Full Text

Turn on search term navigation

ARTICLE

Received 8 May 2014 | Accepted 5 Sep 2014 | Published 16 Oct 2014

Michelle B.M. Pereira1,*, Aline M. Santos1,*, Danieli C. Gonalves1, Alisson C. Cardoso1, Slvio R. Consonni1, Fabio C. Gozzo2, Paulo S. Oliveira1, Ana Helena M. Pereira1, Alana R. Figueiredo2, Ana O. Tiroli-Cepeda2, Carlos H.I. Ramos2, Andr A. de Thomaz3, Carlos L. Cesar3 & Kleber G. Franchini1,4

Focal adhesion kinase (FAK) contributes to cellular homeostasis under stress conditions. Here we show that aB-crystallin interacts with and confers protection to FAK against calpain-mediated proteolysis in cardiomyocytes. A hydrophobic patch mapped between helices 1 and 4 of the FAK FAT domain was found to bind to the b4b8 groove of aB-crystallin. Such an interaction requires FAK tyrosine 925 and is enhanced following its phosphorylation by Src, which occurs upon FAK stimulation. aB-crystallin silencing results in calpain-dependent FAK depletion and in the increased apoptosis of cardiomyocytes in response to mechanical stress. FAK overexpression protects cardiomyocytes depleted of aB-crystallin against the stretch-induced apoptosis. Consistently, load-induced apoptosis is blunted in the hearts from cardiac-specic FAK transgenic mice transiently depleted of aB-crystallin by RNA interference. These studies dene a role for aB-crystallin in controlling FAK function and cardiomyocyte survival through the prevention of calpain-mediated degradation of FAK.

1 Brazilian National Laboratory for Biosciences, Center for Research in Energy and Materials, Campinas, Sao Paulo 13084-971, Brazil. 2 Chemistry Institute, University of Campinas, Campinas, Sao Paulo 13083-970, Brazil. 3 Gleb Wataghin Physics Institute, University of Campinas, Campinas, Sao Paulo 13083-859, Brazil. 4 Department of Internal Medicine, School of Medicine, University of Campinas, Campinas, Sao Paulo 13081-970, Brazil. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to K.G.F. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

DOI: 10.1038/ncomms6159

aB-crystallin interacts with and prevents stress-activated proteolysis of focal adhesion kinase by calpain in cardiomyocytes

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159

Exposure to stress increases the risk of intracellular protein unfolding and aggregation, thus compromising cell viability. To reduce this danger, cells rely on an

elaborate arsenal of molecular chaperones and clearance mechanisms that utilize proteolytic systems, which act to maintain protein homeostasis1. The failure of these systems is emerging as a major mechanism underlying several pathological conditions, such as cancer, neurodegenerative and cardiovascular diseases2,3.

Among protein chaperones, the small heatshock proteins (sHSPs, ranging from 12 to 42 kDa) have unique capacities to bind and prevent the aggregation of partially unfolded proteins under conditions of cellular stress4. sHSPs share a conserved acrystallin domain (ACD), a sequence of seven to eight beta strands that are anked by a diverse amino-terminal arm and a moderately conserved carboxy-terminal extension. The ACD is essential for the assembly of dimers and higher-order oligomers, but it is also required for the function of sHSPs57. The prototypical sHSP, aB-crystallin (22 kDa), is constitutively expressed in the eye lens and in several other tissues, notably, the cardiac and skeletal muscles and the brain4. In muscles, aB-crystallin binds to intermediate laments and sarcomeric myobrils, preventing their irreversible aggregation during stress8,9. It is therefore not surprising that inherited defects and the experimental gene targeting of aB-crystallin result in protein deposition diseases of cardiac and skeletal muscles10,11. However, in addition to its primordial duty as a molecular chaperone, aB-crystallin plays critical roles in essential cellular processes, including differentiation and survival12,13, suggesting that it may function as a key component in signal transduction networks. Accordingly, aB-crystallin is known to bind and modulate the activity of several signalling proteins in their native state, including Bax, Bcl-2 and p53 (refs 14,15); the knowledge of this regulation is rapidly emerging.

