Identifying the Putative Metal Ion-dependent Adhesion Site in the (cid:98) 2 (CD18) Subunit Required for (cid:97) L (cid:98) 2 and (cid:97) M (cid:98) 2 Ligand Interactions*

We have previously demonstrated that Asp 134 and Ser 136 of the (cid:98) 2 subunit are essential for (cid:97) L (cid:98) 2 and (cid:97) M (cid:98) 2 ligand recognition. It has been proposed that these residues may be part of a metal ion-dependent adhesion site (MIDAS) within the (cid:98) subunit homologous to the (cid:97) M I domain MIDAS structure (Lee, J.-O., Rieu, P., Arnaout, M. A., and Liddington, R. (1995) Cell 80, 631–638). In the present study, we evaluated the role of additional can- didate metal ion-coordinating residues in the (cid:98) 2 subunit in ligand interactions. Cells bearing the recombinant (cid:97) L (cid:98) 2 or (cid:97) M (cid:98) 2 mutant(s) were tested for the ability to bind to immobilized ligands. Alanine substitution at Asp 232 in (cid:98) 2 produced a complete loss in the capacity of both (cid:97) L (cid:98) 2 and (cid:97) M (cid:98) 2 to support cell adhesion and sup- pressed the expression of a divalent cation-dependent conformation recognized by mAb 24. Alanine substitu- tion at Glu 235 differentially affected receptor function dependent upon the co-transfected (cid:97) subunit. Cells expressing (cid:97) L (cid:98) 2 with a substitution at Glu 235 failed to adhere to intercellular adhesion molecule 1 (ICAM-1) but did retain the capacity to bind mAb 24.

Integrins are a family of structurally and functionally related adhesion receptors that participate in cell-cell and cellextracellular matrix interaction in a diverse range of biological functions (1). The ␤ 2 leukocyte integrin subfamily mediates a range of adhesive interactions essential for normal immune and inflammatory responses (2,3). The importance of the leukocyte integrins is underscored by the hereditary disease leukocyte adhesion deficiency, which results in profound immunodeficiency due to an absence of cell surface expression of the ␤ 2 integrins (4,5). The ␤ 2 leukocyte integrin subfamily consists of three heterodimeric glycoproteins consisting of a common ␤ subunit noncovalently linked to a distinct ␣ subunit, ␣ L ␤ 2 (CD11a/CD18, LFA-1), ␣ M ␤ 2 (CD11b/CD18, Mac-1), and ␣ X ␤ 2 (CD11c/CD18, p150,95) (6). More recently, a fourth member of the ␤ 2 leukocyte integrins has been identified, ␣ d ␤ 2 (CD11d/ CD18) (7). Each of the four ␣ subunits contains a 200-amino acid inserted or "I" domain (originally designated the A domain) located close to the three EF-hand-like putative divalent cation binding repeats (8 -12). The I domain is also present in three other integrin ␣ subunits, ␣ 1 (13) and ␣ 2 (14) of the ␤ 1 subfamily and ␣ E (15) of the ␤ 7 subfamily.
A number of protein ligands for each of the ␤ 2 integrins have been identified that mediate leukocyte adhesive interactions. ␣ L ␤ 2 , which is expressed on all leukocytes, binds to three distinct members of the immunoglobulin superfamily: ICAM-1 1 (16), ICAM-2 (17), and ICAM-3 (18). ␣ M ␤ 2 , whose expression is more restricted to predominantly myeloid cells and natural killer cells, has been demonstrated to interact with several cell surface and soluble ligands including complement C3 fragment iC3b (19,20), ICAM-1 (21), fibrinogen (22), and factor X (23). Immunological, biochemical, and mutational approaches have identified the presence of multiple ligand contact points in both the ␣ and ␤ subunits (reviewed in Ref. 24). Current indications are that these multiple sites in both subunits cooperate in the recognition of ligands.
