Association of the tetraspanin CD151 with the laminin-binding integrinsα3β1, α6β1, α6β4 and α7β1 in cells in culture and in vivo

CD151 is a transmembrane molecule that has been recently characterized as a member of the evolutionary conserved tetraspanin superfamily(Wright and Tomlinson, 1994;Maecker et al., 1997). Tetraspanin molecules, so called because they cross the plasma membrane four times, have two extracellular loops and three cytoplasmic domains including the C- and N-terminus. Many interactions among tetraspanins and between tetraspanins, integrins and other molecules have been described(Berditchevski et al., 1996;Mannion et al., 1996;Hasegawa et al., 1998;Fitter et al., 1999;Sincock et al., 1999). Hence it is thought that one function of tetraspanins is to facilitate the coupling of signaling pathways by assembling and connecting specific cell surface proteins into large molecular complexes.

On the basis of the stability of the formed complexes, interactions of tetraspanins can be classified into three categories(Class et al., 2001). CD151-α3β1 complexes represent the exceptionally strong, highly stable and selective level 1 association. Most interactions between tetraspanins and integrins, for example, the association of CD151 withα6β4, are less strong and considered to be level 2; they are numerous, fairly stable and selective. CD151-α6β1 interactions are either level 1 or level 2, probably depending on the cell type in which these molecules are expressed. Level 3 interactions, which occur mostly between tetraspanins and molecules other than tetraspanins or integrins, are indirect,numerous and are the least stable of the interactions. The level 1 interaction of CD151 with α3 occurs between the large extracellular loop of CD151 and the extracellular membrane proximal region of the α3 subunit(Yauch et al., 2000). A CD151 epitope recognized by the mAb TS151R is located at this site of interaction(Serru et al., 1999;Yauch et al., 2000). Recently,it was shown that the binding site on CD151 for tetraspanins is different from the one involved in the binding to integrins(Berditchevski et al.,2001).

Integrins are heterodimeric cell surface receptors consisting of a non-covalently associated α and β subunit, which link the extracellular matrix to the cytoskeleton(Hynes, 1992;Schwartz et al., 1995;van der Flier and Sonnenberg,2001). The integrins α3β1, α6β1 andα7β1 bind preferentially to laminin and contain sequential and structural homologies. During maturation, the α3, α6 and α7 integrin subunits are post-translationally cleaved into a heavy and a light chain. Furthermore, the mRNAs for these integrin subunits can be alternatively spliced, leading to extracellular and cytoplasmic variants (Collo and Quaranta, 1993; Hogervorst et al.,1993,Hogervorst et al.,1993; de Melker et al.,1997). The ligand binding and signaling function of these integrin variants is potentially different, which during development may lead to changes in function when variants are switched(Ziober et al., 1993;Thorsteindóttir et al.,1995; Schober et al.,2000).

While much is known about the function and biochemical properties of integrins, the exact function of the associated tetraspanins is still elusive but is gradually becoming unraveled. Previously, we have shown that CD151 is a component of hemidesmosomes, specialized structures in the basal epidermal cell layer that link intermediate filaments to the extracellular matrix(Sterk et al., 2000). Others have shown the involvement of CD151 in cell adhesion(Hasegawa et al., 1998;Fitter et al., 1999), (tumor)cell migration and metastasizing(Yáñez-Mó et al.,1998; Yauch et al.,1998; Sincock et al.,1999; Sugiura and Berditchevski, 1999;Berditchevski and Odintsova,1999; Testa et al.,1999), vesicular transport of integrins(Sincock et al., 1999),signaling (Berditchevski et al.,1997; Yauch et al.,1998; Sugiura and Berditchevski, 1999;Berditchevski and Odintsova,1999; Zhang et al.,2001), cell polarization(Yáñez-Mó et al.,2001), wound healing(Peñas et al., 2000)and neurite outgrowth (Stipp and Hemler,2000).

In vivo, CD151 is expressed by a large variety of cells of epithelial and mesenchymal origin. The tissue distribution of CD151 shows a striking overlap with that of the combined patterns of expression of α3β1,α6β1, α6β4 and α7β1 - all laminin-binding integrins (Sincock et al.,1997). This suggests that CD151 preferentially interacts with the laminin-binding integrins.

Laminin is a basement membrane component that, by interaction with its cell surface receptors, influences cell shape, movement and differentiation(Colognato and Yurchenko, 2000). Besides representing a ligand for integrins,laminin also interacts with the recently discovered Lutheran (LU) molecule expressed on erythrocytes (El Nemer et al., 1998; Parsons et al.,2001; Udani et al.,1998) and with dystroglycan (DG), an integral component of the dystrophin glycoprotein complex (DGC) originally detected in skeletal muscle and subsequently in epithelia (Ibraghimov-Breskovnaya et al., 1992;Durbeej and Campbell,1999).

