Differences in Regulation of Drosophila and Vertebrate Integrin Affinity by Talin

Abstract

Integrin-mediated cell adhesion is essential for development of multicellular organisms. In worms, flies, and vertebrates, talin forms a physical link between integrin cytoplasmic domains and the actin cytoskeleton. Loss of either integrins or talin leads to similar phenotypes. In vertebrates, talin is also a key regulator of integrin affinity. We used a ligand-mimetic Fab fragment, TWOW-1, to assess talin's role in regulating Drosophila αPS2βPS affinity. Depletion of cellular metabolic energy reduced TWOW-1 binding, suggesting αPS2βPS affinity is an active process as it is for vertebrate integrins. In contrast to vertebrate integrins, neither talin knockdown by RNA interference nor talin head overexpression had a significant effect on TWOW-1 binding. Furthermore, replacement of the transmembrane or talin-binding cytoplasmic domains of αPS2βPS with those of human αIIbβ3 failed to enable talin regulation of TWOW-1 binding. However, substitution of the extracellular and transmembrane domains of αPS2βPS with those of αIIbβ3 resulted in a constitutively active integrin whose affinity was reduced by talin knockdown. Furthermore, wild-type αIIbβ3 was activated by overexpression of Drosophila talin head domain. Thus, despite evolutionary conservation of talin's integrin/cytoskeleton linkage function, talin is not sufficient to regulate Drosophila αPS2βPS affinity because of structural features inherent in the αPS2βPS extracellular and/or transmembrane domains.

INTRODUCTION

Integrin adhesion receptors couple the extracellular matrix with the actin cytoskeleton, allowing transmission of both mechanical force and biochemical signals across the plasma membrane (Hynes, 2002). The cytoskeletal protein talin, a product of the talin 1 (TLN1) gene, provides a key physical linkage between integrin β cytoplasmic domains and F-actin (Critchley, 2004; Wiesner et al., 2005). Talin's N-terminal head region contains a FERM domain with a PTB subdomain capable of interacting with several proteins, including integrin β cytoplasmic domains (Wegener et al., 2007). Talin's C-terminal rod domain contains binding sites for several cytoskeletal proteins, including actin (Calderwood, 2004; Critchley, 2004). Loss of talin in worms, flies, and mice leads to lethal phenotypes that closely resemble those caused by loss or mutation of integrins, indicating that talin's function overlaps that of integrins in the developing organism (Monkley et al., 2000; Brown et al., 2002; Brower, 2003; Cram et al., 2003; Wiesner et al., 2005). In fact, talin is essential for the normal development of organisms ranging in complexity from the multicellular slime mold Dictyostelium discoideum to vertebrates (Nuckolls et al., 1992; Albiges-Rizo et al., 1995; Bolton et al., 1997; Priddle et al., 1998; Tsujioka et al., 1999; Monkley et al., 2000; Cram et al., 2003). In vertebrates, it is also clear that talin plays a key role in regulating the affinity of integrins for adhesive ligands (Calderwood et al., 1999, 2002; Tadokoro et al., 2003; Ginsberg et al., 2005; Wegener et al., 2007). Although it is conceivable that the integrin–actin linkage and integrin affinity modulation functions of talin have evolved in concert, it is also possible that they developed independently.

In Drosophila melanogaster, talin null mutant embryos demonstrate normal localization of integrin αPS2βPS to muscle ends, but failure of αPS2βPS clustering into focal adhesion-like structures (Brown et al., 2002). At the onset of contraction, the muscles detach from their tendon cell anchorage points and αPS2βPS detaches from the actin cytoskeleton but remains attached to extracellular matrix ligands (Brown et al., 2002). This suggests that talin is essential for maintenance of an integrin-actin linkage strong enough to resist mechanical force above a certain threshold. This idea is supported by experiments with optical tweezers showing that talin is required for maintenance of a 2 pN slip bond between fibronectin and the cytoskeleton (Giannone et al., 2003; Jiang et al., 2003). A role for talin in affinity modulation of αPS2βPS has been speculated upon, but not measured directly (Brown et al., 2002; Tanentzapf and Brown, 2006).

Until recently, integrin affinity modulation in Drosophila has only been inferred from sophisticated genetic manipulation and indirect measurements using cell adhesion and spreading assays. We have developed a ligand-mimetic antibody Fab fragment, TWOW-1, that is selective for high-affinity αPS2βPS and provides a facile means to assess αPS2βPS affinity in cultured cells (Bunch et al., 2006). The binding selectivity of TWOW-1 is due in part to the fact that the H-CDR3 region of this Fab fragment contains a 53-amino acid RGD tract derived from tiggrin, a natural Drosophila matrix ligand for αPS2βPS (Fogerty et al., 1994). Using TWOW-1, we have shown that αPS2βPS affinity is indeed increased or decreased by certain integrin mutations that would be predicted to do so by structure-function analyses of vertebrate integrins (Bunch et al., 2006; Devenport et al., 2007). Given that the heterodimeric structure of integrins and the major amino acid sequence motifs of both integrins and talin are highly conserved between Drosophila and mammals (Burke, 1999; Hynes and Zhao, 2000; Senetar and McCann, 2005), it is reasonable to hypothesize that, as in vertebrates, talin modulates the affinity of Drosophila integrins. Here TWOW-1 was used to test this hypothesis and we demonstrate that this is not the case. Unlike vertebrate integrins, αPS2βPS appears to be relatively resistant to affinity modulation by talin, even when its cytoplasmic domains containing putative talin-binding sites or its transmembrane domains are replaced with those of human αIIbβ3. On the other hand, overexpression of the Drosophila talin head domain is capable of activating αIIbβ3 in a manner similar to that of the human talin head domain. Taken together, these results suggest that, in contrast to talin's linkage function, the regulatory role of talin in integrin affinity modulation may be a relatively more recent evolutionary development in higher organisms.