A ligand and cation binding site has been identified in a discrete region in the integrin ␤ 3 subunit (25)(26)(27). Alanine substitutions of highly conserved oxygenated residues within this discrete region, Asp 119 , Ser 121 , and Ser 123 , resulted in deficits in ligand binding of ␣ IIb ␤ 3 (25). This region is highly conserved among the ␤ subunits, suggesting it may play a role in a common feature of ligand recognition by the integrins. This hypothesis is supported by the demonstration that substitution of the corresponding Asp 119 in ␤ 1 (Asp 130 ) abolished ligand recognition of ␣ 5 ␤ 1 (28). Recently, we have demonstrated that Asp 134 and Ser 136 of the ␤ 2 subunit, homologous to Asp 119 and Ser 121 in ␤ 3 , are essential for ligand recognition of ␣ L ␤ 2 and ␣ M ␤ 2 (29). Alignment of the conserved sequences of the ␤ subunits illustrated the presence of a similar DXSXS motif in the I domains of the ␣ subunits (25). Mutations of the highly conserved residues Asp 140 , Ser 142 , Ser 144 , or Asp 140 /Ser 142 in ␣ M I domain abolished ␣ M ␤ 2 cation-dependent ligand recognition of iC3b (30,31). These data have led to the proposal that the I domain and the ␤ subunit ligand interactive site share a common structural motif essential for integrin receptor function (25). Recent crystallization of the ␣ M I domain has identified this motif as part of a three-dimensional metal binding site coordinated by Asp 140 , Ser 142 , Ser 144 , Thr 209 , and Asp 242 (32). This structure is referred to as the metal ion-dependent adhesion site (MIDAS) and appears to be critical for binding divalent cations as well as cation-dependent ligand recognition. Mutations at any of these residues in ␣ M I domain (30,31,33), residues homologous to Asp 242 in ␣ 1 ␤ 1 (34), and residues homologous to Asp 140 , Thr 209 , and Asp 242 in ␣ 2 ␤ 1 (35,36) and ␣ L ␤ 2 (33, 37) abolish cation-dependent ligand binding. Hydro-* 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  phobicity comparison between the ␣ I domains and ␤ subunits suggest that a similar MIDAS motif within a three-dimensional fold may exist in the integrin ␤ subunits (32). Moreover, Lee et al. (32) proposed that the previously identified ␤ subunit DXSXS sequences (25,29) were part of the ␤ subunit MIDAS structure.
In the present study, we attempted to determine other potential metal ion-coordinating residues within the putative I domain of ␤ 2 (Fig. 1). Candidate residues were identified through alignment of either the ␣ M I domain with the ␤ 2 putative I domain or all eight ␤ subunits using a computer program from the Genetics Computer Group, Inc. (GCG). Additional residues within the ␤ 2 region Ser 228 -Glu 235 were selected through alignment with a previously implicated ligandinteractive site in ␤ 3 (41,42). Candidate residues were evaluated for their role in cation-dependent ligand recognition of recombinant ␣ L ␤ 2 and ␣ M ␤ 2 . Our results indicate that Asp 232 and Glu 235 play an integral role in the regulation of ligand and cation binding function of both ␣ L ␤ 2 and ␣ M ␤ 2 . However, the effects of Glu 235 on integrin receptor function, which is influenced by the co-transfected ␣ subunit, may be indirect due to its proximity to the cation/ligand interactive site. We hypothesize that these residues, along with Asp 134 and Ser 136 , are part of the putative MIDAS structure of the ␤ 2 subunit.
Mutagenesis and Transfection-The full-length wild-type ␤ 2 cDNA (39) cloned into Bluescript (Stratagene, La Jolla, CA) was used to introduce nucleotide base substitutions by site-directed mutagenesis (48). Mutant clones were confirmed by nucleotide sequencing of the mutated region. A 2.8-kilobase insert containing portions of the 3Ј-and 5Ј-untranslated sequences was isolated by digestion with HindIII and NotI and subcloned into the HindIII and NotI sites of the expression vector pcDNA1/neo or pcDNA3 (Invitrogen, San Diego, CA). Mutant clones were sequenced to verify the absence of any other substitutions. COS-7 cells (a monkey kidney fibroblastoid cell line from ATCC) were transfected by electroporation with the wild-type ␤ 2 or the mutant ␤ 2 constructs with the wild-type ␣ L subunit (12) or ␣ M subunit (9 -11) cDNAs subcloned into the expression vector pCDM8 (Invitrogen). Cells were evaluated for surface expression 48 h after electroporation.
