Adhesion receptors play diverse roles during animal development and require precise spatiotemporal regulation, which is achieved through the activity of their binding partners. Integrins, adhesion receptors that mediate cell attachment to the extracellular matrix (ECM), connect to the intracellular environment through the cytoplasmic adapter protein talin. Talin has two essential functions: orchestrating the assembly of the intracellular adhesion complex (IAC), which associates with integrin, and regulating the affinity of integrins for the ECM. Talin can bind to integrins through two different integrin-binding sites (IBS-1 and IBS-2, respectively). Here, we have investigated the roles of each in the context of Drosophila development. We find that although IBS-1 and IBS-2 are partially redundant, they each have specialized roles during development: IBS-1 reinforces integrin attachment to the ECM, whereas IBS-2 reinforces the link between integrins and the IAC. Disruption of each IBS has different developmental consequences, illustrating how the functional diversity of integrin-mediated adhesion is achieved.
Adhesion receptors play many roles throughout animal development. During dynamic morphogenetic events, transient cell–cell and cell–extracellular matrix (ECM) contacts provide traction for cell migration and facilitate cell rearrangements (Bökel and Brown, 2002). Once three-dimensional tissue architecture is established, it is maintained by stable, long-lasting adhesion. Diverse types of cell adhesion can be obtained through the use of different adhesion receptors or by utilising the same adhesion receptor but changing the composition of the intracellular complex with which it associates. For example, integrins – the principal mediators of cell–ECM attachment in metazoans – associate with different intracellular components over the lifetime of focal adhesions (Geiger et al., 2009), a process referred to as ‘maturation’. This can profoundly affect how the integrin adhesion complex (IAC) behaves. For instance, integrin binding to different cytoplasmic adapter proteins can act to switch on and off dynamic adhesive processes associated with cell migration (Calderwood, 2004).
The IAC is a vast network of interacting proteins (Zaidel-Bar et al., 2007), but there are only a relatively small number of adapter proteins that link integrins and the IAC (Liu et al., 2000). Among these, the large cytoplasmic adapter protein talin provides an essential link between integrins and the cytoskeleton by directly binding actin (Franco-Cea et al., 2010) and by recruiting downstream components of the adhesion complex (Giannone et al., 2003; Tanentzapf and Brown, 2006). In addition, talin plays an important role in regulating the affinity of integrins for their ECM ligands, a process known as inside-out activation (Shattil et al., 2010; Tadokoro et al., 2003). Talin contains two known integrin-binding sites (IBSs): IBS-1 is located in the N-terminal end of the protein, whereas IBS-2 is in the C-terminus. The IBS-1 domain of talin has been studied extensively for its ability to confer conformational changes in the integrin molecule resulting in inside-out activation. Structural studies of IBS-1 have revealed key residues essential for the binding of talin IBS-1 to the integrin cytoplasmic tail (Anthis et al., 2009; Calderwood et al., 2002; Garcia-Alvarez et al., 2003; Wegener et al., 2007). In vitro and in vivo studies have further shown that the binding of talin IBS-1 to the membrane proximal NPxY sequence in the integrin cytoplasmic tail is required for integrin activation (Calderwood et al., 2002; Tadokoro et al., 2003; Wegener et al., 2007; Tanentzapf and Brown, 2006; Tanentzapf et al., 2006). The crystal structure of the talin IBS-2 domain has also been solved, and residues have been identified in both talin and integrin that mediate their interaction (Cheung et al., 2009; Gingras et al., 2008; Moes et al., 2007; Rodius et al., 2008). Importantly, IBS-2 has been shown to interact with a membrane-proximal part of the integrin cytoplasmic tail, distinct from the NPxY motif; this interaction does not induce inside-out activation (Gingras et al., 2008; Moes et al., 2007; Rodius et al., 2008; Tremuth et al., 2004). However, to date the roles of the IBS-2 domain of talin have not yet been explored in the context of a whole developing organism.
Drosophila is a useful model for analysis of integrin function, as different types of adhesion mediated by integrins can be studied throughout embryogenesis. For example, integrins are essential for transient adhesive processes during development, such as during dorsal closure and germband retraction, two morphogenetic events that involve large-scale movement and fusion of epithelial sheets (Brown et al., 2000). Later in development, stable and long-term integrin-mediated adhesion is required to form and maintain the attachment of muscles to tendon cells at myotendinous junctions (MTJs) (Brown et al., 2000). Analysis of integrin function in flies is simplified by the fact that there is only one widely expressed β-integrin orthologue, βPS, encoded by myospheroid (mys). Mutations in more than a dozen genes encoding cytoplasmic factors involved in integrin function have been isolated and characterised inDrosophila, including mutations in the conserved focal adhesion components PINCH (Clark et al., 2003), tensin (Torgler et al., 2004), ILK (Zervas et al., 2001), FAK (Grabbe et al., 2004) and paxillin (Yagi et al., 2001). Moreover, flies have a well-conserved orthologue of talin, encoded by rhea (Brown et al., 2002). Whereas the human and mouse genomes contain multiple partially redundant talin genes, the Drosophila genome encodes only one talin gene; thus, the rhea mutant phenotype represents a true talin-null phenotype. Genetic analysis of rhea has demonstrated that talin is a key component of integrin-mediated adhesions and is essential for integrin-mediated attachment to the ECM, as rhea-null mutants show an identical phenotype to that seen in loss-of-function integrin mutants (Brown et al., 2002).
We have previously shown that targeted disruption of the talin IBS-1 domain does not completely abrogate talin function but leads to a specific defect: detachment of integrins from the ECM late in embryonic development (Tanentzapf and Brown, 2006). This suggests that the talin IBS-2 domain is partially redundant with the IBS-1 domain and also raises the possibility that each IBS can have different and specialised roles during development. Here, we extend our previous analyses to address the specific developmental roles of IBS-2. Our results confirmed that IBS-1 and IBS-2 act redundantly in some contexts but mediate specialised functions over the course of development. In particular, we observed that IBS-1 had a more pronounced role in linking integrins to the ECM, whereas IBS-2 was more important for linking integrins to the intracellular IAC. Consequently, we propose that developmental processes that depend on ECM and cytoskeletal linkage are more sensitive to disruption in the IBS-1 and the IBS-2 domains of talin, respectively. Our work defines the relative importance of each IBS domain of talin during development and provides insight into how the functional diversity of integrin-mediated adhesion is attained.
