1167
Journal of Cell Science 113, 1167-1176 (2000) Printed in Great Britain © The Company of Biologists Limited 2000 JCS0892
Interaction between the cytodomains of the α3 and β1 integrin subunits regulates remodelling of adhesion complexes on laminin Emmanuel Laplantine1, Laurent Vallar2, Karlheinz Mann3, Nelly Kieffer2 and Monique Aumailley1,* 1Institut II für Biochemie, Faculty of Medicine, University of Cologne, 50931 Cologne, Germany 2Laboratoire Franco-Luxembourgeois de Recherche Biomédicale (CNRS and CRP-Santé), L-1511
Luxembourg, Grand Duchy of
Luxembourg 3Max-Planck-Institut für Biochemie, 82152 Martinsried, Germany *Author for correspondence (e-mail:
[email protected])
Accepted 28 January; published on WWW 7 March 2000
SUMMARY The first step of laminin 1-induced signal transduction is β1 integrin-specific initiated by the formation of α6β adhesion complexes. In contrast, on other laminin β1 integrinisoforms the adhesion complexes are α3β β1 specific due to a transdominant regulation of the α6β β1 integrin. To determine the integrin by the α3β mechanism of this regulation, peptides representing the cytoplasmic domain of the α3 or α6 integrin subunits were microinjected together with recombinant enhanced green fluorescence protein into live fibroblasts. Microinjection of the α3 integrin peptide to laminin 1-adherent cells β1 integrin-specific adhesion complexes displaying α6β β1 integrin, while resulted in the disengagement of the α6β microinjection of green fluorescence protein alone or in
combination with the α6 integrin cytodomain had no effect. Further surface plasmon resonance studies revealed that the cytodomain of the β1 integrin subunit interacts with low affinity with the cytoplasmic tail of the α3 integrin subunit, but not with that of several other α subunits including α6. These results imply that the cytoplasmic tails of the integrin α subunits play a critical role in the regulation of integrin-induced signal transduction. In particular, the intracellular tail of the α3 integrin subunit controls the formation of adhesion complexes in cells adhering to laminins.
INTRODUCTION
nucleation sites for the association of signalling and cytoskeletal-linker proteins. Although it is not known whether and when all these intracellular interactions are required for correct integrin signalling, some at least should be strictly regulated since expression of a transgene coding for the cytoplasmic and transmembrane domains of the β1 integrin chain results in dominant negative functional alterations of several integrins of the β1 family (LaFlamme et al., 1994; Lukashev et al., 1994; Chen et al., 1994; Faraldo et al., 1998). Less is known about potential binding partners for the cytoplasmic tail of the integrin α subunits, although it can be intuitively assumed that α subunit-specific regulation mechanisms are necessary for integrin function. All integrin α subunits contain a highly conserved KXGFFKR motif adjacent to the transmembrane domain and differ in their carboxyterminal tail. A potential modulator of integrin activity is calreticulin, which requires the presence of the GFFKR conserved sequence for integrin binding (Rojiani et al., 1991). Using the cytodomain of the α3 integrin subunit as bait in a yeast two-hybrid screen, we have identified several other intracellular proteins that interact with the conserved region of diverse integrin α subunits (Wixler et al., 1999). Transmembrane proteins may also regulate integrin activity, and a stable and specific association between the α3β1 integrin
Cellular interactions with the extracellular matrix lead to integrin activation and consequent triggering of intracellular signalling. This results in the transfer of information between the cell’s environment and interior (LaFlamme et al., 1997). The mechanisms of these events are not well understood. Since integrin cytoplasmic domains are devoid of enzymatic activity, signal transmission presumably requires the propagation of integrin conformational changes and their association with intracellular partners such as signalling and cytoskeleton-linker proteins (Yamada and Geiger, 1997). The quiescent or activated state of integrins may involve different intracellular interactions and in vitro binding studies or interaction cloning in yeast have revealed that a number of proteins interact with different parts of the cytoplasmic tail of the β1 integrin subunit. These include talin (Horwitz et al., 1986; Burn et al., 1988), α-actinin (Otey et al., 1990, 1993), FAK (focal adhesion kinase) and paxillin (Schaller et al., 1995), ILK (integrinlinked kinase; Hannigan et al., 1996), ICAP-1 (integrin cytoplasmic domain associated protein-1; Chang et al., 1997), Rack1 (Liliental and Chang, 1998), and filamin (Loo et al., 1998). Moreover, the integrin β cytoplasmic domain can form oligomers (Zage and Marcantonio, 1998), which could be
Key words: Adhesion complex, Integrin, Laminin, Microinjection
1168 E. Laplantine and others and CD151, a protein of the transmembrane-4 superfamily, has been observed (Yauch et al., 1998) while association of α2β1 or αIIbβ3 integrins with CD47 has been reported (Chung et al., 1997; Wang and Frazier, 1998). Caveolin also appears to associate with certain integrins and to modulate their functions (Wary et al., 1996). Laminins induce activation of specific integrins, including integrin α6β1 for laminin 1 and integrins α3β1 and α6β1 for other isoforms (for review see Aumailley and Rousselle, 1999). Activation of the α6β1 integrin alone or the α6β1 together with α3β1 integrin leads to the formation of adhesion complexes that differ in morphology. More specifically, human fibroblasts and several transformed epithelial cells develop on laminin 1 distinct adhesion complexes in which integrins and cytoskeletal linker proteins are clustered at the end of actin stress fibers (Dogic et al., 1998; Sondermann et al., 1999). In contrast, upon adhesion to other laminin isoforms, the same cells form smaller and more diffuse adhesion complexes distributed in a punctuate pattern along actin microfibrils, possibly due to a transdominant regulation of α6β1 by α3β1 (Dogic et al., 1998, 1999). The regulatory role of the latter may also apply to fibronectin- or collagen IV-binding integrins, as shown by comparing the cytoskeleton arrangement and the functional phenotype of keratinocytes derived from wild-type or α3 integrin subunit-deficient mice (Hodivala-Dilke et al., 1998). To address the mechanism by which the α3 integrin subunit could negatively regulate integrin clustering and the formation of adhesion complexes, we have used peptides representing the cytoplasmic domains of several integrin subunits for in vivo microinjection experiments in fibroblasts as well as for in vitro interaction studies performed by surface plasmon resonance.