MATERIALS AND METHODS

Antibodies

TWOW-1 is a ligand-mimetic Fab specific for Drosophila αPS2βPS that was produced and purified as reported (Bunch et al., 2006). It was used at final concentrations of 2–120 μg/ml as indicated. PAC-1 IgM ascites (Shattil et al., 1985) and PAC-1 Fab (Abrams et al., 1994) are ligand-mimetic antibodies specific for human αIIbβ3. Surface expression of integrins was quantified by flow cytometry using antibodies CF.2C7 (sometimes biotinylated) for Drosophila αPS2 (Brower et al., 1984), CF.6G11 for Drosophila βPS (Brower et al., 1984) or D57 for αIIbβ3 (O'Toole et al., 1994). Talin expression was detected using the mAb J10 for Drosophila talin (Brown et al., 2002) or 8d4 for Chinese hamster ovary (CHO) cell and human talin (Sigma, St. Louis, MO). The secondary antibody for all flow cytometry experiments was either R-phycoerythrin conjugated goat anti-mouse IgG (H+L; Biosource, Camarillo, CA) or Alexa-488–conjugated goat anti-mouse IgG (H+L; Molecular Probes, Eugene, OR), except for PAC-1 IgM, which was detected using R-phycoerythrin–conjugated goat anti-mouse IgM (Biosource). Biotinylated antibodies were detected using R-phycoerythrin-streptavidin (Molecular Probes).

Cell Lines and Culture

Drosophila S2/M3 cells stably expressing the αPS2 and βPS integrin subunits under the control of the Drosophila HSP70 heat-shock promoter have been described (Bunch and Brower, 1992; Zavortink et al., 1993). For all cell culture experiments, the αPS2C (canonical) and βPS4A isoforms were used (Graner et al., 1998). In some experiments, these integrin-expressing S2/M3 cells also stably expressed Drosophila talin head-GFP chimeras (wild-type or R367A mutants) under the control of the yeast Gal4 UAS (Tanentzapf et al., 2006). All Drosophila cells were grown in Shields and Sang M3 medium supplemented with 12.5% heat-inactivated fetal bovine serum and 2 × 10⁻⁷ M methotrexate. For all experiments other than cell-spreading assays, S2/M3 cells were first cleared of accumulated matrix and other surface proteins by dispase/collagenase (Roche Applied Science, Indianapolis, IN). The cells were simultaneously heat shocked at 37°C to induce expression of the integrin transgenes (Jannuzi et al., 2002). CHO cells stably expressing human αIIbβ3 or constitutively active αIIbβ3 (D723R) have been described (Frojmovic et al., 1991; Hughes et al., 1996). CHO cells stably expressing Drosophila αPS2βPS were generated as described below. All CHO cells were grown in DMEM supplemented with 10% fetal bovine serum.

Plasmids and Transfection

αPS2 and βPS cDNAs were excised from the Drosophila HSP70 promoter-driven expression vectors used in the S2/M3 cell system and subcloned into the mammalian expression vector pcDNA3.1. These constructs were cotransfected into CHO cells in a 1:1 weight ratio using Lipofectamine (Invitrogen, Carlsbad, CA). After selection with G418, stable transformants expressing high levels of αPS2βPS integrin were obtained by single-cell sorting using antibody CF.2C7. The βPS/pcDNA3.1 expression vector was used as the template for creation of a βPS double mutant (I830A/K832A) by standard PCR-based site-directed mutagenesis. This vector was transiently transfected along with wild-type αPS2 into CHO cells. After 72 h, surface expression of the mutant integrin was measured by flow cytometry using anti-αPS2 antibody CF.2C7 and was found to be comparable to that of wild-type αPS2βPS; in contrast, several βPS cytoplasmic domain truncation mutants failed to express in the CHO cell system.

Mammalian expression vectors encoding chimeric integrins with the extracellular domains of Drosophila αPS2 or βPS and the intracellular (and in some cases the transmembrane) domains of human αIIb or β3, respectively, were generated using standard overlap extension PCR. The following oligonucleotide primers were used to create the chimeras containing the fly transmembrane domain: αPS2 forward 5′-ACG AGA AGC TGG TGA AGA AGT CCT ATC TGC-3′, αPS2 reverse 5′-GAA GCC GCA CTT GTA GAG CAG CCA GAC-3′, αPS2-αIIb forward 5′-GTC TGG CTG CTC TAC AAG GTC GGC TTC TTC AAG CGG AAC-3′, αIIb reverse 5′-AGA GCG GCC GCG CAC CAT CAC TCC CCC TCT TCA TCA TCT TC-3′, βPS forward 5′-CAA TAT TAT CTT CGC CGT CAC TGC CAG-3′, βPS reverse 5′-GAG CAG CTT CCA CAG CAG GAG AAT G-3′, βPS-β3 forward 5′-C ATT CTC CTG CTG TGG AAA CTC CTC ATC ACC ATC CAC GAC-3′, and β3 reverse 5′-AGA GCG GCC GCA AGA TCT TAA GTG CCC CGG TAC GTG ATA TTG-3′. For the constructs with the human transmembrane domains, the primers are the same as above other than the following: αPS2 reverse 5′-ATC GGG CAC CTG AAG CGG TTC CGG TTC-3′, αPS2-αIIb forward 5′-CCG CTT CAG GTG CCC GAT GCC ATT CCA ATC TGG TGG GTG C-3′, βPS forward 5′-TCC GGT CAT GGT ACC TGC GAA TGC GGT-3′, βPS reverse 5′-ATG AAA ACC TTG GCC GGA CAC TCC TT-3′, and βPS-β3 forward 5′-AGA GCG GCC GCA AGA TCT TAA GTG CCC CGG TAC GTG ATA TTG-3′. Chimeric integrins with the extracellular and transmembrane domains of human αIIb or β3 and the intracellular domains of Drosophila αPS or βPS were constructed with following oligonucleotide primers: αIIb forward 5′-CCT CCT GTC AAC CCT CTC AA-3′, αIIb reverse 5′-CCA CAT GGC CAG GAC CAG G-3′, αIIb-αPS forward 5′-CTG GTC CTG GCC ATG TGG AAG TGC GGC TTC TTT AAC CGC-3′, αPS reverse 5′-TTT CTT AAG CTG GCA CTC TAC AGG TGC TCG TC-3′, β3 forward 5′-GCA ATG GGA CCT TTG AGT GT-3′, β3 reverse 5′-CCA GAT GAG CAG GGC GGC-3′, β3-βPS forward 5′-GCC GCC CTG CTC ATC TGG AAG CTG CTC ACT ACG ATC CAC G-3′, βPS reverse 5′-TTT CCT GCA GGG CGA ATC TAT TTG CCC GCA TAC ATG-3′. These constructs were placed under the control of the CMV promoter in pcDNA3.1 and sequences were confirmed by direct DNA sequencing.