The conserved IBS domains act redundantly to recruit talin to sites of integrin-mediated adhesion downstream of integrin
The IBS-2 domain of talin comprises two bundles of five helices: IBS-2A (residues 1983–2130 in Drosophila talin) and IBS-2B (residues 2131–2297 in Drosophila talin). Although the entire IBS-2 domain is well-conserved between fly and vertebrate talin (66% protein sequence similarity, 47% sequence identity; supplementary material Fig. S1) IBS-2A is much more conserved (80% similarity, 71% identity) than IBS-2B (57% similarity, 41% identity). We have previously expressed a protein fragment corresponding to the IBS-2A domain of Drosophila talin showing that it was stable, soluble and readily crystallized, thus allowing its structure to be elucidated (Cheung et al., 2009). We tagged the IBS-2A domain with GFP (UAS-GFP–IBS2, Fig. 1A), ectopically expressed it in muscles and studied its localisation. In Drosophila embryos, major integrin-containing adhesive junctions are formed at MTJs, where muscle ends attach via the ECM to the epidermis. We have previously developed a robust assay to measure the relative enrichment of proteins at MTJs by quantifying the signal at the MTJ and expressing it as a ratio over the signal in the cytoplasm (Franco-Cea et al., 2010; Tanentzapf and Brown, 2006; Tanentzapf et al., 2006). Talin is known to specifically localise to MTJs (Brown et al., 2002; Tanentzapf and Brown, 2006). When the tagged IBS-2 domain (UAS-GFP–IBS2) was expressed in muscles using the Mef2-GAL4 driver, it showed a twofold enrichment at MTJs compared with that in the cytoplasm (Fig. 1B; supplementary material Fig. S3). This localisation was dependent on integrin because in the absence of βPS integrin, UAS-GFP–IBS2 was not enriched at MTJs. (Fig. 1B; supplementary material Fig. S3). Previous in vitro studies identified two sets of mutations, K2085D and K2089D (KS>DD), and L2094A and I2095A (LI>AA) in vertebrate talin (equivalent to K2094D and S2098D, and L2103A and I2104A in Drosophila talin, respectively) that disrupt the binding of IBS-2 to the integrin cytoplasmic tail (Moes et al., 2007; Rodius et al., 2008). We introduced these mutations into UAS-GFP–IBS2 and found that the mutated IBS-2 domain was not enriched at MTJs (Fig. 1B; supplementary material Fig. S3). Similarly, and consistent with our published data, when the IBS-1 domain was tagged with GFP (UAS-IBS–GFP) and ectopically expressed in muscles using the Mef2-GAL4 driver, it localised to MTJs in an integrin-dependent fashion (Tanentzapf and Brown, 2006) (Fig. 1A,B; supplementary material Fig. S3). Moreover, introduction of a mutation, designed to disrupt its interaction with the integrin cytoplasmic tail, into UAS-IBS–GFP [R367A; equivalent to R358A in vertebrate talin (Garcia-Alvarez et al., 2003)] abolished the enrichment of the tagged IBS-1 domain at the MTJs (Fig. 1B; supplementary material Fig. S3). In addition, in accordance with our previous studies, we observed enrichment of UAS-IBS1–GFP in the nucleus. Our previous work strongly suggests that this nuclear staining is unlikely to be functionally significant and instead is an artefact of overexpression of the GFP-tagged fusion protein (Tanentzapf et al., 2006).
To assess further the importance of each IBS for talin localisation, mutations designed to disrupt the talin IBS domains were introduced into full-length talin. We utilized a previously established protocol and introduced point mutations into a ubiquitously expressed full-length talin rescue construct (Tanentzapf and Brown, 2006) (see Materials and Methods). Five GFP-tagged rescue constructs were made: wild-type talin–GFP, IBS-1 mutant talin R367A, IBS-2 mutant talin KS>DD, IBS-2 mutant talin LI>AA, and IBS1-IBS2 double mutant talin R367A; LI>AA. We ensured that our rescue constructs were expressed at similar levels to the wild-type rescue construct (supplementary material Fig. S2). We measured the relative enrichment of the full-length GFP-tagged talin at the MTJ by determining the ratio of MTJ signal to cytoplasmic signal. The wild-type talin–GFP was enriched at MTJs, similar to the distribution of the endogenous protein (Brown et al., 2002; Tanentzapf and Brown, 2006) as in a wild-type background, GFP signal at the MTJs was over fourfold greater than in the cytoplasm (Fig. 1C,E). In comparison, the IBS-1 mutant talin R367A and the two IBS-2 mutants, talin LI>AA and talin KS>DD, showed reduced enrichment at MTJs (Fig. 1C,G,I,K). Moreover, the IBS-1 and IBS-2 double mutant R367A; LI>AA was not enriched at MTJs in a wild-type background, as the ratio of MTJ to cytoplasmic signal was ~1:1 (Fig. 1C,M). This result is consistent with previous data (Tanentzapf and Brown, 2006; Tanentzapf et al., 2006) suggesting that, in a wild-type background, ectopically supplied talin competes with endogenous talin for IBSs. Moreover, GFP-tagged full-length rescue constructs containing mutations in either the IBS-1 or IBS-2 domain compete less efficiently with the endogenous protein when compared with the wild-type full-length rescue construct.