MATERIALS AND METHODS Peptides representing the cytoplasmic domains of the α3, α6, αv, αIIb and β1 integrin subunits The intracellular domain of the α3A (1016C-1052Y) and α6A (1038C-1073A) integrin subunits were automatically synthesised (Automated synthetiser model 433A, Applied Biosystems, Weiterstadt, Germany) using fast-moc (TM) chemistry in a 0.1 mM scale. Peptide purification was performed by preparative reverse phase HPLC on a Resource RPC column (Pharmacia Biotech, Uppsala, Sweden) using a linear gradient of 0-100% acetonitrile in 0.1% TFA at a flow rate of 1 ml/minute. After analysis by SDS-PAGE, the peptide-containing fractions of several runs were pooled, lyophilised and kept at −20°C. The synthetic peptide corresponding to the αv cytoplasmic domain (987R-1018T) was kindly provided by Dr G. Lippens (Institut de Biologie de Lille, Lille, France), and that corresponding to the αIIb cytoplasmic domain (989K-1008Q) was purchased from Neosystem (Strasbourg, France). The cDNA coding for the β1 integrin cytodomain (752K-799K) was generated by RT-PCR from total RNA of HBL100 cells with sense (5′-CCGGAATTCTGCAAGCTTTTAATGATAATTCATGAC3′) and antisense (5′-AAATGTCCTGCTAGCTAGTTGGATCCCTCGAGCGGA-3′) oligonucleotides used as amplification primers. The underlining indicates the nucleotides added for cloning and the italics denote an inserted codon for an additional cysteine. Briefly, 100 ng of total RNA were reverse transcribed with 1 U M-MLV reverse transcriptase (Appligen) in the presence of the appropriate primers (20 pM) for 30 minutes at 37°C. After deactivation of the transcriptase (2 minutes at 95°C), the cDNAs were amplified using 1 U AmpliTaq
DNA polymerase (Perkin Elmer) for 30 cycles of 30 seconds at 95°C, 1 minute at 56°C, 1 minute at 72°C and finally extended for 10 minutes at 72°C. The resulting cDNAs were digested by EcoRI and XhoI, purified and cloned into the pGEX-4T1 procaryotic vector in frame with the GST coding sequence (Amersham-Pharmacia Biotech). The constructs were confirmed to be mutation free and inframe by nucleotide sequence analysis (Perkin-Elmer, ABI Prism 377). Transformed bacteria were cultured at 37°C in ampicillincontaining medium (50 µg/ml) and induced by 0.1 mM isopropyl βD-thiogalactopyranoside (IPTG) for 2 hours at 30°C. The bacteria were recovered by centrifugation, resuspended in PBS and submitted to lysozyme treatment (1 mg/ml) and sonication. After lysis of the bacteria, the fusion proteins were first extracted with 1% Triton X100 for 30 minutes at 4°C. The lysates were cleared by centrifugation and the pellets were re-extracted with 7 M urea in TBS for 3 hours at 4°C. The proteins in the urea extracts were renatured by successive dialysis against 5 M, 3 M, 1 M and no urea in 1 mM DTT-containing TBS. Solubilised GST-fusion proteins were affinity purified on glutathione-Sepharose beads (Amersham-Pharmacia Biotech) in TBS, washed and either used as GST fusion proteins or cleaved by thrombin (20 units/mg protein) to remove GST. The resulting digests were fractionated by molecular sieve chromatography on a Sephadex G50 superfine column (Pharmacia Biotech). The fractions containing the peptides were pooled, lyophilised and kept at –20°C. Analytical methods Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDSPAGE) was performed on 15% acrylamide gels under reducing (5% β-mercapthoethanol) or non-reducing conditions. Bands were detected using silver nitrate staining. Low molecular mass standard were from Fluka (Deisenhofen, Germany). Purity and correct peptide sequences were confirmed by automatic sequencing (Model 473, Applied Biosystems). For alkylation, peptides were resuspended in 0.1 M Tris-HCl, pH 8, 1 mM EDTA, reduced with 20 mM DTT for 2 hours and alkylated with 40 mM iodoacetamide for 1 hour. The alkylated peptides were re-purified by HPLC as above. Cell cultures, adhesion supports and microinjection The human lung fibroblast line Wi26 (American Type Culture Collection), kindly provided by Dr Roswitha Nischt (University of Cologne, Cologne, Germany), was cultivated in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 2 mM glutamine, antibiotics, and 10% fetal calf serum. Cells were subcultured using 0.05% trypsin, 0.02% EDTA in phosphate-buffered saline (PBS), pH 7.4. All cell culture fluids were from Seromed/Biochrom (Berlin, Germany) and plastic ware from Falcon (distributed by Faust, Cologne, Germany). Freshly suspended fibroblasts (3-5×105 cells/ml) were seeded either in the presence of serum-containing culture medium on uncoated glass coverslips and further cultivated for 24 hours prior to microinjection, or incubated for 90 to 120 minutes in serum-free, 20 mM Hepes (pH 7.2)-containing medium on glass coverslips (Faust) precoated with laminin 1 or fibronectin (10-20 µg/ml). Laminin 1 (laminin-nidogen complex extracted from the Engelbreth-Holm-Swarm tumor of the mouse according to Paulsson et al., 1987) or bovine plasma fibronectin (Miekka et al., 1982) were kindly provided by, respectively, Dr R. Timpl (Max-Plank Institut, Martinsried, Germany) and Dr M. Paulsson (University of Cologne, Cologne, Germany). Purified peptides and recombinant enhanced green fluorescent protein (rEGFP, Clontech) were resuspended at a final concentration of 0.33 mg/ml each in 140 mM potassium glutamate, 20 mM NaCl, 2 mM MgCl2, 4.5 mM CaCl2, 5 mM EGTA, 10 mM Hepes, pH 7.2. The solution was loaded by capillarity into needles prepared in the laboratory (Flaming/Brown Micropipette Puller, Sutter Instruments, distributed by Bachoffer, Reutliengen, Germany) and injected into cells at a constant flow with a pneumatic Pico Pump (PV 830, Helmut Saur Laborbedarf, Reutliengen, Germany) connected to a