Stable Cell Lines-Stably transfected Chinese hamster ovary (CHO from ATCC) cells were generated by electroporation using wild-type ␣ L (12) or ␣ M (9 -11) cloned into the expression vector pCDM8 and wildtype or mutant ␤ 2 cloned into the expression vector pcDNA1/neo. Fortyeight hours after electroporation cells were placed into selection medium containing 700 g/ml G418 (Geneticin, Life Technologies, Inc.). Resistant colonies were isolated after 2 weeks, and positive colonies were identified by flow cytometry using subunit-specific antibodies. Clonal lines were then established by single cell sorting in an Epics 753 Coulter instrument (Coulter Cytometry, Hialeah, FL). Clonal lines were then maintained in medium containing 700 g/ml G418.
Flow Cytometric Analysis-Flow cytometry was carried out as described previously (31) except as modified for detection of the mAb 24 epitope under different cation conditions (47). Briefly, cells were harvested with 3.5 mM EDTA and 0.01% L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Worthington) in Chelex-treated TBS. Cells were washed in the presence of 0.05% soybean trypsin inhibitor (Sigma) and resuspended in TBS. For cation binding experiments, cells were treated with 5 mM EDTA for 15 min at 37°C and then washed twice in Chelex-treated TBS. 50-l aliquots of cells (1 ϫ 10 7 cells/ml) were then resuspended in TBS containing purified mAb 24 (20 g/ml) diluted in the appropriate cation condition (1 mM CaCl 2 , 1 mM MgCl 2 , 1 mM CaCl 2 /MgCl 2 , or 0.5 mM MnCl 2 ). Following incubation for 30 min at 37°C, cells were pelleted by centrifugation and washed with 100 l of TBS containing the appropriate divalent cations. Cells were pelleted by centrifugation and resuspended in 50 l of fluoresceinconjugated goat F(ab)Ј 2 anti-mouse immunoglobulin (Biosource International, Camarillo, CA) for 30 min at 37°C. Cells were pelleted by centrifugation and resuspended in TBS. Antibody binding to the transfected cell lines was analyzed by flow cytometry on a FACScan (Beckman Instruments).
Statistics-Data are represented as mean Ϯ S.D. Comparisons between multiple groups were performed with one way analysis of variance followed by Bonferroni t tests. p Ͻ 0.05 was considered significant.  (32). Boxed residues were mutated to alanine. Asterisks indicate previously identified residues in ␤ 1 , ␤ 2 , and ␤ 3 critical for ligand recognition (25,26,28,29). The filled box represents residues 211-222 of ␤ 3 previously implicated in ligand binding (41).

Construction and Expression of Recombinant
Thr 212 , Ser 228 , Gly 229 , Asp 232 , Glu 235 , Asp 265 , and Asp 266 into wild-type full-length ␤ 2 cDNA by site-directed mutagenesis ( Fig. 1). COS-7 cells were transiently co-transfected with the wild-type or mutant ␤ 2 with the wild-type ␣ L or ␣ M . Cell surface expression was analyzed by immunofluorescence flow cytometry utilizing a panel of previously characterized anti-␣and anti-␤-specific antibodies (Table I). Alanine substitutions of ␤ 2 at Ser 228 , Gly 229 , Asp 232 , or Glu 235 co-transfected with ␣ L or ␣ M were recognized by all antibodies tested, three anti-␣ L , five anti-␣ M , respectively, and three anti-␤ 2 antibodies. The antibodies bound approximately the same percentage of cells and with similar FACS profiles as the wild-type receptors ( Fig. 2 and data not shown). In addition, alanine substitution of ␤ 2 at Ser 228 , Gly 229 , Asp 232 , or Glu 235 did not have any detectable effect on ␣ L ␤ 2 or ␣ M ␤ 2 heterodimer formation as determined by immunoprecipitation of detergent-lysed surface-labeled cells. Anti-␣ antibodies immunoprecipitated both the ␣ and ␤ subunit on COS-7 cells, indicating that both subunits were associated on the cell surface (Fig. 3). These results suggest that alanine substitution of ␤ 2 at Ser 228 , Gly 229 , Asp 232 , or Glu 235 did not greatly affect ␣ L ␤ 2 or ␣ M ␤ 2 heterodimer formation or overall conformation on the cell surface.
