|Home | About | Journals | Submit | Contact Us | Français|
During muscle development, the laminin-specific α7 integrin is alternatively spliced in the putative ligand-binding domain to yield either the α7X1 or the α7X2 variant. The relative level of α7X1 and α7X2 is developmentally regulated. Similarly, the partner β1 integrin cytoplasmic domain is converted from the β1A to the β1D splice variant. To determine whether β1D modulates the activity of the α7 receptor, cells were transfected with α7X1 and β1D cDNA. α7X1 coupled with β1A failed to adhere to laminin-1, whereas cotransfectants expressing α7X1 and β1D showed strong adhesion. Interestingly, α7X1 complexed with β1A and β1D displayed the same level of poor adhesion to laminin-2/4 or strong adhesion to laminin-10/11. These findings indicate that α7 function is regulated not only by X1/X2 in its extracellular domain but also by β1 cytoplasmic splice variants. It is likely that expression of β1D alters α7X1 binding to laminin isoforms by a process related to ligand affinity modulation. Functional regulation of α7β1 by developmentally regulated splicing events may be important during myogenic differentiation and repair because the integrin mediates adhesion, motility, and cell survival.
The dynamic interaction between myoblasts and laminins is important for skeletal muscle development, regeneration, and myotube stability and survival (Vachon et al., 1997 ; Gullberg et al., 1999 ; Colognato and Yurchenco, 2000 ). The laminin-binding α7β1 integrin, expressed in mouse skeletal muscle as early as E10.5 d of development, mediates myoblast motility on laminin substrates (Kaufman et al., 1980 ; Kaufman et al., 1991 ; Yao et al., 1996a ,b ; Crawley et al., 1997 ; von der Mark et al., 2002 ). In mature skeletal muscle, the α7 receptor is associated with costameres and the myotendinous and neuromuscular junctions (Bao et al., 1993 ; Belkin et al., 1996 ; Martin et al., 1996 ; van der Flier et al., 1997 ). The physiological importance of α7 is demonstrated in α7-null mutant mice, which develop a form of muscular dystrophy with myotendinous junction (MTJ) defects (Mayer et al., 1997 ), and in humans with mutations in α7 gene, which develop congenital myopathies (Hayashi et al., 1998 ; Pegoraro et al., 2002 ).
The α7 integrin subunit is alternatively spliced in both the extracellular (X1 and X2) (Ziober et al., 1993 ) and the cytoplasmic (A and B) domains (Song et al., 1993 ; Ziober et al., 1993 ). During the differentiation of myoblasts to myotubes, the cytoplasmic domain of the β1 integrin is converted from isoform A to D (van der Flier et al., 1995 ; Zhidkova et al., 1995 ; Belkin et al., 1996 ). β1D begins to be expressed in late fetal life and eventually displaces β1A in mature muscle (Brancaccio et al., 1998 ). Previous studies have shown that the presence of β1D is correlated with altered adhesion and is associated with enhanced interaction with actin cytoskeleton (Belkin et al., 1996 ; Belkin et al., 1997 ; Pfaff et al., 1998 ). This complex differentiation-dependent pattern of α7 and β1 splicing allows the generation of a unique set of variant α7β1 complexes that are structurally, and presumably functionally, distinct.
Laminins, the only known ligand for α7, also undergo a complex pattern of expression during muscle development (Gullberg et al., 1999 ; Colognato and Yurchenco, 2000 ; Pedrosa-Domellof et al., 2000 ). Laminin-1 is expressed during embryonic and fetal stages. Interestingly, laminin-1 is concentrated at the MTJ of fetal human skeletal muscle and persists until birth. As development of muscle continues, laminin-2 and 4 become the dominant isoforms.
The X1/X2 alternative splicing of α7 is developmentally regulated (Ziober et al., 1993 ). In early muscle differentiation of rodent limb, α7X1 shows a relative increase in expression but postnatally the X2/X1 ratio increases (Hodges et al., 1997 ). Previously, we transfected mouse α7X1 and α7X2 isoforms into human MCF-7 cells, which normally adhere poorly to laminin-1 (Yao et al., 1996a ; Ziober et al., 1997 ). In the transfectants, α7X2 bound laminin-1 readily, but α7X1 bound only when activated by the β1-activating monoclonal antibody (mAb) TS2/16. However, both X1 and X2 were functional when expressed in the HT-1080 cell line. These results indicated that alternative splicing regulates α7β1 ligand-binding competence in a cell-specific manner. Recently, von der Mark and collaborators reported that the X1/X2 variants have different binding specificity and affinities for laminin isoforms (von der Mark et al., 2002 ).
We proposed that the X1 isoform is important during dynamic adhesion related to muscle development and repair (motility, fusion, and matrix assembly), whereas the X2 variant performs more stable adhesion functions (mature junctional assembles) (Ziober et al., 1997 ). This is consistent with the dynamics of the expression of α7 and β1 alternatively spliced variants during development (Ziober et al., 1993 ) and muscle regeneration (Kaariainen et al., 2001 ; Kaariainen et al., 2002 ). Laminin isoform expression also changes during muscle development and repair (reviewed in Colognato and Yurchenco, 2000 ) and so the expression of integrin and potential ligand is coordinately coupled.