Integrin cytoplasmic domains 1169 micromanipulator (Leica, Heidelberg, Germany). The cells were maintained at 37°C for 20 to 90 minutes and further processed for immunofluorescence staining. Immunofluorescence staining The cells were fixed with 2% paraformaldehyde in PBS for 15 minutes, permeabilised with 0.2% Triton X-100 for 1 minute and incubated with either mAb F-VII against human vinculin (a gift from Dr M. Glukhova, Institut Curie, Paris, France) mAb K20 (Immunotech, Marseille, France) against the β1 integrin subunit, followed by Cy3-conjugated secondary antibodies against mouse immunoglobulins (Jackson, distributed through Immunotech). Fibrillar actin was stained with phalloidin coumarin phenyl isothiocyanate (CPITC) from Sigma (Deisenhofen, Germany). The samples were mounted in 9:1 (v/v) glycerol/PBS and observed with an Axiophot microscope (Zeiss, Oberkochen, Germany) equiped with epifluorescence optics. Photomicrographs were taken on Kodak TMax 400 film for black and white prints or on Fujichrome Sensia II 400 for color slides. Quantification of immunofluorescence observations Printed micrographs of cells immunolabeled for vinculin were used to count the number of cells with typical or abnormal stainings by two independent observers. Typical vinculin staining was defined as thick and distinct patches of fluorescence as shown in Fig. 3 (filled arrowheads). Absence, dot-like or thin streaks of fluorescence as shown in Fig. 3 (open arrowheads and thin arrows, respectively) were classified as abnormal stainings. After immunostaining of the β1 integrin subunit and photography, color slides were scanned with a silver-fast Nikon equipment and screen-projected using Photoshop and Corel Draw softwares. The length of fluorescence patches was measured using the internal ruler provided with the software. Surface plasmon resonance Real time biomolecular analysis was performed using the Biacore XTM instrument (Biacore, Uppsala, Sweden). Peptides corresponding to the integrin β1 cytoplasmic tail were covalently attached by thiol coupling to carboxymethyldextran (CM5) chips (Biacore) activated with a mixture of N-hydroxysuccinimide and N-ethyl-N′dimethylaminopropyl carbodiimide followed by pyridinyldithioethaneamine according to the manufacturer’s instructions. Alternatively, the cytodomain of the integrin β1 subunit in the form of a GST fusion protein was captured by polyclonal antibodies against GST (Biacore) immobilised onto a CM5 sensor chip through amine coupling using EDC/NHS chemistry. The experiments were performed at 25°C in either Biacore buffer (150 mM NaCl, 3.4 mM EDTA, 0.005% surfactant P20, 10 mM Hepes, pH 7.4) or TBS-Bia (150 mM NaCl, 0.005% P20, 50 mM Tris-HCl, pH 7.4) with or without 2 mM CaCl2. The buffers were passed over the sensor chip at a constant flow-rate, and were always used to prepare sample dilutions in order to reduce bulk effect. At the end of each cycle of analysis, the sensor was regenerated with a short pulse of 10 mM glycine-HCl, pH 2.2. The amount of analyte bound to the immobilised ligand was monitored by measuring the variation of the plasmon resonance angle as a function of time. Results were expressed in Resonance Units (RU), an arbitrary unit specific for the Biacore instrument (1000 RU correspond to approximately 1 ng of bound protein/mm2 and are recorded for a change of 0.1 degree in resonance angle). The background signal was systematically subtracted from the recorded value by simultaneous injection of samples over a blank surface (reference cell) where dextran was substituted with 50 mM lcysteine, 1 M NaCl solution for β1 peptide-coated chip, or 1 M ethanolamine for anti-GST coated surface. The transformation of crude data, the preparation of overlay plots, and the determination of kinetic parameters were performed using the Biaevaluation 3.0 software.
RESULTS Peptide synthesis, purification and sequencing After chemical synthesis, peptides corresponding to the cytoplasmic domains of the α3 and α6 integrin subunits were purified by preparative reverse phase HPLC (Fig. 1). Those representing the β1 integrin subunit were expressed recombinantly and purified by affinity and molecular sieve chromatographies (not shown). SDS-PAGE analysis and staining of the gels by silver nitrate revealed that under reducing conditions, the α3, α6 and β1 peptides displayed migration mobilities corresponding to 5.5, 4.7, and 3.8 kDa, respectively (Fig. 2). These molecular masses agree with the theoretical values calculated for the cytodomains of α3 and α6 integrin subunits. For unknown reasons, the peptide corresponding to the cytodomain of the integrin β1 subunit migrated faster than expected. Nevertheless, as for the α peptides, the amino acid sequence was found to be correct by automated sequencing (Table 1). In agreement with the presence of a cysteine residue at the N terminus of the peptides, as in the authentic cytoplasmic domain sequences, the integrin
Fig. 1. Peptide purification by preparative reverse phase HPLC. The α3A and α6A integrin subunits were synthesised using fast-moc (TM) chemistry and purified by preparative reverse phase HPLC on a Resource RPC column using a linear gradient of 0-100% acetonitrile in 0.1% TFA at a flow rate of 1 ml/minute. The eluted fractions were analysed by SDS-PAGE and the bands were revealed by silver staining of the gels. The fractions containing a single band migrating with the expected electrophoretic mobility were pooled as shown by the bar over the peak and used for further studies. (A and B) α3 peptides; (C and D) α6 peptides; before (A and C) and after reduction and alkylation (B and D).
1170 E. Laplantine and others Table 1. Peptide sequencing for the intracellular domains of the human α3A and α6A integrin subunits α3A: XGFFKRARTRALYEAKRQKAEMKSQPSETERLTDDY α6A: XGFFKRNKKDHYDATYHKAEIHAQPSDKERLTSDA β1A: gspef(X)KLLMIIHDRREFAKFEKEKMNAKWDTGENPIYKSAVTTVVNPKYEGK
Fig. 2. SDS-PAGE of purified peptides corresponding to the intracellular domains of the α3, α6, and β1 integrin subunits. Synthetic (α3, α6) or recombinant (β1) peptides were purified by chromatography and loaded on 15% acrylamide gels under reducing and non reducing conditions as indicated on the figure. ‘alk’ denotes peptides which were reduced and alkylated. The bands were revealed by silver nitrate staining of the gel.