Plasmids expressing a chimeric integrin subunit were transfected into CHO cells in a 1:1 weight ratio with either the corresponding Drosophila wild-type or chimeric integrin subunit. For talin head overexpression experiments, chimeric and wild-type integrin constructs were transiently transfected into CHO cells along with a green fluorescent protein (GFP)-murine talin head F2–F3 domain chimera (or empty vector) and harvested 48 h later for analysis of TWOW-1 binding by flow cytometry. For talin knockdown experiments, chimeric integrins were stably transfected into the CHO cells. After selection with G418, stable transformants expressing high levels of the chimeric integrins were obtained by single-cell sorting using antibody CF.2C7.

A mammalian expression vector encoding GFP-Drosophila talin head domain chimera was created by cloning the Drosophila talin head domain (amino acid residues 1-470) using reverse-transcribed RNA obtained from S2/M3 cells. The talin head coding sequence was subcloned in-frame into pEGFP-N1 (Clontech, Palo Alto, CA) to create GFP-Drosophila talin head under the control of the CMV promoter. A GFP-murine talin head F2-F3 fragment (amino acids 206-405) was described previously (Calderwood et al., 2002). GFP-talin head chimeras were transfected into CHO cells expressing either αIIbβ3 or αPS2βPS using Lipofectamine. Expression was confirmed by GFP fluorescence and Western blotting of cell lysates with a monoclonal anti-enhanced GFP (EGFP) antibody (Clontech).

RNA Interference

Knockdown of talin and βPS integrins by RNA interference (RNAi) in Drosophila S2/M3 cells was performed as described (Worby et al., 2001) with exceptions as noted. Total genomic DNA was isolated from S2/M3 cells (QIAamp DNA minikit, Qiagen, Chatsworth, CA) and used as the template for PCR amplification of single-stranded DNA fragments 400-600 nucleotides (nt) in length that correspond to exons of the rhea (talin) and myospheroid (βPS) genes as follows: DT273 exon 5 of talin, DT996 exon 9 of talin, and Mys6010 exon 5 of βPS. The oligonucleotide primers used to amplify these segments of DNA are as follows, where the first 20 nt of each is the T7 RNA polymerase recognition sequence: DT273 forward 5′-TAA TAC GAC TCA CTA TAG GGC GCA CCA AGG GAA TCG AGA-3′; DT273 reverse 5′-TAA TAC GAC TCA CTA TAG GGC CAC TGA CTC TTC CAC CAT-3′; DT996 forward 5′-TAA TAC GAC TCA CTA TAG GGC GCT CTA TTG AAT GGC GTG-3′; DT996 reverse 5′- TAA TAC GAC TCA CTA TAG GGC GTT GGT GCC ACA ACT TTG-3′; Mys6010 forward 5′-TAA TAC GAC TCA CTA TAG GGC GAG GTG AAG AAT GCC ACA G-3′; Mys6010 reverse 5′-TAA TAC GAC TCA CTA TAG GGC AAC CAC ATT GGA TGA ATC G-3′. Cells were harvested 5 d after the addition of double-stranded RNA to the cell culture medium because preliminary experiments showed maximum reduction in levels of the target protein at this time, as determined by Western blotting or flow cytometry. For cell-spreading experiments using Drosophila S2/M3 cells, double-stranded RNAs targeting talin were prepared using the following oligonucleotide primer pairs, where the first 27 nt of each is the T7 RNA polymerase recognition sequence: talin exon 2 forward 5′-GAA TTA ATA CGA CTC ACT ATA GGG AGA CCA GCG AAT ATG GAC TGT TTA-3′; talin exon 2 reverse 5′-GAA TTA ATA CGA CTC ACT ATA GGG AGA TTT CAT CGT CCG TCT TTA GTT TC-5′; talin 3′ untranslated region forward 5′-GAA TTA ATA CGA CTC ACT ATA GGG AGA CTA TAT GCC TCT AC-3′; talin 3′ untranslated region reverse 5′-GAA TTA ATA CGA CTC ACT ATA GGG AGA GCA ACT GCA TAC ACG ACT CG-3′; GFP forward 5′-GAA TTA ATA CGA CTC ACT ATA GGG AGA ATG GTG AGC AAG GGC-3′; GFP reverse 5′-GAA TTA ATA CGA CTC ACT ATA GGG AGA CAG TTA TTA CTT GTA CAG-3′. For these constructs, near complete talin knockdown was confirmed by Western blotting at 72 h after transfection. Knockdown of talin in CHO cells was performed using short hairpin RNA (shRNA; Paddison et al., 2002). The plasmid encoding shRNA targeting murine talin1 (MT749), in contrast to its mismatched control shRNA, has been shown to effectively knockdown CHO cell talin, with a maximum effect at 72 h (Tadokoro et al., 2003). Therefore, CHO cells stably expressing αIIbβ3 (D723R) were cotransfected with αPS2, βPS, pEGFP-C1 (Clontech) and either MT749 or its mismatched control using Lipofectamine. In experiments with CHO cells stably expressing chimeric or wild-type integrins, cells were similarly transfected with MT749 or mismatched control shRNA and analyzed 72 h later by flow cytometry or Western blotting.

