The development of three-dimensional tissue architecture requires precise control over the attachment of cells to the extracellular matrix (ECM). Integrins, the main ECM-binding receptors in animals, are regulated in multiple ways to modulate cell–ECM adhesion. One example is the conformational activation of integrins by extracellular signals (‘outside-in activation’) or by intracellular signals (‘inside-out activation’), whereas another is the modulation of integrin turnover. We demonstrate that outside-in activation regulates integrin turnover to stabilize tissue architecture in vivo. Treating Drosophila embryos with Mg2+ and Mn2+, known to induce outside-in activation, resulted in decreased integrin turnover. Mathematical modeling combined with mutational analysis provides mechanistic insight into the stabilization of integrins at the membrane. We show that as tissues mature, outside-in activation is crucial for regulating the stabilization of integrin-mediated adhesions. This data identifies a new in vivo role for outside-in activation and sheds light on the key transition between tissue morphogenesis and maintenance.
During animal development integrins play essential roles in numerous contexts by mediating diverse cellular processes (Bokel and Brown, 2002; Brown et al., 2000; Thiery, 2003). Integrin-based adhesions that form during development can be transient, especially during cell migration, or very long lasting, such as when tissues reach terminal differentiation (Bulgakova et al., 2012; Yuan et al., 2010). The strength and duration of integrin-based cell–extracellular-matrix (ECM) adhesion is thus carefully regulated depending on the type of adhesion the animal requires in a given tissue at a specific time. Drosophila has proven to be an excellent system in which to analyze integrin regulation during development (Brown et al., 2000). Integrin-based adhesions in flies are generally very well conserved to those in vertebrates at the molecular level but the simpler genome of the fly ensures reduced redundancy and simplifies genetic analysis. Genetic screens in the fly have identified many essential components of the integrin adhesion complex that are structurally and functionally similar to their vertebrate counterparts (Brown et al., 2000; Gotwals et al., 1994). Fly muscles have proven to be a particularly useful model to study integrin function during tissue development and maintenance as they contain prominent integrin-based adhesions at myotendinous junctions (MTJs). MTJs are specialized sites at the surface of muscles where force is transmitted to epidermal tendon cells and their proximity to the surface of the embryo makes them very suitable for in vivo imaging and treatment with chemical reagents (Schweitzer et al., 2010; Yuan et al., 2010).
In animals there are multiple strategies to control the strength and duration of integrin-mediated attachments to the ECM. The two main ways to control the strength of integrin adhesion to the ECM are through receptor clustering and conformational changes. In the process of integrin clustering, integrins that are diffusely distributed throughout the membrane aggregate in particular regions, where they form large complexes with attachments to the cytoskeleton (Margadant et al., 2011; Stewart and Hogg, 1996). Regulation by conformational changes in integrin, also known as integrin activation, involves reorganization of the integrin extracellular domain such that its binding affinity for ECM ligands is substantially increased (Campbell and Humphries, 2011; Shattil et al., 2010). Integrin activation can be induced by interactions between the integrin cytoplasmic domain and cytoplasmic factors, particularly Talin, (inside-out activation) or by interactions with extracellular ligands (outside-in activation) (Liddington and Ginsberg, 2002).
A great deal has been learned about the mechanism of activation through detailed structural, biophysical and molecular studies (Campbell and Humphries, 2011; Shattil et al., 2010). A comprehensive set of structural biology approaches using multiple integrin isoforms has shown that the extracellular domain of the integrin heterodimer is composed of the ‘head’ domain, containing the ECM-ligand-binding pocket, resting on two long ‘legs’ with flexible ‘knees’. This arrangement allows integrin to exist in multiple conformations with differing affinities for its ECM ligands. In its lowest affinity state, the ‘knees’ are bent and the head region is closed, whereas in its highest affinity state, the entire extracellular domain is extended and the head is open; many intermediate states also exist (Campbell and Humphries, 2011; Ye et al., 2012; Zhang and Chen, 2012). Integrin activation is proposed to involve the switchblade-like extension of the extracellular homodimer that extends the ligand-binding head region away from the plasma membrane (Shattil et al., 2010; Takagi et al., 2002). The divalent cations Mn2+ and Mg2+ are known to be potent activators of integrin, and induce an increase in affinity for ECM ligands across multiple integrin heterodimers (Ye et al., 2012). It is known that for at least certain integrin heterodimers, divalent cations mimic ligand binding and induce activation by inducing wholesale conformational change (Takagi et al., 2002; Ye et al., 2012). Mn2+ is particularly effective at activating integrins, whereas, at least in culture, Mg2+ has been suggested to activate integrin only at very low Ca2+ concentrations (Dransfield et al., 1992). Intriguing evidence from Drosophila, using a sensor for integrin activation, has suggested that Mn2+ might be able to activate integrins in vivo, in larval imaginal discs (Helsten et al., 2008). Moreover, reverse and forward genetic approaches in the fly have identified point mutations in Talin and the β-integrin subunit (known as βPS integrin) that have been proposed to modulate activation (Ellis et al., 2014, 2011; Jannuzi et al., 2002, 2004; Pines et al., 2012, 2011; Tanentzapf and Brown, 2006).