α3 and α6 subunit peptides migrated as dimers under non reducing conditions, and reduction and alkylation was necessary to obtain a preparation of peptide monomers (Fig. 2). As judged by chromatographic profiles, silver-stained SDSPAGE gels and sequencing data contaminating species were not present in the peptide preparations. Microinjection of integrin cytoplasmic domains in live fibroblasts Human Wi26 fibroblasts were plated on laminin 1, which is not a ligand for the α3β1 integrin, and incubated on the coat to allow adhesion and spreading. Once spreading was achieved, the cells were co-injected with rEGFP and dimers of the cytodomain of the α3 or α6 integrin subunits. After 30 to 40 minutes, the fibroblasts were fixed and processed for
The sequences obtained by peptide sequencing as described in Materials and Methods are identical to the predicted sequences deduced from the corresponding cDNA. X indicates the cysteine which is not visible during sequencing. For β1, the residues in low case letters refer to the additional amino acids coming from the multicloning site of the pGEX-4T1 procaryotic vector.
indirect immunofluorescence staining of vinculin, a marker of focal adhesion complexes. Microinjected cells were detected by the green fluorescence of the rEGFP, and their adhesion complexes were carefully examined after selecting the appropriate filter to observe the pattern of the red labelling of vinculin. In cells microinjected with dimers of the cytodomain of the α3 integrin subunits, adhesion complexes were dramatically altered (Fig. 3A). They were either absent or reduced to a barely visible punctuate staining (open arrowheads in Fig. 3A) or very thin (small arrows in Fig. 3A). In contrast, cells microinjected with the cytoplasmic domain of the α6 integrin subunits or unmicroinjected cells displayed strong vinculin-positive adhesion complexes localised at the cell margins (Fig. 3A, filled arrowheads). To exclude the possibility of an effect caused by peptide dimerisation or by the presence of a reactive cysteine, the activity of reduced and alkylated peptides was tested and similar results were obtained, i.e. distinct adhesion complexes were not seen in cells microinjected with monomers of the α3 peptides (Fig. 3C), but
Fig. 3. Microinjection of peptides representing the cytoplasmic domain of the integrin α3 subunit induces disruption of vinculin association with adhesion complexes. Wi26 fibroblasts adhering to laminin 1 for 90 minutes were co-injected with rEGFP and with monomers (alk as indicated on the figure) or dimers of the α3 (A,C) or α6 (B,D) peptides. Adhesion complexes were revealed 25-35 minutes post-injection by immunofluorescence staining with antibodies against vinculin followed by Cy3-conjugated secondary antibodies. Microinjected cells (asterisk) were detected by the green fluorescence of rEGFP and after switching to the appropriate filter the pattern of vinculin staining (red) was analysed. Panels show vinculin stainings only. Note that non injected cells are decorated by numerous thick patches of fluorescence as indicated by the filled arrowheads in the upper cell shown in A, while such thick fluorescent patches are absent (open arrowheads in A) or thinner (three small arrows in A) in cells micoinjected with the α3 peptides. Bar, 5 µm.
Integrin cytoplasmic domains 1171 were present in cells microinjected with monomers of the α6 peptides (Fig. 3D). In addition, in cells microinjected with monomers or dimers of the α3 peptides, most of the fluorescence signal was diffusely distributed over the entire cell, with an increased intensity in the peri-nuclear region, as seen on comparing microinjected and not microinjected cells in Fig. 3A and C. To quantify the effect of the peptides, microinjected cells with or without typical vinculin-positive adhesion complexes were counted (Fig. 4). Absence or abnormalities of adhesion complexes were observed in 62% of the cells microinjected with the cytoplasmic domain of the integrin α3 subunit. In contrast, after injection of rEGFP alone or together with the α6 peptides, adhesion complexes were
altered in only 18 to 25% of the cells, a percentage similar to that found for non-injected cells (20%). The effect of the α3 peptides was investigated in more detail by processing the cells for immunofluorescence staining at different time points after microinjection or by injecting the cells at different time points following adhesion to laminin 1 and onset of spreading. In the first experimental design, the fibroblasts were microinjected after a fixed period of adhesion of 90 minutes. Five to ten minutes after injection there was no alteration of the cell adhesion complexes (Fig. 5A). Fifteen to twenty minutes after injection the adhesion complexes started to vanish (Fig. 5B) and the effect was best seen between 25 to 35 minutes (Fig. 5C). Ninety minutes later typical adhesion
α3 + rGFP 20
0
α6 + rGFP
40
Ø
rGFP
60
(in % of total)
Cells with altered adhesion complexes
80
20 135 42 59
Fig. 4. Quantification of the effect observed after microinjection of integrin peptides. Fibroblasts were microinjected and processed for indirect immunofluorescence staining of vinculin as described in Materials and Methods. Cells with vinculinpositive staining (defined by thick patches of fluorescence as those indicated by closed arrowheads in Fig. 3A) or with disruption of vinculin staining (absence or reduction of staining as indicated by open arrowheads or thin arrows in Fig. 3A, respectively) were counted in the population of microinjected cells detected by rEGFP fluorescence and in a randomly selected population of non-injected cells (Ø). The number of cells counted are indicated at the bottom of each column.
Fig. 5. The effect of the integrin α3 subunit peptides on the organisation of adhesion complexes is transient and specific. Wi26 cells were seeded either on laminin 1 and incubated on the coats for 90 (A-D) or 180 (E) minutes or on fibronectin for 90 minutes (F) before being microinjected with rEGFP together with α3 peptides. The cells were processed 10 (A), 20 (B), 35 (C,E,F) and 90 (D) minutes post-injection for vinculin staining as described in Fig. 2. Microinjected cells (asterik) were detected and observed as in Fig. 3. Bar, 5 µm.