Flow Cytometry

Surface expression and affinity of integrins was monitored by flow cytometry as described for S2/M3 cells (Bunch et al., 2006) and CHO cells (Tadokoro et al., 2003). Analyses were gated on high-GFP–expressing cells as a marker for transfection. Nonspecific binding of ligand-mimetic antibodies to αPS2βPS (TWOW-1) or αIIbβ3 (PAC-1) was taken as binding in the presence of 10 mM EDTA or 2 mM RGDS and was subtracted from total binding to obtain specific binding. In experiments designed to test the effect of metabolic energy depletion on integrin affinity, harvested cells were incubated in buffer containing 0.2% sodium azide and 4 mg/ml 2-deoxy-d-glucose (Sigma) for 30 min at room temperature before staining with ligand-mimetic antibodies. Staining and washes were performed in the absence of glucose, whereas controls were processed in glucose-containing buffer. For measurement of intracellular talin expression, cells were fixed for 30 min in 0.5% paraformaldehyde, which was then neutralized for 5 min with 0.2 M glycine in 0.5 M Tris, and 0.2% sodium azide, pH 7.4. After two washes with PBS, cells were permeabilized with 0.1% saponin in PBS/1.0% bovine serum albumin (BSA), pH 7.4, and talin staining was achieved with antibody J10 (2%) or 8d4 (10 μg/ml). Nonimmune Ig of appropriate isotype was used as a negative control. After addition of fluorophore-conjugated secondary antibody and further washing with buffer containing 0.025% saponin, talin expression was quantified by flow cytometry (Tadokoro et al., 2003).

Cell Spreading

S2/M3 cells, including those stably expressing αPS2βPS, were treated with RNAi constructs targeting talin or GFP (a negative control) in Ultra Low Attachment 24-well plates (Corning Glass Works, Corning, NY). Two different double-stranded RNAs targeting talin were used: one corresponding to exon 2 and the other corresponding to the 3′-untranslated region. After 3 d, the cells were plated on tissue culture plates in usual growth medium containing serum and allowed to spread for 4–5 h before being photographed to assess cell spreading. This spreading is integrin-mediated (parental S2/M3 cells without αPS2βPS integrins do not spread under these conditions) and depends upon ligands provided by the serum in the medium or synthesized by the cells (Bunch and Brower, 1992; Jannuzi et al., 2002). All photographs were coded, and mixed, and the percentages of spread cells was determined by a blinded observer. For each experiment, three representative fields were photographed and scored. Results are the averages of the three experiments ± SEM.

Affinity Chromatography with Recombinant Integrin Cytoplasmic Tails

Expression and purification of recombinant human integrin tail constructs bound to His-bind resin (Novagen, Madison, WI) and affinity chromatography using these constructs have been described (Pfaff et al., 1998; Arias-Salgado et al., 2003). The recombinant Drosophila βPS tail and its IYK/AYA mutant were similarly prepared using the appropriate βPS/pcDNA3.1 vectors as templates for PCR amplification of the coding sequence corresponding to residues K-800 to K-846. The following oligonucleotide primers were used: forward 5′-GCT ATC TGG AAG CTT CTC ACT ACG ATC-3′; and reverse 5′-CTT TTC ATC GGA TCC AAT CTA TTT GCC CGC-3′. As a source of talin, human platelets or S2/M3 cells were solubilized in buffer containing 1% NP-40, 150 mM NaCl, 50 mM Tris, pH 7.4, 1 mM sodium vanadate, 0.5 mM sodium fluoride, 1 mM leupeptin, and complete protease inhibitor cocktail (Roche Applied Science). After clarification, 750 μg of platelet lysate or 2.5 mg of S2/M3 lysate were incubated with 100 μl of resin for 4 h at room temperature, and after washing, bound proteins were subjected to SDS-PAGE. Then talin was detected on Western blots with antibody 8d4 or J10. The amount of cytoplasmic tail proteins bound to the resin was monitored by Coomassie brilliant blue staining of the polyacrylamide gels (Pfaff et al., 1998; Arias-Salgado et al., 2003).

TWOW-1 Binding to Imaginal Discs

TWOW-1 binding to heads from late third instar larvae was carried out in BES-buffered Tyrode's (137 mM NaCl, 2.9 mM KCl, 20 mM BES, 0.1% glucose, pH 7.5) containing 1 mg/ml BSA, 0.01 mM Ca²⁺, and 1 mM Mg²⁺. Where indicated, the buffer also contained either 1 mM MnCl₂ or 5 mM EDTA. The heads were blocked in BES-Tyrode's plus BSA for 10 min before incubation with TWOW-1 (120 μg/ml) for 20 min at room temperature. Samples were washed three times in buffer then incubated with secondary antibody (0.02 mg/ml AlexaFluor 568 goat anti-mouse immunoglobulin; Molecular Probes) for 20 min at room temperature. After three washes, they were transferred in steps of increasing glycerol concentration to 30% 0.1 mM Tris, pH 8.2, and 70% glycerol. Wing imaginal discs were dissected and mounted in VectaShield Mounting Medium (Vector Laboratories, Burlingame, CA) before visualization on a Zeiss Universal Microscope using an AxioCam MRm digital camera with AxioVisionAC v4.2 software (Zeiss, Thornwood, NY).

RESULTS AND DISCUSSION

In vertebrates, talin modulates the adhesive ligand binding affinity of many β1, β2, and β3 integrins (Ginsberg et al., 2005). Here we sought to determine whether talin is also capable of modulating the affinity of Drosophila αPS2βPS, whose β cytoplasmic domain is closely related to vertebrate β cytoplasmic domains, β1 in particular (Burke, 1999; Hynes and Zhao, 2000). The affinity state of αPS2βPS was monitored with TWOW-1, a ligand-mimetic Fab fragment selective for high-affinity αPS2βPS (Bunch et al., 2006). For comparison, in some experiments the affinity of human αIIbβ3 was monitored with the ligand-mimetic PAC-1 Fab (Abrams et al., 1994). Because both TWOW-1 and PAC-1 Fabs are monovalent, they are sensitive to changes in integrin affinity but not to changes in integrin valency or clustering (Abrams et al., 1994; Hato et al., 1998).