Another important strategy for regulating cell–ECM adhesions is by modulating their turnover. Integrin-based adhesions undergo dynamic remodeling through endocytic and exocytic trafficking of integrin receptors (De Franceschi et al., 2015; Paul et al., 2015; Valdembri and Serini, 2012). The general model of integrin turnover involves constitutive clathrin- or caveolin-mediated endocytosis of surface integrins and their subsequent endosomal sorting back to the membrane or to the degradation machinery (Bridgewater et al., 2012; Paul et al., 2015). This canonical process provides many opportunities for regulating the stability of integrin-based adhesion by controlling the transport, removal and sorting of integrins to and from the membrane (Bridgewater et al., 2012; Paul et al., 2015). Diverse cell behaviors including cell migration, invasion and cytokinesis are regulated by turnover either by direct changes in cell–substrate adhesion or by altering downstream cell signaling pathways (Bridgewater et al., 2012; De Franceschi et al., 2015; Paul et al., 2015). The Drosophila MTJs serve as a powerful model to study the turnover of integrins and their adhesion complex in the context of living intact animals. The turnover of wild-type and point-mutated tagged components of the integrin adhesion complex is analyzed at the MTJ using fluorescence recovery after photobleaching (FRAP) and mathematical modeling (Hakonardottir et al., 2015; Pines et al., 2012; Yuan et al., 2010). Previous work has established that, in fly MTJs, FRAP recovery is mostly due to endocytosis and exocytosis of integrin rather than their lateral diffusion (Yuan et al., 2010). Furthermore, clathrin-mediated endocytosis is essential for turnover and consequently for maintaining adhesions at the MTJs (Yuan et al., 2010). Experiments using inducible mutations that either increased or decreased the mechanical force applied on MTJs have shown that force is a potent regulator of integrin adhesion complex turnover (Hakonardottir et al., 2015; Pines et al., 2012). Increased force results in decreased endocytosis of integrins and an increase in the rate of assembly of new adhesions, which together combine to stabilize integrin-based adhesions (Hakonardottir et al., 2015; Pines et al., 2012). Furthermore, analysis of point mutations in integrin or Talin that impinged on activation hinted that outside-in and inside-out activation may be involved in the transmission of mechanical cues that regulate turnover (Hakonardottir et al., 2015; Pines et al., 2012). However, we currently have only limited understanding of whether and how activation regulates integrin-mediated cell–ECM adhesion, in particular its turnover in vivo in the context of the whole organism.
Here, we investigated whether the activation state of integrins affects the stability and turnover of integrin-based cell–ECM adhesion in live intact Drosophila embryos and larva. We developed an approach combining permeabilization of embryos to allow the introduction of divalent cations followed by FRAP and mathematical modeling to analyze integrin turnover. We show that outside-in integrin activation using divalent cations leads to stabilization of integrin-mediated adhesions by modulating the rate of endocytosis and exocytosis of integrins from the membrane. This conclusion was bolstered by a genetic approach, using targeted point mutations in the βPS integrin protein to either induce or block its activation. Reducing the amount of ECM ligands had the opposite effect, elevating the rate of integrin endocytosis as well as overall turnover. Moreover, we identified a role for outside-in integrin activation in stabilizing integrin-based adhesions at the MTJs over the course of development. In particular, mutations that blocked the regulation of outside-in integrin activation were not able to stabilize integrins at MTJs during development. Finally, we show that Rap1 signaling is essential for the transmission of outside-in signals that stabilize integrins at the membrane.