Surface expression of alanine substitution of ␤ 2 at Thr 212 , Asp 265 , or Asp 266 co-transfected with ␣ L was not detected on COS-7 cells as determined by flow cytometry (Table I and Fig. 2). In contrast, these same substitutions of ␤ 2 were expressed on the cell surface when co-transfected with ␣ M . However, FACS analysis utilizing a panel of antibodies revealed lack of binding of several relevant antibodies to cells expressing these mutant ␣ M ␤ 2 receptors. Only one of the three anti-␤ 2 antibodies and four of the five anti-␣ M antibodies recognized the recombinant ␣ M ␤ 2 mutant receptor(s). Moreover, we observed a moderate reduction in cell surface expression of ␣ M ␤ 2 (T212A), ␣ M ␤ 2 (D265A), or ␣ M ␤ 2 (D266A) (20 -25%) bearing cells compared with the wild-type ␣ M ␤ 2 (30 -40%; Fig. 2 and data not shown). Immunoprecipitation utilizing anti-␣ M antibody of detergent-lysed surface-labeled cells demonstrated that these ␤ 2 mutants did not form heterodimers with co-transfected ␣ M at the cell surface (Fig. 3). The anti-␣ M antibody precipitated the transfected ␣ subunit only, although in one experiment the anti-␣ M antibody did precipitate ␣␤ heterodimers (data not shown). These results indicate that alanine substitution of ␤ 2 at Thr 212 , Asp 265 , or Asp 266 caused significant changes in the global secondary structure of ␣ L ␤ 2 and ␣ M ␤ 2 that resulted in the loss of surface expression or unstable conformation on the cell surface of the complex, respectively.
Interaction of Cells Bearing Recombinant Mutant(s) ␣ L ␤ 2 or ␣ M ␤ 2 with ICAM-1 or iC3b, Respectively-To determine whether these residues of ␤ 2 were involved in ␣ L ␤ 2 /ICAM-1 interaction, the capacity of the transfected COS-7 cells to adhere to microtiter wells coated with immunopurified ICAM-1 was examined. COS-7 cells expressing the recombinant wildtype ␣ L ␤ 2 bound to affinity-purified ICAM-1 immobilized on microtiter wells (Fig. 4A). Likewise, cells bearing ␣ L ␤ 2 (S228A) or ␣ L ␤ 2 (G229A) adhered to immobilized ICAM-1 in a manner similar to cells expressing the wild-type ␣ L ␤ 2 , although adhesion was slightly increased or decreased, respectively. Adhesion was dependent on ␣ L ␤ 2 , since it was completely inhibitable Cells were harvested 48 h after transfection, stained with a subunitspecific antibody followed with a fluorescein-conjugated goat F(ab)Ј 2 anti-mouse immunoglobulin, and analyzed on a FACScan flow cytometer. The log of the fluorescence intensity is shown on the abscissa and the cell number on the ordinate; 5,000 (wild-type or mutant ␣ L ␤ 2 ) or 10,000 events (wild-type or mutant ␣ M ␤ 2 ) were collected.

TABLE I
Reactivity of anti-␣ L , ␣ M , and ␤ 2 antibodies to COS-7 cells transfected with wild-type or mutant ␣ L ␤ 2 or ␣ M ␤ 2 as determined by flow cytometry COS-7 cells co-transfected with wild-type or mutant ␣ L ␤ 2 or ␣ M ␤ 2 receptors were incubated with mAbs for 30 min at room temperature. Cells were then washed and further incubated for 30 min with fluorescein isothiocyanate-conjugated goat anti-mouse F(ab)Ј 2 fragments and analyzed on a FACScan. Mock cells were transfected with the vector pCDM8 alone. ϩ, positive staining similar to that seen in Fig. 2, Ϫ, staining not greater than that seen for the mock-transfected cells. Results represent three or more separate experiments.
by a blocking antibody to the ␤ subunit (mAb 8H1; data not shown) or ICAM-1 (mAb 8.4; Fig. 4A). In contrast, alanine substitution of ␤ 2 at position Asp 232 or Glu 235 abrogated the ability of recombinant ␣ L ␤ 2 to attach to immobilized ICAM-1.