In initial studies, TWOW-1 was found to bind specifically to Drosophila S2/M3 cells stably expressing αPS2βPS, as assessed by flow cytometry. In contrast, S2/M3 cells expressing no significant amounts of αPS2βPS had negligible TWOW-1 binding (not shown). Addition of 1 mM MnCl₂ caused a further increase in TWOW-1 binding to S2/M3 cells expressing αPS2βPS (Figure 1A, compare the black and hatched bars on the far left). MnCl₂ was used here because it is an extrinsic activator of Drosophila and vertebrate integrins (Takagi et al., 2002; Litvinov et al., 2004; Bunch et al., 2006) and because cellular agonists that activate αPS2βPS have yet to be identified, unlike the case of platelets where physiological agonists, such as thrombin, activate αIIbβ3 (Shattil and Newman, 2004). With vertebrate integrins, both the basal activation state and agonist-induced activation are blocked by RNAi-mediated talin knockdown (Tadokoro et al., 2003). Therefore, talin in the S2/M3 cells was knocked down with either of two independent double-stranded RNAs derived from Drosophila talin (DT273 and DT996). DT273 or DT996 caused a specific 85–90% reduction in talin expression, as assessed by intracellular staining with an anti-talin antibody and flow cytometry (Figure 1B) or Western blotting (not shown). Despite talin knockdown, neither basal nor MnCl₂-induced TWOW-1 binding were affected (Figure 1A). As a control, a double-stranded RNA (Mys6010) designed to knock down βPS caused a marked reduction in αPS2βPS expression and TWOW-1 binding (Figure 1, A and B). Although it is theoretically possible that the small amount of residual talin in the talin knockdown cells was sufficient to support the basal activation state of αPS2βPS, it is notable that knockdown of vertebrate talin by only 70% abolishes vertebrate integrin activation (Tadokoro et al., 2003), and talin knockdown significantly affected the ability of the cells to spread. Thus, Drosophila talin does not appear to be required for regulation of basal αPS2βPS affinity in Drosophila S2/M3 cells.

Figure 1.

Talin knockdown does not affect αPS2βPS affinity in Drosophila S2/M3 cells. S2/M3 cells stably expressing αPS2βPS were transfected with double-stranded RNA constructs targeting talin (DT273 and DT996) or the βPS integrin subunit (Mys6010). Cells were harvested 5 d later and analyzed by flow cytometry. (A) Specific binding of the ligand-mimetic Fab fragment, TWOW-1 (10 μg/ml). In some cases, TWOW-1 binding was performed in the presence of 1 mM MnCl₂. Results are expressed as means ± SEM; n = 3 independent experiments. (B) Knockdown of talin significantly reduces intracellular talin expression, but does not affect surface expression of αPS2βPS, and vice versa. Protein expression, measured by flow cytometry as detailed in Materials and Methods, is reported as percentage of that observed in control cells not subjected to knockdown. Data represent means ± SEM; n = 8–14.

To determine whether cellular context can account for this apparent difference between vertebrate integrins and Drosophila αPS2βPS, we coexpressed αPS2βPS and a high-affinity, talin-regulatable form of αIIbβ3, αIIbβ3 (D723R), in CHO cells. CHO cell talin can be specifically knocked down using an shRNA (MT749) that targets murine talin1 (Tadokoro et al., 2003). Although MT749, but not a mismatched control shRNA, decreased the binding of ligand-mimetic PAC-1 Fab to αIIbβ3 (D723R), it had no significant effect on TWOW-1 binding to αPS2βPS (Figure 2). Together, these results indicate that the ligand-binding affinity of αPS2βPS is relatively unaffected by depletion of Drosophila or mammalian talin, independent of cellular context.

Figure 2.

Talin knockdown does not affect αPS2βPS activation in CHO cells. CHO cells stably expressing the constitutively active human integrin mutant αIIbβ3 (D723R) were transiently cotransfected with plasmids encoding GFP, αPS2, βPS, shRNA targeting talin (MT749) or a mismatched shRNA control. Seventy-two hours later, transfected cells were identified by GFP fluorescence and specific binding of TWOW-1 Fab or PAC-1 Fab was measured by flow cytometry. MT749, in contrast to its mismatched control, significantly reduced intracellular talin expression, but did not affect surface expression of αPS2βPS (not shown). Results are means ± SEM; n = 4.

One question raised by these results is whether the activation state of αPS2βPS is subject to cellular regulation at all. Affinity regulation of mammalian integrins is dependent on metabolic energy (O'Toole et al., 1994). To determine if the same is true for αPS2βPS, TWOW-1 binding to αPS2βPS was measured in S2/M3 (Figure 3A) and CHO cells (Figure 3B) incubated in the presence or absence of 2-deoxyglucose and sodium azide to deplete metabolic ATP. Basal TWOW-1 binding was significantly reduced by 2-deoxyglucose and sodium azide in both cell types, although perhaps not to the same degree as the reduction in PAC-1 Fab binding to αIIbβ3 (D723R) in CHO cells (Figure 3B). As previously observed for PAC-1 binding to αIIbβ3 (O'Toole et al., 1994; Tadokoro et al., 2003), the effect of energy depletion on TWOW-1 binding was minimized by extrinsic activation of αPS2βPS with MnCl₂ (Figure 3, A and B). In the case of αIIbβ3, a mutation in the membrane-proximal region of the αIIb cytoplasmic tail (GFFKR > GFFKA) leads to increased PAC-1 binding that is prevented by cellular energy depletion with 2-deoxyglucose and sodium azide (our unpublished observations). We found that similar mutations within the membrane-proximal region of αPS2 (GFFNR > GFANA) also led to increased TWOW-1 binding (Bunch et al., 2006); however, this binding was not reduced by energy depletion. Thus, the affinity state of wild-type αPS2βPS may be subject to cellular regulation, but there appear to be differences between αPS2βPS and αIIbβ3 in this context, including a differential requirement for talin.

Figure 3.