In the present study we investigated the expression of various tetraspanins and laminin receptors in tissues that strongly express one or more of the laminin-binding integrins, such as kidney (α3β1 in glomerular cells; α6β1 in tubular cells), skin (α3β1 andα6β4) and muscle (α7β1). At sites of contacts between cells and the matrix, CD151 is the only one of the tetraspanins tested that is always colocalized with the laminin-binding integrins. The intimate relation between CD151 and the laminin-binding integrins is also stressed by their high stoichiometric interaction and the increase of CD151 cell surface expression in K562 cells transfected with one of these integrins. By immunoprecipitation it was demonstrated for the first time that CD151 is associated withα7β1, and that their firm association is comparable to that of the association between α3β1 and CD151. Cytoplasmic or extracellular splice variants or mutations preventing cleavage of the integrin αsubunit were found to have no influence on the formation of integrin-CD151 complexes. Although there is a considerable overlap in expression between the laminin receptors from the three different receptor families and CD151, no interaction between CD151 and either LU or DG could be identified in vitro. Finally, we provide evidence that the same site on CD151 interacts with all the different laminin-binding integrins. Furthermore, the exposure of a CD151 epitope involved in this binding is cell type specific and may vary depending on the (patho) physiological status of the tissues involved.

Mouse mAbs J8H against the extracellular domain of the human α6 integrin subunit and 29A3 against the cytoplasmic domain of the humanα3A subunit have been described previously(Hogervorst et al., 1993,Hogervorst et al., 1993;de Melker et al., 1997). The hybridomas producing the mouse mAbs J143, recognizing an extracellular epitope on the human α3 integrin subunit(Kantor et al., 1987) and TS2/16 against β1 (Hemler et al.,1984), were obtained from the American Type Culture Collection(Rockville, MD). Sam-1 is a mouse mAb against α5(Keizer et al., 1987), and GoH3 is a rat mAb against α6(Sonnenberg et al., 1987). The mouse mAbs CA25 and 13C2 against α7 and αv were generously provided by R. H. Kramer (University of California San Francisco, San Francisco, CA) and M. Horton (Imperial Cancer Research Fund, London, UK),respectively. Rat mAb 439-9B against β4(Falcioni et al., 1988) was purchased from Pharmingen (San Diego, CA). The rabbit polyclonal antibodies against the cytoplasmic domains of α3A(DiPersio et al., 1995),αv (Defilippi et al.,1991) and α7B were kind gifts from M. DiPersio (Albany Medical College, Albany, NY), G. Tarone (Università di Torino, Torino,Italy) and U. Mayer (University of Manchester, Manchester, UK), respectively. Mouse mAbs against α-dystroglycan were obtained from Novocastra(Newcastle, UK). Polyclonal antibody AP83 against β-dystroglycan was kindly provided by K. P. Campbell and M. Durbeej (Howard Hughes Medical Institute, Iowa, USA). Mouse mAb LU4F2 (workshop 70597) against Lutheran blood group antigens was a kind gift from E. van der Schoot (Central Laboratory for the Blood Transfusion Service, Amsterdam, The Netherlands). The mouse mAbs P48 and P49, also known as 11B1.G4 and 14A2.H1, respectively, were clustered as CD151 in the VI International Leukocyte Typing Workshop(Ashman et al., 1997;Sincock et al., 1997).

The other mouse mAbs against CD151 were all gifts: 5C11 from M. E. Hemler(Dana-Farber Cancer Institute, Boston, MA), 8C3 from K. Sekiguchi (Osaka University, Osaka, Japan), Sfa-1 from H. Hasegawa(Hasegawa et al., 1996) (Ehime University, Ehime, Japan), LIA1/1 from F. Sánchez-Madrid (Hospital de la Princesa, Madrid, Spain), IIG5 from M. Humphries (The University of Manchester, Manchester, UK) and TS151 and TS151R(Yauch et al., 2000;Serru et al., 1999) from E. Rubenstein (Hôpital Paul Brousse, Villejuif Cedex, France). The mouse mAbs MEM62 (CD9;(Hořejší and Vlček, 1991)), MEM53 (CD53(Angelisová et al.,1994)), M38 (CD81 (Imai and Yoshie, 1993)) and C33 (CD82(Imai and Yoshie, 1993)), were generously provided by V. Hořejší (Institute of Molecular Genetics, Prague, Czech Republic). F. Berditchevsky (The University of Birmingham, Birmingham, UK) kindly supplied the mouse mAb 6H1 (CD63(Berditchevsky et al., 1995)). The sheep anti-mouse and donkey anti-rabbit horseradish-peroxidase-coupled secondary antibodies were purchased from Amersham (Harlington Heights, IL). FITC-or Texas Red-conjugated goat anti-mouse, goat anti-rat or goat anti-rabbit secondary antibodies were obtained from Rockland (Gilbertsville, PA). The FITC- and TRITC-conjugated goat anti-mouse isotype specific antibodies were obtained from Nordic Immunochemical Laboratory (Tilburg, The Netherlands). Phycoerythrin-conjugated goat anti-mouse or anti-rat antibodies were obtained from Jackson laboratories(Westgrove, PA).