As endogenous talin can compete with the GFP-tagged full-length talin rescue constructs, we studied their localisation in talin-deficient embryos (see Materials and Methods). In the absence of endogenous talin, there was no significant difference (P<0.5935) in enrichment at MTJs between the GFP-tagged full-length wild-type talin rescue construct and GFP-tagged full-length rescue constructs containing mutations in either the IBS-1 or IBS-2 domains (Fig. 1D,F,H,J,L). However, the enrichment of the double IBS mutant talin was reduced by a statistically significant amount (P<0.0058) in the absence of endogenous talin (Fig. 1D,N). These results suggest that introducing mutations that impinge on integrin binding in either IBS domain confer only a small reduction in talin recruitment to integrins, whereas mutating both IBS domains confers a larger reduction in talin recruitment.
In addition to the MTJs, we studied the localisation of GFP-tagged full-length talin rescue constructs in ectodermal tissues that require talin for normal morphogenesis. Wild-type talin-GFP localised to the cell cortices of ectodermal epithelia in stage 12 embryos undergoing germband retraction, and also to the leading edge of the epidermis in stage 14 embryos undergoing dorsal closure (supplementary material Fig. S4). Introduction of mutations into either IBS-1 or IBS-2 in the GFP-tagged full-length rescue constructs did not significantly affect cortical enrichment in either stage 12 ectodermal epithelia or stage 14 leading edge cells (supplementary material Fig. S4). However, consistent with our MTJ studies, the double IBS mutant was not enriched at cell cortices in either cell type (supplementary material Fig. S4).
The two IBS domains of talin are recruited to integrins by different mechanisms
Previous studies exploring how the IBS-1 domain of talin interacts with integrins have implicated the distal half of the integrin cytoplasmic tail, including the first of its two NPxY motifs (Kaapa et al., 1999; Pfaff et al., 1998; Tanentzapf and Brown, 2006; Tanentzapf et al., 2006). There are conflicting reports as to whether talin IBS-2 also requires the distal part of the integrin cytoplasmic tail (Gingras et al., 2008; Rodius et al., 2008). To elucidate the mechanism that recruits the IBS-2 domain of talin to integrin we used an allele of the βPS integrin subunit that lacks the second half of the integrin cytoplasmic tail (supplementary material Figs S1, S3). This allele, mysG1, carries a mutation in a splice acceptor site, resulting in truncation of the cytoplasmic tail, but does not affect protein stability or surface expression (Jannuzi et al., 2002). The GFP-tagged IBS-2 domain concentrated at MTJs to the same extent in muscles in mysG1 mutant embryos as in wild-type embryos (supplementary material Fig. S3). By comparison, GFP-tagged IBS-1 domain failed to concentrate at MTJs in mysG1 mutant embryos (supplementary material Fig. S3).
We have previously shown that full-length talin is recruited by integrin lacking the distal half of the integrin cytoplasmic tail (Tanentzapf and Brown, 2006). We confirmed this observation using GFP-tagged full-length talin, which indeed concentrated at MTJs in mysG1 mutant embryos (supplementary material Fig. S3). The GFP-tagged full-length IBS-1 R367A mutant was enriched at MTJs (supplementary material Fig. S3). However, the GFP-tagged full-length IBS-2 KS>DD and LI>AA talin mutants were not enriched at MTJs (supplementary material Fig. S3). This suggests that integrin-mediated recruitment of talin is mediated by interactions of IBS-1 with membrane distal regions, and IBS-2 with proximal regions, of the integrin cytoplasmic tail. Moreover, the IBS-2 domain can recruit talin to MTJs independently of IBS-1-mediated interactions with the integrin cytoplasmic tail.
Both IBS domains of talin contribute to talin function but have different roles during development
To assess the functions of its IBS domains, wild-type talin was replaced in developing Drosophila embryos with rescue transgenes containing mutations in the IBS sequences. The ability of wild-type and mutant talin rescue constructs to compensate for the absence of talin was studied by removing talin [using the dominant female sterile germline clone technique (Chou and Perrimon, 1996)] and replacing it with ubiquitously-expressed rescue transgenes. Three major talin-dependent developmental processes were quantitatively assayed: dorsal closure, germband retraction and muscle attachment at the MTJs (Fig. 2). The wild-type talin–GFP rescue transgene, on which all of the mutant rescue transgenes were based, rescued all the null mutant phenotypes we assayed (supplementary material Fig. S5). Although the rescue obtained with the wild-type talin transgene was nearly complete, a small percentage (1%–5%) of embryos exhibited defects (Fig. 2A,G–I; supplementary material Fig. S5A–E). In accordance with our published data, we found that the talin IBS-1 mutant R367A substantially rescued both the integrin-dependent morphogenetic processes of germband retraction and dorsal closure (Tanentzapf and Brown, 2006) (Fig. 2C,G,H). However, we observed muscle detachment at late stages of development in a significant proportion of embryos rescued with the talin R367A transgene (Fig. 2C,I). This phenotype is different from the complete muscle detachment and rounding observed in talin-null mutant flies (Fig. 2B) and we have previously shown that this phenotype can be rescued by expression of an activated integrin mutant (Tanentzapf and Brown, 2006). By comparison, neither mutant IBS-2 rescue transgene, talin LI>AA nor talin KS>DD, substantially rescued the germband retraction and dorsal closure defects in talin-deficient embryos (Fig. 2D,E,G,H). However, both mutant IBS-2 transgenes partially rescued MTJ defects (Fig. 2D,E,I). Furthermore, the double IBS mutant rescue construct, talin R367A; LI>AA conferred very poor rescue of all talin mutant phenotypes assayed (Fig. 2F,G–I). These results show that although the IBS domains of talin act partially redundantly during development, they have specific roles during germband retraction, dorsal closure and muscle attachment.
