Slik phosphorylation of Talin T152 is crucial for proper Talin recruitment and maintenance of muscle attachment in Drosophila

Cell-matrix adhesion is generally mediated by integrins, transmembrane receptors connecting the extracellular matrix to the actin cytoskeleton with the help of intracellular linker proteins (Sun et al., 2019). First among these is talin, which is crucial for all integrin adhesive functions in metazoans (Goult et al., 2018; Klapholz and Brown, 2017). Talin is a large 270 kD adaptor protein, consisting of a 400 amino acid N-terminal head domain connected by an unstructured linker region to a long rod domain. The talin head comprises an atypical FERM (band 4.1, ezrin, radixin, moesin) domain, which contains, in addition to the F1, F2 and F3 domains, an F0 domain and an unstructured loop in the F1 domain. The F3 domain contains the major integrin binding site. The C-terminal rod contains binding sites for vinculin, integrin and, most importantly, actin, as well as a dimerization domain at the end. The vinculin-binding sites are mechanosensitive and are exposed when forces pull on talin. Talin is typically a dimer, and exists in a globular, inactive conformation, or an extended, active conformation (Goult et al., 2013). Both Rap1 and lipid binding play a role in talin activation (Camp et al., 2018; Ellis et al., 2013; Goult et al., 2018; Klapholz and Brown, 2017). In a classical model, active, extended talin binds with its head domain to integrin, and with its rod domain to the actin cytoskeleton (Kanchanawong et al., 2010). Pulling forces exerted by this connection recruit vinculin and other adhesion complex proteins.

In Drosophila, a single Talin (Rhea – FlyBase) mediates all integrin functions, from muscle attachment to wing adhesion and germ band retraction (Brown et al., 2002). Both structurally and functionally, Drosophila Talin is highly conserved with vertebrate talins (Ellis et al., 2013). The Talin head plays a particularly important role in adhesion complex stabilization and integrin clustering (Camp et al., 2018; Ellis et al., 2014). In addition, a mutagenesis study revealed functions for additional integrin-binding sites and orientations of Talin with respect to the membrane; however, at muscle attachment sites, Talin orientation is canonical (Klapholz et al., 2015). Vertebrate talins are heavily phosphorylated (Ratnikov et al., 2005), which may play a role in talin localization during platelet activation (Bertagnolli et al., 1993), but only one kinase has been identified up to now. Cdk5 phosphorylates talin-S425 affecting talin stability and increasing cell migration and metastasis (Huang et al., 2009; Jin et al., 2015).

Here, we identify phosphorylation sites of Drosophila Talin and show that phosphorylation coincides with terminal stages of muscle differentiation. We identify the Ste20-like kinase Slik as being responsible for phosphorylating threonine 152 in the unstructured loop of the Talin head domain. Talin T152 phosphorylation is crucial for proper recruitment of Talin to muscle attachment sites and for maintaining strong muscle attachments able to withstand full contractile load.

We first determined whether Drosophila Talin is post-translationally modified in late embryos, when myofibril assembly occurs (Katzemich et al., 2013). In 2D gels, we observed two main spots of Talin, one at the expected isoelectric point of 6.14 and one migrating to a more acidic point (Fig. 1A, asterisk), possibly indicative of Talin phosphorylation at late embryonic stages. We subsequently analyzed a Talin-GFP transgene that rescues talin mutants (Tanentzapf and Brown, 2006). We observed a similar migration pattern (Fig. 1B), indicating we could use this Talin fusion protein for affinity purification and mass spectrometry to identify the nature of these post-translational modifications.

We then wanted to identify the post-translational modifications on Talin in wild-type embryos. For this, we purified Talin-GFP from late-stage wild-type embryos by affinity purification in 3M urea to maintain post-translational modifications. Mass spectrometry revealed 15 phosphorylation sites in wild-type Talin, all on serine or threonine residues, suggesting that the acidic spots correspond to phosphorylation events in Talin (Fig. 1C). Most of these sites are located in the large rod domain. However, one residue, T152, is situated in the FERM domain. A recent phosphoproteomics analysis of all embryonic proteins identified 21 serine/threonine phosphorylation sites for full-length Talin, 11 of which overlap with our results, demonstrating the validity of our approach (Hu et al., 2019). We also purified Talin from adult thoraces but identified only three phosphorylation sites by mass spectrometry (S815, T816, S2700). This suggests that Talin phosphorylation may play a regulatory role in muscle attachment, which first occurs in late embryonic stages.