The full-length c-DNA encoding the X1 and X2 splice variants of the extracellular domains of human α7B were a kind gift from D. Mielenz and K. von dem Mark (University of Erlangen, Erlangen, Germany). α7BX1 and-X2 c-DNA was released from pUC-18 or pcDNA3 respectively, by digesting with EcoRI. The resulting fragments were ligated into the retroviral LZRS-IRES-zeo expression vector, a modified LZRS retroviral vector conferring resistance to zeocin (Kinsella and Nolan,1996; van Leeuwen et al.,1997), to give an LZRS-IRES-zeo-α7BX1 or -α7BX2 vector. These constructs were then introduced into the Phoenix packaging cells(Kinsella and Nolan, 1996) by the calcium phosphate precipitation method and virus containing supernatant was collected. K562 cells were infected with the recombinant virus by the DOTAP method (Boehringer Mannheim Corp.). After incubation for 8 hours at 37°C, infected cells were grown in medium as described below for the wild-type K562 cells. Cells expressing α7 at their surface were analyzed and sorted using FACScan® flow cytometer (Becton Dickinson, Mountain View,CA).

The human erythroleukemic cell line K562 was maintained in RPMI-1640 supplemented with 10% heat-inactivated fetal calf serum (Life Technologies,Paisly, UK), 100 U/ml penicillin and 100 μg/ml streptomycin (both from Life Technologies). K562 cells stably expressing α3β1, α6β1 and α6β4 were established as described previously(Delwel et al., 1993;Niessen et al., 1994). An established cell line of human glomerular visceral epithelial cells(Krishnamurti et al., 1996)was cultured in medium made up of 1 volume of each Dulbecco's modified Eagle's medium (DMEM) (ICN, Costa Mesa, CA) and Hams F-10 (Life Technologies),supplemented with 5% Nu serum (Becton Dickinson, Bedford, MA), 25 ng/ml prostaglandin E₁, 0.5 nM tri-iodothyronine, 10 nM sodium selenite,5 μg/ml transferrin, 50 nM hydrocortisone, 5 μg/ml insulin (all hormones from Sigma, St. Louis, MO), 100 U/ml penicillin and 100 μg/ml streptomycin. Human proximal tubular epithelial cells (PTEC) also known as HK-2 cells are HPV 16-immortalized cells purchased from the American Type Culture Collection. Cells were grown in conditioned medium consisting of 1 volume of DMEM and 1 volume Ham F-12, supplemented with 5% heat-inactivated FCS, 100 U/ml penicillin and 100 μg/ml streptomycin, 2 mM L-glutamine, 5 μg/ml insulin, 5 μg/ml transferrin, 5 ng/ml sodium selenite, 20 ng/ml tri-iodothyronine, 5 ng/ml hydrocortisone, 5 ng/ml prostaglandin E₁and 5 ng/ml epidermal growth factor (all from Sigma, St. Louis, MO). Cells were grown at 37°C in a humidified, 5% CO₂ atmosphere.

Transfected K562 cells, stably expressing α3A, α3B, α6A,α6B or α6A^RGGD mutant, were surface labeled with ¹²⁵I by the lactoperoxidase method as previously described(Sonnenberg et al., 1987). Cells were solubilized in lysis buffer containing 1% (w/v) CHAPS (Sigma), 5 mM MgCl₂, 25 mM Hepes, pH 7.5, 150 mM NaCl and proteinase inhibitors(1 mM phenylmethanesulphonyl fluoride, 10 μg/ml soybean trypsin inhibitor and 10 μg/ml leupeptin). Lysates were clarified at 20,000 gand precipitated with protein A-sepharose beads CL-4B (Amersham Pharmacia Biotech Inc., Uppsala, Sweden), which had previously been sequentially incubated with rabbit anti-mouse IgG and the precipitating antibody. After incubation for 1.5 hour at 4°C, the beads carrying the immune complexes were washed and treated with sodium dodecyl sulfate (SDS) sample buffer. Precipitated proteins were analyzed by SDS-polyacrylamide gel electrophoresis.