1172 E. Laplantine and others complexes were again observed (Fig. 5D). In the second experimental design, the cells were microinjected 90, 120 (not shown) or 180 (Fig. 5E) minutes after seeding and processed for immunofluorescence staining 25-35 minutes later. In all cases, injection of the α3 peptides was followed by the disruption of vinculin-positive adhesion complexes. This indicated that the α3 peptides were active on newly formed (90-minute-old, Fig. 5C) as well as on mature (180-minute-old, Fig. 5E) adhesion complexes. In parallel experiments, injection of peptides representing the cytoplasmic domain of the integrin α6 subunit or of rEGFP alone (not shown) did not perturb the pattern of adhesion complexes and vinculin staining was identical in microinjected and uninjected cells. Finally, to test whether the observations were specific for cells adhering to laminin 1, fibroblasts were seeded on fibronectin coats or cultured for 24 hours in the presence of fetal calf serum prior to microinjection. Under the later condition, adhesion of the fibroblasts to the culture surfaces is mediated by both adhesive proteins (e.g. fibronectin, vitronectin) of the serum and by the cell’s own synthesised extracellular matrix. Microinjection of the cytodomain of the α3 integrin subunit in cells plated on fibronectin (Fig. 5F) or in cells cultivated for 24 hours in the presence of serum (not shown) had no noticeable effect and distinct vinculin-positive adhesion complexes were observed. In similar experiments, immunofluorescence detection of the integrin β1 subunit showed that the size of the aggregates was reduced after microinjection of the integrin α3 peptide (Fig. 6A,E,F), while both parameters were not affected upon
Fig. 6. Double immunofluorescence staining of integrin β1 subunit and fibrillar actin after microinjection of integrin α3 or α6 peptides. Wi26 fibroblasts were seeded on laminin 1 coats for 90 minutes prior to microinjection with α3 (A,B) or α6 (C,D) peptides together with rEGFP. Thirty minutes later, the cells were fixed and stained with mAb K20 (A,C) against the β1 integrin subunit and with CPITCconjugated phalloidin (B,D). Images corresponding to vinculin staining (red) and rEGFP (green) were superimposed in A and C and the microinjected cells are those appearing with a yellowish nucleus. The inserts at the bottom of the figure show magnified (×3) adhesion complexes from α3- (E,F) and α6-microinjected (G,H) cells and from non-injected (I,J) cells. Arrows in A and C indicate areas chosen for magnification (filled arrows for non-injected cells and open arrows for injected cells). Bar, 5 µm (A-D).
Table 2. Size of β1 integrin-containing adhesion complexes Cell status
Length (nm ± s.d.)
Number of patches measured
Non-injected α6-injected α3-injected
1202±162 1222±56 700±95
223 (20 cells) 75 (7 cells) 114 (10 cells)
The lengths of fluorescent patches as shown in Fig. 6 were measured on color slides with the help of a computer-assisted programme as described in Materials and Methods. The number given in parenthesis indicates the total number of cells used for the measurements. s.d.: standard deviation.
microinjection of the integin α6 peptide (Fig. 6C,G,H). Measuring the lengths of the fluorescent aggregates decorated by the antibody against the β1 integrin subunit indicated that microinjection of the α3 peptide induced a size reduction by ~40% in comparison to controls or to α6-microinjected cells (Table 2). Visualisation of fibrillar actin with CPITCconjugated phalloidin did not reveal obvious alteration of the actin cytoskeleton in α3 or α6 microinjected cells (Fig. 6B,D). Altogether, these results indicate that the observed effect is specific for the cytoplasmic domain of the integrin α3 subunit and it suggests that direct interactions between integrin intracellular tails could be involved. Interactions between the cytoplasmic domains of the α3 and β1 integrin subunits In vitro complex formation of α3 and β1 integrin cytoplasmic
Integrin cytoplasmic domains 1173 A
50
Relative Response (RU)
Relative Response (RU)
A
40 30
1 2 3 4
20 10 0
1
50 40
2
30 20 10 0 -10
-10 -20
80
180
280
380
-20
80
Time (s)
B
80
B 1
50
2
40
3
30
4
Relative Response (RU)
Relative Response (RU)
70 60
180
280
380
Time (s)
20 10 0 -10
80 70 60 50 40 30
1
20 2
10 0 -10
-20
80
180
280
380
Time (s)
Fig. 7. Surface plasmon resonance analysis of the binding of α3 and β1 cytoplasmic tails. The experimental sensorgrams were recorded on a Biacore X apparatus using a sensorchip with ~220 RU of covalently immobilised β1 peptides, TBS-Bia as running buffer, and a flow rate of 5 µl/minute. Data are given as relative responses after subtraction of the background signal monitored on a reference cell made up of ethanolamine-substituted dextran. (A) Response curves recorded during and after the injection of a solution containing either (1) 100 µM α3, (2) 200 µM αIIb, (3) 100 µM α6, or 300 µM αv cytoplasmic peptide. (B) Overlaid-dose curves obtained with (1) 200, (2) 150, (3) 100 or (4) 30 µM α3 peptide solutions.
tails was investigated using surface plasmon resonance. Purified recombinant peptides corresponding to the intracellular domain of the β1 subunit and containing an additional N-terminal cysteine residue, were immobilised on a carboxymethyl dextran sensorchip through thiol coupling. The immobilised β1 peptides were exposed to soluble peptide monomers corresponding to the cytoplasmic domain of different α integrin subunits and the binding profiles were recorded. A characteristic binding signal of low magnitude was observed when the β1 peptide was brought into contact with the α3 peptide (Fig. 7A). Formation of the complex was reversible, as flushing the flow cell with running buffer alone led to a marked decrease in resonance units, indicating a fast dissociation of α3 peptides from immobilised β1. In contrast, no binding was observed with either αv, αIIb or α6 cytoplasmic tail peptides (Fig. 7A). Here, a rapid change in the resonance signal was observed which was attributed to a dilution buffer-induced nonspecific change in the bulk refractive index. Binding of α3 peptides to an uncoated reference cell made up of ethanolamine-substituted dextran was negligible, indicating the absence of nonspecific
-20
60
140
220
300
380
Time (s)
Fig. 8. Effect of Ca2+ and ionic strength on the α3/β1 heterodimerisation as monitored by surface plasmon resonance. The experiments were carried out on a Biacore X instrument using a sensorchip with ~220 RU of immobilised β1 peptides, TBS-Bia as running buffer, and a flow rate of 5 µl/minute. (A) Sensorgrams collected during and after the injection of a 100 µM α3 peptide solution in the presence (1) or absence (2) of 2 mM CaCl2 in the running buffer. (B) Binding curves obtained during and after the perfusion of a 200 µM α3 peptide solution containing 1 M (1) or 0.15 M (2) NaCl.