The most promising of the identified phosphorylation sites was T152, because it is located in an exposed loop in the F1 domain of the Talin head, and both the F1 domain and the T152 phosphoresidue are conserved in mammalian talin (T144 in human talin 1; Ratnikov et al., 2005) (Fig. 1D). Neither of these sites has been analyzed in vivo. To assess directly the function of Talin phosphorylation at T152, we generated three Talin transgenes: wild-type full-length Talin (UAS-Talin), the non-phosphorylatable alanine mutation, which changes both T152 and the neighboring T150 to alanine (UAS-Talin-T150/T152A), and the phosphomimetic glutamic acid mutation (UAS-Talin-T152E). We changed both T150 and T152 to alanine because we wanted to ensure an absence of phosphorylation in the exposed loop of the F1 domain. In addition, T150 was identified in the recent phosphoproteomics screen as a phosphorylation site in addition to T152 (Hu et al., 2019). We expressed these transgenes with Mef2-Gal4 in muscles in the background of the zygotic talin null mutant rhea^13-8 (Levi et al., 2006) to determine their localization and rescue ability. Zygotic rhea^13-8 mutants regularly display ‘rounded-up’ muscles at embryonic stage 16, when most maternal Talin is depleted (Fig. 2A-B″). All three transgenes rescued muscle attachment defects at stage 16 (Fig. 2C-E). Wild-type Talin localized well to muscle attachment sites showing twice the amount of muscle attachment site staining compared with cytoplasmic staining (Fig. 2C,F). Talin-T150/T152A localization to muscle attachment sites was significantly reduced (Fig. 2D,F). Surprisingly, Talin-T152E showed the strongest reduction with almost equal localization at muscle attachment sites and in the cytoplasm (Fig. 2E,F).

Next, we tested how well these constructs rescue embryonic lethality. Intriguingly, both wild-type Talin and Talin-T152E fully rescue embryonic lethality, whereas Talin-T150/T152A cannot rescue embryonic lethality at all (Fig. 2G). UAS-Talin expressed with Mef2-Gal4 rescued rhea^13-8 zygotic mutants to viability, demonstrating full functionality of the wild-type Talin transgene. It also suggests either that zygotic Talin is mainly required in muscles, or that Mef2-Gal4 mediates sufficient leaky expression in other tissues to allow full rescue. Surprisingly, despite low localization at muscle attachment sites UAS-TalinT152E also rescued to viability, indicating that constitutive phosphorylation at T152 is not detrimental. The first Talin-T150/T152A transgene we analyzed was expressed weakly compared with the others. Therefore, we generated another stock expressing two Talin-T150/T152A transgenes (2xTalin-T150/T152A) in a talin mutant background. 2xTalin-T150/T152A was expressed at the same level as wild-type Talin and Talin-T152E (Fig. 2H). Nevertheless, embryonic lethality was unchanged (Fig. 2G) and also muscle attachment site localization compared with Talin-T150/T152A was not altered (Fig. 2F). Thus, Talin-T150/T152A shows reduced localization at muscle attachment sites and cannot rescue the embryonic lethality of a talin null mutant, regardless of expression levels.

Finally, we employed phalloidin staining of the sarcomeric cytoskeleton to assess the phenotype of Talin-T150/T152A embryos at late stage 17. Embryos expressing Talin-T150/T152A in one or two copies exhibited muscle detachment from the cuticle at late stage 17 (Fig. 3). Again, there was no difference in phenotype between rescue with one or two copies of Talin-T150/T152A (Table 1). This phenotype is very similar to that of hypomorphic rhea mutant embryos (Prout et al., 1997). In addition, striations were lost, suggesting additional functions for Talin-T152 phosphorylation in myofibril assembly. Together, these results indicate that T150 and T152 are crucial for full Talin functionality.