Outside-in integrin activation using divalent cations stabilizes cell–ECM adhesion by decreasing integrin turnover
We previously reported our adaptation of a published permeabilization protocol that allows small molecules to be introduced to intact fly embryos without compromising the vitelline membrane (Hakonardottir et al., 2015; Schulman et al., 2013). We used this protocol to introduce Mn2+ and Mg2+ cations to fly embryos (stage 16) expressing a tagged genomic integrin rescue construct; turnover of integrins at MTJs was then analyzed using FRAP (Yuan et al., 2010). To obtain greater mechanistic insight into integrin turnover following cation-induced outside-in integrin activation, we employed a mathematical model of integrin turnover at MTJs that was previously implemented and validated (Pines et al., 2012). This model allows us to determine the reaction rates for endocytosis (kendo) and exocytosis (kexo) of integrins. By calculating the ratio of kendo to kexo (kendo/kexo) we are able to see shifts that favor increased turnover, build-up or breakdown of the adhesion complex.
To control for possible effects on turnover due to changes in osmotic conditions in the embryo, we treated embryos with NaCl. The addition of Na+ did not substantially affect integrin turnover; mobile fraction and the rate of integrin endocytosis, kendo, did not differ statistically from untreated controls (Fig. 1A,D). Although a very small increase in the rate of exocytosis kexo was observed the overall kendo/kexo remained for the most part unchanged (Fig. 1E). In comparison, the introduction of Mg2+ or Mn2+ cations achieved by the addition of MnCl2 or MgCl2 significantly affected turnover. In both cases the mobile fraction was substantially lower than untreated controls (Fig. 1B,C). Moreover, kendo was lower after treatment with either Mg2+ or Mn2+; and in the case of Mn2+ kexo was also slightly higher (Fig. 1B–D). The overall result of these changes in rate constants was that the kendo/kexo was nearly 34% lower than that in controls (from ∼0.32 in buffer and Na+ to ∼0.21 in Mn2+ and Mn2+ conditions) (Fig. 1E). This change would favor the accumulation and stabilization of integrins at the membrane and thus likely explain the lower mobile fraction. These results suggest that activation of integrin through outside-in signaling stabilizes integrin at the membrane largely by reducing their removal from the membrane by endocytosis.
Point mutations in integrin that induce activation affect turnover similarly to treatment with divalent cations
If outside-in activation stabilizes integrin at the membrane, we would predict that mutations in integrins that activate outside-in signaling would phenocopy treatment with Mg2+ or Mn2+. We previously described the production and characterization of such mutations (Pines et al., 2011). We chose two point mutations that would be expected to induce conformational changes consistent with integrin activation. These point mutations were made as fluorescently tagged rescue constructs and were shown to efficiently rescue loss-of-function mutations in the fly βPS integrin (Pines et al., 2011). Importantly these rescue constructs express at levels that are similar to the endogenous wild-type integrin (Pines et al., 2011). FRAP analysis was performed in stage 16 embryos. The first mutation, L211I, is a mutation in the I-like domain (L138 in β3 integrin; Luo et al., 2009), which promotes vertebrate β3 activation by stabilizing the extended extracellular domain conformation (Luo et al., 2009). The second mutation G792N (G708 in β3 integrin; Li et al., 2003), disrupts the packing of the transmembrane helices of α and β integrin to induce activation. Our previous analysis has shown that integrin rescue transgenes containing the L211I and G792N mutations rescued the tissue development and maintenance defects of integrin-null mutants as efficiently as a wild-type integrin rescue transgene (Pines et al., 2011). However, both the L211I and G792N integrin mutants showed significant changes in turnover; in both cases the mobile fraction was substantially lower than wild-type integrin controls (Fig. 2A,B). Moreover, kendo was lower for both mutants whereas kexo was higher (Fig. 2A″,B″,C). The overall result of these changes in rate constants was that the kendo/kexo was nearly 70% and 30% lower than that of wild-type controls for L211I and G792N, respectively (Fig. 2D). As in the case of Mn2+ and Mg2+ treatment, this change favors the accumulation and stabilization of integrins at the membrane and likely explains the lower mobile fraction. Taken together, these data link outside-in activation to reduced integrin endocytosis and increased integrin exocytosis, which together act to stabilize integrins at the membrane and reduce their mobile fraction.