Integrins are capable of changing their function dynamically (Diamond and Springer, 1994 ; Humphries, 1996 ; Mould, 1996 ; Hughes and Pfaff, 1998 ). It has been proposed that certain intracellular events lead to alterations in integrin conformation, resulting in enhanced adhesion to ligand (inside-out signaling) (reviewed in Liddington and Ginsberg, 2002 ). Alternatively, the process of ligand binding may regulate activity of the integrin receptor (outside-in signaling). In the case of inside-out signaling, there is strong evidence that the actin cytoskeleton and its link proteins (e.g., talin) are one mechanism that can regulate integrin activation. This process may represent changes in integrin affinity or integrin-clustering events leading to increased avidity (Zent et al., 2000 ). However, it remains controversial whether events such as lateral rearrangements that may modulate receptor avidity can play a role in integrin function (reviewed in Shimaoka et al., 2002 ). As stated above, after myoblast differentiation into myotubes and mature muscle, the β1 integrin cytoplasmic alternative splice shifts from A to D. Given the consensus that the cytoplasmic domains of β integrins can regulate the conformation and activity of integrin subunits, it is possible that the β1D isoform controls α7X1/X2 function. To investigate this possibility, we examined the functionality of α7 integrin in the presence of the β1A or β1D cytoplasmic domain.
For cell adhesion studies, several different ligands were used as described previously. Laminin-1 isolated from the mouse EHS tumor was obtained from Invitrogen (Carlsbad, CA). Human placental merosin, consisting of laminin-2/4 isoforms purified by EDTA extraction and followed by ion-exchange chromatography was purchased from Invitrogen. Human laminin-10/11, isolated from placenta by mild pepsin digestion and by affinity chromatography on mAb 4C7-coupled Sepharose, was also from Invitrogen. A sample of recombinant laminin-10 was kindly provided by Masayuki Doi (University of Okayama, Okayama, Japan) and Karl Tryggvason (Karolinska Institute‡, Stockholm, Sweden). Collagen was obtained from Cohesion (Palo Alto, CA). Human vitronectin and fibronectin were gifts from Caroline Damsky (University of California, San Francisco, San Francisco, CA).
Antibodies against integrin subunits, including rat anti-human β1 (mAb AIIB2), mouse anti-human α2 (mAb VM1), rat anti-human α6 (mAb GoH3), and rabbit anti-mouse β1D cytodomain (polyclonal antibody anti-β1D), were kindly provided by Caroline Damsky and Vera Morhenn (SRI International, Menlo Park, CA), Arnond Sonnenberg (Netherlands Cancer Center, Amsterdam, The Netherlands), and Eva Engvall (Burnham Institute, La Jolla Cancer Research Center, CA), respectively. Mouse anti-α3 (mAb J143) was from the American Type Culture Collection (American Type Culture Collection, Manassas, VA), rat anti-mouse β1 (mAb MB1.2) and mouse anti-human β1 (mAb 2000) were from Chemicon International (Temecula, CA), rat anti-mouse α7 (mAb CY8), mouse antihuman α7 (mAb 9.1), rabbit anti-human β1 cytodomain (polyclonal antibody [pAb] 22778), and rabbit anti-mouse α7 light chain (pAb 1211) were from our laboratory, as described previously (Yao et al., 1996b , 1997 ; Vizirianakis et al., 2001 ). Fluorescein-conjugated secondary antibodies were obtained from Jackson Immunoresearch Laboratories (West Grove, PA).
The full-length mouse β1D cDNA was prepared by reverse transcription followed by polymerase chain reaction (PCR) from a mouse skeletal muscle cDNA library (BD Biosciences Clontech, Palo Alto, CA). The two primers for PCR were reverse primer, 5′-CTAGTCTAGAATTCAGAGACCAGCTTTACGTCCATAG and forward primer, 5′-CGCGGATCCGAATTCAAGATGAATTTGCAACTGGTTTCCTG. The sequence of full-length mouse β1D was verified at the Biomolecular Resource Center (University of California, San Francisco). The mouse β1D cDNA was then ligated into BamHI and XbaI sites of pcDNA3.1/Hygro (Invitrogen). Full-length mouse β1A cDNA was obtained from Louis Reichardt (University of California, San Francisco) and was cloned into the regulatable expression vector in the same manner as β1D (see below).