interaction of the analyte with the matrix (not shown). Further studies performed with increasing concentrations of soluble α3 peptide showed that the binding was dose-dependent (Fig. 7B). The maximum response monitored at the end of the peptide injection phase was approximatively 70 RU for 220 RU of immobilised β1 peptide. Considering the signal to mass relationship of 1 pg of protein.mm−2.RU-1, usually used in SPR studies (Schuck, 1997), the corresponding molar ratio was estimated to be ~0.40 mole of bound α3 peptide per mole of immobilised β1. Taken together, these data indicate that the α3 cytoplasmic domain specifically interacts with β1 cytoplasmic tail to form a heterodimeric complex. We next investigated the influence of calcium and of ionic strength on the α3/β1 complex formation. The curves obtained in the presence or absence of 2 mM CaCl2, during both the association and the dissociation phases, were similar (Fig. 8A), demonstrating that the α3/β1 dimerisation was Ca2+independent. In contrast, α3 binding experiments performed using a running buffer containing variable NaCl concentrations showed that the interaction was totally abolished at high ionic strength (Fig. 8B). It indicates that formation of the α3/β1 complexation is most likely supported by ionic, rather than
1174 E. Laplantine and others hydrophobic, interactions. Similar results were obtained using the integrin β1 subunit fused to GST and captured by antibodies against GST immobilised on the sensorchip (not shown). DISCUSSION Numerous experiments indicate that specific signalling pathways are induced upon integrin-ligand interactions (LaFlamme et al., 1997; Yamada and Geiger, 1997). However, the mechanism of integrin signal induction is as yet not understood. Integrin activation could involve conformational changes and association/dissociation of the integrin cytoplasmic domains (Haas and Plow, 1997), resulting in the exclusive binding of intracellular proteins to the integrin tails and induction of specific signalling pathways. The nature of these modifications in integrin structure and association, as well as the mechanism by which they could be promoted are not known. Many proteins have been shown to bind the cytoplasmic domain of the β1 integrin subunit, although it is not known whether the interactions take place with free β1 integrin subunits or when they form dimers with integrin α subunits. Both scenarios would be relevant, corresponding to different situations under which the integrins are either clustered and active at the cell membrane or present as monomeric subunits in the reticulum. We have investigated the functional role of the cytoplasmic domains of two laminin receptors, integrins α3β1 and α6β1. Integrin α6β1 binds to all tested laminin isoforms, while α3β1 interacts with all laminin variants except laminin 1 (Delwel et al., 1993, 1994; Rousselle and Aumailley, 1994; Dogic et al., 1999). Activation of the α6β1 integrin by solely laminin 1 leads to distinct clustering of integrins and cytoskeletal linker proteins into typically thick and well individualised adhesion complexes (Dogic et al., 1998). In contrast, punctuate and dispersed adhesion complexes are observed when both the α3β1 and α6β1 integrins are activated by extracellular interactions with either laminin 5 or a mixture of laminin isoforms, or alternatively by antibodies against α3β1 integrins in cells adhering to laminin 1 (Dogic et al., 1998, 1999). These data suggested that the α3β1 integrin negatively regulates α6β1 integrin-mediated clustering of cytoskeletal linker proteins into macroaggregates. Studies of keratinocytes deficient in the α3 integrin subunit showed that the transdominant inhibitory role of the α3β1 integrin on cytoskeletal organisation could be extended to those integrins which bind fibronectin or collagen IV (Hodivala-Dilke et al., 1998). Together these results suggested that the regulatory role of integrin α3β1 could involve the cytoplasmic domain of the α3 subunit. The role of integrin cytoplasmic domains has been previously approached by overexpressing functional or mutated β cytoplasmic tail (LaFlamme et al., 1994; Lukashev et al., 1994; Chen et al., 1994; Faraldo et al., 1998; Mastrangelo et al., 1999). In such experiments, it is assumed that integrin function is altered by either depleting the cytosolic proteins that associate with the β subunits or activating inhibitory signalling pathways. In the microinjection experiments reported here, a similar approach was used with α subunit peptides to perturb in vivo integrin interactions with cytosolic
proteins under conditions allowing the formation of α6β1 integrin-specific adhesion complexes, i.e. following cell adhesion to laminin 1. Under these conditions, both a disruption of vinculin from the adhesion complexes and a reduction in the lengths of β1 integrin-positive clusters were observed following injection of the cytoplasmic domain of the α3 integrin subunit. The effect was specific of the α3 peptides and not observed after injection of peptides corresponding to the α6 integrin subunit. It occured relatively rapidly and was best seen 25-35 minutes following injection. The effect was transient, since α6β1 integrin-specific adhesion complexes were re-established at about 90 minutes after injection. Finally, it was integrin-specifc since it was observced on cells adhering to laminin1 but not on cells adhering to fibronectin or cultivated in the presence of fetal calf serum. These data indicate that free α3 integrin peptides, but not free α6 peptides, were able to specifically disrupt interactions between the cytoplasmic tails of integrin β1 subunits and proteins associated with adhesion complexes, such as vinculin, in cells adhering to laminin 1. Further, surface plasmon resonance studies indicated that peptides representing the intracellular portion of the integrin β1 subunit interact with peptides corresponding to the α3 subunit, but not with α6, αv, or αIIb peptides. The cytoplasmic tail of all integrin α subunits contains the conserved membrane-proximal KXGFFKR sequence and differs in the distal portion. Although we cannot exclude that the conserved KXGFFKR motif could participate in the interaction between the α3 and β1 cytoplasmic tails, the carboxy-terminal sequence of the α3 subunit is probably responsible for the specificity. Interestingly, apart from the conserved region, the carboxy-terminal sequence of the α3 and α6 integrin subunits have 63% homology and 43% identitity, so that the specificity is presumably given by a few amino acid residues only. The signals observed by surface plasmon resonance were dose-dependent for the α3 integrin peptides and were abolished at high salt concentration, underlining the ionic nature of the interaction. We have recently observed a similar selective interaction between the cytodomains of the αIIbβ3 integrin (Vallar et al., 1999). The association reaction proceeded at approximately the same rate (kon~2-4×102 M−1.s−1) for the αIIb/β3 and the α3β1 interactions. The dissociation reaction of the α3/β1 complex was markedly faster than that of the αIIb/β3 complex and it could not be measured with reliability since it is close to the limit specified for the instrument. It indicates, however, that the affinity between the cytoplasmic domains of the α3 and β1 integrin subunits is weaker than that between the αIIb and β3 integrin tails. Moreover, surface plasmon resonance results obtained for αIIb/β3 complexation indicated that cations were not required for the interaction to occur, but stabilised the heterodimer by decreasing the dissociation rate, in agreement with spectroscopic data showing the presence of a Ca2+ binding site within the αIIb cytoplasmic tail (Cierniewski et al., 1994; Haas and Plow, 1996). In contrast, Ca2+ had no effect on the α3/β1 interaction which could explain the faster dissociation observed. Together our data suggest that a specific spatial relationship between the α and β intracellular domains may be crucial in regulating receptor function of a particular integrin. Finally, the kinetic parameters are at an order of magnitude compatible with transient binding such as that expected in signal transduction.