Wild-type and transfected K562 cells stably expressing αvβ1,α3β1, α6β1, α6β4, α7X1β1 orα7X2β1, and podocytes and PTEC were lysed in 1% (w/v) CHAPS buffer with proteinase inhibitors. Alternatively, cells were lysed in 1% (v/v)Nonidet P-40, 50 mM Tris-HCl, pH 7.5, 150 mM NaCl and 1 mM EDTA. Lysates were incubated for 1 hour at 4°C with Gamma Bind-G Sepharose beads or protein A-Sepharose beads (Amersham Pharmacia Biotech, Uppsala, Sweden), previously incubated overnight at 4°C with the precipitating antibodies. The beads carrying the immune complexes were washed three times with lysis buffer and two times with PBS at 4°C. The immune complexes were eluted by the addition of sample buffer, heated at 95°C or 65°C and separated by SDS-PAGE under reducing (2% β-mercaptoethanol) or non-reducing conditions. After electrophoresis, gels were electrophoretically transferred to a PVDF membrane (Immobilon-P, Bedford). Blots were stained with Coomassie blue to indicate the markers, destained in 45% methanol, 5% acetic acid in demineralized water and blocked for 30 minutes with 5% dry milk in TBST-buffer(10 mM Tris, pH 7.5, 150 mM NaCl, 0.3% Tween-20). Subsequently blots were incubated for 1 hour with primary antibodies in 0.5% dry milk in TBST. Primary antibodies were mAbs 29A3 against α3 (neat supernatant) or 8C3 against CD151 (1:1,000), or polyclonal antibodies against α3 (1:100), α7B(1:100), αv (1:100) or β-dystroglycan (1:100). After washing three times with TBST/0.5% dry milk, blots were incubated for an additional hour with secondary sheep anti-mouse, donkey anti-rabbit or goat anti-rat Ig-horseradish-peroxidase-coupled, diluted 1:5,000 or 1:1,000 respectively, in 0.5% dry milk in TBST. After washing three times with TBST, proteins were visualized by enhanced chemiluminescence (ECL, Amersham Pharmacia Biotech.),as described by the manufacturer.

The surface expression of integrin subunits and CD151 on wild-type and transfected K562 cells was assessed by flow cytometry. 5×10⁵cells were washed three times with PBS containing 2% FCS (PBS/FCS), followed by incubation with the primary antibody or antibodies for 1hour. The cells were washed three times with PBS/FCS and then incubated with phycoerythrin-conjugated goat anti-mouse or anti-rat antibodies for another hour. After another three washes with PBS/FCS, cells were resuspended in PBSA and analyzed with a FACScan® flow cytometer.

Cryostat sections (3 μm thick) from human kidney, skin, smooth, striated or heart muscle were fixed in acetone for 10 minutes at 4°C. For immunoperoxidase staining, endogenous peroxidase was blocked by 0.1% NaN3 and 0.3% H₂O₂ in PBS for 10 minutes. For blockage, PBS containing 2% BSA (PBSA) was used for 10 minutes, followed by overnight incubation with the primary antibody at 4°C. Anti-rabbit or anti-mouse horseradish-peroxidase-conjugated secondary antibodies (PowerVision,Technologies Co., Daly City, CA) were added for 30 minutes and visualized by hydrogen peroxide and 3′,3′-amino-9-ethyl carbazole (Sigma Chemical Co.). Sections were counterstained with hematoxylin and mounted in Depex (Britisch Drug House Chemicals, Poole, UK). In between steps, sections were washed three times with PBS. For double-labeled immunofluorescence analyses, sections were blocked in PBSA for 30 minutes and incubated with the primary antibody diluted in PBSA overnight at 4°C. The sections were washed three times with PBSA and then, depending on the primary antibodies used, incubated with different combinations of goat anti-mouse, goat anti-rat or goat anti-rabbit FITC- or Texas-Red-conjugated antibodies for 45 minutes. After washing twice with PBS, sections were mounted in Vectashield (Vector Laboratories Inc., Burlingame, CA) and viewed under a Leica TCS NT confocal laser-scanning microscope equipped with an Ar/Kr laser (Leica Microystems,Heidelberg, Germany). Unless otherwise mentioned, both procedures were performed at room temperature.

Interactions between tetraspanins and integrins have been defined in biochemical assays, but there is limited information on the associations of these proteins in tissues. The tetraspanin CD151 has been shown to form tight complexes with α3β1 (Yauch et al., 1998), but in addition it may also associate with other laminin-binding integrins (Fitter et al.,1999; Sincock et al.,1999; Sterk et al., 2001). To assess its distribution relative to that of laminin-binding integrins in vivo, we determined the expression in tissues that express different laminin binding integrins; kidney (glomeruli:α3β1; tubuli: α6β1), skin (α3β1 andα6β4) and muscle (α7β1). In these tissues the distribution of other tetraspanins and of the laminin receptors DG and LU was also assessed.