Focal adhesion kinase (FAK) is a mediator of the hyper-trophic growth and survival of cardiomyocytes during mechanical stress1619. However, the regulation of FAK in this particular cell type remains unclear. One current model suggests that following its activation, FAK Y397 becomes autophosphorylated, creating a binding site for Src, which then phosphorylates other tyrosine residues of the catalytic and the C-terminal FAK domains20. The phosphotyrosines function to enhance FAK catalytic activity and to activate downstream signalling molecules and scaffolding proteins21,22. In particular, the Src-dependent phosphorylation of Y925 is linked to the regulation of FAK subcellular localization and downstream signalling by ERK1/2 (refs 22,23). Furthermore, the level of FAK activity can be regulated by calpain-mediated proteolysis. In fact, this method seems to be a mechanism for inhibiting FAK activity at focal adhesions, which is required to promote cell migration24.

Since both FAK and aB-crystallin play active roles in modulating cellular response to stress, we investigated whether they mechanistically integrate into a signalling pathway. We found that mechanical stress triggers a functionally important interaction of FAK with aB-crystallin in cardiomyocytes. A combination of structural, imaging and biochemical tools was used to determine that the FAKs C-terminal focal adhesion-targeting (FAT) domain mediates its interaction with aB-crystallin. Through this interaction, FAK is protected from degradation by calpain and contributes to reduce the cardiomyocyte susceptibility to apoptosis during mechanical stress. Thus, the recruitment of FAK by aB-crystallin meets independent requirements for the regulation of FAK signalling and impacts the ability of cardiomyocytes to cope with mechanical stress.

ResultsFAK associates with aB-crystallin in response to stress. To determine whether aB-crystallin interacts with FAK during mechanical stress, we initially studied ventricular cardiomyocytes derived from 2-day-old neonatal rats (neonatal rat ventricular myocytes (NRVMs)) cultured on a silicone-based membrane system. The controlled biaxial cyclic stretch (10% from the slack length; 60 Hz; 112 h) of the NRVMs increased the FAK Y397 phosphorylation (Fig. 1a), which is a hallmark of FAK activation and functions in Src recruitment and activation20. FAK Y925 phosphorylation, which has been recognized as an Src-dependent process22, was also enhanced in stretched cells (Fig. 1a). No obvious change in the FAK protein level was noticed in the stretched compared with the non-stretched cultures. A parallel immunoblot analysis revealed increases in the levels of aB-crystallin protein and chaperone-activating S59 phosphorylation in the stretched versus non-stretched NRVMs (Fig. 1a). To evaluate a potential interaction between FAK and aB-crystallin, we immunoprecipitated NRVM extracts with an anti-aB-crystallin antibody. An association between FAK and aB-crystallin was detected in the extracts of stretched, whereas only a minor association was expressed in the extracts of the non-stretched NRVMs (Fig. 1b). The ratio between the amounts of FAK and aB-crystallin precipitated by the anti-aB-crystallin antibody following cell stretch increased up to twofold over the non-stretched controls (Fig. 1b). Furthermore, the aB-crystallin immunoprecipitates obtained from the stretched cells were enriched with Y397- and Y925-phosphorylated FAK, suggesting that aB-crystallin associates with the active form of FAK (Fig. 1c,d). Changes in FAK and aB-crystallin in NRVMs were further evaluated by immunocytochemistry. In non-stretched myocytes, the distribution of FAK was sarcomeric and cytoplasmic, concentrated in the perinuclear region, while aB-crystallin displayed a perinuclear distribution, with no obvious association with the sarcomeres (Fig. 1e). In these cells, perinuclear FAK and aB-crystallin barely overlapped in merged images. Cell stretch enhanced aB-crystallin signal and induced its localization to the sarcomeres, where it overlaps with FAK (Fig. 1e).