Full-length human α7X2 cDNA (Vizirianakis et al., 2001 ) was used as a template for the generation of full-length human α7X1. The X2 exon was replaced with the X1 exon by uracil DNA glycosylase excision to generate cohesive ends on the PCR product of the X1 exon. Specifically, we designed primers containing uracil that matched the corresponding T in the splicing regions. We then used these primers to generate the X1 exon by PCR from human α7 cDNA by using human fetus limb muscle RNA as template (forward primer, 5′-AAGGGUACUGCCAGGGTGGAGCTCTGTG; reverse primer, 5′-AGAAGCCAAAGUA GCUGTTGGCAGGGA). The full-length cDNA proximal to the splicing site was also generated by PCR (forward primer, 5′-CCGGGATCCATGGCCGGGGCTCGGAGCCG; reverse primer, 5′-AGTACCCUUCCAATTATAGGTTCCTGGGG). The primers for full-length cDNA distal to the splicing site were as follows: forward primer, 5′-AGCTACTTTGGCUUCUCT ATTGACTCGGGGAAAGGTCTG; reverse primer, 5′-CCGCTCGAGCTAGGCGG TGCCTGGCCCT. The three pieces of PCR products were first treated with uracil DNA glycosylase according to the manufacturer's protocol (Invitrogen) and then ligated. The ligation product was purified by gel electrophoresis followed by cloning into pCDNA3.1/Hygro and verified by sequencing.
MCF-7 human carcinoma cells and the α7-transfected cells were cultured as described previously (Yao et al., 1996a ; Ziober et al., 1997 ). MCF-7 cells do not adhere well to laminin-1 even though they express moderate levels of integrins α2, α3, and α6, which are potential receptors for this ligand. Previously, we generated MCF-7 cells expressing full-length mouse α7 cDNA containing either the X1 or X2 splice form, which formed heterodimers with the partner β1A subunit. MCF-7 cells expressing the splicing isoform α7X2 were active in binding laminin-1, whereas cells expressing α7X1 lacked ligand-binding activity (Ziober et al., 1997 ). The original X1/A-p expressing population was further enriched by fluorescence-activated cell sorting (FACS) by using mAb CY8 against mouse α7 to yield the X1/A cell line used in the current experiments. The X2/A cell line was derived from MCF-7 cells transfected with mouse α7X2, which heterodimerizes with the endogenous human β1A (Yao et al., 1996a ).
Expression of mouse β1D in MCF-7 cells was generated in two different systems. In the first, the X1/A-p cells were stably transfected with mouse β1D cDNA by the calcium phosphate procedure (Mammalian Transfection kit; Stratagene, La Jolla, CA). After selection with hygromycin, the double transfectants (α7X1/β1D) were enriched by FACS after labeling with anti-mouse β1 antibody (mAb MB1.2; Chemicon International). The enriched population was thereafter called X1/D. In addition, a total of 20 single cell clones were isolated either by single cell sorting by FACS or by limiting dilution into individual wells of 96-well plates. The expression of mouse β1D was verified by immunoblot analysis with β1D-specific pAb from Eva Engvall and by flow cytometry with mAb MB1.2. A number of clones with a range of α7 and β1D levels, as determined by FACS, were chosen for further study.
In the second approach, MCF-7 cells were transfected with pTet-Off (pUHD15-1 neo; BD Biosciences Clontech), and the expression of tTA-vp16 was verified. The MCF-7 Tet-Off cells were then transfected with the full-length cDNA for either human α7X1 integrin (as described above) or human α7X2 integrin (Vizirianakis et al., 2001 ). We used anti-human α7 mAb 9.1 to enrich the α7-expressing population by FACS. Retroviral transduction was then used to introduce the pRevTRE plasmids (BD Biosciences Clontech) with mouse β1A or mouse β1D into the human α7 stable transfectants. In brief, Phoenix Ampho retrovirus packaging cells from American Type Culture Collection were transfected with pRevTRE β1A/β1D via calcium phosphate. The mouse β1D or mouse β1A cDNA was cloned downstream of the tetracycline-responsive element and the minimal immediate early promoter of cytomegalovirus (PminCMV). Retroviral supernatants were then used to infect α7-expressing cells. To test the regulation of expression, cells were cultured with various concentrations of doxycycline. After retroviral induction, cells were screened by FACS with mAb MB1.2 against mouse β1 ectodomain, and cell lines with high expression of mouse β1 integrin in the absence of doxycycline and low background in the presence of doxycycline were selected for further study. MCF-7 Tet-Off cells with human α7X1/mouse β1A, human α7X1/mouse β1D, or human α7X2/mouse β1A were designated HuX1/A, HuX1/D, and HuX2/A, respectively. For adhesion assays, cells were typically cultured for 5–7 d in the presence of doxycycline (0.3 ng/ml or less) to stabilize integrin expression levels.
Standard procedures for flow cytometry were followed (Yao et al., 1996a ). Briefly, cells (106/ml) were incubated with predetermined optimal concentrations of primary antibodies, washed, and incubated with secondary FITC-conjugated fluorescein-labeled antibodies (affinity-purified goat anti-mouse or anti-rat antibodies; Jackson Immunoresearch Laboratories). After washing, the cells were stained with propidium iodide and processed for flow cytometry on an FACScan (BD Biosciences, San Jose, CA). Samples without primary or secondary antibody were always included as a control. Data are expressed as the mean fluorescence intensity after subtraction of background staining produced by secondary antibody alone (<5% of signal).