Mutations that block or strongly induce outside-in integrin activation are insensitive to treatment with divalent cations
We hypothesized that mutations that block or strongly induce outside-in integrin activation should be less sensitive to treatment with Mn2+. FRAP analysis on stage 16 embryos revealed that the mobile fraction of the activated integrin mutant L211I remained largely unchanged after exposure to divalent cations (Fig. 3A). Treatment with Mn2+ caused no significant change in the rate of integrin exocytosis kexo but a slight change in the rate of endocytosis kendo, though this effect was smaller than that observed for the wild-type integrin (Fig. 3A,F). The mobile fraction of the activated integrin mutant G792N also remained unchanged after exposure to divalent cations although both reaction rates, kendo and kexo, decreased (Fig. 3B,F). Importantly, and in striking contrast to the wild-type integrin, Mn2+ treatment did not give dramatic change in the ratio of kendo/kexo in either activated integrin mutants (Fig. 3G). Next, mutations in the βPS integrin protein that were predicted to block outside-in activation were tested. Three such mutations were chosen, two in the cytoplasmic domain and one in the extracellular domain: a mutation in the proximal, integrin-binding NPxY motif (N828A; Calderwood et al., 2002; Ellis et al., 2011; Tadokoro et al., 2003; Wegener et al., 2007), a mutation in distal, Kindlin-binding NPxY motif (N840A; Harburger et al., 2009; Moser et al., 2008; Pines et al., 2011), and a mutation in the ligand-binding pocket (S196F; Pines et al., 2011). The S196F mutation (equivalent to S123F in β3 integrin or S134F in β1 integrin) is expected to bind to a subset of ECM molecules but prevent the resulting conformational changes essential to transmit outside-in signals (Bajt and Loftus, 1994; Chen et al., 2006a,,b; Hogg et al., 1999; Pines et al., 2011). Previous analysis of these activation-blocking mutations has shown that the integrin containing the S196F mutation effectively behaved like a null allele. In comparison, integrin containing the N828A mutation gave rise to severe defects but was slightly weaker than a null whereas the N840A mutation gave rise to a strong-to-intermediate phenotype. In addition, all three activation-blocking mutations have been shown to be unable to regulate turnover in response to modulation of the mechanical force placed on MTJs (Pines et al., 2012). For all three mutations, Mn2+ treatment did not affect the mobile fraction or kendo, and kexo was unchanged in either N828A or N840A (Fig. 3C,D,F). The S196F mutant did exhibit a very slight increase in kexo but this did not result in a notable change in the ratio of kendo/kexo, which was also the case for the N828A or N840A mutants (Fig. 3E–G). These results support the notion that divalent cations affect turnover by inducing outside-in signaling in vivo in Drosophila. Exposure to divalent cations has little effect on activating mutations as they are already mostly active. Furthermore, cation treatment fails to activate mutations that cannot transduce outside-in activation. Furthermore, this data illustrates the effectiveness and specificity of the integrin mutations used in this study.
Reducing the availability of ECM ligands increases integrin turnover
Given that increasing outside-in integrin activation lowers turnover, it should follow that reducing outside-in activation would have the opposite effect. One way to accomplish this is by lowering the availability of ECM ligands as these induce outside-in activation. To test this hypothesis, we reduced the availability of the ECM ligands Laminin, specifically the β-Laminin subunit encoded by the gene LanB1, and Collagen IV, encoded by the gene viking (vkg) (Broadie et al., 2011; Brown, 2011; Urbano et al., 2009) in stage 17 embryos. To reduce ECM ligands, flies heterozygous for alleles of vkgK00236 and landef were combined with integrin–YFP in order to carry out FRAP experiments. This analysis showed that the mobile fraction of integrin was substantially higher upon reduction of ECM ligands (Fig. 4A,B). Importantly, mathematical modeling revealed that this higher mobile fraction was largely the result of an increase in the rate of integrin endocytosis, kendo (Fig. 4A″,B″,C). As a result, there was an increase in kendo/kexo, consistent with more turnover of integrins at the membrane (Fig. 4D). Overall, these data suggest that ECM ligands induce outside-in activation in embryonic fly MTJs and that this is important to stabilize integrin-based adhesions.
