Critical residues for ligand binding in an I domain-like structure of the integrin beta1 subunit.

Several integrin α subunits have an inserted sequence of about 200 residues (the I or A domain) that is critical for ligand interactions. The presence of an I domain-like structure within the integrin β subunit has been proposed based on the similarity of the hydropathy profiles and the homology of sequences between the α and β subunits. This study was designed to determine whether the region of the β1 subunit that includes residues 101-335 has the characteristics of an I domain. We found novel critical residues for ligand binding (Ser-132, Asn-224, Asp-226, Glu-229, Asp-233, Asp-267, and Asp-295, in addition to the previously reported Asp-130) using site-directed mutagenesis. The critical residues for ligand binding are located in several of loop structures of the region (or in a potential loop between an α helix and a β strand), which have been predicted using multiple secondary structure prediction methods. The data suggest that the β subunit has multiple disrupted critical oxygenated residues for ligand binding similar to those found in the α I domain.

The integrins are a family of ␣/␤ heterodimers of cell adhesion receptors that mediate cell-extracellular matrix and cellcell interactions (1)(2)(3)(4)(5). Several ␣ subunits (␣1, ␣2, ␣L, ␣M, ␣X, and ␣E) have an inserted sequence of about 200 amino acid residues (the I or A domain) that is located close to three cation binding sites (6 -14). These I domain-like structures are found in a number of other proteins including von Willebrand factor (15), type VI collagen (16,17), complement factors B and C2 (18,19), and cartilage matrix proteins (20). There is growing evidence that the I domains of ␣1, ␣2, ␣L, and ␣M are important in ligand binding and receptor activation (21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31). Conserved oxygenated residues Asp-140, Thr-209, and Asp-242 of the ␣M I domain and the corresponding residues of ␣L and ␣2 are critically involved in ligand binding (30 -32). The crystal structures of the ␣M I domain (33) and the ␣L I domain (34) show that the domain adopts a classic ␣/␤ "Rossmann" fold and contains an unusual Mg 2ϩ coordination site on its surface (the metal ion-dependent adhesion site "MIDAS"). Interestingly, Asp-140, Asp-242, and Thr-209 of ␣M, as well as the corresponding residues of ␣L, are all involved in the coordination of Mg 2ϩ at MIDAS, underscoring the importance of these residues in ligand binding.
Lee et al. (33) suggest the presence of an I domain-like structure within the ␤ subunit based on the similarity in hydropathy profiles between the I domain and part of the ␤ subunit. There is sequence homology between a region of the ␤ subunit (around Asp-130 in ␤1) and a region of the I domain (around Asp-140 in the ␣M I domain) (35), suggesting the possibility that Asp-130 of the putative ␤1-I domain corresponds to Asp-140 of the ␣M I domain. Asp-130 of ␤1 (36,37) and the corresponding Asp residues of ␤2 (38), ␤3 (39), and ␤6 (40) have been shown to be critical for ligand binding to integrins. However, the presence of the I domain-like structure in the ␤ subunit has not been proven. The present study was designed to determine whether the region spanning residues 101 to 335 of the ␤ subunit has disrupted critical residues for ligand binding. By introducing multiple mutations of oxygenated residues conserved among ␤ subunits into ␤1, we identified novel critical residues for ligand binding: Ser-132, Asn-224, Asp-226, Glu-229, Asp-233, Asp-267, and Asp-295, in addition to the previously reported Asp-130. The critical residues of ␤1 were located in the predicted loop structures of the domain. The data suggest that the putative ␤ I domain has multiple disrupted oxygenated residues critical for ligand binding in several predicted loop structures. FN 110K Fragment-Sepharose Affinity Chromatography-Cells were harvested with 3.5 mM EDTA in PBS (10 mM phosphate buffer and 0.15 M NaCl, pH. 7.4) and washed with PBS. Cells (about 5 ϫ 10 6 ) were then surface labeled with 125 I by using iodogen (Pierce) (46), washed three times with PBS, and solubilized in 0.4 ml of 100 mM octylglucoside in 10 mM Tris-HCl/0.15 M NaCl (TBS), 1 mM MnCl 2 , 1 mM phenylmethylsulfonyl fluoride (Sigma), pH 7.4 at 4°C for 1 h. The insoluble materials were removed by centrifugation at 15,000 ϫ g for 10 min. The supernatant was then incubated with a small amount of underivatized Sepharose 4B at 4°C for 1 h to remove nonspecific binding material. The supernatant was incubated at 4°C overnight with 1 ml of packed FN 110K fragment-Sepharose that had been equilibrated with TBS containing 1 mM MnCl 2 , 1 mM phenylmethylsulfonyl fluoride, and 25 mM octylglucoside, pH 7.4 (washing buffer). The unbound materials were washed with 20 ml of washing buffer, and the bound materials were eluted with 20 mM EDTA instead of 1 mM MnCl 2 in washing buffer; then 2.5-ml fractions were collected. Aliquots (0.25 ml) of the peak fraction (2.5 ml) of the eluted radioactivity were used for immunoprecipitation. The samples were precleared by incubating with * This work was supported by National Institutes of Health Grants GM47157 and GM49899. This is publication 9766-VB from The Scripps Research Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. protein G-Sepharose and then were immunoprecipitated with polyclonal antibody 172 (specific to human ␤1) or mAb 7E2 (specific to hamster ␤1). Immunoprecipitated material was analyzed by SDSpolyacrylamide gel electrophoresis with 7% gel under nonreducing conditions.