Activation of αPS2βPS is dependent on cellular metabolic energy. S2/M3 cells stably expressing αPS2βPS (A) or CHO cells transiently expressing αPS2βPS and stably expressing human αIIbβ3 (D723R) (B) were depleted of metabolic ATP by incubation with 2-deoxy-d-glucose and sodium azide for 30 min. Specific TWOW-1 or PAC-1 Fab binding was then measured. Results are means ± SEM; n = 4.

In CHO cells, overexpression of talin's isolated head domain or its F2-F3 subdomain activates αIIbβ3 (Calderwood et al., 1999, 2002). To test whether Drosophila αPS2βPS integrins behave similarly, GFP-talin head chimeras were expressed in S2/M3 and CHO cells. In neither cellular context did the fly talin head domain activate αPS2βPS to any substantial degree, although αPS2βPS could still be activated by Mn²⁺ (Figure 4, A and B). In S2/M3 cells, the relative lack of response of αPS2βPS to talin head overexpression was similar to that obtained with a talin head construct containing an arginine to alanine substitution (R367A) that is predicted to prevent talin binding to β integrin tails (Garcia-Alvarez et al., 2003) and that shows reduced localization to sites of muscle attachment in vivo (Tanentzapf and Brown, 2006; Figure 4A). Despite failing to activate αPS2βPS, the wild-type talin head domain was capable of reducing αPS2βPS-mediated cell spreading under normal growth conditions, whereas the R367A mutant was not (not shown). Next, we compared the effects of overexpression of the Drosophila talin head domain with overexpression of murine talin head (F2-F3 fragment) in CHO cells. Neither talin head construct activated αPS2βPS (Figure 4B, left panel), whereas each one activated human αIIbβ3 (Figure 4B, right panel). Recent data suggest that activation of vertebrate β1 integrins requires the presence of the complete talin head, including the N-terminal subdomains (Bouaouina et al., 2007). Though βPS has significant homology with vertebrate β1 integrins (Brower, 2003), these findings cannot fully explain the inability of talin head to activate αPS2βPS because, in our experiments, we used a full-length Drosophila talin head construct. Bouaouina et al. (2007) also found that the full talin head domain (F0–F3) was able to activate vertebrate αIIbβ3 to a greater extent than did the F2–F3 fragment alone, perhaps explaining the relatively higher PAC-1 binding we saw with the Drosophila full talin head versus the murine F2–F3 fragment (Figure 4B, right panel). The fact that the Drosophila talin head domain has the ability to activate vertebrate integrins suggests that it is the αPS2βPS integrin that is relatively more resistant to activation by talin, rather than an intrinsic inability of fly talin to activate integrins.

Figure 4.

Talin head overexpression does not increase activation of αPS2βPS. (A) TWOW-1 (30 μg/ml) binding to S2/M3 cells stably expressing αPS2βPS and GFP-Drosophila talin head (TH wild-type or TH (R367A) mutant). Where indicated, TWOW-1 binding was assessed in the presence of 1 mM MnCl₂. GFP was used as a marker for transfection and an analysis gate for TWOW-1 binding was set on cells with high or no GFP expression to assess transfected (GFP-TH) or untransfected (No TH) cells, respectively. Data are means ± SEM; n = 3. (B) CHO cells stably expressing αPS2βPS or αIIbβ3 were transiently transfected with either a GFP Drosophila talin head chimera or a GFP murine talin head (F2-F3 fragment) chimera. After 48 h, specific TWOW-1 or PAC-1 IgM binding was measured. Preliminary experiments showed that transient expression of the Drosophila or murine talin head constructs in CHO cells caused a 70–100% increase in the surface expression of αPS2βPS, as determined using two separate anti-integrin antibodies, but there was no such effect on surface expression of αIIbβ3. Although the mechanism for this effect is not known, TWOW-1 binding was therefore normalized for αPS2βPS expression in this series of experiments. Data are means ± SEM; n = 5–6.

Because activation of mammalian integrins involves the direct binding of talin to integrin β tails (Tadokoro et al., 2003; Wegener et al., 2007), interpretation of the above TWOW-1 binding data requires knowledge as to whether Drosophila talin is capable of binding to mammalian β tails and whether Drosophila βPS can bind to mammalian talin. To address this, recombinant integrin cytoplasmic tail model proteins, or “tail mimics” (Pfaff et al., 1998; Arias-Salgado et al., 2003) were used in an attempt to pull down talin from cell lysates. As shown previously (Tadokoro et al., 2003), human talin from platelet lysates bound to integrin β1 and β3 cytoplasmic tails, but not to a scrambled, random peptide derived from β3 or to the αIIb cytoplasmic tail. Under these same conditions, human talin also bound to the βPS cytoplasmic tail (Figure 5A). Conversely, Drosophila talin from S2/M3 cells bound not only to the βPS tail, but also to the β1 and β3 tails (Figure 5B). A double mutation of β3 tail residues L-746 and K-748 to alanine abolishes talin interaction with and activation of αIIbβ3 (Tadokoro et al., 2003). When alanines were substituted for the equivalent conserved residues in βPS (I830A/K832A), talin binding in vitro was similarly abolished (Figure 6A), but TWOW-1 binding to αPS2βPS in CHO cells was not affected (Figure 6B). Thus, the observed species differences in talin's effects on integrin affinity cannot be readily explained by major differences in the way that the human and Drosophila integrin β tails bind to talin.

Figure 5.

Talin binding to β integrin cytoplasmic tails is not species-specific. Affinity chromatography was performed as described in Materials and Methods using recombinant cytoplasmic tail mimics of human β1, β3, a random peptide derived from β3 (rβ3), αIIb, and Drosophila βPS. Lysates from human platelets (A) or Drosophila S2/M3 cells (B) were used as a source of human talin (hTalin) or Drosophila talin (dTalin). Binding of talin to integrin tails was assessed by Western blotting. Input amounts of tail mimics were monitored by staining with Coomassie brilliant blue. These experiments are representative of three so performed.

Figure 6.