Cell attachment was measured using a published protocol (Yao et al., 1996a ; Ziober et al., 1997 ). Briefly, microtiter plates (96-well Immulon 1B plates; Thermo Labsystems, Franklin, MA) were coated with matrix proteins at the indicated concentrations in phosphate-buffered saline for 1 h at 37°C. Single cell suspensions were prepared and assayed in triplicate in 96-well plates with an incubation period of 30–60 min at 37°C. Normally, cell adhesion assays were performed using a short time of incubation (30 min) at the lower range of ligand-coating concentrations to preferentially study early stages of adhesion. Adherent cells were fixed, stained with crystal violet, and solubilized in 2% SDS. Absorbance was read at 562 nm. Background cell adhesion to 1% bovine serum albumin-coated wells (usually <5% of value) was subtracted from all readings. Values for total cell input per well to represent 100% attachment were determined by seeding cells on a separate microtiter plate coated with collagen type I or polylysine (10 μg/ml) followed by incubation at 37°C for 120 min. The 100% attachment value was then estimated by fixation on the monolayer and staining with crystal violet as described above. The effect of specific blocking antibodies was tested by preincubating the cells with an optimal blocking concentration of mAb on ice for 30 min before the assay. Optional concentrations of the blocking antibodies were predetermined and used as described previously (Yao et al., 1996a ; Ziober et al., 1997 ). The concentrations of integrin function-blocking mAb were as follows: 10 μg/ml purified GoH3 (anti-α6) and J143 (anti-α3). VM1 (anti-human α2) ascites were used at a dilution of 1:400. Both CY8 (anti-mouse α7) and AIIB2 (anti-human β1) ascites were used at a dilution of 1:300. Cytochalasin D was purchased from Sigma-Aldrich (St. Louis, MO) and dissolved in dimethyl sulfoxide (DMSO). Adhesion assays included control cells treated with DMSO alone (final concentration, 0.2%). The cells were processed as described above except that they were pretreated with 20 μM cytochalasin D or DMSO alone for 20 min on ice before the assay.
Cell lysates were prepared by extraction of cells with lysis buffer (1% Triton × 100 in 50 mM Tris-HCl, pH 7.5, 1 mM CaCl2, with 2 mM phenylmethylsulfonyl fluoride, 1 mM N-ethylmaleimide, and 2 mM leupeptin as protease inhibitors) for 90 min and centrifuged. The supernatants were precleared with agarose beads coupled with secondary antibodies overnight. For the purpose of depletion analysis, aliquots of cell lysates (600 μg of protein) were immunoprecipitated with antibody against α2, α3, or α7 for three consecutive rounds. The beads were pooled and washed with wash buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM MgCl2, 0.5% Nonidet P-40, and 0.1% bovine serum albumin) three times and heated at 100°C with SDS sample buffer. Samples including controls for whole cell lysates (10 μg protein/lane) were processed for SDS-PAGE under nonreducing conditions and then transferred to Immobilon-P membranes (Millipore, Bedford, MA). The membranes were incubated with rabbit antibody against human integrin β1 (pAb 22778) (Martin et al., 1996 ; Yao et al., 1997 ) or mouse β1 (mAb 2000; Chemicon International). In a different approach to assess the α7 preference to β1A or β1D, equal amounts of cell lysate (500 μg of protein) prepared from X1/D-27 cells (a clone of X1/D cells) were immunoprecipitated with human β1A or mouse β1D antibody for three consecutive rounds. The samples were then processed as described above except under reducing conditions for immunoblotting and probed with rabbit antibody (pAb 1211) against mouse α7 (Yao et al., 1997 ).
To test the hypothesis that alternatively spliced amino acid sequences in the β1 cytoplasmic domain may regulate the activity of the α7 integrin, we transfected human MCF-7 cells with mouse α7X1 alone (X1/A cell line) or with both mouse α7X1 and β1D cDNA (X1/D cell line). After selection, the transfectants expressed high levels of α7X1 and β1D at their surface, as detected by flow cytometry with mAb directed against α7 (CY8) or mouse β1 (MB1.2) (Table 1). The level of α7 expression in the X1/A AND X2/A cell lines were found to be similar. In the case of the double transfectant X1/D, the X1 level was somewhat lower than that of the X1/A cell line from which it was derived.
To assess whether X1/D and X1/A cells displayed different adhesion profiles, we used standard adhesion assays to examine their ligand binding to the extracellular matrix components vitronectin and type I collagen. On these immobilized ligands, X1/A and X1/D cells showed a similar ligand dose-response pattern (Figure 1, A and B) over a range of ligand-coating concentrations. Thus, expression of β1D did not alter the adhesion of α7X1-expressing cells to these components. Previous reports showed that in MCF-7 cells, adhesion to vitronectin and type I collagen was mediated primarily by αvβ1 (Maemura et al., 1995 ) and α2β1 integrins (Jones et al., 1995 ; Maemura et al., 1995 ), respectively. This result indicated that there was no change in the general adhesion activity toward type I collagen or vitronectin in the transfectants.