Antibodies
Binding of Fluorescein Isothiocyanate (FITC)-labeled FN to Cells expressing Human ␤1 Mutants-FN was labeled with FITC essentially as described by Rinderknecht (47). Briefly, FN (0.5 mg) and 1.5 mg/ml FITC isomer I on celite (Sigma) in 1 ml 0.1 M NaHCO 3 , pH 9.3, were incubated for 1 h at room temperature in the dark. Free Prediction of Secondary Structure of the ␤ Subunits using Multiple Prediction Methods-Predictions of secondary structures were generated for human ␤1-8 (49 -60) and Drosophila ␤ (Myospheroid) (61) using the Gibrat method (62), the Levin method (63), the DPM method (64), the SOPMA method (65,66), and the PhD method (67). The consensus of predicted secondary structures for each ␤ subunit was aligned according to the sequence alignment of the 9 ␤ subunits (60), and consensus predictions were calculated for each position again.
Other Methods-Site-directed mutagenesis of the ␤1 cDNA in a pBJ-1 vector (68, 69) was carried out using unique restriction site elimination (70). The presence of mutations was confirmed by DNA sequencing. Transfection into CHO cells by electroporation, selection of the transfected cells with G418, immunoprecipitation, and flow cytometric analyses were carried out as described by Takada and Puzon (71).

RESULTS
Alanine-scanning Mutagenesis of the ␤1 Subunit-To identify critical residues of ␤1 for ligand binding, we introduced mutations into Ser-132, Thr-157, Ser-198, Thr-206, Asn-207,  (60). ‫,ء‬ the residues whose functions in ligand binding were examined in this study. The Asp-259 to Ala mutation was omitted because the expression level of the mutant was not high enough to allow us to obtain reliable data about its ligand binding function. Predictions (Pred.) of secondary structures were generated using the Gibrat method (62), the Levin method (63), the DPM method (64), the SOPMA method (65,66), and the PhD method (67). H, helix; E, sheet; C, coil; T, turn. Consensus of the predictions was calculated as described under "Materials and Methods." Gln-219, Ser-222, Asn-224, Asp-226, Glu-229, Asp-233, Asn-249, Asp-267, Asp-278, Asp-295, and Asn-310, which are conserved oxygenated residues within the region spanning residues 101-335, except for Ser-198 and Asn-207, which served as controls (Fig. 1). This region was chosen because: 1) the region around Asp-130 of ␤1 has homology to the region of Asp-140 of the ␣M I domain (35); 2) the region of ␤3 corresponding to residues 107-326 of ␤1 contains 4 exons (exons C-F or III-VI) (72,73); and 3) the size of the ␣ I domain is about 200 amino acid residues. Mutant ␤1 cDNAs were transfected into CHO cells together with a neomycin-resistant gene. After selection with G418, 20 -50% of the cells stably expressed mutant ␤1. Using cells stably expressing human ␤1 mutants, we examined the ligand binding function of ␤1 mutant/hamster ␣5 by FN 110K affinity chromatography. Since CHO cells express endogenous wild-type ␤1, human ␤1 was detected by immunoprecipitation of the materials eluted from FN 110K-Sepharose using an antibody specific to human ␤1. Endogenous hamster ␣5␤1 was used as a positive control. We detected human ␤1 in the eluate with T157A (Thr-157 to Ala), S198A, T206A, N207A, Q219A, S222A, N249A, D278A, and N310A ␤1 mutants, which are associated with endogenous hamster ␣5. We did not, however, detect human ␤1 in the eluate with the S132A, N224A, D226A, E229A, D233A, D267A, and D295 ␤1 mutants ( Fig.  2A). Cells expressing these negative ␤1 mutants were cloned by cell sorting to obtain high expressors. We repeated the experiments with cloned cells expressing the negative ␤1 mutants at a high level. Essentially the same results were obtained: the S132A, N224A, D226A, E229A, D233A, D267A, and D295 ␤1 mutants did not bind to FN 110K-Sepharose ( Fig. 2A). The absence of human ␤1 in the eluate with the negative ␤1 mutants was confirmed by overexposure of the gels (data not shown).