To investigate whether the association between FAK and aB-crystallin occurs in response to distinct forms of cellular stress, we treated H9c2 cardiac myoblasts with hydrogen peroxide (H2O2). After the treatment with H2O2 (1 mM, 10120 min), the abundance of aB-crystallin and Y397-phosphorylated FAK were signicantly increased (Supplementary Fig. 1a). Treatment with H2O2 also markedly enhanced the association of FAK with aB-crystallin in H9c2 cells, expressed by the ratio between the amounts of FAK and aB-crystallin precipitated by the anti-FAK antibody (Supplementary Fig. 1b). We then examined FAK and aB-crystallin distribution in H9c2 cells. In untreated cells, FAK localizes prominently to adhesions and less intensely along the stress bres and in the perinuclear region, while aB-crystallin distributes in the cytoplasm, concentrated at the perinuclear region (Supplementary Fig. 1c). In these cells, FAK co-localizes with aB-crystallin in the perinuclear region, but not at focal adhesions. Cells treated with H2O2 reveal no substantial change in

FAK localization, but show an increase in the cytoplasmic aB-crystallin, which corresponds to the increased expression detected in the western blots. Consistent with the results of coimmunoprecipitation assays, FAK and aB-crystallin co-localization increases in cells treated with H2O2, mainly in the cytoplasm and along the stress bres (Supplementary Fig. 1c).

Because aB-crystallin translocates to mitochondria under stress conditions25, we investigated whether the FAK-aB-crystallin complexes formed in response to oxidative stress in H9c2 are localized in mitochondria. Cells were probed with anti-FAK and

2 NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159 ARTICLE

ST NS

pY397FAK

pY925FAK FAK

B-crystallin pS59 B-crystallin

GAPDH

1 2 6 12 (h)

125 kDa

125 kDa 125 kDa23.5 kDa23.5 kDa

37 kDa

FAK

250 200 150

FAK/B-crystallin

(a.u.)

100

125 kDa23.5 kDa

B-crystallin

IP: B-crystallin

0 NS 1 2 6 12

(h)

IP: B-crystallin pY925FAKB-crystallin

125 kDa

23.5 kDa

B-crystallin

pY397FAK

400

* * *

125 kDa23.5 kDa

200 *

* *

pY397FAK/B-crystallin

(a.u.)

300 200

pY925FAK/B-crystallin

(a.u.)

250 100

100

IP: B-crystallin

0 NS 1 2 6 12 (h)

NS FAK

B-crystallin FAK B-crystallin

Merged Co-localization Merged Co-localization

Figure 1 | FAK interacts with aB-crystallin in cardiomyocytes. (a) Representative immunoblots of six independent experiments showing extractsfrom NS or ST (1, 2, 6 and 12 h) NRVMs probed as indicated. GAPDH served as loading control. (bd) Representative (n 6) anti-FAK, -pY397-FAK and

anti-pY925-FAK immunoblots of aB-crystallin immunoprecipitates (IP) from extracts of NS or ST NRVMs (1, 2, 6 and 12 h). Bar graphs show the densitometric readings of FAK, pY397-FAK and pY925-FAK normalized to aB-crystallin. *Po0.0001 versus NS NRVMs (P values were determined by analysis of variance and Bonferronis post hoc comparisons). Data are presented as means.e.m. (e) Confocal immunouorescence imaging of

NS or ST (2 h) NRVMs stained with anti-FAK and anti-aB-crystallin. FAK distribution is shown in green and aB-crystallin is shown in red. Merged images are composed of anti-FAK, anti-aB-crystallin and DAPI stained. Regions of co-localization are shown as separate images. Scale bar, 20 mm.

IP, immunoprecipitation; NS, non-stretched; ST, stretched.

anti-aB-crystallin antibodies and the mitochondria marker Mitotracker (Supplementary Fig. 1d). In cells treated with H2O2, aB-crystallin partially co-localized with mitochondria, but FAK did not, indicating that the fraction of aB-crystallin bound to FAK in stressed cells is not associated with mitochondria.

FAK associates with aB-crystallin in overloaded rat heart. To explore whether the association between FAK and aB-crystallin occurs under acute pathophysiological conditions in adult hearts, we examined samples of rat left ventricle subjected to 2 h of pressure overload induced by transverse aortic constriction