IBS-2 is required to maintain integrin linkage to talin and other IAC components
As the proposed function of the IBS domains of talin is to link integrin and talin, we analysed whether talin colocalised with integrin in confocal z-stacks of the embryonic musculature in talin-deficient embryos expressing the talin IBS mutant transgenes. In the absence of maternal and zygotic talin expression, talin was not detected at MTJs (Fig. 3B). Although muscle tissue architecture was severely disrupted as a result of loss of talin function, integrin localised to MTJs at levels that were comparable to those observed in wild-type MTJs (Fig. 3B) (Brown et al., 2002). When a wild-type talin rescue transgene was introduced into talin-deficient embryos, talin and integrin overlapped at MTJs (Fig. 3A,P). In talin-deficient embryos rescued with the IBS-1 mutant transgene talin R367A, talin and integrin also appeared to overlap at MTJs (Fig. 3C). However, analysis of colocalisation by the Pearson correlation coefficient showed a slight decrease of 12.5% in colocalisation compared with that for the wild-type rescue transgene; this is probably due to a slight reduction in talin levels following muscle detachment (Fig. 3P). Strikingly, in talin-deficient embryos rescued with the IBS-2 mutant transgenes talin LI>AA or talin KS>DD, we observed a substantial reduction in colocalisation between talin and integrin at MTJs (Fig. 3D,E; arrows in 3D′,E′). Compared with wild-type rescued embryos, Pearson correlation coefficients were reduced by 38% and 32% in talin LI>AA and talin KS>DD rescued embryos, respectively (Fig. 3P). Importantly, even though MTJ architecture was disrupted in talin-deficient flies rescued with either the IBS-1 mutant or with the two IBS-2 mutant constructs, the Pearson correlation coefficients were significantly lower in the IBS-2 mutants (P=0.0009). These data suggest that the IBS-2 domain is essential for maintaining the link between talin and integrin.
Because talin forms a crucial bridge between integrins and the IAC at MTJs (Tanentzapf and Brown, 2006), we sought to determine whether other IAC components remained associated with integrin when the talin IBS domains were mutated. The distribution of the IAC marker paxillin (Yagi et al., 2001) was used to visualise the IAC in talin-deficient embryos rescued with talin IBS domain mutant transgenes (Fig. 3F–O). In the absence of maternal and zygotic talin, talin protein was absent and paxillin was severely reduced at the MTJ, although integrin localisation was comparable to that observed in wild-type MTJs (Fig. 3G,L). By comparison, when a wild-type talin rescue transgene was introduced into embryos lacking endogenous talin, paxillin, integrin and talin colocalised at MTJs (Fig. 3F,K). In talin-deficient embryos rescued with the IBS-1 mutant transgene talin R367A, paxillin localisation largely overlapped with both integrin and talin, even when MTJs were disrupted (Fig. 3H,M). Moreover, analysis of colocalisation by use of the Pearson correlation coefficient showed that there was no statistically significant difference in colocalisation between paxillin and talin, and paxillin and integrin at MTJs in the IBS-1 mutant compared with that for the wild-type rescue transgene (Fig. 3Q,R). By comparison, clear separation was observed between paxillin and integrin at MTJs in talin-deficient embryos rescued with the IBS-2 mutant transgenes talin LI>AA or talin KS>DD (Fig. 3I,J; arrows in I′,J′), but the paxillin signal still overlapped with that of talin (Fig. 3N,O). Analysis of colocalisation using the Pearson correlation coefficient showed that there was a statistically significant reduction in colocalisation in MTJs between paxillin and integrin in the IBS-2 mutants compared with that for the wild-type rescue transgene (47% and 38% for KS>DD and LI>AA, respectively; Fig. 3Q). By comparison, colocalisation of paxillin and talin in the IBS-2 mutants was similar to that seen with the wild-type transgene (Fig. 3R). To confirm further these results, distribution of the protein PINCH (Clark et al., 2003), an additional IAC component, was examined at MTJs. As was the case for paxillin, when IBS-2 was mutated, PINCH showed reduced colocalisation with integrin (supplementary material Fig. S6D,E), but it overlapped and colocalised with talin at MTJs (data not shown). Although we observed considerable disruption of MTJ morphology in talin-deficient embryos rescued with the mutant IBS-1 rescue transgene, PINCH and paxillin remained overlapped and colocalised with integrin. We tested the additional IAC markers, phosphorylated FAK (Grabbe et al., 2004) and phosphorylated tyrosine (Yagi et al., 2001), and observed identical results to those shown for PINCH and paxillin for both the IBS-1 and IBS-2 mutant transgenes (data not shown). Altogether, these results show that the IBS-2 domain of talin is not only required to maintain linkage of talin to integrin but also to maintain association between integrins and the rest of the IAC.
IBS-1 but not IBS-2 is required to maintain integrin linkage to the ECM
Tiggrin, the RGD-motif-containing ECM ligand for αPS2βPS integrins, accumulates at MTJs in the space between adjacent muscles and tendon cells, and colocalises with integrin (Brown et al., 2000; Fogerty et al., 1994). In talin-deficient embryos rescued with the wild-type talin transgene, tiggrin distribution completely overlapped with integrin (Fig. 4A). Furthermore, and similar to what we have previously shown, we found that in talin-deficient embryos rescued with the IBS-1 mutant transgene talin R367A, integrin showed substantial separation from the ECM (Fig. 4B, arrows in 4B‴) (Tanentzapf and Brown, 2006). However, in talin-deficient embryos rescued with the IBS-2 mutant transgenes talin LI>AA or talin KS>DD, integrins and the ECM remained colocalised at MTJs (Fig. 4C,D). These results indicate that the IBS-1 domain but not the IBS-2 domain of talin is essential for maintaining association between integrins and the ECM.