Immunoprecipitation with anti-␤1 antibodies of a lysate of surface 125 I-labeled cells expressing these negative ␤1 mutants detected both human and hamster ␤1 (M r 110,000) (Fig. 2B). Since ␣5␤1 is a major ␤1 integrin on CHO cells, mutant ␤1 mainly associates with endogenous ␣5. In some cases (the D233A, D267A, and D295A mutants), we did not detect associated endogenous ␣ subunit under the detergent conditions used (1% Triton X-100 and 0.05% Tween 20), suggesting that some mutations induce adverse effects on the ␣/␤ association of the ␤1 subunit.
Binding of FITC-labeled FN to ␤1 Mutants-Ligand binding function of the S132A, N224A, D226A, E229A, D233A, D267A, and D295 ␤1 mutants was further examined using a soluble FN binding assay (Fig. 3A). The binding of FITC-labeled FN to cells expressing these ␤1 mutants was measured in the presence of activating (8A2) or inhibiting (4B4) mAbs that are specific to human ␤1. The difference in binding in the presence of 8A2 and in the presence of 4B4 reflects binding to human ␤1/␣5 but not to hamster ␤1/␣5. Significant binding of FITC-labeled FN specific to human ␤1/␣5 was observed with cells expressing wildtype human ␤1 (Fig. 3A) but not with parent CHO cells or cells expressing the previously reported dominant-negative D130A ␤1 mutant (37) as a negative control. No detectable FN binding to human ␤1/␣5 was observed in cells expressing these ␤1 mutants, which is consistent with the data acquired by FN 110K affinity chromatography. Cells expressing the T157A and D278A ␤1 mutants, as positive controls, showed significant FN binding function, in contrast to cells expressing the N224A, D226A, D233A, D267A, or D295A ␤1 mutants, which have similar expression levels to cells expressing the T157A or D278A mutants (Fig. 3B). The addition of Mg 2ϩ (10 mM) to the assay medium did not have any effect on the binding of FN to FIG. 2. Effects of ␤1 mutations on binding of ␣5␤1 to immobilized FN 110K fragment. A, immunoprecipitation of eluates from FN 110K fragment-Sepharose affinity chromatography. Lysates of surface 125 I-labeled CHO cells expressing ␤1 mutants were incubated with FN 110K-Sepharose, and bound materials were eluted with 10 mM EDTA in buffer. Eluates (ELU) were immunoprecipitated with polyclonal antibody 172 (specific to human ␤1), control serum (C), or mAb 7E2 (specific to hamster ␤1). Immunoprecipitated materials were analyzed by SDSpolyacrylamide gel electrophoresis. ϩ or Ϫ, the presence or absence, respectively, of significant amounts of protein bands precipitable with antibody 172 (human ␤1) in the eluates from FN 110K-Sepharose. Cells homogeneously expressing human ␤1 were used for wild-type, S132A, N224A, D226A, E229A, D233A, D267A, D278A, and D295A ␤1. Expression of ␤1 (%) with mAb A1A5 using FACScan (Beckton-Dickinson) is 13 (T157A), 44 (S198A), 32 (T206A), 41 (N207A), 32 (Q219A), 33 (S222A), 25 (N249A), and 29 (N310A), respectively. The background level is 2-3% using A1A5 and parent CHO cells. Mean fluorescent intensities for human ␤1 mutants on the cloned CHO cells are shown in Fig. 3. We are able to see a faint band precipitable with 172 in the eluate from FN 110K-Sepharose using E229A mutant upon overexposure. This is not significant, however, if we consider the very high expression level of this mutant (Figs. 2B and 3). B, immunoprecipitation of human ␤1 