Mutations that abrogate talin binding to integrin β cytoplasmic tails do not affect αPS2βPS integrin activation state. (A) Affinity chromatography was performed as in Figure 5 using human β1, Drosophila wild-type βPS, and the βPS (I830A/K832A) mutant (designated IYK/AYA), all incubated with Drosophila S2/M3 cell lysates. For clarity, the amino acid sequences of the wild-type and mutant βPS tails are shown, with the positions of the mutated amino acids underlined. This experiment demonstrates that the IYK to AYA mutations abolish talin binding to βPS. (B) CHO cells were transiently transfected with GFP, αPS2, and either wild-type βPS or βPS (I830A/K832A). Then TWOW-1 (10 μg/ml) binding was determined. Differences in the results obtained for βPS and βPS (I830A/K832A) were not statistically significant. Also, there were no differences in integrin surface expression as measured by flow cytometry (not shown). Results are mean ± SEM; n = 3.

Despite this evidence that Drosophila talin can bind to a βPS integrin tail mimic in vitro, there still remains the possibility that fly talin does not interact productively with αPS2βPS integrins within the context of living cells. This idea is contradicted by observations in the whole organism that integrins are required for recruitment of Drosophila talin to focal adhesion-like structures, that absence of talin abolishes clustering of integrins into these structures (Brown et al., 2002) and that overexpression of αPS2βPS increases recruitment of the talin head (Tanentzapf and Brown, 2006). Furthermore, in S2/M3 cells expressing αPS2βPS, we observed that talin knockdown by RNAi reduced the percentage of spread cells in normal growth conditions from 41.5 ± 0.7% (in the presence of control RNAi targeting GFP) to 12 ± 0.6%. Another RNAi construct targeting the 3′-untranslated region of talin also significantly reduced cell spreading (21.7 ± 5.7%). These results are consistent with the report that knockdown of talin by RNAi in cultured Drosophila S2R+ cells, which are normally adherent and well spread on extracellular matrix, causes them to round up and detach from culture plates (Kiger et al., 2003). Analogously, a truncated form of βPS (mys^XR04) recruits reduced levels of talin (Tanentzapf et al., 2006) and decreases cell spreading under normal growth conditions (Jannuzi et al., 2002). These observations indicate that Drosophila talin indeed interacts with αPS2βPS to maintain linkage between the extracellular matrix and the actin cytoskeleton in flies and in our cell culture system.

Amino acid sequences within both the α (Knezevic et al., 1996; Yuan et al., 2006) and β (Patil et al., 1999; Vinogradova et al., 2002; Garcia-Alvarez et al., 2003; Wegener et al., 2007) cytoplasmic domains of vertebrate integrins appear capable of supporting talin interactions. These sequences include the β subunit membrane-proximal region and NPxY motifs, which are highly conserved in Drosophila βPS (Figure 7). In addition, the transmembrane domains of vertebrate integrin subunits are involved in transmitting talin-mediated conformational changes to the integrin extracellular domains (Li et al., 2003; Ginsberg et al., 2005; Luo et al., 2005; Partridge et al., 2005). In flies, these structure–function relationships are less well understood. However, mutations in either the membrane-proximal region or the first NPxY motif of a dimeric βPS mimic abolish both recruitment of the talin head and the dominant negative muscle detachment phenotype of the βPS mimic (Tanentzapf et al., 2006). Moreover, mys^XR04, which lacks NPxY motifs (Jannuzi et al., 2002), is unable to recruit talin head, but retains the ability to recruit full-length talin, albeit at reduced levels (Tanentzapf and Brown, 2006). To determine whether differences between Drosophila and vertebrate extracellular, transmembrane or cytoplasmic domains might help to explain the relative resistance of αPS2βPS to affinity modulation by talin, we generated chimeric Drosophila/human integrin α and β subunits. The first set of chimeras studied contained the αPS2 and βPS extracellular domains and either the transmembrane or cytoplasmic domains of αPS2βPS or αIIbβ3 (Figure 7, A and B). Chimeric integrins were expressed in CHO cells with or without the murine GFP-talin head F2-F3 fragment, and TWOW-1 binding was quantified. Although expression of GFP-talin F2-F3 increased PAC-1 binding to αIIbβ3 in CHO cells as expected (not shown), GFP-talin F2-F3 failed to increase TWOW-1 binding to any of these chimeric integrins, even ones that contained the transmembrane and cytoplasmic domains of αIIb and β3 (Figure 7D). Furthermore, shRNA-mediated knockdown of talin had no effect on TWOW-1 binding to these chimeric integrins (Figure 7E).

Figure 7.

Influence of integrin domains on talin activation of chimeric Drosophila/human integrins. (A) The amino acid sequences of the cytoplasmic domains of αPS2βPS and αIIbβ3. Clustal W (version 1.83) multiple sequence alignments of the cytoplasmic tail domains of the respective fly and human integrin subunits are shown. Regions in gray boxes are thought to be important for talin binding (to varying degrees), in particular the membrane-proximal portions of both α and β integrin subunits and the region in β involving the first NxxY (NPXY) motif (see text for references). Clustal W symbols: * = identical, = conserved, = semiconserved. (B and C) The chimeric integrins studied in this experiment. For clarity, domains from αPS2βPS are denoted D (open bars) and those from αIIbβ3 are denoted H (gray bars). For example, a chimeric integrin designated D/D/H contains α and β subunits with the fly extracellular and transmembrane domains and the human cytoplasmic domains. In D and F, CHO cells were transiently transfected with wild-type αPS2βPS (designated D/D/D), wild-type αIIbβ3 (H/H/H), or a chimeric integrin as indicated. They were cotransfected with either GFP as a marker of transfection (No TH) or a murine GFP-tagged talin head F2–F3 fragment (TH). After 48 h, cells were incubated in the absence or presence of MnCl₂ and TWOW-1 (D) or PAC-1 (F) binding was measured by flow cytometry (n = 3). In E and G, CHO cells were cotransfected with the indicated wild-type or chimeric integrins and with MT749 to knockdown talin or a mismatch control shRNA as in Figure 2. Then TWOW-1 (E) or PAC-1 (G) binding was quantified by flow cytometry (E, n = 3; G n = 4). TWOW-1 and PAC-1 binding are expressed as specific binding normalized to integrin expression, the latter measured in parallel by flow cytometry. In E and G, MT749, but not its mismatched control, significantly reduced intracellular talin expression (not shown). Data represent means ± SEM.