(TAC). FAK phosphorylation at Y397 and Y925 were increased, while the FAK protein levels remained unaltered in this model (Fig. 2a). Acute pressure overloading also increased the levels of aB-crystallin protein and Ser-59 phosphorylation and the association with FAK in the rat left ventricle (Fig. 2a). The ratio between the amounts of FAK and aB-crystallin precipitated by the anti-aB-crystallin antibody increased twofold in the overloaded over the left ventricle of sham-operated (SO) mice (Supplementary Fig. 2a). Additional analyses demonstrated increased levels of Y397- and Y925-phosphorylated FAK in the anti-aB-crystallin precipitates obtained from the lysates of the overloaded left ventricle (Fig. 2a). We also demonstrated that FAK phosphorylation, aB-crystallin expression and FAK

NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159

TAC2 h

IP: B-crystallin

pY397FAK

125 kDa

TAC2 h

pY925FAK

FAK

pY397FAK

125 kDa

FAK

125 kDa 125 kDa

125 kDa

TAC2 h

125 kDa

23.5 kDa

23.5 kDa 23.5 kDa

pY397FAK

FAK

B-crystallin

125 kDa

23.5 kDa

37 kDa

B-crystallin

23.5 kDa 125 kDa

pY925FAK

B-crystallin

FAK

IP: FAK

pS59

B-crystallin GAPDH

B-crystallin

GAPDH

37 kDa

c d

TAC

FAK

B-crystallin

Co-localization

7 Days

30 Days

60 Days

90 Days

FAK

125 kDa

B-crystallin

23.5 kDa

GAPDH

37 kDa

TAC2 h

FAK

B-crystallin

Co-localization

IP: FAK

30 Days

7 Days

60 Days

90 Days

Figure 2 | Mechanical stress increases the association of FAK with aB-crystallin in overloaded left ventricle of rodents. (a) Representativewestern blots of anti-FAK, -pY397-FAK, -pY925FAK, -aB-crystallin, -p59aB-crystallin and -GAPDH antibodies obtained with extracts of the left ventricle from SO (n 6) or TAC (2 h; n 6) rats. GAPDH served as loading control; at right, anti-FAK, -pY397-FAK and anti-pY925-FAK immunoblots of anti-aB-

crystallin immunoprecipitates (IP). aB-crystallin was used as loading control. (b) Shown are specic immunoblots of extracts from isolated adult cardiomyocytes obtained from SO (n 4) or TAC (2 h; n 4) rats, as indicated. Representative anti-aB-crystallin and -FAK immunoblots of anti-FAK

immunoprecipitates (IP) are shown on the right. FAK was used as loading control. (c) Confocal imaging of anti-FAK and anti-aB-crystallin double-labelled ventricular tissue from SO and TAC (2 h) rats. FAK distribution is shown in green and aB-crystallin is shown in red. Regions of co-localization between anti-FAK and anti-aB-crystallin stainings are indicated in white. Scale bar, 20 mm. (d) Representative western blottings showing FAK, aB-crystallin and

GAPDH expression in samples of the left ventricle from SO (n 5) or TAC (7, 30, 60 and 90 days; n 5 for each group) mice. GAPDH served as

loading control. Lower panel shows representative anti-aB-crystallin and anti-FAK immunoblottings of the anti-FAK immunoprecipitated from extracts of left ventricle from SO and TAC mice. FAK served as loading control. IP, immunoprecipitation.

association with aB-crystallin were increased in adult cardiomyocytes isolated from overloaded rat left ventricle compared with those isolated from control rats (Fig. 2b; Supplementary Fig. 2b). Furthermore, immunohistochemistry of the rat left ventricle revealed an enhanced co-localization of FAK and aB-crystallin at the sarcomeres of cardiomyocytes from overloaded compared with the control hearts (Fig. 2c). Collectively, these results indicate that FAK associates with aB-crystallin in cardiomyocytes. This association accompanies FAK activation, suggesting that aB-crystallin may participate in the regulation of

FAK signalling in cardiomyocytes and in the heart of rat.

