the anti-␤1 immunoprecipitates from 4M7, QP(101-102)HL, and I108M cells (Fig. 3). Together with the flow cytometry analyses, these results indicate that ␣4 residues Arg 89 -Asp 90 are contained within the region corresponding to ␣4 epitope A.

Absence of Homotypic Aggregation by RD(89 -90)SA Transfectants in Response to Epitope A Anti-␣4
Antibodies-To characterize the effect of the mutations in the ␣4 subunit on VLA-4-mediated functions, we first analyzed the homotypic aggregation in the various ␣4 transfectants in response to different anti-␣4 mAb recognizing distinct ␣4 epitopes (39). We found that 4M7, QP(101-102)HL, and I108M cells aggregated equally well in response to epitope A anti-␣4 mAb HP1/3 and HP1/7, as well as to the HP2/4 (epitope B2) and to the anti-␤1 mAb Lia 1/2 and Lia 1/5 (Table II and Fig. 4). However, the RD(89 -90)SA transfectants showed a severely impaired capability to aggregate in response to HP1/3 and HP1/7 mAb, while displaying high cell aggregation in response to the HP2/4 and anti-␤1 mAb (Table II and Fig. 4). These effects were evident as early as 3 h after the addition of the antibodies, and were maintained for 20 h. The epitope C anti-␣4 mAb B-5G10 inhibited the homotypic aggregation in all transfectants. On the other hand, the different transfectants, as well as the untransfected K562 cells aggregated similarly in response to the anti-CD43 mAb TP1/36 (Table II) (Fig. 5). These results indicate that the potential of the ␣4␤1 heterodimer carrying substitutions at residues Arg 89 and Asp 90 in adopting a high affinity conformation is not affected by these mutations. When we analyzed cell adhesion to sVCAM-1, no significant differences among the 4M7, RD(89 -90)SA, QP(101-102)HL, and I108M transfectants were observed, both in the absence and in the presence of TS2/16 (Fig. 6). In addition, the increment of cell adhesion to sVCAM-1 in the presence of TS2/16 was lower than that observed in the case of FN-40, and was observed mainly at lower sVCAM-1 concentrations (Fig. 6).

DISCUSSION
In the present study we show that residues Arg 89 and Asp 90 form part of the integrin ␣4 subunit epitope A. Substitutions at these residues resulted in a diminished capability of three anti-␣4 epitope A mAb (HP1/1, HP1/3, and HP1/7) to recognize the ␣4 subunit expressed by RD(89 -90)SA transfectants, which correlated with a decrease in cell adhesion to the FN-40 fragment of fibronectin and lack of homotypic cell aggregation of these transfectants. Anti-␣4 mAb specific to epitopes B1, B2, and C recognized the ␣4 molecule in the RD(89 -90)SA transfectants in a similar fashion to wild type ␣4 transfectants, and complete homotypic cell aggregation in the RD(89 -90)SA transfectants was obtained using either anti-␣4 epitope B2 or anti-␤1 mAb, indicating that the mutations did not affect the overall conformation of ␣4. These results indicate that the ␣4 residues Arg 89 and Asp 90 are involved in the interaction of VLA-4 with FN-40, as well as in homotypic cell aggregation triggered by epitope A anti-␣4 mAb. It is unlikely that substitutions at Arg 89 and Asp 90 modified nearby cysteine disulfide bonds, since these residues do not appear to be in the middle of such a bond, according to the ␣4 disulfide bond pairing disposition shown earlier (54), based in a comparison with the ␣IIb disulfide bond pairing previously assigned (55). Substitutions at residues Arg 89 and Asp 90 did not affect the adhesion of RD(89 -90)SA transfectants to VCAM-1, indicating that these residues do not appear to be essential in the interaction between VLA-4 and VCAM-1.
Interestingly, the Arg 89 ␣4 residue posses the highest surface index probability of the ␣ 4/80 proteolytic N-terminal fragment. The ␣4 proteolytic cleavage site located between residues Lys 557 -Arg 558 and Ser 559 (27), also shows a very high surface index probability, probably reflecting a surface exposure to proteases. This suggests that the Arg 89 and Asp 90 residues indeed may be on the surface of the ␣4 subunit, which could account for their functional relevance.