SLK and LOK are members of the Ste20 family of kinases, belonging to the germinal center kinase-V subfamily (Dan et al., 2001), and have been implicated in phosphoregulation of FERM domain proteins. The ortholog and only member of this subfamily in Drosophila is Slik (SLK and LOK-like kinase). Slik is best known for phosphorylating two FERM domain proteins: Moesin, phosphorylation of which by Slik is crucial for stable cytoskeletal organization of cells in different contexts, and the tumor suppressor protein Merlin (Carreno et al., 2008; Hipfner et al., 2004; Hughes and Fehon, 2006; Kunda et al., 2008). Interestingly, in extracts from Slik¹ mutants, the more acidic Talin spot is absent, suggesting that Talin post-translational modifications are strongly reduced (Fig. 4A,B). We therefore tested the ability of Slik to phosphorylate Talin at T152, using S2R+ cells as a heterologous system. Compared with cells transfected with a myc-tagged Talin FERM domain alone, co-expression of Slik led to a substantial increase in phosphorylation (as detected with a phospho-specific antibody raised against a phospho-T152 peptide) (Fig. 4C). This effect was largely blocked upon mutation of T152 to alanine, demonstrating the specificity of the antibody. To determine whether Slik directly phosphorylates T152, we conducted in vitro kinase assays using Slik immunopurified from S2R+ cells and bacterially expressed purified Talin-FERM domain proteins. Compared with the negative control, we observed substantial Talin-FERM phosphorylation in the presence of Slik (Fig. 4D). The absence of phosphorylation of the T152A variant Talin-FERM confirmed the specificity of the signal. Taken together, these results suggest that Slik directly phosphorylates T152 of Talin in vivo.

Talin localizes to muscle attachment sites to carry out its functions in muscle attachment. We therefore investigated whether Slik localizes similarly and shows a similar phenotype in vivo when mutated. Antibody staining of βPS integrin, a marker of muscle attachment sites, and Slik demonstrated that Slik is present throughout the embryo, but is enriched at muscle attachment sites (Fig. 5A, Fig. S1A-A″). Slik also colocalizes with Talin at muscle attachment sites of wild-type embryos, but not Slik¹ mutant embryos (Fig. S1A-B″). The enrichment of Slik at muscle attachment sites was more pronounced when UAS-Slik was overexpressed in muscles with Mef2-Gal4 (Fig. S1C-C″). A Slik¹ null mutant showed no muscle detachment at stage 16 of embryogenesis, when talin zygotic null mutants first show muscle detachment (Fig. 5B). The lack of muscle detachment at stage 16 in Slik¹ null mutants is consistent with the absence of Talin-T150/T152A phenotypes at that stage. Importantly, Talin localization was reduced at muscle attachment sites in Slik¹ mutant embryos (Fig. 5B). Quantification demonstrated a significantly higher muscle attachment site staining than cytoplasmic staining in wild type compared with Slik¹ mutants (Fig. 5C). We also observed a similar reduction in Talin staining at muscle attachment sites in Slik¹/Df(2R)BSC603 embryos (Fig. 5C). Thus, Slik phosphorylation regulates Talin activation or recruitment to muscle attachment sites. Furthermore, the majority of Slik¹ mutants died at late stage 17, just before or during hatching, similar to 2xTalin-T150/T152A embryos. We therefore analyzed these late-stage non-hatched Slik¹ mutants carrying ZaspGFP (Zasp52[zcl423]), which labels Z-discs and muscle attachment sites, to assess whether they show muscle detachment phenotypes. In wild-type embryos at hatching, all larval muscles were properly attached (Fig. 6A). Importantly, in Slik¹ mutants we detected muscle detachment at these late stages (Fig. 6A, asterisks). We also analyzed late-stage 17 Slik¹/Df(2R)BSC603 embryos by myosin heavy chain and Talin antibody staining. We observed similar muscle detachments indicating that this defect is specific to a mutation in Slik (Fig. 6B). Furthermore, the presence of Talin and the absence of myosin heavy chain at the site of muscle detachment suggest a detachment of the cytoskeleton from integrins, consistent with a failure of Talin function (Fig. 6B). Slik is therefore required for the maintenance of muscle attachments and shows a similar, albeit weaker and slightly later-occurring, phenotype compared with Talin-T150/T152A.