Integrin cytoplasmic domains 1175 We have recently shown that integrin α3β1 transdominantly regulates integrin α6β1 and the associated formation of adhesion complexes. The morphology of the complexes that were previously observed when the α3β1 integrin is engaged in interactions with extracellular ligands (Dogic et al., 1998, 1999) is reminiscent of that observed after microinjection of the α3 peptides in cells seeded on laminin 1. Under these conditions there is a morphological shift of the adhesions from integrin α6β1-specifc macroaggregates to integrin α3β1specific microaggregates. Thick vinculin-positive clusters corresponding to α6β1 integrin activation are lost in the α3microinjected cells and the size of β1 integrin aggregates is reduced. Given these results several hypothesis can be drawn. Upon microinjection, the α3 peptides interfere directly, by competition or by inducing a conformational change of the β1 cytoplasmic domain, with binding of the latter to intracellular proteins as it occurs in typical adhesion complexes. Alternatively, the α3 peptides may interfere indirectly with the formation of typical adhesion complexes by segregating and redistributing focal adhesion-associated proteins to another sub-cellular localisation. Biochemical analyses of adhesion complexes that differ in morphology, such as those formed by fibroblasts adhering to laminin 1 or to fibronectin (Sondermann et al., 1999) or that developed by wild-type or α3 integrindeficient keratinocytes (Hodivala-Dilke et al., 1998) indicated that the difference is due to a sub-cellular redistribution rather than to changes in the level of expression of cytoskeletalassociated proteins. These results together with our observation that in the cells microinjected with the α3 peptides, the vinculin-associated fluorescence signal is more concentrated in the peri-nuclear region, favor the hypothesis of a redistribution of cytoskeletal linker proteins. Absence of interactions between the cytodomain of the α6 and β1 integrin subunits would allow the latter to interact with focal adhesionassociated proteins and the transmission of an α6 integrinspecific signal. In contrast, when the α3 integrin cytodomain is interfering, different, or lack of, interaction partners for the cytosolic portion of the integrin β1 subunit will allow either the release of a signal specific for the integrin α3 subunit or the inhibition of the signal specific for the α6 integrin subunit. We thank Drs M. Glukhova, G. Lippens, M. Paulsson and R. Timpl for kind gifts of reagents, M. Pesch for maintaining the cell culture facilities, the central facilities of the ZMMK, and Drs J. W. Fox, D. Heinegård, N. Smyth, L. Tunggal and M. Paulsson for valuable suggestions. We acknowledge the financial support from the Deutsche Forschungsgemeinschaft (Kr 558/10-1 and AU 86/5-1) and from the University of Cologne. This work was carried in the context of the CNRS european network ‘Integrins and Transfer of Information’ (GDRE-ITI). M.A. and N.K. are researchers of the Centre National de la Recherche Scientifique.
REFERENCES Aumailley, M. and Rousselle, P. (1999). Laminins of the dermo-epidermal junction. Matrix Biol. 18, 19-28. Burn, P., Kupfer, A. and Singer, S. (1988). Dynamic membrane-cytoskeletal interactions: specific association of integrin and talin arises in vivo after phorbol ester treatment of peripheral blood lymphocytes. Proc. Nat. Acad. Sci. USA 85, 497-501. Chang, D. D., Wong, C., Smith, H. and Liu, J. (1997). ICAP-1, a novel β1 integrin cytoplasmic domain-associated protein, binds to a conserved and
functionally important NPXY sequence motif of β1 integrin. J. Cell Biol. 138, 1149-1157. Chen, Y. P., O’Toole, T. E., Shipley, T., Forsyth, J., LaFlamme, S. E., Yamada, K. M., Shattil, S. J. and Ginsberg, M. H. (1994). ‘Inside-out’ signal transduction inhibited by isolated integrin cytoplasmic domains. J. Biol. Chem. 269, 18307-18310. Chung, J., Gao, A. G. and Frazier, W. A. (1997). Thrombospondin acts via integrin-associated protein to activate the platelet integrin αIIbβ3. J. Biol. Chem. 272, 14740-14746. Cierniewski, C. S., Haas, T. A., Smith, J. W. and Plow, E. F. (1994). Characterization of cation-binding sequences in the platelet integrin GPIIbIIIa (αIIbβ3) by terbium luminescence. Biochemistry 33, 12238-12246 Delwel, G. O., Hogervorst, F., Kuikman, I., Paulsson, M., Timpl, R. and Sonnenberg, A. (1993). Expression and function of the cytoplasmic variants of the integrin α6 subunit in transfected K562 cells. Activation-dependent adhesion and interaction with isoforms of laminin. J. Biol. Chem. 268, 25865-25875. Delwel, G. O., de Melker, A. A., Hogervorst, F., Jaspars, L. H., Fles, D. L. A., Kuikman, I., Lindblom, A., Paulsson, M., Timpl, R. and Sonnenberg, A. (1994). Distinct and overlapping ligand specificities of the α3Aβ1 and α6Aβ1 integrins: recognition of laminin isoforms. Mol. Biol. Cell 5, 203-215. Dogic, D., Rousselle, P. and Aumailley, M. (1998). Cell adhesion to laminin 1 or 5 induces isoform-specific clustering of integrins and other focal adhesion components. J. Cell Sci. 111, 793-802. Dogic, D., H¸lsmann, H., Sherman, N., Fox, J. W., Broermann, R., Paulsson, M. and Aumailley, M. (1999). Cell adhesion to a population of laminin isoforms isolated from normal renal tissue. Matrix Biol. 18, 433444. Faraldo, M. M., Deugnier, M. A., Lukashev, M., Thiery, J. P. and Glukhova, M. A. (1998). Perturbation of β1-integrin function alters the development of murine mammary gland. EMBO J. 17, 2139-2147. Haas, T. A. and Plow, E. F. (1996). The