Regulation of Expression of mAb 24 Epitope to Recombinant
␣ L ␤ 2 and ␣ M ␤ 2 Mutants-Since we hypothesized that these ␤ 2 residues may be part of a metal ion-dependent adhesion site (MIDAS motif) in the ␤ subunit, we examined the expression of a cation-sensitive epitope present on all three ␤ 2 integrin ␣ subunits recognized by mAb 24 (43). Expression of the mAb 24 epitope correlates with the Mg 2ϩ -or Mn 2ϩ -occupied form of the ␤ 2 integrins (47). The mAb 24 epitope was fully expressed on cells bearing the wild-type ␣ L ␤ 2 or ␣ M ␤ 2 in the presence of 0.5 mM Mn 2ϩ , as determined by flow cytometry (Fig. 5, A and B).
Similarly, mAb 24 bound to cells expressing ␤ 2 mutation at Ser 228 or Gly 229 when co-transfected with ␣ L or ␣ M . mAb 24 binding was completely suppressed in the presence of 10 mM EDTA (data not shown). In contrast, the binding of mAb 24 was not expressed in the presence of 0.5 mM Mn 2ϩ on cells expressing ␤ 2 mutation at Asp 232 when co-transfected with ␣ L or ␣ M . Surprisingly, cells expressing the ␤ 2 mutation at Glu 235 gave different results dependent upon which ␣ subunit was co-transfected. Cells expressing ␣ L ␤ 2 (E235A) did not have any detectable effects on mAb 24 binding compared with the wild-type ␣ L ␤ 2 . In contrast, mAb 24 binding was significantly attenuated in cells expressing ␣ M ␤ 2 (E235A) compared with the wild-type receptor. In addition, mAb 24 binding was absent in cells ex- Fluorescently labeled COS-7 cells expressing either wild-type or mutant ␣ L ␤ 2 or ␣ M ␤ 2 were allowed to attach to microtiter wells coated with immunoaffinity-purified ICAM-1 (A) or iC3b (B), respectively, for 30 min at 37°C in the presence or absence of a blocking antibody. Nonadherent cells were removed, and bound cells were quantitated using a Pandex fluorescence concentration analyzer. The data are expressed as the percentage bound, where 100% equals the total number of cells bound to wells coated with the anti-␤ 2 antibody mAb 8H1 to correct for the levels of integrin expressed by the different transfectants. Bars represent the mean Ϯ S.D. of three determinations. #, p Ͻ 0.05 (control versus antibody-treated); *, p Ͻ 0.05 (wild-type control versus mutant control). not shown). These results suggest that the ␤ 2 mutation at Asp 232 abolished a divalent cation-dependent conformation of ␣ L ␤ 2 and ␣ M ␤ 2 , which correlates with suppression of ligand binding function. Moreover, the ␤ 2 mutation at Glu 235 , which abolished ligand binding function of ␣ L ␤ 2 and ␣ M ␤ 2 , differentially alters mAb 24 cation-dependent binding dependent upon the presence of the ␣ subunit.
The Effects of ␤ 2 Mutations on ␣ L ␤ 2 and ␣ M ␤ 2 Ligand Interactions following Antibody-induced Activation-To further examine the effects of the ␤ 2 mutation at Asp 232 or Glu 235 on ligand recognition, clonal stable cell lines expressing a mutant ␤ 2 subunit with the wild-type ␣ L or ␣ M were established in CHO cells as described under "Materials and Methods." As seen with the transient expression in COS-7 cells, the substitutions did not affect surface expression, since all cell lines exhibited strong immunostaining with anti-␣ or anti-␤ 2 antibodies (data not shown).
We next examined the ability of the wild-type and mutant ␣ L ␤ 2 cell lines to support cellular interaction with ICAM-1 following antibody-induced activation with mAb 2D8 (Fig. 6). Cells expressing wild-type ␣ L ␤ 2 adhere constitutively to ICAM-1. However, adhesion of the ␣ L ␤ 2 transfectant cells to immobilized ICAM-1 was enhanced by mAb 2D8. This adhesion was blocked by anti-␤ 2 mAb 8H1 (Fig. 6) or anti-ICAM-1 mAb 8.4 (data not shown). In contrast, cells expressing ␣ L ␤ 2 (D232A) or ␣ L ␤ 2 (E235A) failed to adhere to ICAM-1 in the presence or absence of mAb 2D8.