Mutations in integrin that disrupt binding to talin phenocopy mutations in the talin IBS domains
To examine further whether the talin IBS domains have different roles in integrin-mediated adhesion, mutations designed to disrupt IBS binding were introduced into the βPS integrin cytoplasmic tail. The ability of wild-type and mutant rescue constructs to compensate for the absence of βPS integrin was studied by removing endogenous βPS integrin [using the dominant female sterile germline clone technique (Chou and Perrimon, 1996)] and replacing it with ubiquitously-expressed wild-type and mutant YFP-tagged βPS integrins. We ensured that mutant βPS–YFP transgenes were expressed at similar levels to the wild-type transgene (supplementary material Fig. S2). The wild-type βPS–YFP transgene rescued the embryonic defects associated with loss of integrin function and recruited both talin and the IAC component paxillin at levels similar to those observed in wild-type MTJs (Fig. 5A,B,E–H,I; Fig. 6A,D,E; supplementary material Fig. S7). Previous work has identified the membrane proximal NPxY motif in the integrin cytoplasmic tail as essential for interaction with the talin IBS-1 domain in vertebrates and flies (Calderwood et al., 2002; Kaapa et al., 1999; Tanentzapf and Brown, 2006). We generated a βPS integrin transgene containing the point mutation N828A in the proximal NPxY motif (equivalent to N785A in β1 integrin and N744A in β3 integrin; supplementary material Fig. S1) and found that it failed to rescue the germband retraction and dorsal closure defects of integrin-deficient embryos, but partially alleviated MTJ defects (Fig. 5C,E–G). Moreover, the MTJ defects were similar to those observed in talin IBS-1 domain mutants, as the ECM separated from integrin and analysis of Pearson correlation coefficients showed that there was a statistically significant reduction in colocalisation between tiggrin and integrin (by ~30% in the N828A mutant MTJs compared with those of embryos rescued with the wild-type βPS transgene; Fig. 5J,M, arrowhead in J‴). There was some variability in this hypomorphic phenotype and some MTJs appeared more disrupted than others (compare Fig. 6B with Fig. 5J). Additionally, the N828A mutant βPS integrin was able to recruit, albeit at reduced levels, both talin and paxillin to MTJs and there was no statistically significant change in colocalisation between integrin and talin, integrin and paxillin, or talin and paxillin (Fig. 5H,J,L; Fig. 6B,D,E; supplementary material Fig. S7).
Mutations have been identified in the integrin cytoplasmic tail that interfere with its binding to the IBS-2 domain of talin (Rodius et al., 2008). We generated a βPS integrin rescue transgene containing point mutations in the corresponding residues in Drosophila talin, E810Q and E817Q (equivalent to E726Q and E733Q in vertebrate β3 integrin; supplementary material Fig. S1). This βPS integrin EE>QQ transgene failed to rescue the germband retraction and dorsal closure defects of integrin-deficient embryos (Fig. 5D–F) but partially rescued MTJ defects (Fig. 5G). Importantly, the MTJ defects identified were very similar to those observed with the talin IBS-2 mutant rescue transgenes (compare Fig. 5K with Fig. 3D,E). Although ECM attachment to βPS integrin EE>QQ was maintained (Fig. 5K,M), both talin and paxillin were recruited to MTJs at reduced levels compared to wild-type βPS rescue embryos (see Fig. 5H,L; Fig. 6C,D; supplementary material Fig. S7). Although paxillin and talin remained colocalised with one another (Fig. 6E), they separated away from the integrin, similar to the IBS-2-mutant-rescued talin-deficient embryos. Analysis of βPS EE>QQ mutant embryos by Pearson correlation coefficients revealed ~20% reduction in colocalisation between talin and integrin, and a loss in colocalisation of nearly 60% between paxillin and integrin compared with MTJs in embryos rescued with the wild-type βPS transgene (Fig. 5K,M; Fig. 6C,D). These data show that in general, the N828A and EE>QQ mutations in the integrin cytoplasmic tail give rise to phenotypes similar to those caused by mutations in talin that disrupt the functional interaction between integrin and talin.
IBS-1 and IBS-2 confer different, opposing roles in adhesion dynamics at the MTJ
We wanted to test the hypothesis that different interactions between talin and integrin, through IBS-1 or IBS-2, comprise a mechanism by which the overall stability of integrin-mediated adhesions can be regulated. We have recently shown that the turnover of the IAC can be studied in vivo in live Drosophila embryos using fluorescence recovery after photobleaching (FRAP) to determine the mobile fraction of IAC components (Yuan et al., 2010) (see Materials and Methods). We performed FRAP experiments on GFP-tagged mutant talin rescue constructs expressed in a wild-type background and measured their turnover rates at MTJs by determining the mobile fraction (Fig. 7). We found that mutating IBS-1 or IBS-2 conferred opposite effects on the mobile fraction of talin: the GFP-tagged IBS-2 mutant talin KS>DD had a mobile fraction that was 43% lower than that of wild-type talin-GFP (Fig. 7). Conversely, the IBS-1 mutant talin R367A was more dynamic at MTJs, with a mobile fraction ~25% higher than wild-type talin–GFP (Fig. 7). These results suggest that mutating IBS-1 causes talin to undergo higher turnover, whereas mutating IBS-2 leads to lower talin turnover.
The work presented here is the first comprehensive in vivo analysis of both IBSs of talin. Studies of the IBS domains of talin are confounded in vertebrates because there are two talin genes that can act redundantly (Monkley et al., 2000). However, as the Drosophila genome encodes a single talin gene, we were able to overcome this problem. Our results demonstrate that the C-terminal IBS-2 domain of talin is crucial for its function, but that IBS-1 and IBS-2 share some overlapping functions. Specifically, either IBS domain is sufficient to recruit talin to sites of integrin-mediated adhesion. Furthermore, talin mutants bearing mutations in either domain rescued different subsets of talin-dependent processes; when both IBS domains are disrupted, the resulting phenotype is similar to that observed in a talin-null fly. Detailed analyses of IBS function during various developmental processes revealed that each IBS is specifically required for a discrete subset of talin functions: IBS-1 has a more pronounced role in linking integrins to the ECM, whereas IBS-2 is more important for linking integrins to the intracellular IAC. These results provide important insight into how integrin function is mediated during different developmental processes.