mutants from whole-cell lysates (before FN 110K affinity chromatography). Lysates from surface 125 I-labeled cells homogeneously expressing human ␤1 mutants that do not show binding to FN 110K fragments were immunoprecipitated with antibody 172 (human ␤1), control serum (C), or mAb 7E2 (hamster ␤1). Triton X-100 (1%) and Tween 20 (0.05%) were used throughout. Immunoprecipitated materials were analyzed in SDSpolyacrylamide gel electrophoresis (7% gels) under nonreducing conditions. the mutants (data not shown), suggesting that the high Mg 2ϩ concentration cannot overcome the effects of the mutations. (The effect of the high Ca 2ϩ concentration was not measured because Ca 2ϩ (10 mM) induced high background binding of FN to parent CHO cells.) Cells expressing still-positive mutants (the S198A, T206A, N207A, Q219A, S222A, N249A, and N310A ␤1 mutants) showed significant binding to FITC-FN under the same assay conditions (data not shown), although calculation of levels of FN binding relative to levels of human ␤1 expression is difficult because expression of these mutants is not homogeneous (cells are not clonal). DISCUSSION The present study establishes seven novel critical residues (Ser-132, Asn-224, Asp-226, Glu-229, Asp-233, Asp-267, and Asp-295) of ␤1 for ligand binding using site-directed mutagenesis. These mutations block binding of ␣5␤1 to immobilized FN 110K fragments and soluble FN. The secondary structure prediction of the putative ␤ I domain using multiple prediction methods suggested the presence of multiple ␤ strands and ␣ helices (Fig. 1). Interestingly, the critical residues of the ␤1 subunit for ligand binding are located in several separate predicted loop structures of the region (or in a potential turn between a predicted ␣ helix and a predicted ␤ strand), which may constitute multiple ligand binding sites on the ␤ subunit. Although we have yet to perform detailed functional analysis of each critical residue, we suspect they might be directly involved in interaction with ligands and/or cations. Although some mutations (the D233A, D267A, and D295A) induce adverse effects on ␣/␤ association on immunoprecipitation under the detergent conditions used, more detailed studies will be required to determine whether the effects of these substitutions are due to weak ␣/␤ association.
The importance in ligand binding of the Asp-130 (36, 37) and Ser-132 (in this study) residues in ␤1 and the corresponding residues in ␤2, ␤3, and ␤6 has been shown (36,39,40,74). D'Souza et al. (75) recently reported that a synthetic peptide from the corresponding region of ␤3 (MDLSYSMKDDLWSI, residues 118 -131) produced a ternary complex with cations and a ligand. Also, recently the sequence DDLW (residues 126 -129 of ␤3) was shown to be critical for interaction with the RGD sequence, using a phage display system (76). Therefore, the region containing the conserved Asp-130 and Ser-132 residues of ␤1 may be involved in the ligand binding of integrins. The critical Asp-130 and Ser-132 residues of ␤1 are located between the predicted ␤-strand ␤B and ␣-helix ␣A. There is homology between the amino acid sequence surrounding Asp-130 of ␤1 and that surrounding Asp-140 of ␣M (35). This suggests that the Asp-130 and Ser-132 residues of ␤1 may be in a topologically identical position to the critical Asp-140 and Ser-142 of ␣M (33).