We next examined the ability of the wild-type and mutant ␣ M ␤ 2 receptors to support cellular interactions with three known macromolecular ligands: iC3b, fibrinogen, and ICAM-1 (Fig. 7). Cells expressing wild-type ␣ M ␤ 2 adhere constitutively to iC3b, and no further increase is observed in the presence of the activating mAb 2D8. In contrast, ␣ M ␤ 2 (D232A) transfectants failed to adhere to immobilized iC3b in the presence or absence of mAb 2D8, consistent with our previous results. Surprisingly, cells expressing ␣ M ␤ 2 (E235A) were able to adhere to iC3b in the presence of mAb 2D8, although adhesion was significantly lower than that observed in cells expressing the wild-type ␣ M ␤ 2 . Adhesion was specific since it was abolished by mAb 8H1 (anti-␤ 2 ).
While cells expressing wild-type ␣ M ␤ 2 do not constitutively adhere to immobilized fibrinogen or ICAM-1, activation with mAb 2D8 significantly induced adhesion to both substrates. Adhesion was dependent upon the transfected ␣ M ␤ 2 , since it was completely inhibitable by a mAb 8H1 (anti-␤ 2 ). In contrast, FIG. 6. Adhesion of CHO cells stably transfected with wildtype or mutant ␣ L ␤ 2 to ICAM-1. Fluorescently labeled cells were allowed to adhere to ICAM-1-coated wells in the absence or presence of the indicated monoclonal antibody. Cells were pretreated with antibodies for 30 min prior to addition to the ICAM-1 coated wells. Unbound cells were removed, and total fluorescence was determined on a Pandex fluorescence concentration analyzer. Data are presented as the percentage bound, where 100% is the fluorescence of cells bound to mAb 8H1 (anti-␤ 2 )-coated wells in order to normalize to the total number of receptors per cell line. Results represent the mean Ϯ S.D. of three separate experiments. #, p Ͻ 0.05 (control versus 8H1-treated); f, p Ͻ 0.05 (control versus 2D8 treated); *, p Ͻ 0.05 (wild-type control versus mutant control); ϩ, p Ͻ 0.05 (wild-type 2D8-treated versus mutant 2D8-treated). ␣ M ␤ 2 (D232A) or ␣ M ␤ 2 (E235A) transfectants failed to adhere to immobilized fibrinogen or ICAM-1 in the absence or presence of mAb 2D8. The expression of mAb 2D8 epitope on recombinant mutant ␣ L ␤ 2 and ␣ M ␤ 2 receptors was confirmed by flow cytometry (data not shown).
Expression of mAb 24 by divalent cations (Ca 2ϩ , Mg 2ϩ , Ca 2ϩ / Mg 2ϩ , or Mn 2ϩ ) to the recombinant wild-type or mutant ␣ L ␤ 2 and ␣ M ␤ 2 was also examined by flow cytometry (Fig. 8). Similar results to those reported above were obtained utilizing these stable cell lines. mAb 24 exhibited the expected defined divalent cation dependence as previously reported (43). Wild-type ␣ L ␤ 2 and ␣ M ␤ 2 transfectants stained brightly with mAb 24 in the presence of Mn 2ϩ , and to a lesser extent with Mg 2ϩ . Similarly, mAb 24 bound to cells expressing ␣ L ␤ 2 (E235A) in the presence of the same divalent cations (Mn 2ϩ or Mg 2ϩ ) as compared with the wild-type ␣ L ␤ 2 . In contrast, minimal binding of mAb 24 in the presence of Mn 2ϩ or Mg 2ϩ to cells expressing ␣ L ␤ 2 (D232A), ␣ M ␤ 2 (D232A), or ␣ M ␤ 2 (E235A) was observed. In the presence of Ca 2ϩ or EDTA, mAb 24 binding was attenuated or absent, respectively, on cells expressing the wild-type ␣ L ␤ 2 or ␣ M ␤ 2 receptors.