This study utilises previously identified mutations in talin to disrupt the integrin-binding capacity of the IBS domains of talin. These experiments rely on two assumptions: first, that the mutations do in fact abrogate integrin binding, and second, that the mutations do not disrupt other protein–protein interaction domains within talin. Several lines of evidence from our previously published results and the work presented here strongly support these assertions. First, when the mutations we used were introduced in the context of the isolated IBS domains (UAS-IBS1–GFP and UAS-GFP–IBS2), they blocked the recruitment to integrins at MTJs. Second, full-length talin transgenes carrying these mutations were less efficiently recruited to MTJs in a wild-type background compared with that of the wild-type transgenes, indicating a reduced ability to compete with endogenous talin for IBSs. Further evidence can be derived from experiments using the truncated βPS-integrin mysG1, which contains only the putative IBS-2 binding site and eliminates the IBS-1 binding site in the distal part of the β-integrin tail. This mutant integrin protein recruited wild-type talin or the IBS-1 mutant talin R367A, but not talin containing mutations in the IBS-2 domain, suggesting specific disruption of IBS-1-dependent talin association with integrin. Altogether, our in vivo analyses of talin R367A, KS>DD and LI>AA are in line with published biochemical studies that have described the ability of these mutations to specifically abrogate integrin binding. Our observations also suggest that the mutations we used did not interfere with other talin functions. For example, we incorporated the GFP reporter at the C-terminal end of the talin molecule and observed normal GFP fluorescence with all IBS domain mutations, indicating proper protein folding downstream of the mutation. Furthermore, full-length talin mutants containing single IBS mutations localised correctly in the absence of endogenous talin and retained functionality, as illustrated by their ability to recruit IAC components and to partially rescue muscle attachment at MTJs. In addition, two different sets of mutations in the IBS-2 domain, including those that affect two surface residues (K2094 and S2098) (Cheung et al., 2009; Moes et al., 2007), gave rise to similar phenotypes. Nonetheless, even though we found no evidence to suggest this, we cannot completely discount the possibility that the mutations we utilised affect other functions of talin other than integrin binding.
Although full-length talin mutant containing a double IBS mutation (talin R367A; LI>AA) was not enriched at MTJs in a wild-type background, in a talin-null background a small amount of the double IBS mutant was unexpectedly detected at MTJs. This observation can be explained by recent biochemical studies that have shown that the N-terminal head domain of talin interacts with negatively charged lipids in the plasma membrane and might therefore be weakly stabilised in the cell membrane independently of the IBS domains (Elliott et al., 2010; Kalli et al., 2010; Anthis et al., 2009; Wegener and Campbell, 2008).
Interestingly, the phenotypes observed when the NPxY motif of βPS integrin, which binds to the talin IBS-1 domain, was disrupted were stronger than those seen with the mutant talin IBS-1 rescue transgene. This can explained by previous studies showing that mutations in the NPxY motif can also disrupt the conformation of the other regions of the integrin cytoplasmic tail, including those that interact with the IBS-2 domain (Rodius et al., 2008). Indeed, the overall phenotype of the N828A mutation fits somewhere in the spectrum between the phenotypes observed with the IBS-1 and the IBS-2 mutant talin transgenes and thus might represent the combined effects of a severe disruption of binding to the IBS-1 domain with a partial disruption of binding to the IBS-2 domain. Nevertheless, the characteristic defect seen in MTJs in embryos rescued with the N828A integrin mutant was highly specific (i.e. detachment from the ECM and not from the IAC) a defect that is unique to flies in which the IBS-1 domain of talin has been mutated.
The functional redundancy of the IBS-1 and IBS-2 domains of talin was principally manifested in two ways: first, when either one of the IBS domains was disrupted, talin was still recruited to integrins; second, with only one functionally active IBS, talin recruited downstream components of the IAC. A third manifestation of this redundancy was that the knockout of either IBS domain gave rise to weaker phenotypes at MTJs than knockout of both domains. The principal implication of these observations is that either IBS domain is capable of mediating the function of talin as an adaptor protein, enabling talin to bind to integrins and to provide linkage to the cytoskeleton. Another potential implication of this finding is that, on many occasions, the precise mechanism by which integrins recruit talin is not important as long as the recruitment occurs. This leads to the question of what mechanism determines whether IBS-1 or IBS-2 binds to integrin. Our data shows that IBS-1 and IBS-2 are recruited by separate mechanisms involving independent interactions with different parts of the integrin cytodomain. For example, the IBS-2 but not the IBS-1 domain mediates talin recruitment in the absence of the entire distal part of the integrin cytoplasmic tail. It is therefore probable that the choice between the two IBS domains of talin is regulated by events that occur at the integrin cytoplasmic tail, perhaps linked to changes in the conformation or activation states of both talin and integrin.
The fact that talin has two different domains that carry out a seemingly identical function, binding to the integrin cytoplasmic tail, is somewhat curious. Previous reports have shown that the IBS-1 domain of talin is essential (Calderwood et al., 1999; Tadokoro et al., 2003; Tanentzapf and Brown, 2006), and our finding that the IBS-2 domain is also required for talin function helps explain why both binding sites have been maintained over hundreds of millions of years of evolution. The essential role of the IBS-1 domain of talin is to regulate inside-out activation of integrins; however, talin has another vital function in integrin-mediated adhesion, which is to link integrins to the IAC. Our data indicate that this function appears to be heavily dependent on the IBS-2 domain. Importantly, ablation of IBS-2 does not prevent the initial recruitment of IAC components but instead disrupts the attachment between integrins and the assembled IAC. These observations are also supported by a recent study that used FRET-FLIM analysis in cultured cells to demonstrate that IBS-2, but not IBS-1, forms a stable complex with integrin (Parsons et al., 2008). This corroborates our finding that IBS-2 is required to stably link talin to integrins, because although IBS-1 efficiently localises talin to adhesion sites, the interaction between IBS-1 and integrin was not sufficient to stabilise talin in the complex.