This study has shown that residues Asn-224, Asp-226, Glu-229, and Asp-233 of the ␤1 subunit, which are critical for ligand binding, are clustered in a large predicted loop structure (residues 222-231) between the predicted ␣B and ␣C helices. A synthetic peptide of ␤3, DAPEGGFDAIMQATV (residues 217-231 of ␤3, corresponding to residues 226 -240 of ␤1), has been shown to bind to immobilized fibrinogen, von Willebrand factor, and fibronectin (77,78). A synthetic peptide of ␤3, SVS-RNRDAPEG (residues 211-221 of ␤3), has been reported to block binding of fibrinogen to ␣IIb␤3 (79,80). These findings suggest that the predicted loop structure may be a major ligand binding site on the ␤ subunit. It is possible that the region containing Asn-224, Asp-226, Glu-229, and Asp-233 corresponds to that of the I domain containing Thr residues (Thr-221 of ␣2, Thr-209 of ␣M, and Thr-206 of ␣L) critical for ligand FIG. 3. Binding of FITC-labeled FN to cells expressing human ␤1 mutants. In A, cloned CHO cells expressing mutant human ␤1 were first incubated with mAbs 4B4 or 8A2 in Dulbecco's modified Eagle's medium at the saturating concentration (1000 times dilution of ascites) for 30 min on ice. After washing once with PBS, cells were incubated with FITC-FN (at a final concentration of 25 g/ml) in PBS for 30 min on ice. After washing once with PBS, cells were suspended in PBS and analyzed using FACScan. OO, ϩ 8A2 (activating anti human ␤1); (----), ϩ 4B4 (inhibiting anti human ␤1). FN binding without any antibody (⅐⅐⅐⅐) is almost identical to that with 4B4 in each case. The data suggest that FN binds to wild-type ␣5␤1 but not to the N224A mutant ␤1/␣5 in the presence of 8A2. All of the mutant ␤1 reacted with 8A2 or 4B4, as detected by flow cytometric analysis (data not shown). In B, FITC-FN binding was determined from the difference between median fluorescence intensities with 8A2 and those with 4B4. Mean fluorescent intensity of human ␤1 was determined with mAb A1A5. The data indicated that FITC-FN does not bind to the S132A, N224A, D226A, E229A, D233A, D267A, and D295A ␤1 mutants in the presence of 8A2. The dominant-negative D130A mutant (37) was included as a negative control. Cloned cells expressing the T157A and D278A ␤1 mutants work as positive controls. Uncloned cells expressing the other ␤1 mutants (the S198A, T206A, N207A, Q219A, S222A, N249A, and N310A mutants) showed significant binding to FITC-FN under the same assay conditions (data not shown). This is consistent with the data obtained using FN 110K fragment affinity chromatography.
binding to the I domain (30,31). It will be interesting to learn whether the residues of other integrins corresponding to Asn-224, Asp-226, Glu-229, and Asp-233 of ␤1 are critical for ligand binding.
Mutation of Asp-242 in ␣M blocks ligand binding and divalent cation binding to the ␣M I domain (32). Also, Asp residues of the ␣2 or ␣L I domains corresponding to Asp-242 in ␣M are critical for ligand binding to ␣2␤1 or ␣L␤2 (25, 31). The predicted loop structure containing Asp-267 of ␤1, which may correspond to that containing Asp-242 of the ␣M I domain, is critically involved in ligand binding. Although there is another conserved Asp residue at position 259 in the same predicted loop structure of ␤1, we could not show whether Asp-259 is critical for ligand binding, since the expression level of the D259A mutant was not high enough to obtain reliable data on ligand binding (data not shown). We identified another critical residue, Asp-295 of ␤1, which is located in a large loop structure between predicted ␤G and ␣D. However, there is no report that the corresponding oxygenated residues of the I domain in the ␣ subunit are involved in ligand binding functions. It will be interesting to determine whether mutations of the ␣ I domain corresponding to Asp-295 of ␤1 affect the function of the I domain.
We previously identified a small region of ␤1 (residues 207-218, a regulatory epitope) that is recognized by both activating (e.g. 8A2, TS2/16) and inhibiting (e.g. 4B4, AIIB2) anti-␤1 antibodies (71). This region is located in the predicted loop between ␤-strand ␤D and ␣-helix ␣B of the region. These activating and inhibiting anti-␤1 antibodies probably induce high or low affinity states, respectively, by changing the conformation of the ␤1 subunit (71). Based on these observations, we speculate that the regulatory epitope might be located in the nonligand binding site of the region.
In summary, we identified seven additional critical residues for ligand binding within the predicted ␣/␤ structure of the putative I domain in the ␤1 subunit. The putative ␤ I domain may be structurally and functionally homologous to the I domain of the ␣ subunit. The predicted ␣/␤ structure and disrupted critical residues for ligand binding constitute multiple binding sites for ligands or cations. The present information will be useful for future structural and functional studies using functional recombinant fragments. Studies in this direction are in progress.