Integrin turnover is developmentally regulated by integrin activation
A central question in developmental biology is how the dynamic adhesion characteristic of tissue morphogenesis is replaced by the stable long-lasting adhesion required to maintain tissues over the life of the organism. Over the course of development the mobile fraction of integrins decreases (Fig. 5A). Specifically, there is a large drop in turnover shortly after the embryos transition to larval stages (Figs 5A and 6A; Pines et al., 2012; Yuan et al., 2010). To understand how this reduction in the mobile fraction is achieved, integrin turnover was studied using FRAP in stage 16 and 17 embryos as well as third-instar larva. Stage 16 occurs towards the end of embryogenesis at a time when major morphogenetic processes are undergoing completion. In comparison, stage 17 is the final stage of embryogenesis and represents the transition from embryo to larva. The third-larval instar occurs 72 h after hatching and is the culmination and final stage of larval life before metamorphosis. Importantly, there is gradual increase in the mechanical force applied to the MTJs between embryonic stages 16 to 17 as well as between the embryonic and larval stages as muscles grow and muscle contractions increase in frequency and strength. The mathematical model for integrin turnover was used to analyze turnover at each of the three stages. This analysis revealed that stabilization of integrins during development occurs through two mechanisms, a large reduction in the rate of integrin endocytosis, kendo, occurs between stages 16 and 17 (Figs 5A″ and 6A″). Moreover, there is an increase in the rate of integrin exocytosis, kexo, between every developmental stage (Figs 5A″ and 6A″). As a result of these changes the kendo/kexo is decreased by over 57% between stages 16 and 17, which gives rise to more stable adhesion.
We hypothesized that integrin activation is involved in stabilizing integrin at MTJs during the transition from embryo to larva, as they experience greater mechanical force. To test this hypothesis the turnover of the activated integrin mutants L211I and G792N was studied using FRAP in stages 16 and 17 embryos as well as third-instar larva. The L211I mutation failed to show the characteristic drop in mobile fraction between stages 16 and 17 that is indicative of stabilization of integrin at the MTJs and the corresponding decrease in kendo and increase in kexo also did not take place (Fig. 5B,D). Moreover, the G792N mutation gave a noticeable, but somewhat weaker, effect (Fig. 5C,D). The drop in the mobile fraction between stages 16 and 17 was smaller than that observed for wild-type integrin, and the changes in kendo and kexo over development were less pronounced. As a result the kendo/kexo stayed constant between stages 16 and 17 for L211I and decreased less dramatically than wild-type for G792N (Fig. 5E). Similarly, we found that mutations that rendered integrin unable to transmit outside-in activation also failed to stabilize integrin during development. All three activation-defective mutations were tested: N828A and S196F showed no changes in the mobile fraction during the transition from embryo to larva, whereas N840A showed a smaller reduction compared to wild-type (Fig. 6B–E). Similarly, N828A, and S196F showed abnormal changes in kendo or kexo during development when compared to wild-type (Fig. 6C–E). As a result, the kendo/kexo was not substantially lower between stages 16 and 17 for N828A, N840A and S196F accounting for the failure to stabilize integrin at the MTJs (Fig. 6F). Taken together, this analysis of activating and activation-defective integrin mutants at different stages suggests that modulation of activation is important in regulating turnover during animal development.