We previously reported that MCF-7 cells expressing α7X1 with the endogenous human β1A partner adhered poorly to laminin-1 (Ziober et al., 1997 ). However, after transfection with cDNA of the mouse β1D subunit, X1/D cells acquired the capacity to strongly bind laminin-1 (Figure 1C). This increase in adhesion was evident at coating concentrations above 8 μg/ml. We also generated 20 single cell clones by limiting dilution and by single cell sorting (see below). A number of representative clones were chosen for the same assay, and the results were similar (our unpublished data). Because the stimulatory effect of β1D on adhesion was restricted to laminin-1 and not vitronectin or collagen I, this suggested that laminin-binding integrins such as α7X1, but not other potential β1 partner subunits, were involved. The poor adhesion of X1/A cells to laminin-1 was not related solely to the level of α7X1 expressed at the surface, because this cell line had a significantly higher expression level of α7X1 than did X1/D cells (Table 1).
To establish that the β1D-induced increase in laminin-1 adhesion was attributable specifically to α7X1, we performed standard adhesion assays with blocking mAb against other expressed α subunits (Figure 1D). The α7X1-expressing cells (X1/A) showed poor adhesion to laminin-1, and adhesion could be diminished further by treatment with a cocktail of blocking mAb to α2, α3, and α6 integrins. In contrast, the X1/D cells expressing β1D showed strong adhesion to laminin-1, and most of this activity was resistant to the mixture of antibodies to α2, α3, and α6 integrins. Importantly, blocking mAb CY8 to α7 completely ablated adhesion to laminin-1.
To determine what fraction of the α7 subunit was partnered with endogenous β1A or transfected β1D, two different strategies were used. In the first, lysates prepared from X1/D cells were exhaustively immunoprecipitated with mAb specific to either human β1 or mouse β1, followed by immunoblotting for partner α7 (Figure 2A). The results show that the fraction of α7 that heterodimerized with either the mouse β1D (lane 5) or the human β1A (lane 6) isoform was approximately equal. Thus, neither subunit seems to have any preferential affinity for α7 during integrin heterodimerization.
In the second approach to estimate the relative pairing of α7 integrin subunit with endogenous β1A or transfected β1D, we used the X1/D-27 cell line, which expresses high levels of both α7X1 and β1D (Figure 2B). Cells were first processed for immunoprecipitation with anti-α7 antibody, followed by immunoblotting analysis with antibody specific for either human β1(A isoform) or mouse β1(D isoform). Again, similar levels of β1A and β1D were detected in the immunoprecipitates (Figure 2B, lane 3, left and right panels). There seemed to be an abundance of β1D precursor (Figure 2B, lane 1; right) compared with the mature, fully processed form of the subunit. In contrast, nearly all of the endogenous β1A subunit was present in the mature form (Figure 2B, lane 1; left). Similarly, we found that α2 and α3 integrin subunits were associated with equivalent proportions of β1A and β1D (Figure 2C). In other studies, we examined the amount of integrin subunits associated with β1A and β1D by immunoprecipitation of [35S]methionine-labeled cells; similar results were observed (our unpublished data). On the basis of these studies, we concluded that α7 as well as α2 and α3 subunits were able to complex with similar efficiency to β1A and β1D subunits.
To further define the relationship between β1D and α7 activity, we compared the relative adhesion to laminin-1 of a large series of α7/β1D-expressing clones and cell lines. A three-dimensional scatter plot summarizing these results indicates that adhesion to laminin-1 strictly correlated not only with the expression of α7, but also with the relative level of expression of β1D (Figure 3). The parental cells with endogenous β1A (MCF-7) or cells transfected with α7X1 alone (X1/A) showed poor adhesion to laminin-1. The α7X1 and β1D double transfectants (X1/D and clones 10, 12, 13, 17, 21, and 27) showed a strong correlation between levels of α7 and β1D expression. Thus, the expression of mouse β1D in MCF-7 cells transfected with α7X1 consistently enhanced the expression of α7 in this system. The level of α7 increased in parallel with that of the transfected mouse β1D and correlated with correspondingly greater adhesion to laminin-1. Furthermore, transfection of cells with mouse β1A or β1D in the absence of α7 to yield the β1A or β1D cell lines did not result in enhanced adhesion to laminin-1. This result indicated that the overexpression of mouse β1A or β1D was unable to augment adhesion activity of endogenous α2, α3, or α6 integrins for laminin-1. In other studies, we generated human α7X1- and α7X2-expressing MCF-7 Tet-Off cells and found that these transfectants showed a similar adhesion profile to laminin-1, with α7X1 showing poor binding to laminin-1 and α7X2 showing strong binding (our unpublished data).