It is striking that dynamic morphogenetic processes rely so heavily on the IBS-2 but not on the IBS-1 domain of talin. During dynamic adhesive processes, adhesion complexes must be rapidly assembled and disassembled. For such processes it is probable that the conversion of transient integrin-mediated cell–ECM adhesion into strong and stable cell–ECM adhesion, which is the process regulated by IBS-1 (Tanentzapf and Brown, 2006) (this study), is not a crucial factor. Furthermore, our FRAP studies suggest that the binding of integrin to talin through IBS-1 or IBS-2 affects talin turnover in opposite ways: turnover is higher with the IBS-1 mutation and lower with the IBS-2 mutation. We hypothesise that this occurs because the reduced integrin activation that results from mutating IBS-1 leads to a less stable and more dynamic adhesion complex. By contrast, when IBS-2 is mutated, talin can only link to integrin through the IBS-1 domain, leading to increased integrin activation compared to that elicited by wild-type talin. We speculate that increased integrin activation acts to stabilise the integrin adhesion complex. During morphogenesis, it is probable that dynamic adhesion plays a larger role, and, as a result, IBS-2 plays a greater role in linking talin to integrin. We propose a model whereby when the talin IBS-2 domain is mutated, talin links to integrin exclusively through IBS-1, which impinges upon the ability of the adhesion complex to be rapidly turned over and leads to defects in morphogenesis.
Another potential reason for the importance of IBS-2 during morphogenesis is that the cytoskeletal and cellular rearrangements that occur during dynamic morphogenetic processes probably require strong linkage between integrins and the IAC. Our data showing separation between IAC markers and integrin at MTJs upon disruption of the IBS-2 domain suggests that IBS-2 maintains the integrin–IAC connection. It is unclear why IBS-2 is particularly important in maintaining the integrin–IAC link. One possibility is that the interaction between integrins and the talin IBS-2 domain frees IBS-1 from integrin binding so that the talin head can associate with other cytoplasmic components. The talin head domain could play a pivotal role in IAC assembly as it contains a multitude of protein binding domains that interact with a variety of IAC components, including Wech, FAK, Ha-RAS, phosphatidylinositol phosphate kinase (PIPK1g90) and others (reviewed by Critchley, 2009; Zaidel-Bar et al., 2007). Consistent with this assertion, deletion of the talin head domain severely disrupts talin function (G.T. and M.P., unpublished data).
On the basis of our results, we propose that differential binding by each IBS domain of talin to the integrin cytoplasmic tail comprises an important regulatory mechanism in integrin-mediated adhesion. The ability of integrin to differentially interact with two distinct sites in talin may comprise a regulatory switch between two modes of integrin-mediated adhesion (Fig. 8). Our study suggests that IBS-1 function supports stable linkage of integrins to the ECM. By comparison, association of talin with integrin, through IBS-2, stabilised connections to the intracellular IAC. It is probable, under certain circumstances, that either IBS domain is equally capable of mediating the role of talin in integrin mediated-adhesion. However, during crucial periods in animal development, specific functions may be preferentially mediated through integrin binding to either IBS-1 or IBS-2, and it is only then that the choice between the IBS domains becomes important. This differential interaction illustrates one mechanism for how diverse adhesion complex functions can be achieved over the life of an organism.
Materials and Methods
UAS-IBS1-GFP and UAS-GFP-IB1-R367A are as described previously (Tanentzapf and Brown, 2006). To make the UAS-GFP-IBS2, sequence, corresponding to amino acids 1981–2113, IBS-2 was PCR-amplified from genomic rhea (using primers 5′-TATACTGCAGTCGCGTGGAACTCAGGCGTG-3′ and 5′-TATACCGCGGTCACTTGCCGGAGGCCAACTTAG-3′) and cloned using PstI and SacII into the pUAS-mGFP6 plasmid (Franco-Cea et al., 2010). The LI>AA and KS>DD mutant IBS2 domains were made using the QuikChange Lightning mutagenesis kit (Stratagene) and the following mutagenesis primers (only forward is shown, changes indicated in bold): LI>AA, 5′-GTAGCCTCTGCACTTGGTGATGCGGC TAACTGCACTAAGTTGGCCTCC-3′ and KS>DD, 5′-GTCATCAATGCCGTCGAC-GATGTAGCCGATGCACTTGGTGATTTG-3′. Full-length talin–GFP rescue constructs were all based on the pUbi-Talin[EGFP] construct (Yuan et al., 2010), except talin R367A which was adapted from pUbi::Talin(R367A) (Tanentzapf and Brown, 2006) to include a GFP as outlined in (Yuan et al., 2010). The IBS double mutant talin R367A; LI>AA transgene was constructed by sub-cloning a 2207 bp section with KpnI and StuI from the pUbi-talin(R367A) construct into the talin LI>AA transgene. pUbi-βPS-YFP was made as described previously (Yuan et al., 2010). Mutations were introduced using site-directed mutagenesis with the following primers: N828A, 5′-TGGGATACGGGCGAGGCTCCCATCTACAAGCAG-3′; E810Q, 5′-CACGATCGGCGGGCGTTCGCTCGCTTC-3′; E817Q, 5′-CGCTTCGAGAAGCGCGCATGAACGCC-3′. All transgenes were generated by Bestgene (Chino Hills, CA).
Fly stocks and genetics
In order to obtain embryos maternally and zygotically null for talin or βPS, germline clones (GLCs) were generated according to the dominant female sterile technique (Chou and Perrimon, 1996). For rhea GLCs, males of the genotype y,w,hsFlp/Y;;OvoD1,FRT2A/TM3 were crossed to rhea79,FRT2A/TM2 females carrying each of the different talin rescue transgenes so the mutant protein was supplied maternally and zygotically. Progeny of this cross was heat shocked in larval stages for 2 hours at 37°C for 3 consecutive days, and transgene/+;rhea79,FRT2A/OvoD1,FRT2A females were crossed to rhea79/TM3 males. Embryonic progeny were collected for fixation every 24 hours. For mys GLCs the same procedure were followed using the lines OvoD1FRT101/Y;hsFlp38 and mysXG43FRT101/FM7, Kr::GFP. Flies of the genotype OvoD1,FRT101/mysXG43,FRT101 were selected for and then crossed to males carrying each of the βPS rescue transgenes of interest. For analysis of talin localisation in mysG1 background, y,w,mysG1,FRT18A/FM7,Kr::GFP females were crossed to males bearing each talin transgene and embryos lacking the Kr::GFP balancer were selected for analysis.