Rap1 regulates integrin turnover downstream of outside-in activation
Next, we sought to determine how outside-in signaling was transmitted inside the cell to induce changes in integrin turnover. One important signaling molecule that regulates cell–ECM adhesion and is known to act downstream of outside-in activation is the small GTPase Rap1 (Franke et al., 2000; Uemura and Griffin, 1999). To determine whether Rap1 was involved in transducing outside-in signaling in the MTJ, a dominant-negative mutant of Rap1 (Rap1-DN) was co-expressed with fluorescently tagged integrin in fly muscles, which were then treated with Mn2+. First, we determined the effects of expression of the dominant-negative Rap1 transgene on its own in MTJs. It did not alter integrin turnover in stage 16 embryos (Fig. 7A). However, the characteristic changes in integrin turnover induced by Mn2+ treatment were blocked upon expression of the dominant-negative Rap1 construct. Specifically, we did not observe either a drop in mobile fraction after Mn2+ treatment or a corresponding decrease in kendo and increase in kexo (Fig. 7A,B). Furthermore, the kendo/kexo was not substantially lower after Mn2+ treatment in flies expressing Rap1-DN (Fig. 7C). Taken together this data suggests that Rap1 signaling is important for regulating turnover downstream of outside-in activation.
In this work, we have employed chemical and genetic approaches to alter the activation state of integrin and subsequently applied quantitative imaging techniques to analyze integrin turnover in an intact living organism. Specifically, we analyzed embryos where outside-in integrin activation was increased or decreased by using treatment with divalent cations, the introduction of point mutations in βPS integrin or by reducing the levels of ECM ligands. Our analysis identifies a role for outside-in signaling in regulating integrin turnover. Moreover, our experiments identify an important physiological role for the regulation of integrin turnover by outside-in activation. Specifically, that it is required for stabilizing cell–ECM adhesion during development as animals transition from morphogenesis to tissue consolidation and long-term maintenance. Outside of hematopoietic cells there are currently only few examples of a role for integrin activation in regulating cell–ECM adhesion, and even fewer studies that provide mechanistic insight into this process (see, for example, Theodosiou et al., 2016). Our findings constitute another such example in an important in vivo context, that of animal development.
Links between activation and integrin trafficking have been previously investigated by a number of groups. Tiwari et al. have shown that Ca2+ binds to newly synthesized integrin as they are trafficked to the membrane in order to prevent their ectopic activation and signaling (Tiwari et al., 2011). Once at the membrane, Mg2+ and Mn2+ displace Ca2+ to facilitate activation (Tiwari et al., 2011). Valdembri et al. investigated the endocytosis of ECM-ligand-bound integrins and found that they are trafficked through the endosomal machinery differently from non-ligand-bound integrins (Valdembri et al., 2009; Valdembri and Serini, 2012). Sandri et al. have shown that a complex composed of R-Ras, RIN2 and Rab5 specifically controls the recycling of active β1 integrins (Sandri et al., 2012). Of particular interest for our study is the work of Arjonen and colleagues (2012). This work used conformation-specific antibodies and a quenching-based assay to visualize and compare the trafficking of active and inactive β1 integrin through the endocytoic machinery. Of note, the study showed that despite initial overlap in the early endosome between the localization of active and inactive integrin, they eventually sorted into different compartments. Specifically, inactive integrins were sorted to a faster recycling compartment (Arjonen et al., 2012). These studies lend support to, and provide mechanistic insight into, our observation that integrin activation can modulate the stability of the adhesion complex.
Our evidence suggests that regulation by outside-in activation might also involve increase in the rate of integrin exocytosis, or kexo. This raises the possibility that the activation state of integrin regulates its delivery to the membrane. Work on sorting nexin 17 (Snx17) has suggested that inactive or non-ligand-bound integrins are, at least initially, targeted for degradation (Bottcher et al., 2012). When Snx17 binds to the integrin cytoplasmic tail, it retrieves ligand-free and/or inactive β1 integrins from compartments targeted for lysosomal degradation and instead re-routes them from early/recycling endosomes to the plasma membrane. Less is known about why activating integrin promotes exocytosis, but this might be a result of signaling downstream of integrin. This idea is supported by our observations showing that a disruption of Rap1 signaling downstream of outside-in activation blocks changes in turnover following treatment of embryos with Mn2+. Importantly, it has been previously shown that Rap1 regulates αLβ2 integrin trafficking to the membrane (Medeiros et al., 2005).