In a parallel approach to establish the role of integrin β1 cytoplasmic variants on the expression and function of human α7, we created stable cell lines constitutively expressing either the human α7X1 or α7X2 isoform and tetracycline-regulatable mouse β1A or β1D integrin. These cell lines were designated as HuX1/A, HuX1/D, and HuX2/A were selected for further study. To regulate the expression of mouse β1 integrin, cells were cultured in medium containing increasing concentrations of doxycycline, and integrin expression was then measured by FACS by using mAb to α7 integrin and mouse β1 integrin. Doxycycline suppressed the expression of mouse β1 integrin in a dose-dependent manner (Figure 4A) and at the lower range of doxycycline shown the integrin expression was linear. The sensitivity to regulation by doxycycline repression of HuX1/A and HuX1/D differed slightly; the HuX1/D required about twice the doxycycline concentration to suppress mouse β1 integrin expression compared with that required for the HuX1/A. Expression of mouse integrin was completely suppressed with a doxycycline concentration of >1 ng/ml for β1D and 0.65 ng/ml for β1A. By modulating the levels of doxycycline, comparable levels of mouse β1 integrin expression were reached for both cells (Figure 4A). Due to the difference in sensitivity to doxycycline repression, we chose to use 0.1 and 0.3 ng/ml doxycycline for HuX1/A and HuX1/D, respectively, for subsequent adhesion assays on laminin-1. HuX1/A, HuX1/D, and HuX2/D cells were cultured at this low concentration of doxycycline, the integrin profile as detected by FACS is summarized in Table 2. Importantly, for the Tet-Off cells, high levels of human α7 and the mouse β1 level are maintained at increasingly higher levels as the doxycycline concentration was lowered. The mouse β1 seems to be the dominant subunit compared with the endogenous human β1A. As the level of mouse β1A or β1D integrin was increased after induction, we observed a corresponding increase in α7 surface expression (Figure 4B). The levels of induced expression of α7 were 5–6 times that of the background expression at high doxycycline concentrations. If the concentration of doxycycline was further reduced, the increase in integrin expression eventually reached a plateau (our published data). Analysis of integrin profiles revealed that, besides α7 and β1D subunits, other major endogenous integrins (human α2, α3, and β1A) remained relatively stable as the regulatable β1D was modulated (Figure 4C and Table 2). It is interesting that as the level of regulatable mouse β1D expression was increased there was no significant increase in the surface expression of α2, α3 integrins (Figure 4C). This may be due to the high level of α7 precursor that is able to compete for available β1. Presumably, as more mouse β1 is synthesized, the α7 heterodimers continue to be preferentially assembled and expressed at the surface.
We also analyzed the influence of relative α7X1 and β1 expression on the adhesion of doxycycline-regulatable cells to laminin-1 in standard attachment assays (Figure 4D). Mouse β1A-expressing cells showed poor adhesion at all levels of β1A integrin expression even though α7X1 was increased several fold over the range of doxycycline treatment (Figure 4B). In contrast, induction of β1D at low doxycycline concentrations resulted in a strong adhesive response to laminin-1 that corresponded to elevated β1D and α7X1 expression levels. In fact, the elevation in adhesion to laminin-1 correlated closely to the increase in α7X1 expression. For comparison, we also analyzed the adhesion efficiency of Tet-Off cells expressing α7X2/A and α7X2/D. We found no significant difference in adhesion to laminin-1 between α7X2 cells expressing either β1A or β1D (our unpublished data). Both double transfectants adhered strongly to laminin-1 (70–85%). This result indicated that α7X2 is fully functional for adhesion to laminin-1 and that exogenous β1D did not enhance the adhesion further.
The above-mentioned results indicate that β1D was able to modify the binding of α7X1 to laminin-1. We compared the relative adhesion efficiency of α7X1-expressing cells in the presence of β1A or β1D to available laminin isoforms. We first tested the adhesion of α7-expressing cells on preparations of human placental merosin. Previous analyses have indicated that these preparations contain primarily human laminin-4 (α2β2γ1, S-merosin) with lesser amounts of laminin-2 (α2β1γ1, merosin) (Delwel et al., 1994 ; Spinardi et al., 1995 ; Yao et al., 1996a ). Thus, we tested the relative adhesion of HuX1/A, HuX1/D, and HuX2/A to human merosin (laminin-2/4) (Figure 5A). Interestingly, cells expressing α7X1/β1A (HuX1/A) or α7X1/β1D (HuX1/D) adhered poorly. In contrast, cells expressing α7X2/β1A (HuX2/A) showed strong attachment. Adhesion by HuX2/A cells to laminin-2/4 was not sensitive to a mixture of mAbs to α2, α3, and α6 integrins, but adhesion was completely blocked by anti-α7 mAb 9.1 (Figure 5B).
In a similar manner, we examined the adhesion capacity of these three cell lines to laminin-10/11 (Figure 6A). In contrast to adhesion on laminin-2/4 substrates, HuX2/A adhered poorly to laminin-10/11. Whereas HuX1/A and HuX1/D both attached well to this substrate, HuX2/A displayed poor adhesion even at high ligand coating concentrations. Treatment of cells with anti-human α7 blocking mAb effectively inhibited adhesion confirming that this receptor is mediating binding (Figure 6B). Similar results were obtained using recombinant human laminin-10 (our unpublished data). Together, these results show that replacing β1A with β1D confers activity of α7X1 for laminin-1 but does not alter binding of the double transfectants to laminin-2/4 or laminin-10/11 substrates.