On the basis of these results, an additional chimera was constructed consisting of the extracellular and transmembrane domains of αIIb and β3 and the cytoplasmic domains of αPS2 and βPS, respectively (Figure 7C). This chimera, or wild-type αIIbβ3, was expressed in CHO cells with or without GFP-talin head F2-F3, and PAC-1 binding was quantified. Compared with αIIbβ3, this chimera exhibited higher basal binding of PAC-1 compared with αIIbβ3, consistent with a degree of constitutive activation. Furthermore, in contrast to αIIbβ3, this chimera showed no further PAC-1 binding in response to talin head overexpression (Figure 7F). However, talin knockdown with MT749 shRNA caused a significant reduction in PAC-1 binding to this chimera. A different chimera with the human extracellular and fly transmembrane and cytoplasmic domains could not be evaluated because it was not expressed when transfected into cells. Whereas data with cross-species integrin chimeras must be interpreted cautiously, these results suggest that the differences in talin's ability to regulate αPS2βPS and αIIbβ3 affinity is not likely due to differences in talin's interaction with the integrin subunits. Rather, the differences appear to be explained by structural features inherent in the αPS2βPS extracellular and/or transmembrane domains, features that remain to be identified.

Certain lethal muscle or wing phenotypes in the fly resulting from αPS2βPS gain- or loss-of-function mutations have been attributed to an increase or decrease in integrin affinity for ligands (Martin-Bermudo et al., 1998; Brower, 2003; Tanentzapf and Brown, 2006; Devenport et al., 2007). More recently TWOW-1 binding experiments in S2/M3 cells have supported these contentions, demonstrating that αPS2βPS integrin affinity is influenced by activating mutations in the αPS2 cytoplasmic tail or the βPS I-like domain (Bunch et al., 2006). However, it must be emphasized that changes in αPS2βPS affinity in fly embryos have been inferred rather than measured directly. In an attempt to evaluate αPS2βPS affinity in a relevant Drosophila tissue, TWOW-1 binding to wing imaginal discs was determined. Imaginal discs (and other tissues) from late third instar larvae were dissected and incubated with TWOW-1 in BES-Tyrode's buffer with Ca²⁺ and Mg²⁺, or in the same buffer plus Mn²⁺ to activate integrins. Imaginal discs from this stage express αPS2βPS primarily on the ventral epithelium of the presumptive wing, making this an excellent tissue to assess binding specific for αPS2βPS. Analysis of cDNAs from imaginal discs suggests that they express primarily the αPS2m8 isoform at this time (Brown et al., 1989), and experiments with S2 cells have shown that this variant has low affinity for TWOW-1 in the absence of activation (Bunch et al., 2007). Wing discs incubated without MnCl₂ showed very little TWOW-1 binding (Figure 8A). However if MnCl₂ was present during the incubation, increased TWOW-1 binding was observed on the ventral cells (Figure 8B). In tissues such as imaginal discs, some TWOW-1 binding to αPS2βPS may be obscured by the presence of a competing endogenous ligand, such as tiggrin. Nonetheless, the current results suggest that at least a portion of αPS2βPS on this epithelium is in a low-affinity state just before metamorphosis.

Figure 8.

Drosophila αPS2βPS is not constitutively active in imaginal discs. Posterior halves of Drosophila wing imaginal discs from late third instar larvae, stained for TWOW-1 binding in the absence (A) or presence of MnCl₂ (B). TWOW-1 binding to αPS2βPS expressing ventral cells (arrows) is seen only when integrins are activated by MnCl₂, except for slight staining seen in the wing pouch (asterisk in A), where integrin density is greatest. For both images the initial exposures and any contrast and brightness adjustments are identical. Bar, ∼50 μm.

Here we have shown that talin is not sufficient to modulate αPS2βPS affinity in cultured cells. Whether this is true in the intact organism remains to be determined. Although it is logical to think that TWOW-1 Fab might be a useful reagent to address this question, the use of this ligand-mimetic reporter in flies could be limited to the extent to which αPS2βPS is constitutively bound to tiggrin or other cogante ligands in vivo. Large proteins like talin containing several functional domains frequently play diverse roles within cells. Talin likely has multiple functions in flies, including regulation of gene expression (Becam et al., 2005) and perhaps promotion of integrin clustering, which could account for the observed talin null phenotype (Brown et al., 2002). Furthermore, expression of a full-length talin mutant (R367A) appears to cause a much weaker phenotype than does complete absence of talin, possibly because it is still capable of clustering αPS2βPS into focal adhesion-like structures (Tanentzapf and Brown, 2006). Consequently, it is not necessary to posit a role for talin in affinity regulation of αPS2βPS to explain these phenotypes. It is possible that there exist modes of integrin activation that are independent of, or supplementary to, talin. For example, recent studies of mammalian integrins suggest that kindlin or MIG-2, proteins with a split FERM domain, may play such a role (Kloeker et al., 2004; Shi et al., 2007). However in our preliminary studies, overexpression of kindlin or MIG-2 had no effect on TWOW-1 binding to αPS2βPS in CHO cells (H. Kato, M. H. Ginsberg, C. Wu, and S. J. Shattil, unpublished observations). Altogether, the current results with Drosophila αPS2βPS lead us to speculate that talin's direct role in integrin activation may have developed later during vertebrate evolution to provide exquisite regulation of integrin affinity in highly motile cells. Additional studies will be required to test this hypothesis.

Abbreviations used:

CHO

Chinese hamster ovary

GFP

green fluorescence protein

shRNA

short hairpin RNA

BSA

bovine serum albumin

PBS

phosphate-buffered saline

RNAi

RNA interference.