If the absence of Slik-mediated phosphorylation of Talin at T152 is the cause of the muscle detachment observed in Talin-T150/T152A and Slik¹ mutants, then the phosphomimetic mutant Talin-T152E should be able to rescue aspects of the Slik¹ mutant phenotype. We therefore expressed Talin-T152E as well as Slik in muscles with Mef2-Gal4 in a Slik¹ mutant background and compared embryonic lethality and larval viability with those of Slik¹ mutant controls. Slik¹ mutant embryonic lethality was significantly rescued in the presence of Talin-T152E, very similar to rescue by Slik itself (Fig. 6C). Furthermore, Slik¹ mutants that managed to hatch had a median age of survival of 84 h, whereas Talin-T152E-expressing Slik¹ mutants lived 102 h (Fig. 6D). These data indicate that the Slik phenotype in muscles and at this developmental stage can be fully rescued by Talin-T152E.

We have identified phosphorylation of the Talin FERM domain by Slik as being important for Talin function in muscles. This pathway is likely conserved in vertebrates, because SLK and talin colocalize at focal adhesions and a conditional Slk knockout in skeletal muscles results in progressive myopathy (Pryce et al., 2017). In platelets, the high stoichiometry phosphorylation sites of Talin are T144 and T150. The effect of their phosphorylation is somewhat inconclusive, but it appears that T144/T150A mutations in tissue culture reduce cell adhesion (Li et al., 2016) and increase focal adhesion turnover (Elliott et al., 2010). A recent structural study indicates that the unstructured F1 loop does not interact with positively charged membrane phospholipids, suggesting that F1 loop phosphorylation should not disrupt membrane recruitment (Chinthalapudi et al., 2018). This is consistent with the full rescue we observed with T152E. Talin-E1777A, a mutant keeping Talin always in the extended, active conformation, fully rescues muscle detachment, and localizes more strongly to muscle attachment sites than does wild-type Talin (Ellis et al., 2013). We propose a multi-step mechanism of Talin activation, in which phosphorylation of threonine 150/152 is one step contributing to Talin activation and separation of the head and rod domains of Talin. Although the detailed mechanism and the sequence of events remain to be uncovered, our work identifies the first kinase involved in Talin F1 loop phosphorylation and demonstrates that this phosphorylation is crucial for the maintenance of muscle attachment.

The following stocks were used: rhea^13-8 (tendrils^13-8, provided by Mark Krasnow, Stanford University School of Medicine, CA, USA), Slik¹, pUBItalinEGFP (provided by Guy Tanentzapf, University of British Columbia, BC, Canada), Df(2R)BSC603/SM6a, Zasp52[zcl423] and Mef2-Gal4 (Bloomington Drosophila Stock Center). UAS-Talin, UAS-Talin-T150/T152A, UAS-2xTalin-T150/T152A, and UAS-Talin-T152E transgenic lines were generated by P-element transformation.

The following recombinants were generated by standard genetic crosses: rhea^13-8 Mef2-Gal4, rhea^13-8 UAS-talin, rhea^13-8 UAS-talin-T150/T152A, rhea^13-8 UAS-talin-T152E, and UAS-talin-T150/T152A; rhea^13-8 UAS-talin-T150/T152A. For the analysis of Talin transgenes in a talin mutant background, rhea^13-8 Mef2-Gal4 females were crossed to the different recombinant rhea^13-8 UAS-talin lines and set up for embryo collection. For the rescue assays, we generated the stocks Slik¹/CyO, Kr-GFP; UAS-talin-T152E Mef2-Gal4, Slik¹/CyO, Kr-GFP; Mef2-Gal4, and UAS-Slik; Slik¹/CyO, Kr-GFP by standard genetic crosses. For embryonic lethality assays, late non-green dead embryos were counted. For the larval survival assay, non-green first instar larvae were selected and age of death was recorded. Egg lays and aging were determined at 18°C. For analysis of the Slik¹ mutant phenotype, the recombinant Slik¹ Zasp52[zcl423]/CyO was generated. For imaging, this stock was back-crossed to Slik¹/Cyo-Kr-GFP flies and Slik¹/Slik¹ Zasp52[zcl423] embryos were selected.