Integrin–cytoskeletal interactions have been shown to influence both integrin clustering and extracellular domain conformation. It is well established that integrins connect to the cytoskeleton through the cytoplasmic domain of the β subunit and evidence indicates that talin may play an important role in regulating integrin activity by binding to the β1 cytoplasmic domain (Liddington and Ginsberg, 2002 ). Furthermore, it has been shown that the β1D tail binds talin better than the β1A cytoplasmic domain (Belkin et al., 1997 ; Pfaff et al., 1998 ). To test the potential importance of cytoskeleton linkage to the adhesion activity of α7X1-expressing cells, we assayed binding of MCF-7 Tet-Off double transfectants to laminin-1 substrates in the presence of cytochalasin D, an inhibitor of actin polymerization. Treatment of cells with 20 μM cytochalasin D did not affect the adhesion of the HuX1/A and HuX2/A cell lines to laminin-1 substrates (Figure 7). However, the presence of the drug reduced HuX1/D cell adhesion to laminin-1 by nearly 70%. Adhesion of all cell lines to type I collagen was unaffected by cytochalasin D (our unpublished data). These results suggest that the formation of polymerized actin during the adhesive process was necessary for the β1D to confer enhanced binding of α7X1 to laminin-1.
We demonstrate that when the α7X1 alternatively spliced integrin heterodimerizes with the muscle β1D subunit, the ligand specificity is enhanced. If α7X1 is coupled to β1A, it is unable to induce cell adhesion to laminin-1 but can efficiently mediate adhesion to laminin-10/11. After the exchange of β1A for β1D, the receptor can then also promote cell binding to laminin-1. These results show that the replacement of β1A with β1D may induce a higher order of activation of α7, thereby permitting the receptor to bind not only to laminin-10/11 but also to laminin-1. This effect seems to represent an example of ligand affinity modulation of an integrin by the companion β cytoplasmic domain.
The alternative splicing of α7X1/X2 occurs between the III and IV homology repeat domains, a region implicated in ligand specificity, and affinity. Springer originally proposed that the extracellular portion of integrin α chain is organized in a cyclic-β propeller structure (Springer, 1997 ), and this has been confirmed by x-ray crystallography (Xiong et al., 2002 ). In this model, the alternatively spliced segment of α7, which is predicted to be located outside the “blade” structure joined by the two-homology repeat domains, may modulate folding of these adjacent β sheets, thereby controlling integrin conformation and function. We speculate that α7X1, when bound to the β1D variant, forms a modified active ligand-binding pocket capable of binding laminin-1 and may partially resemble the structure formed by the constitutively active α7X2/β1A. Yet both integrins are functionally distinct because they differentially bind laminin-2 and -10/11.
Exogenous β1D expression in MCF-7 cells did not enhance cell adhesion to laminin-1 in the absence of α7 and, in the presence of α7X1 and β1D, the enhancement of adhesion was mainly attributed to α7 and not to the other moderately expressed α2, α3, or α6 integrins that could potentially bind laminin-1. Because there is no preference in the coupling of these α subunits to β1A or β1D (Belkin et al., 1997 ; Pfaff et al., 1998 ; this study), our results indicate that β1D-induced functional augmentation in MCF-7 cells is specific for the α7X1 integrin. MCF-7 cells have β1 integrin receptors for other ligands, including α2β1 for collagen 1 (Jones et al., 1995 ; Maemura et al., 1995 ) and αvβ1 for vitronectin (Wong et al., 1998 ). Adhesion to these two ligands was not altered by the expression of β1D. In addition, the forced expression of β1D in the fully functional α7X2 cells did not further enhance the adhesion to laminin-2/4 or -10/11 (our unpublished data). These results suggest that the β1D partner subunit can selectively alter α7X1 functionality. However, it is possible that in these cases that the integrin (e.g., α2 or α7×2) may already be at a fully activated level and that binding activity cannot be further enhanced by coupling with β1D.
Affinity modulation can occur by inside-out signaling via transmembrane conformational transitions (Faull et al., 1994 ; Schwartz et al., 1995 ; Hughes and Pfaff, 1998 ; Longhurst and Jennings, 1998 ; Liddington and Ginsberg, 2002 ). Previous studies with cells transfected with β1D indicated that the subunit induced a number of cellular alterations that included decreased spreading and migration, increased binding of fibronectin, and incorporation into matrix (Belkin et al., 1997 ). These results are consistent with stabilized integrin–cytoskeleton interaction that seems to be mediated by increased binding of β1D with talin and elevated levels of integrin activation. In the context of the current studies, α7X1/β1A seems to exist in a partially active conformation as indicated by the finding that the TS2/16 mAb could switch this integrin to a functional laminin-1 receptor (Ziober et al., 1997 ). Formation of the α7/β1D complex seems to induce a similar conversion from an intermediate level of activation to a fully active state. There is precedence for intermediate affinity states (Takagi et al., 2002 ).