DISCUSSION
These studies establish the following. 1) Alanine substitution of Ser 228 , Gly 229 , Asp 232 , or Glu 235 of ␤ 2 does not affect heterodimer formation or surface expression when co-transfected with ␣ L or ␣ M . 2) Alanine substitution at Asp 232 results in the complete loss of the capacity of both ␣ L ␤ 2 and ␣ M ␤ 2 to recognize ligand(s), which correlates with the suppression of a divalent cation-dependent conformation recognized by mAb 24. 3) Alanine substitution at Glu 235 results in differential deficits in ligand recognition and mAb 24 binding when co-transfected with ␣ L or ␣ M . 4) Substitution of Ser 228 or Gly 229 to alanine does not affect ligand recognition or mAb 24 expression of ␣ L ␤ 2 or ␣ M ␤ 2 . 5) Substitution of Thr 212 , Asp 265 , or Asp 266 to alanine results in loss of surface expression or abnormal surface expression when co-transfected with ␣ L or ␣ M in COS-7 cells, respectively. These studies implicate Asp 232 and Glu 235 in the ligand binding function of ␣ L ␤ 2 and ␣ M ␤ 2 , although the effects of Glu 235 on integrin receptor function, which is influenced by the co-transfected ␣ subunit, may be indirect due to its proximity to the cation/ligand interactive site. We hypothesize that these residues, along with Asp 134 and Ser 136 , are part of the putative MIDAS like structure of the ␤ 2 subunit that is essential for cation-dependent ligand binding.
Individual alanine substitutions at Thr 212 , Asp 265 , or Asp 266 grossly affected the global secondary structure of the ␤ 2 subunit when co-transfected with ␣ L or ␣ M . This is based on the finding that these substitutions within ␤ 2 impede heterodimer formation and surface expression with the co-transfected ␣ L . In addition, these same ␤ 2 mutations when co-transfected with ␣ M significantly alter the conformation of the receptor based on the lower levels of surface expression, unstable heterodimer formation, and disruption of subunit-specific epitopes. Thus, because integrin structural changes induced by the mutations could affect functional assays, the role of Thr 212 , Asp 265 , and Asp 266 in ␤ 2 ligand recognition could not be interpreted. Previous studies have identified natural point mutations within a criti- cal region of ␤ 2 (residues 128 -361), which impair heterodimer formation and surface expression (reviewed in Ref. 51). This suggests that this region of ␤ 2 represents critical contact points required for proper ␣␤ precursor association and biosynthesis. Our results further implicate this region in leukocyte integrin heterodimer complex formation and surface expression.
In contrast, individual alanine substitution at Ser 228 , Gly 229 , Asp 232 , or Glu 232 in ␤ 2 did not affect the capacity of ␤ 2 to associate with ␣ L or ␣ M . This is based on the finding that anti-␣ antibodies co-precipitated both ␤ 2 and ␣ L or ␣ M . Furthermore, these residues did not significantly disrupt the secondary structure of the receptors based on the reactivity of a panel of anti-␣ or ␤ 2 antibodies. However, the present work suggests that residue Asp 232 of ␤ 2 is essential for cation-dependent ligand recognition of both ␣ L ␤ 2 and ␣ M ␤ 2 . This conclusion is based on 1) the failure of cells expressing ␣ L ␤ 2 to adhere to immobilized ICAM-1 and 2) the failure of cells expressing ␣ M ␤ 2 to adhere to immobilized iC3b, fibrinogen, or ICAM-1. In addition, our results suggest that Asp 232 alters the cooperative interaction between receptor-bound divalent cation and ligand binding. This conclusion is based on the decreased binding of Mn 2ϩ -induced mAb 24 to cells expressing ␣ L ␤ 2 (D232A) or ␣ M ␤ 2 (D232A). Moreover, mutations at Asp 134 or Ser 136 in ␤ 2 that we have previously implicated in ␣ L ␤ 2 and ␣ M ␤ 2 ligand binding (29) also completely abolish the binding of mAb 24 in the presence of Mn 2ϩ (data not shown).