Our results hint an intriguing link between integrin turnover, outside-in signaling, and mechanical force. Our previous work has shown that increasing mechanical force regulates integrin turnover by lowering the rate of endocytosis thus reducing the mobile fraction (Pines et al., 2012). Here, we report that outside-in activation also lowers the mobile fraction and that this is at least in part due to a reduction in kendo. This is particularly striking when outside-in activation was induced by treatment with Mg2+ as this lowered mobile fraction solely through a reduction in kendo. Links between outside-in activation and mechanical force are well documented (Friedland et al., 2009; Kong et al., 2013; Wang and Ha, 2013). Moreover, we have already shown that mutations in integrin that are defective in ligand binding or in outside-in signaling cannot modulate turnover in response to force (Pines et al., 2012). It is therefore likely that the stabilization of integrin in response to force is mediated, at least in part, by activation of outside-in signaling. However, unlike mechanical force, outside-in activation also regulates integrin turnover through changes in the kexo indicating that these two regulatory mechanisms diverge at certain points. Moreover, the observation that activating mutations in integrin affect adhesion stabilization during development but do not give rise to dramatic phenotypes suggest the existence of diverse redundant mechanisms that ensure tissue maintenance under conditions of increased force.
To date there have been a number of studies looking at integrin activation in flies, and from these an overall picture of this process emerges. Initial results using mutations that disrupted the binding of the Talin head domain to integrin, a key step in integrin activation, suggested that this mode of activation might be important in flies (Tanentzapf and Brown, 2006). Subsequent analysis by Helsten et al. provided convincing evidence that inside-out activation in flies and vertebrates might work through divergent mechanisms (Helsten et al., 2008). Specifically, whereas divalent cations were able to induce changes in integrin affinity to its ligands, Talin-mediated inside-out signaling failed to do so (Helsten et al., 2008). Further work has shown that the introduction of a point mutation in Talin that specifically blocked the ability of its head domain to induce inside-out activation but did not impinge on its ability to bind to the integrin cytoplasmic tail did not give rise to a detectable phenotype (Ellis et al., 2014). Taken together, these studies point to a model whereby outside-in integrin activation but not Talin-mediated inside-out integrin activation is present in flies. Our current study expands on this analysis by identifying a specific role for outside-in activation in flies in regulating turnover and suggests that this function is particularly important during the stabilization of tissue architecture that occurs at the conclusion of embryogenesis.
Overall, our data identifies a clear role for outside-in integrin activation in consolidation of tissue architecture in the fly MTJs. Based on the data shown here, we propose the following model: during the late stages of fly development, once the MTJs form, they are exposed to increasingly more powerful muscle contractions (Crisp et al., 2008). This applies more mechanical force on ligand-bound integrins at the plasma membrane and results in conformational changes in the integrin that trigger an increase in outside-in signaling. As a result of outside-in signaling, Rap1 is activated, which can increase integrin trafficking to the membrane and leads to a higher rate of integrin exocytosis (Caswell and Norman, 2006; Medeiros et al., 2005). The increased number of activated integrins also slows down integrin endocytosis, perhaps because active integrins are sorted to the slower cycling endosomal compartments and the rate of integrin endocytosis decreases (Arjonen et al., 2012). As a result of these changes in the rates of integrin endocytosis and exocytosis, they become stabilized at the membrane and this is manifested through a lower mobile fraction. This is consistent with our previous finding showing that towards the end of fly embryogenesis, when the MTJs are exposed to increasing mechanical force and require integrin activation (Tanentzapf and Brown, 2006), Rap1 actively stabilizes the adhesion complex (Ellis et al., 2013). Taken together, this mechanism provides a framework for translating environmental cues such as the availability of ECM ligands or the mechanical forces present in a developing tissue into changes in the stability of integrin-based cell–ECM adhesion.
MATERIALS AND METHODS
For FRAP analysis, we visualized integrin using flies that had a single copy of the wild-type or point-mutated pUBI-βPS-integrin-YFP transgenes in a background that was heterozygous for the integrin-β-subunit-null mutant mysXG43, to ensure the overall integrin level was similar to wild type (Pines et al., 2011; Yuan et al., 2010). The βPS integrin mutants used were as follows: L211I (Pines et al., 2011), G792N (Pines et al., 2011), N840A (Pines et al., 2011), N828A (Pines et al., 2012) and S196F (Pines et al., 2011).