The diminished adhesion of RD(89 -90)SA cells to FN-40 in the absence of TS2/16 was in average 40% with respect to wild type ␣4 transfectants. This result is in accordance with a previous report showing that adhesion to FN-40 could be partially inhibited (up to 60%) by anti-␣4 epitope A mAb (39). The same studies indicated that ␣4 epitopes B1 and B2 were involved in the interaction of VLA-4 with FN-40, which could be completely blocked by a panel of anti-␣4 mAb. Using ␣4 murine/human chimeric constructs expressed in mammalian cells, it has recently been shown that epitope A mAb mapped to the most ␣4 N-terminal 100 amino acids in one case (56), whereas another report found that residues 1-38 contained the epitope A (57). Moreover, they reported that B1 and B2 epitopes mapped to ␣4 residues 152-203 (56), while the other obtained different sites for the B1 (␣4 195-268) and B2 (␣4 108 -182) epitopes (57). In spite of these differences which could be solved by performing point mutations, it is clear that a region C-terminal to residue 108 contains the main site of ␣4 involved in the interaction with the CS-1 domain of fibronectin, and that, as demonstrated in this report, additional binding sites are found in a region which includes residues Arg 89 and Asp 90 . Our data confirm that these residues do not form part of the B1 and B2 epitopes. The possibility of the presence of several sites in ␣4 capable to interact with fibronectin could result in a potential strengthening of the VLA-4-mediated adhesion. It is conceivable that ␣4 residues Arg 89 and Asp 90 might form part of an ␣4 domain interacting with another site contained inside the FN-40 fragment of fibronectin, such as the H1 site (58), and that perhaps these interactions might be differentially regulated.
The finding that ␣4 residues Arg 89 and Asp 90 are included in the region promoting homotypic cell aggregation agrees with the ␣4 epitope mapping studies mentioned earlier (56). Homotypic cell interaction mediated by VLA-4 might play a relevant role in processes such as extracellular matrix invasion by melanoma cells during metastasis, where homotypic interactions between these cells results in a reduction of invasion (20). In this regard, it has been reported that the ␣4 integrin chain can directly interact with ␣4␤1 and ␣4␤7 (59), suggesting that cells could use the ␣4 integrins to interact in a homotypic manner. The results from the present work do not distinguish between a role of Arg 89 and Asp 90 ␣4 residues in the initial steps leading to cell aggregation, or as part of the ␣4 molecule directly interacting with other ␣4 integrins or with unknown ligands.
Residues Gln 101 and Pro 102 are also included in an ␣4 region with high surface probability index. However, substitutions at these residues did not affect binding of any anti-␣4 mAb to the ␣4 subunit expressed by the QP(101-102)HL transfectants, nor altered the adhesive functions of these cells. Transfectants expressing the ␣4 subunit with mutations at residue Gly 130 showed only minor reactivity with anti-␣4 epitope C mAb, while no stable transfectants were obtained with mutations at residues Asp 138 and Leu 139 . Both Asp 138 and Gly 130 residues are highly conserved among integrin ␣ subunits, which could suggest that alteration at this region of the ␣4 subunit might result in an improper folding of the molecule and/or an inability to associate with the ␤1 subunit.
Previous studies have shown that mutations at the ␣4 putative divalent cation sites (included in the region between residues 281 and 414) resulted in a reduction of VLA-4 interaction with both CS-1/fibronectin and VCAM-1 (60). Our results show that a region included within the first N-terminal 100 amino acids of the integrin ␣4 subunit participates in certain VLA-4mediated cell adhesion functions, namely cell aggregation and binding to fibronectin. The ␣4 residues involved in the interaction of VLA-4 with VCAM-1, as well as those residues forming part of the epitope B1 and B2 involved in adhesion to CS-1, remain to be identified. The results from the present study provide an insight into the integrin ␣4 regions involved in the interaction of VLA-4 with its ligands. This information could be useful in future designing of products aimed at interrupting these interactions in several inflammatory and autoimmune pathologies.