In addition, alanine substitution of residue Glu 235 abolished ligand recognition function of both ␣ L ␤ 2 and ␣ M ␤ 2 . This conclusion is based on 1) the failure of cells expressing ␣ L ␤ 2 (E235A) to bind ICAM-1 and 2) the failure of cells expressing ␣ M ␤ 2 (E235A) to bind iC3b, fibrinogen, or ICAM-1. An unexpected finding was the ability of cells expressing ␣ M ␤ 2 (E235A) to adhere to iC3b following antibody-induced receptor activation, although not to the same levels as that observed with cells expressing the wild-type ␣ M ␤ 2 . One possible explanation is that alanine substitution at Glu 235 alters the conformation of the "resting" receptor, since cells expressing wild-type ␣ M ␤ 2 constitutively adhere to iC3b. Alanine substitution at Glu 235 may not be as important in the "activated" conformation of the receptor. Conceivably, the inability of the cells to adhere to fibrinogen or ICAM-1 following antibodyinduced activation may be due to the lower affinity of these ligands compared with iC3b for ␣ M ␤ 2 . Alternatively, alanine substitution at Glu 235 may moderately change the structure of the ligand binding pocket, consequently exposing additional and/or different interactive sites. These interactive sites may be differentially influenced by the ␣ subunit. This is further supported by the observations that cells expressing ␣ L ␤ 2 (E235A) bound mAb 24, while cells expressing ␣ M ␤ 2 (E235A) do not. Taken together, this suggests that the ␣ subunit may regulate ligand and/or cation interactions by altering the conformation of the ␤ subunit.
The principal aspect of ␣ M I domain MIDAS structure is the metal binding site coordinated directly or indirectly by certain oxygenated residues. We hypothesize that Asp 232 and Glu 235 may contribute to the cation coordination site in the ␤ 2 subunit MIDAS structure. However, we have not provided direct evidence to support this hypothesis. It is possible that these ␤ 2 mutations directly effect ligand or mAb 24 binding alone. Conceivable, a mutation of a cation binding coordinating residue could result in reduced affinity for the cation. It could be argued that higher cation concentrations could restore ligand or mAb 24 binding to the Asp 232 and Glu 235 ␤ 2 mutants. However, 4-fold higher cation conditions (4 mM Ca 2ϩ /Mg 2ϩ ) were ineffective in restoring ligand or mAb 24 binding to cells expressing these ␣ L ␤ 2 or ␣ M ␤ 2 mutants (data not shown). These mutations may have resulted in a substantial higher reduction in cation affinity, 50 -100-fold. Similar observations have been made with regard to the ␣ M I domain MIDAS mutation, D242A (30). Half-maximal binding of wild-type ␣ M ␤ 2 to iC3b was observed at approximately 10 M Mn 2ϩ . In contrast, binding of cells expressing ␣ M (D242A)␤ 2 iC3b was absent even in the presence of 1 mM Mn 2ϩ .
In summary, we have previously implicated two ␤ 2 residues (Asp 134 and Ser 136 ) as essential for cation-dependent ligand binding for both ␣ L ␤ 2 and ␣ M ␤ 2 (29). These highly conserved oxygenated residues, Asp 134 and Ser 136 , reside in the DXSXS motif, which is part of the identified metal binding site in the ␣ M I domain (MIDAS) (25,29,32). Hydrophobicity comparison suggests a similar MIDAS-like structure may exist in the integrin ␤ subunit putative I domain (32). In the present study we have identified Asp 232 and Glu 235 of ␤ 2 as critical for ligand binding function and the interactions of bound divalent cations to ␣ L ␤ 2 and ␣ M ␤ 2 . We hypothesize that the highly conserved residues Asp 134 , Ser 136 , and Asp 232 of ␤ 2 are homologous to Asp 140 , Ser 142 , and Asp 242 of ␣ M I domain and support the existence of a ␤ 2 subunit MIDAS structure essential for cationdependent ligand recognition. Furthermore, it is likely that Glu 235 is proximal to or part of the cation/ligand interactive site. Glu 235 of ␤ 2 may contribute to the sixth coordination site in the ␤ subunit MIDAS structure, stabilizing the integrin conformation and allowing access to distinct ligand binding sites within the receptor (31,52). Interestingly, in the ␣ M I domain crystal structure, an additional coordination site, Glu 314 , is provided by an adjacent I domain. Other investigators (52) have hypothesized that an oxygenated residue in the ligand, for example Glu 34 in ICAM-1, contributes to this coordination site.
The present study in association with previous studies indicates the presence of multiple ligand interactive sites in both the ␣ and ␤ subunits. Potentiality, Asp 232 and Glu 235 may contribute to the coordination of a second cation-dependent ligand-interactive site consisting of residues from both the ␣ and ␤ subunits. The mechanisms by which these multiple interactive sites participate in ligand recognition remains to be determined.