Analysis of kidney tissues by immunoperoxidase staining, using mAb P48,revealed strong expression of CD151 in glomeruli, at the basal surface of tubular epithelial cells and in vascular smooth muscle cells(Fig. 1A). Double immunofluorescence analysis revealed a striking colocalization withα3β1 (Fig. 2A) and DG (Fig. 2B) within glomeruli in a predominant, continuous GBM-bound pattern. Along the GBM, DG was only partially co-distributed with α3β1(Fig. 2C), whereas LU was absent from the glomerulus but strongly expressed in the subendothelial matrix and along the basal side of the tubules, colocalized with α6β1(Fig. 2D). Like CD9 and CD63,the LU molecule is also focally distributed throughout the cytoplasm of tubular epithelial cells (not shown). Transport of LU to the apical or basolateral side of polarized cells has been described before(El Nemer et al., 1999). This,together with the presence of LU at the site of cell-cell contacts between keratinocytes (see below), suggests that LU is not merely a laminin receptor but has other functions as well. The tetraspanins CD9 (not shown), CD63(Fig. 2E) and CD81(Fig. 2F) are present in glomerular cells but are not colocalized with α3β1 along the GBM. Glomerular cells do not express CD53 or CD82, which however are present on a subset of the distal tubules together with CD9 and CD81, the latter being weakly expressed by all tubular cells (not shown).

Immunolocalization of CD151 in skin revealed strong reactivity at the basolateral surface of basal keratinocytes(Fig. 1B), where the integrinsα3β1 and α6β4 are also concentrated(Sterk et al., 2000). Only very small amounts of DG are present at the dermo-epidermal junction, whereas LU is focally expressed on all basal keratinocyte surfaces (not shown). The tetraspanins CD9 and CD81 are not only present in the basal cell layer but also in the more superficial epidermal cell layers. They are absent from the dermo-epidermal junction but are concentrated at inter-keratinocytic contacts(not shown). CD63 is only expressed by melanocytes present between the basal keratinocytes (not shown).

Skeletal, cardiac and smooth muscle cells all express CD151 at their plasma membrane (Fig. 1C,Fig. 9A-C). Additionally, CD151 is present at the costamers in skeletal and cardiac muscle and the intercalated discs in the heart. In these cells, it is colocalized withα7Bβ1 (Fig. 3A,C)and DG (not shown). Besides CD151, smooth muscle cells express the tetraspanins CD9 and CD81 and some CD53 and CD82 (not shown). Small amounts of CD63 and CD81 are present at the sarcolemma of skeletal and heart muscle and at the intercalated discs. LU is not expressed in myocytes (not shown). In peripheral nerves, also present in the smooth muscle sections,α6β4, α7β1, LU, CD9, CD63, CD81 and CD151 are all localized at the endo- and perineurium (not shown). In summary, CD151 is extensively colocalized with various laminin receptors, both integrins and non-integrins, at sites of cell-matrix interactions.

To determine whether CD151 is also able to associate with the integrinα3 and α6 subunits, K562 cells stably transfected with α3 orα6 were ¹²⁵I-surface-labeled and lysed in 1% CHAPS. As shown in Fig. 4, antibodies againstα3 precipitated α3β1 and antibodies against α6,α6β1 from lysates of α3- and α6 transfected K562 cells,respectively. The transfected integrins were also present in precipitates prepared with antibodies against CD63, CD81 and CD151. However, although large amounts of α3β1 and α6β1 were coprecipitated with CD151,only small amounts were coprecipitated with CD63 or CD81. This difference is not due to differences in the level of expression of the proteins, as all three tetraspanins proved to be present at substantial levels in K562 cells,as shown by FACS analysis (Yauch et al.,1998) (L.M.T.H. and A.S., unpublished). Rather, it is the result of the high stoichiometric binding of CD151 to the laminin-binding integrins,which indicates the unique and specific role of CD151 in the formation of these stable complexes. Furthermore, immunoprecipitates of CD63 and CD151 from the two transfected K562 cell lines, contained a protein of the size of CD81(23 kDa). Because it is known that tetraspanins can interact with other tetraspanins, we suspect that CD81 is coprecipitated because it is associated with CD63 and CD151, and not, as in the case of CD151, with integrins. CD63,which migrates as a smear between 45-60 kDa on gels, could not be detected in the various immunoprecipitations, probably because it is labeled only poorly with ¹²⁵I. As anticipated, similar amounts of the α3 andα6 subunits were coprecipitated with CD151 from lysates of K562 cells transfected with the cytoplasmic A or B variants of α3 or α6 (not shown). Thus, the formation of CD151-α3β1 or CD151-α6β1 complexes is not influenced by the nature of the splice variants in the integrin molecules.

Association of CD151 with either splice variant of the extracellular domain X1 or X2 of the laminin receptor α7β1 was investigated in coimmunoprecipitation experiments; the strength of the interaction was also tested by using different lysis buffers. CHAPS is considered to be a mild detergent in which weak interactions between different molecules are not disrupted, whereas Nonidet P-40 is more stringent leaving only stable complexes intact. As shown in Fig. 5 (upper panel), α7 is coprecipitated with CD151 from K562 cells stably expressing either the X1 or X2 splice variant, and vice versa CD151 is co-precipitated with α7 (lower panel). The α7 subunit and CD151 were precipitated in comparable amounts irrespective of the lysis buffer used. Thus, CD151 not only associates with the α3β1 andα6β1 integrins but also with α7β1, the strength of this interaction being comparable with that with α3β1.