To combine ECM mutants with Integrin–YFP we introduced a copy of the pUBI-βPS-integrin-YFP transgene into animals heterozygous for the mutants landef (Urbano et al., 2009) and vkgK00236 (Spradling et al., 1999). The version of dominant-negative Rap1 used was Rap1-S17A (Ellis et al., 2014) driven by the muscle driver mef2-GAL4 and combined with one copy of the pUBI-βPS-integrin-YFP transgene.
FRAP analysis was conducted in embryos (stage 16 and 17) and larvae (third instar). To allow proper imaging, embryos were treated with 50% bleach for 4 min to dissolve the chorion, which interferes with imaging, and then washed with water. Larvae required only a rinse to remove debris. Samples were then mounted in phosphate-buffered saline (PBS). FRAP is performed after a waiting period (1–1.5 h) in a diverse population of MTJs as described in Pines et al. (2012).
Images were collected using an inverted confocal microscope (Olympus Fluoview, FV1000) with an UplanSApo 60×/1.35 NA oil objective (Olympus, Tokyo, Japan). Fluorescence intensity was measured for 825 frames, every 0.4 s, and 75 frames were taken prior to the bleaching event. Bleaching was performed using the Tornado scanning tool (Olympus) with a 473-nm laser at 5% power during 2 s at 100 μs/pixel. To control for drift of embryos, 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. The mobile fraction and statistical tests were performed using Prism 5 and MATLAB software.
Integrin activation assay
We used the protocol described in Schulman et al. (2013) in stage 16 embryos to deliver in vivo MnCl2 (20 mM) and MgCl2 (60 mM) to ectopically activate integrins. To control for osmolarity, we delivered NaCl (60 mM). After the treatment, embryos were mounted as previously described in halocarbon oil rather than PBS to prevent desiccation. FRAP was then performed as previously described, without the waiting period.
FRAP recovery curves were fitted to a two-compartment mathematical model similar to the model described in previous work (Pines et al., 2012). In this version of the model, integrins exist in two distinct compartments – at the membrane (with concentration M) and within the cell (vesicular concentration V). Integrins in the membrane are assumed to undergo internalization (through endocytosis) at a fixed-rate kendo and vesicles are assumed to be externalized (through exocytosis) at a fixed-rate kexo. We neglect diffusive transport in the membrane because it was previously shown to be unimportant on the timescale of the experiments we fit here (∼300 s) (Yuan et al., 2010). We applied the same protocol to obtain the FRAP recovery curves as in Hakonardottir et al. (2015) to improve the data collection and the robustness of model parameter fitting. We add a correction for the background photobleaching of fluorescently labeled proteins in the membrane compartment, with fixed-rate parameter δ.
Parameter estimation and bootstrap confidence intervals
During the imaging acquisition, 75 frames were taken prior to the bleaching event and saved as a reference set to assist in estimating the background photobleaching rate. To find the best estimates of the parameters, we minimized the sum of square residuals (SSR) between the analytic solution for m and the observed fluorescence data. In order to achieve robust and consistent fits across the data, we found it was advantageous to weight the residuals for the reference region with weight of a tenth relative to the recovery region. To do the fitting we used a nonlinear least-square (NLLS) minimization algorithm using custom code written in Python as previously described (Pines et al., 2012; available on request from Daniel Coombs). To find the confidence intervals on the parameter values, we synthetically created new samples using the bootstrapping technique previously described (Pines et al., 2012).
The exponential model was fitted to each replicate individually. The mobile fraction for the complete sample was then computed as the average of the obtained estimates for fmax.
We thank Dr Martín-Bermudo [Centro Andaluz de Biología del Desarrollo, (CSIC)-Universidad Pablo de Olavide, Sevilla, Spain] for providing us with numerous fly lines.
G.T. conceived, designed, supervised and participated in data analysis for this project. P.L.-C. carried genetics, FRAP experiments and data analysis. A.D.H.-R. and D.C. performed the mathematical modeling and data analysis. G.T. and P.L.-C. wrote the manuscript.
We were supported by the Canadian Institutes of Health Research (CIHR) [grant number MOP-285391]; and Natural Sciences and Engineering Research Council of Canada (NSERC) [grant number 356502, RGPIN-2015-04611]. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.
The authors declare no competing or financial interests.