Transgene expression constructs UAS-Talin, UAS-Talin-T150/T152A, and UAS-Talin-T152E were cloned in a two-step procedure. Different Talin N-terminal fragments (amino acids 1-1237) containing the desired point mutations (amino acid 150 and/or 152), a 3xFlag sequence at the 5′end, and flanking BglII and EcoRI restriction sites, were generated by Genscript. The remaining Talin C-terminal sequence (aa 1238-2836) was obtained from the Berkeley Drosophila Genome Project (clone SD07969 in pOT2). The N-terminal Talin fragment was cloned into the C-terminal fragment containing pOT2 by BglII and EcoRI digest, resulting in a full-length Talin insert (amino acids 1-2836, corresponding to FlyBase rhea-PB, rhea-PE and rhea-PF). Full-length Talin was excised by BglII and XhoI and cloned into pUASt.

For expression in S2R+ cells, Drosophila Talin sequences encoding the FERM domain (amino acids 1-414) were PCR amplified from a talin cDNA using oligonucleotides that introduced an EcoRI restriction site at the 5′ end and a stop codon followed by a KpnI site at the 3′ end. The resulting fragment was cloned into the EcoRI-KpnI sites of pBluescript. PCR-based site-directed mutagenesis was performed to mutate the codon encoding T152 to alanine. The wild-type and T152A-mutated fragments were cloned into the EcoRI-KpnI sites of a pRmHa3.puro expression construct modified to encode an N-terminal myc epitope tag, downstream of the copper-inducible metallothionein promoter, to create pRmHa.3puro/NTmycTalin-FERM and pRmHa.3puro/NTmycTalin-FERM^TA.

For bacterial expression of the Talin FERM domain proteins, the same wild-type and T152A-mutated FERM domain-coding sequences were PCR amplified with oligonucleotides introducing a 5′ EcoRI site and an in-frame 3′ XhoI site. The resulting fragment was cloned into the EcoRI-XhoI sites of pET-23a for expression as a C-terminal 6xHis fusion protein. The pRmHa.3puro/Slik expression construct has been described previously (Hipfner and Cohen, 2003). Sequence details of the dsRNA targeting the 5′ untranslated region (UTR) of Slik and the method for preparing it have also been described previously (Panneton et al., 2015).

S2R+ cells (Drosophila Genomics Resource Center) were grown in Drosophila Schneider's medium (Lonza) supplemented with 10% fetal bovine serum (Gibco) and 50 U/ml penicillin and streptomycin (Gibco), and cultured at 25°C; 5×10⁵ cells were plated in 24-well plates and allowed to attach for approximately 4 h. Cells were then transfected with 100 ng of each of the appropriate pRmHa.3puro expression constructs, as indicated, using XtremeGENE HP reagent (Roche) according to the manufacturer's protocol. Transgene expression was induced the following day by addition of CuSO₄ to a final concentration of 0.5 mM. For in vitro kinase assays, cells were pre-treated 2 days prior to transfection with 4 µg of Slik 5′ UTR dsRNA, and an additional 4 µg of dsRNA were added at the time of induction. Cells were harvested after 2 days of transgene expression.

Embryos were homogenized and proteins extracted in 2D sample buffer (7 M urea, 2 M thiourea, 4% CHAPS, 30 mM Tris, 1% DTT and 1% IPG buffer). This was followed by passive rehydration of the sample onto 7 cm IPG strips (pH 4-7) for 16 h at 22°C using the Protean IEF System (Bio-Rad). After rehydration, the first dimension was separated for 95,000 total volt hours. Post IEF run, strips were equilibrated for 15 min in SDS equilibration buffer (100 mM Tris, 6 M urea, 30% glycerol, 2% SDS, 0.002% Bromophenol Blue) supplemented with 0.5% DTT, followed by another 15 min in equilibration buffer and 4.5% iodacetamide. For the second dimension, IEF strips were inserted and sealed into 6% SDS-PAGE gels with 0.5% low melting agarose and separated for 2 h at 80 volts.