The extracellular domain of the integrin α3, α6 and α7 subunits is cleaved during biosynthesis into a heavy and a light chain. This has been shown to be required for proper insideout signaling byα6β1 in K562 cells but has no influence on ligand binding(Delwel et al., 1996). When the RKKR sequence (amino acids 876-879) is mutated to RGGR, cleavage into a heavy and a light chain cannot occur. For the formation of stable complexes of tetraspanins and integrins, their extracellular domains are essential(Yauch et al., 2000). To determine whether cleavage influences the binding of integrins to CD151, K562 cells were transfected with the α6A^RGGR-mutant (uncleaved). Cells were ¹²⁵I-surface-labeled, lysed in 1% CHAPS and the lysates precipitated with antibodies against α5, α6, β1, CD81 or CD151. As shown in Fig. 6,α6β1 could be precipitated by antibodies against α6 orβ1 as well as by antibodies against CD81 or CD151. Consistent with previous results (Fig. 4), the amount of α6 precipitated by anti-CD81 is much smaller than that precipitated by anti-CD151. The α5β1 integrin that is endogenously present on K562 cells is only precipitated by antibodies against α5 orβ1, not by antibodies against either of the two tetraspanins, CD81 or CD151.

TS151R is a monoclonal antibody directed against an epitope on CD151 that is located at the binding site for α3, whereas P48 recognizes an epitope outside this binding site (Serru et al.,1999; Yauch et al.,2000). To investigate whether the CD151 epitope recognized by TS151R is also involved in the binding to α7β1, lysates of K562 cells stably expressing the integrin α7β1 were subjected to immunoprecipitation with mAb TS151R, and the presence of α7 was tested by immunoblotting. Immunoprecipitations prepared from lysates of K562 cells that were transfected with cDNAs encoding the α3- or αV-subunits served as positive and negative controls, respectively. As shown inFig. 7, the α3 andα7 integrins were only precipitated by P48 and not by TS151R. TheαVβ1 integrin was not precipitated by either of the antibodies. Thus, the TS151R epitope is not only masked when CD151 is associated withα3β1 but also when it is bound to α7β1.

Next we compared the distribution of the CD151 epitope recognized by TS151R with that by P48 (Table 1). Kidney sections incubated with TS151R (Fig. 8A,C) show hardly any fluorescent staining in the glomerular cells that express α3β1 (Fig. 8B,C), whereas in the tubuli expressing α6β1, staining is comparable to that of mAb P48. In the skin, where CD151 is not only associated with α3β1 but also with α6β4(Fig. 8E,F,H,I), the staining by TS151R (Fig. 8G,I) is weak compared with that by mAb P48 (Fig. 8D,F). Staining of the different muscle tissues expressingα7β1 varied: all three muscle tissues were strongly stained by P48(Fig. 1C,Fig. 9A,B), smooth muscle cells were also strongly stained by TS151R (Fig. 9C), cardiomyocytes only weakly at the intercalated discs(Fig. 8J,L), and striated muscle cells were not stained at all by this antibody(Fig. 9D). Staining of cardiomyocytes for α7 expression is shown inFig. 8K,L. For further comparison and the possible recognition of CD151 epitopes involved in the binding of integrins, all the above tissues were incubated with a panel of other antibodies against CD151. This resulted in staining patterns comparable to those produced by P48 (not shown). Thus, binding of CD151 to laminin-binding integrins leads to masking of the TS151R epitope in many, but not all, types of cells in vivo.

Previously it has been shown that expression of α3β1 in K562 cells results in an increase in the surface expression of CD151(Yauch et al., 1998). To determine whether this ability of α3β1 to induce CD151 expression is a general property of laminin-binding integrins, the surface expression of CD151 was assessed by flow cytometry (FACScan) of transfected K562 cells stained with mAb P48. As shown in Fig. 10, the expression of transfected αV had little or no effect on the expression of CD151. However, expression of α3β1,α6β1, α6β4, α7X1β1 or α7X2β1 in K562 cells increased the expression of CD151 three- to sixfold. Staining of the cells with TS151R revealed that the CD151 epitope, which is recognized by this antibody, is masked, and it is not dependent on the type of laminin-binding integrin that the K562 cells were transfected with. These results are consistent with the results of the above biochemical study showing that the TS151R epitope is not expressed on K562 cells expressing different laminin-binding integrins. However, also no CD151 was detected with TS151R on untransfected K562 cells. We therefore cannot exclude non-reactivity of this antibody in fluorescence studies, because of low affinity when the number of available epitopes is restricted. Furthermore, the results indicate that the surface levels of CD151 on cells are strongly influenced by the expression of the laminin-binding integrins.