Immunohistochemistry, confocal microscopy, and image analysis
Embryos were heat-fixed (Tanentzapf and Brown, 2006) for most experiments, with the exception of phalloidin stainings, which required a methanol-free protocol whereby embryos were fixed for 20 minutes in 4% paraformaldehyde, devitellinated in 2:1 heptane and 90% ethanol, and permeabilised in 0.2% Triton X-100. Antibodies used were against the following proteins: MHC (1:200; Dan Kiehart, Duke University, Durham, NC), PINCH (rabbit, 1:1000; Mary Beckerle, Huntsman Cancer Institute, UT), talin (rabbit, 1:500; E4), talin (mouse, 1:50; E17B), αPS2 (rat, 1:100; 7A10), βPS (mouse monoclonal,1:20; Nick Brown, Gurdon Institute, Cambridge, UK), paxillin (rabbit, 1:1000) (Yagi et al., 2001), tiggrin (mouse, 1:500; Liselotte Fessler, UCLA, CA), GFP (rabbit, 1:1000; A6455, Invitrogen). Rhodamine-conjugated phalloidin (Invitrogen) was used to stain actin filaments (1:400). Fluorescently-conjugated Alexa-Fluor-488, Cy3 and Cy5 secondary antibodies were used at 1:400 dilution (Molecular Probes).
Images were collected using an Olympus FV1000 confocal microscope with an UplanSApo 60× 1.35 NA oil objective and a UplanFL N 40× 1.30 NA oil objective and processed using Adobe Photoshop. All quantification of fluorescence at MTJs was performed using ImageJ (NIH, Bethesda, MD). Statistically significant differences in localisation between wild-type and mutant constructs was assessed by the two-tailed Student's t-tests in all cases, except when we sought to compare between multiple constructs, where one-way ANOVA was used. Statistical analysis was carried out using Prism4 software (GraphPad, La Jolla, CA). For intensity traces across MTJs, the ImageJ plot profile tool was used to determine the average signal intensity across the boxed area indicated on the images. Each channel was independently normalised from unprocessed grey-scale images so that the peak intensity of each channel across the area of interest was set as 100% and the lowest intensity was set to be 0%. In the cases where one of the two signals was mislocalised (as in Fig. 4B‴), 100% was set as the maximum intensity of the localised channel for both channels.
To determine colocalisation, we used the colocalisation feature in Olympus FluoView1000 software to determine the Pearson Correlation Coefficient. A region of interest was drawn around single 1.0 μm section images in the middle plane of each MTJ. Images were taken so that the signal of each marker being measured was set to just below saturation at the MTJ prior to analysis. At least three MTJs in three different embryos were selected for each set of measurements.
Western blot analysis
For the western blot shown in supplementary material Fig. S2, protein was isolated from whole flies and blotted according to standard procedures using primary antibodies against GFP (1:1500; A6455, Invitrogen) and actin (8224, 1:5000, Abcam) and secondary antibodies conjugated to InfraRed (IR) dyes, IR680 and IR800 (LiCor Odyssey, 1:15,000). Blots were scanned on a LiCor Odyssey Imaging System.
For the quantitative real-time PCR (qPCR) data shown in supplementary material Fig. S2C, total RNA was isolated from whole flies using TRIzol (Invitrogen) and treated with DNase (Fermentas). A total of 1000 mg total RNA was converted into cDNA using the qScript cDNA Synthesis Kit (Quanta Biosciences). Subsequently, qPCR was performed using the PerfeCTa SYBR Green FastMI ROX kit (Quanta Biosciences). As all βPS transgenes were YFP-tagged, we could to quantify transgenic mRNA transcript levels by assaying YFP expression using the primer pair: 5′-GGCACAAGCTGGAGTACAAC-3′ and 5′-AGCTCAGGTAGTGGTTGTC-3′. GAPDH mRNA levels were assayed as an internal control using the primer pair: 5′-AAAGCGGCAGTCGTAATAGC-3′ and 5′-GACATCGATGAAGGGATCGT-3′. Expression changes were determined by using the comparative Ct method for relative quantitation.
Stage 17 embryos were collected and prepared for FRAP as described previously (Yuan et al., 2010). Briefly, embryos heterozygous for each transgene were collected from apple juice plates, dechorinated in 50% bleach for 3 minutes, washed with PBS and mounted onto glass slides in PBS. FRAP analysis was performed at room temperature. Photo-bleaching was performed using a 405 nm laser at 30% power with the Tornado scanning tool (Olympus) for 2 seconds at 100 mseconds per pixel. Fluorescence recovery was recorded over 5 minutes at 1 frame every 4 seconds. To control muscle twitching in and out of focus, multiple regions of interest (ROIs) were selected in non-photobleached regions; only samples for which intensities within control ROIs remained steady throughout the FRAP experiment were used. Recovery data were further analysed using Prism software (GraphPad); the mobile fraction was calculated as previously described (Reits and Neefjes, 2001). Statistical tests (Student's t-tests or ANOVA tests) were performed using Prism software.
We thank Dan Kiehart, Mary Beckerle and Lisa Fessler for antibodies, and Tom Bunch and the Bloomington Stock Center for fly lines. This work was funded through the Canadian Institutes of Health Research, operating grant 89835. S.J.E. and M.J.F. hold NSERC Alexander Graham Bell Canada Graduate Scholarships. G.T. is a CIHR New Investigator and Michael Smith Foundation for Health Research Scholar. Research in the laboratory of G.T. is supported by an HFSP Career Development Award.