For samples of pUBITalinEGFP stage 17 embryos, embryos were collected for 24 h at 25°C and aged for another 24 h in the same conditions. Stage 17 embryos/first instar larvae or adult thoraces were washed in 1× PBS and homogenized in 8 M urea in 1× PBS. After centrifugation (14,000 g, 30 min), the supernatant was collected and the buffer was diluted to a new concentration of 3 M urea in 1× PBS, supplemented with protease inhibitors, 10 mM Na₂MoO₄ and 2 mM Na₃VO₄. Protein extracts were incubated with prewashed GFP-Trap beads (Chromotek) for 2 h in 3 M urea at 4°C. Beads were washed in RIPA buffer and boiled in sample buffer. Eluted proteins were loaded onto a 6% SDS-PAGE gel that was not separated to completion (80 volts for 10 min) and Coomassie stained. The entire protein sample was excised, subjected to tryptic digestion and analyzed by nanoliquid chromatography/tandem mass spectrometry (Q Exactive Plus from Thermo Fisher Scientific at Proteomics Core Facility, IRIC, Université de Montréal). Data were processed and post-translational modifications including phosphorylation sites were assigned using Peaks 8.5 from Bioinformatics Solutions (Han et al., 2011). The data were visualized with Scaffold 4.3.0 from Proteome Software (proteins filtered with at least two peptides identified and a final false discovery rate of 1% for peptides). The reference genome used for analysis was Drosophila melanogaster r5. We achieved 73% peptide coverage of full-length Talin (2836 amino acids).

Anti-Slik immunoprecipitations, and immunoblotting methods have all been described previously (Panneton et al., 2015). As a substrate for Slik kinase assays, the wild-type and T152 mutant Talin FERM domains were expressed as 6xHis fusion proteins in Escherichia coli BL21 cells and purified using the B-PER 6xHis Fusion Protein Purification Kit (Thermo Fisher Scientific). Assays were performed as described (Panneton et al., 2015) using 1 µg of the appropriate Talin-FERM fusion protein as a substrate. Phosphorylation was detected by immunoblotting using the anti-phospho-T152 Talin antiserum (generated by Genscript).

Stage 16 embryos were fixed using heat fixation. In brief, embryos were dechorionated in 50% bleach, fixed for 10 s in boiling embryonic wash buffer (70 mM NaCl, 0.1% Triton X-100), and immediately cooled down by adding three volumes of ice-cold embryonic wash buffer. Embryos were devitellinized in heptane/methanol followed by rehydration in PBS, 0.3% Triton X-100. Embryos were blocked for 1 h in PBS, 0.3% Triton X-100, 1% bovine serum albumin followed by primary antibody incubation at 4°C overnight. Secondary antibody incubation was carried out for 1 h at room temperature. Samples were mounted in ProLong Gold antifade solution (Life Technologies). Images were obtained on a Zeiss LSM510 Meta confocal microscope.

All antibodies were validated with mutant controls (this work or references provided). Primary antibodies used were rat anti-myosin (1:100, Babraham Bioscience Technologies, MAC147), mouse anti-Flag (1:500, Sigma-Aldrich, F3165), guinea pig anti-Slik (1:500; Hipfner and Cohen, 2003), rabbit anti-Talin (1:200, provided by Nick Brown, Gurdon Institute, University of Cambridge, UK) and mouse anti-βPS integrin (1:10, Developmental Studies Hybridoma Bank, CF.6G11). Fluorescently labeled secondary antibodies of the Alexa Fluor Series (Invitrogen, A11006, A21428, A11073 and A21235) were used at a dilution of 1:400.

Filamentous actin of stage 17 embryos was visualized using phalloidin-TRITC (1:1000, Sigma-Aldrich). These embryos were fixed in 4% paraformaldehyde/heptane for 60 min and devitellinized by hand. Embryos carrying ZaspGFP were mounted without fixation and imaged immediately.

All measurements described in this article were performed using ImageJ. For the analysis of Talin localization, average signal intensities were measured for entire muscle attachment sites and in the middle of non-overlapping muscle fibers (cytoplasmic staining) using the line tool. Talin localization was then expressed as a ratio. Measurements were carried out in 2-19 attachment sites per embryo for 4-12 embryos of the relevant genotype. The primary data are summarized in Tables S1 and S2. For western blot analysis, band intensities of three independent experiments were measured and normalized to myosin as a loading control.

We thank Nick Brown, Mark Krasnow and Guy Tanentzapf for materials; Beili Hu for injection of transgenic constructs; and Elke Küster-Schöck at the CIAN Imaging facility for help with confocal microscopy. We also thank Eric Bonneil for mass spectrometry services (Proteomics Core Facility at IRIC, Université de Montréal).