The colocalization of CD151 with the different kinds of laminin receptors,DG, LU and integrins, may suggest a role for CD151 in linking these three kinds of laminin receptors to each other. To investigate whether CD151, DG and LU are not only colocalized but also complexed, cultured human podocytes and proximal tubular epithelial cells (PTEC) were used for immunoprecipitation studies (Fig. 11). FACS analysis showed that both cell lines express α3β1, CD151, DG and LU, although LU is only weakly expressed by podocytes. These expression profiles correspond with those of podocytes and PTECs in vivo. Lysates from cells prepared with 1% CHAPS were incubated with antibodies against α3,CD151, LU or β-DG. Subsequently the presence of α3 (light chain, 30 kDa) or β-DG (43 kDa) in the precipitates was assessed by immunoblotting using specific antibodies. Protein bands corresponding to α3 could be detected on the immunoblots of the immunoprecipitates containingα3β1 or CD151 from both cell lysates (left panels). LU or DG could not be coprecipitated with α3β1 with any of the cell lysates. P48 against CD151 and antibodies against α3 or LU did not coprecipitate DG(right panels) either. This demonstrates that in the tested cell lines, CD151 bound to the laminin-binding integrins does not associate with the other laminin receptors LU or DG and therefore cannot serve as a link between them.

In a former study we have shown that CD151 is associated withα3β1 and α6β4 in the basal keratinocyte layer of the skin and in hemidesmosomes (Sterk et al.,2000). Now we have investigated the association of CD151 with various laminin receptors in kidney, skin and muscle tissues selected for the laminin-binding integrins they express.

As indicated by this study, there are different distribution patterns of CD151. In the first pattern most of the CD151 is colocalized with the laminin-binding integrins present and thus is mainly expressed at the site of cell-matrix interactions where there is a basement membrane containing at least one laminin isoform (Miner et al.,1997; Aumailley and Rouselle,1999). Thus, in glomerular cells, CD151 and α3β1 are colocalized at the GBM side of cells. In renal tubular epithelial cells, CD151 and α6β1 are concentrated at the tubular basement membrane, and in skin CD151 and α3β1 and α6β4 are present at the dermo-epidermal junction. Finally in myocytes, CD151 together withα7β1 is expressed at the sarcolemma. The second pattern is seen in cells such as endothelial and most hematopoietic cells in which the amount of laminin-binding integrins is limited. As shown by others, these cells contain,in addition to a considerable pool of unbound CD151, CD151 that is weakly associated with α5β1 and other surface molecules(Hasegawa et al., 1998;Fitter et al., 1999;Sincock et al., 1999). The third distribution pattern is seen in fibroblasts (stromal cells) and K562 cells, which contain CD151 but no laminin-binding integrins.

In addition to the previously reported association between CD151 and the laminin-binding integrins α3β1, α6β1 andα6β4, we demonstrate here that CD151 also interacts withα7β1. Like CD151-α3β1 complexes, CD151-α7β1 complexes remained intact after Nonidet P-40 lysis of transfected K562 cells,which classifies their interaction as level 1. Also using transfected K562 cells, it was shown that CD151-α6β1 and CD151-α6β4(Sterk et al., 2000) interact at level 2, the interaction not being resistant to Nonidet P-40 treatment. A correlation between the presence of the laminin-binding integrins and CD151 expression was established by FACS analysis, which showed an increase of CD151 expressed at the cell surface of K562 cells after transfection of either of the laminin-binding integrins.

Two non-integrin receptors for laminin are DG and LU. DG was originally isolated from skeletal muscle as a component of the large oligomeric DGC,which amongst other proteins contains sarcospan and dystrophin(Durbeej and Campbell, 1999). The third type of laminin receptor is the molecule expressing the LU antigens,which is able to bind to the laminin α5 chain present in the subendothelial matrix (Parsons et al.,2001).

Our in vivo results show a considerable overlap in the distribution of these different types of laminin receptors. Hence, a tight regulation and precise coordination of the signaling mechanisms by the different receptors seems essential. An example of an interaction between the different laminin receptors is shown in glomerulogenesis, during which a switch occurs in expression of the laminin-binding integrins from α6β1 toα3β1 simultaneously with a switch in the deposition of laminin isoforms, that is, from laminin-1 to laminin-11(Sterk et al., 1998). Others have shown that the symptoms induced by the absence of dystrophin, which disturb the function of the laminin receptor DG, can be partly compensated by increasing the expression of α7β1(Burkin et al., 2001). Moreover, myocytes of patients with Duchenne's disease lacking dystrophin and of dystrophin-deficient mice express an increased amount of α7β1(Hodges et al., 1997). This indicates that when one laminin receptor is lacking, another laminin receptor can (partially) compensate for it. A protein present in the DGC that is homologous to tetraspanins is sarcospan(Crosbie et al., 1999). Although sarcospan is a component of the DGC and has a function in maintaining the integrity of muscle tissue, sarcospan-deficient mice have only mild symptoms of muscular atrophy (Lebakken et al., 2000). Moreover, epithelial cells expressing DG do not contain sarcospan (Durbeej and Campbell,1999). These data suggest that in sarcospan-deficient mice and epithelia expressing DG, other tetraspanins may compensate for the absence of sarcospan. We demonstrate that, except in the skin, in all tissues tested, DG,the laminin-binding integrins and CD151 are co-distributed. Furthermore, this study shows that CD151 is the only known tetraspanin in the glomerulus that is present at the side of the basement membrane, together with DG. Therefore and analogous to the sarcolemma, the sarcospan function in the glomerulus could be taken over by CD151, thereby connecting the integrin family with the DGC. However, despite their co-distribution, coimmunoprecipitation studies could not demonstrate a physical association between α3β1-CD151 complexes and DG or LU in two different cell lines obtained from the kidney, that is,human podocytes and proximal tubular epithelial cells.