cytoplasmic domain of αIIbβ3. A ternary complex of the integrin α and β subunits and a divalent cation. J. Biol. Chem. 271, 6017-6026. Haas, T. A. and Plow, E. F. (1997). Development of a structural model for the cytoplasmic domain of an integrin. Protein Eng. 10, 1395-1405 Hannigan, G. E., Leung-Hagersteijn, C., Fitz-Gibbon, L., Copolino, M. G., Radeva, G., Filmus, J., Bell, J. C. and Dedhar, S. (1996). Regulation of cell adhesion and anchorage-dependent growth by a new β1-integrinlinked protein kinase. Nature 79, 91-96. Hodivala-Dilke, K. M., DiPersio, C. M., Kreidberg, J. A. and Hynes, R. O. (1998). Novel roles for α3β1 integrin as a regulator of cytoskeletal assembly and as trans-dominant inhibitor of integrin receptor function in mouse keratinocytes. J. Cell Biol. 142, 1357-1369. Horwitz, A., Duggan, K., Buck, C., Beckerle, M. C. and Burridge, K. (1986). Interaction of plasma-membrane fibronectin receptor with talin – a transmembrane linkage. Nature 320, 531-533. LaFlamme, S. E., Thomas, L. A., Yamada, S. S. and Yamada, K. M. (1994). Single subunit chimeric integrins as mimics and inhibitors of endogenous integrin functions in receptor localization, cell spreading and migration, and matrix assembly. J. Cell Biol. 126, 1287-1298. LaFlamme, S. E., Homan, S. M., Bodeau, A. L. and Mastrangelo, A. M. (1997). Integrin cytoplasmic domains as connectors to the cell’s signal transduction appparatus. Matrix Biol. 16, 153-163. Liliental, J. and Chang, D. D. (1998). Rack1, a receptor for activated protein kinase C, interacts with integrin β subunit. J. Biol. Chem. 273, 2379-2383. Loo, D. T., Kanner, S. B. and Aruffo, A. (1998). Filamin binds to the cytoplasmic domain of the β1-integrin. Identification of amino acids responsible for this interaction. J. Biol. Chem. 273, 23304-23312. Lukashev, M. E., Sheppard, D. and Pytela, R. (1994). Disruption of integrin function and induction of tyrosine phosphorylation by the autonomously expressed β1 integrin cytoplasmic domain. J. Biol. Chem. 269, 1831118314. Mastrangelo, A. M., Homan, S. M., Humphries, M. J. and LaFlamme, S. E. (1999). Amino acid motifs for isolated β cytoplasmic domains to regulate ëin trans’ β1 integrin conformation and function in cell attachment. J. Cell Sci. 112, 217-229. Miekka, S. I., Ingham, K. C. and Menache, D. (1982). Rapid methods for isolation of human plasma fibronectin. Thromb. Res. 27, 1-14. Otey, C. A., Pavalko, F. M. and Burridge, K. (1990). An interaction between α-actinin and the β1 integrin subunit in vitro. J. Cell Biol. 111, 721-729. Otey, C. A., Vasquez, G. B., Burridge, K. and Erickson, B. W. (1993).
1176 E. Laplantine and others Mapping of the α-actinin binding site within the β1 integrin cytoplasmic domain. J. Biol. Chem. 268, 21193-21197. Paulsson, M., Aumailley, M., Deutzmann, R., Timpl, R., Beck, K. and Engel, J. (1987). Laminin-nidogen complex. Extraction with chelating agents and structural characterization. Eur. J. Biochem. 166, 11-19. Rojiani, M. V., Finlay, B. B., Gray, V. and Dedhar, S. (1991). In vitro interaction of a polypeptide homologous to human Ro/SS-A antigen (calreticulin) with a highly conserved amino acid sequence in the cytoplasmic domain of integrin α subunits. Biochemistry 30, 9859-9866. Rousselle, P. and Aumailley, M. (1994). Kalinin, a short laminin isoform, is more efficient than laminin to promote adhesion of epithelial cells and has different requirements in integrin receptors. J. Cell Biol. 125, 205214. Schaller, M. D., Otey, C. A., Hildebrand, J. D. and Parsons, J. T. (1995). Focal adhesion kinase and paxillin bind to peptides mimicking β integrin cytoplasmic domains. J. Cell Biol. 130, 1181-1187. Schuck, P. (1997). Use of surface plasmon resonance to probe the equilibrium and dynamic aspects of interactions between biological macromolecules. Annu. Rev. Biophys. Biomol. Struct. 26, 541-566. Sondermann, H., Dogic, D., Pesch, M. and Aumailley, M. (1999). Targeting of cytoskeletal linker proteins to focal adhesion complexes is reduced in fibroblasts adhering to laminin-1 when compared to fibronectin. Cell Adh. Commun. 7, 43-56.
Vallar, L., Melchior, C., PlanÁon, S., Drobecq, H., Lippens, G., Regnault, V. and Kieffer, N. (1999). Divalent cations differentially regulate integrin αIIb cytoplasmic tail binding to β3 and to calcium- and integrin-binding protein. J. Biol. Chem. 274, 17257-17266. Wang, X. Q. and Frazier, W. A. (1998). The thrombospondin receptor CD47 (IAP) modulates and associates with α2β1 integrin in vascular smooth muscle cells. Mol. Biol. Cell 9, 865-874. Wary, K. K., Mainiero, F., Isakoff, S. J., Marcantonio, E. E. and Giancotti, F. G. (1996). The adaptor protein Shc couples a class of integrins to the control of cell cycle progression. Cell 87, 733-743. Wixler, V., Laplantine, E., Geerts, D., Sonnenberg, A., Petersohn, D., Eckes, B., Paulsson, M. and Aumailley, M. (1999). Identification of novel interaction partners for the conserved region of α-integrin cytoplasmic domains. FEBS Lett. 445, 351-355. Yamada, K. M. and Geiger, B. (1997). Molecular interactions in cell adhesion complexes. Curr. Opin. Cell Biol. 9, 76-85. Yauch, R. L, Berditchevski, F., Harler, M. B., Reichner, J. and Hemler, M. E. (1998). Highly stoichiometric, stable, and specific association of integrin α3β1 with CD151 provides a major link to phosphatidylinositol 4-kinase, and may regulate cell migration. Mol. Biol. Cell 9, 2751-2765. Zage, P. E. and Marcantonio E. E. (1998). The membrane proximal region of the integrin β cytoplasmic domain can mediate oligomerization. Cell Adhes. Commun. 5, 335-347.