Alternatively, β1D-induced activation of α7X1 may be related to postreceptor occupancy events such as receptor clustering and avidity regulation. For example, enhanced tethering of integrins to the cytoskeleton and localized clustering may lead to strengthened adhesion as a result of increased avidity. However, as Springer and colleagues (Shimaoka et al., 2002 ) have summarized, the data supporting this mechanism of increased integrin activity (avidity modulation) remain controversial. Furthermore, the cellular environment can also modulate integrin function (Zhang et al., 1996 ; Hughes et al., 1997 ). Thus, in contrast to MCF-7 cells, when HT-1080 cells that express only β1A were transfected with α7 splice variants, both X1 and X2 isoforms were fully active for binding to laminin-1 (Ziober et al., 1997 ), indicating that factors related to cell-type specificity are important in regulating α7 integrin function. Additional studies are needed to further define the specific mechanism responsible for β1D-induced functional modulation of α7X1.
Differences in the interaction of β1 alternatively spliced cytoplasmic tails with the cytoskeleton have been reported. Previous work by Belkin et al. (1997 ) has shown that overexpression of β1D physically displaces endogenous β1A normally found at adhesion sites with the extracellular matrix. As a result of this displacement of β1A by β1D integrins, cells exhibited a dominant positive phenotype and formed a stronger interaction with the underlying matrix. This activity is apparently related to differences in how β1D and β1A interact with the cytoskeleton. The muscle-specific β1D splice variant has been shown to bind talin with higher affinity than β1A (Belkin et al., 1997 ; Pfaff et al., 1998 ). Conversely, the β1A tail has been reported to bind filamin and α-actinin more strongly than does β1D (Belkin et al., 1997 ; Pfaff et al., 1998 ). It is possible that in the current experiments the effect of β1D on α7X1 activity is a result of this increased interaction with the actin cytoskeleton. In support of this possibility we found that treatment of HuX1/D cells with cytochalasin D resulted in the loss of β1D induced stimulation of adhesion to laminin-1 (Figure 7). Yet the effect seems to be specific to α7 since other integrins, e.g., α2, which were shown to associate with β1D, did not show any enhanced adherence to collagen I. As mentioned above, this may be explained by the possibility that the α2 integrin is already fully activated and adhesion cannot be enhanced further by complexing with β1D. Interestingly, Gimond et al. reported that replacement of β1A with β1D in embryonic stem cells did not lead to either faster rates or a higher extent of cell adhesion to either fibronectin or laminin 1 substrates via the α5β1 and α6β1 integrin, respectively (Gimond et al., 2000 ). In fact, on both substrates, the β1D cells bound much less efficiently than the β1A cells. This may suggest that depending on the α chain that is complexed with β1D, the effect on integrin function may vary.
We found that cells expressing α7X1/β1A (HuX1/A) adhered poorly to laminin-2/4. In contrast, cells expressing α7X2/β1A (HuX2/A) showed strong attachment to this ligand. On laminin-10/11, HuX2/A cells adhered poorly whereas HuX1/A cells showed strong affinity. von der Mark and collaborators recently showed that soluble double truncated forms of α7X1 and α7X2 bind differently to various laminin isoforms (von der Mark et al., 2002 ). Our results are similar in that α7X1 expressing cells bound to laminin-10/11 substrates whereas α7X2 bound preferentially to laminin-1. However, our data differ with regard to α7X1 and laminin-2/4 substrates. In the previous study, both receptors showed intermediate affinities to purified laminin-2. The divergent results may be due to the nature of the laminin preparations used.
We have investigated the mechanism underlying functional regulation of the alternatively spliced α7 variants by its partner β1 integrin. The physiological importance of the α7 and β1 splice variants is related to the varied roles of α7 as the major laminin-binding receptor in skeletal muscle, where it has multiple functions not only during myoblast motility and differentiation but also in maintenance of mature fiber anchorage and assembly of its underlying basement membrane. When myoblasts begin to differentiate into myotubes, the partner integrin switches from the β1A to the β1D cytoplasmic form. Our results indicate that coupling of the α7X1 variant with the β1D isoform modifies the receptor's ligand binding specificity and converts α7X1 to a promiscuous receptor that can mediate adhesion to laminin-1 and other laminin isoforms.
In summary, we have demonstrated that the cytoplasmic domain of the β1 integrin can modulate the functionality and ligand specificity of the α7 receptor. We show that when coupled to the β1D partner, α7X1 splice variant shows a broader spectrum of ligand specificity and can efficiently bind and mediate adhesion to laminin-1. The formation α7X1/β1D in vivo seems likely given that both subunits are present during the late embryonic stage of skeletal muscle development. This suggests that the simultaneous expression of α7X1 along with β1D could facilitate the organization and stability of the MTJ during the early stages of muscle fiber maturation.
This research was supported by National Institutes of Health grants DE13479. We thank Drs. Caroline Damsky and Louis Reichardt for critical review of this manuscript.
Article published online ahead of print. Mol. Biol. Cell 10.1091/mbc.E02–12–0824. Article and publication date are available at www.molbiolcell.org/cgi/doi/10.1091/mbc.E02-12-0824.
Abbreviations used: DMSO, dimethyl sulfoxide; FACS, fluorescence-activated cell sorting; FITC, fluorescein isothiocyanate; mAb, monoclonal antibody; MTJ, myotendinous junction; NMJ, neuromuscular junction; pAb, polyclonal antibody; PCR, polymerase chain reaction.