The binding between tetraspanins and integrins is established via the extracellular domains of the tetraspanin and the integrin α subunit(Yauch et al., 2000). Theoretically, splicing of the mRNA for the extracellular or cytoplasmic domains of integrin subunits or maturation-induced cleavage of the αchain could influence the association of tetraspanins with integrins and thereby alter integrin signaling. However, biochemically no such differences could be demonstrated for the cytoplasmic A and B variants variants of the integrin α3 and α6 subunits or for the extracellular variants X1 and X2 of α7. Furthermore, mutational changes in the extracellular domain of α6 preventing cleavage seemed to have no influence on the formation of a complex with CD151. This implies that switches of splice variants as described during development(Collo et al., 1993;Thorsteindóttir et al.,1995; Sterk et al.,1998) or cleavage of the α subunit during maturation have no effect on CD151-integrin binding.

The reactivity of CD151 with TS151R and P48 in immunohistochemistry indicates that masking of the TS151R epitope detected in the biochemical experiments and FACS analysis also occurs in vivo. Glomerular cells expressingα3β1, basal keratinocytes expressing α3β1 andα6β4 and cardiac and skeletal muscle cells expressingα7β1 either did not or only weakly react with TS151R, although these cell types were strongly stained by mAb P48. A likely interpretation of this finding is that in these tissues most if not all of the CD151 is bound to a laminin-binding integrin, leading to masking of the TS151R epitope. However,renal tubular and smooth muscle cells, which express α6β1 andα7β1, respectively, were stained equally well by the two antibodies. The reason for this observation is not clear, but there are several explanations. First, the strength of the interaction between CD151 and integrins might be regulated from within the cell and/or by the microenvironment, by which the tightness of the binding may be influenced and thus the exposure of the TS151R epitope. Second, CD151 may not directly interact with one of the laminin-binding integrins but indirectly by forming a tetraspanin web with other molecules. In fact, different binding sites have recently been described on CD151 for tetraspanins and integrins(Berditchevski et al., 2001). Third, CD151, in addition to being complexed with a laminin-binding integrin,could also be bound to other cell surface molecules, as has been described for different tetraspanins and MHCII, DRAP27 and HB-EGF(Angelisova et al., 1994;Nakamura et al., 1995). This binding should, in that case, involve another part of the binding site than that which interacts with integrins. Fourth, in kidney sections from HIV patients, we found that the amount of CD151 reactive with TS151R is greatly increased in the glomeruli and comparable with that reactive with P48(L.M.T.S., unpublished). This indicates that under certain pathophysiological conditions the TS151R epitope on CD151, which is normally complexed withα3β1, becomes exposed. Finally, a difference in affinities of the two antibodies for CD151, as suggested by our FACS data, may have influenced their reactivity.

In summary, we demonstrate extensive co-distribution of various kinds of laminin receptors in different tissues. We confirm previous findings that CD151 binds to the laminin-binding integrins α3β1 andα6β1 and show for the first time that it also interacts withα7β1. CD151 does not bind to the laminin receptors DG or LU. Moreover, we show that variants of the cytoplasmic or extracellular domains bind equally well to CD151 and that cleavage of the integrin α subunits does not alter their binding. Studies with the mAb TS151R, which has previously been shown to recognize an epitope at the binding site of CD151 with α3β1, indicate that the same site is involved in the binding to α7β1 and possibly also to α6β1 and α6β4. The finding that in immunohistochemical studies few cells and tissues,irrespective of the laminin-binding integrins they express, reacted with TS151R, suggests that the strength of interaction of CD151 is subject to regulation either from within the cell and/or by the cell's environment.

We thank A. Pfauth and E. Noteboom for their assistance with the FACS analysis of cells. We also thank L. Oomen for advice on the use of the CLSM and E. Danen and P. Engelfriet for critical reading of the manuscript. We are grateful to many of our colleagues for providing us with antibody reagents and would especially like to thank D. Mielenz and K. von dem Mark for the full-length cDNAs for the α7BX1 and α7BX2 splice variants. This work was supported by a grant from the Dutch Kidney Foundation (C 96.1581).