We next examined whether the association between FAK and aB-crystallin in cardiomyocytes persists during the transition of compensated to decompensated hypertrophy. Mice were subjected to 790-day TAC, which results in concentric hypertrophy of the left ventricle by 7 days, followed by progressive left ventricular dilation thereafter, as conrmed by M-mode echo-cardiography (see Supplementary Table 1). Importantly, FAK expression in the left ventricle did not change in response to

TAC. However, the expression of aB-crystallin and its association with FAK, as detected in the immunoprecipitates of anti-FAK antibody, increased (twofold) early and remained increased throughout the observational period (Fig. 2d; Supplementary Fig. 2c). These data suggest that the association between FAK and aB-crystallin may have an important role in controlling FAK signalling and cardiac homeostasis in pathophysiological conditions.

aB-crystallin interacts directly with the C terminus of FAK. To determine whether aB-crystallin can bind FAK directly, we performed pull-down assays using recombinant glutathione S-transferase (GST)-tagged full-length FAK (GST-FAK; Invitrogen) and 6xHis-tagged aB-crystallin (His-CryAB) proteins.

His-CryAB was expressed in Escherichia coli and puried to homogeneity by afnity and size-exclusion chromatography (SEC; Supplementary Fig. 3a). Unlike the GST control, GST-FAK interacted with His-CryAB (Fig. 3a). In separate studies,

4 NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159 ARTICLE

Pull-down

b c

Pull-down

His-CryAB

GST-FERM

GST-KINASE

GST-cTERM

His-CryAB

125 kDa

FAK

GST

GST-FAK

GST

146.5 kDa

23.5 kDa

B-crystallin

Loading control

Total extracts

FAK

His-CryAB

Loading control B-crystallin

B-crystallin

23.5 kDa

125 kDa

B-crystallin 23.5 kDa

Pull-down

Total extracts

1.00.80.60.40.20.0

Molar mass LS

Molar mass LSRI

Molar mass (Da) Molar mass (Da)

Normalized LS and RINormalized LS and RI

GST

GST-FERM

GST-KINASE

GST-cTERM

GST

GST-FERM

GST-KINASE

GST-cTERM

106

105

104

23.5 kDa

10 15 20 25 30 35 40

23.5 kDa

Time (min)

1 2 3

Molar mass LSRI

Molar mass (Da)

106

105

1.00.80.60.40.20.0

Normalized LS and RI

106

105

104

103 20 25

Time (min)

15 20 25 30 35 40

30 35 40

Time (min)

Figure 3 | FAK C terminus mediates the direct interaction of FAK with aB-crystallin. (a) Representative (n 4) anti-FAK immunoblot from His pull-down

assay of FAK (GST-FAK) performed with His-CryAB coupled to Ni2 Sepharose beads. GST alone was used as negative control. Anti-aB-crystallin immunoblots were used as loading control. (b) Representative (n 4) anti-FAK immunoblot from pull-down assays of extracts from NS or ST (2 h) NRVMs

performed with His-CryAB. aB-crystallin was used as loading control. Amounts of FAK in the extracts of the NS or ST NRVMs were analysed by immunoblotting probed with anti-FAK. (c) Representative (n 4) aB-crystallin immunoblot from pull-down assay of His-CryAB performed with

recombinant FAK fragments (GST-FERM, GST-Kinase or GST-cTERM) and GST alone as control. (d) Representative (n 4) anti-aB-crystallin immunoblot

from pull-down assays of extracts from NS or ST NRVMs performed with recombinant FAK fragments (GST-FERM, GST-Kinase or GST-cTERM) and GST alone (upper panel). The lower immunoblot shows the amount of aB-crystallin from NS or ST NRVM total extracts. (e) The SEC-MALS chromatograms show the refractive index (blue line), light-scattering (red line) intensity (90 detector) and calculated molar mass (black squares). His-CryAB elutes as a single peak centred on a maximum population species of 580 kDa, corresponding to an B28-subunit oligomer. (f) SEC-MALS shows His-FAK-cTERM eluting as a single peak of 321 kDa, in agreement with the monomer molar mass. (g) SEC-MALS prole of His-CryAB and His-cTERM mixture. His-CryAB and His-cTERM interaction resulted in shifts in the retention time and molecular weight from the individual components. The complex formed by His-CryAB and His-FAK-cTERM eluted between the peaks of isolated proteins, with an average molar mass of 1141 kDa, in agreement with a 2:2 stoichiometry. The SEC-MALS assays were carried out on a GE Healthcare Superdex200 HR 10/30 column connected to the multi-angle laser light-scattering system and are representative of six independent experiments.

His-CryAB was shown to precipitate greater amounts of FAK from extracts of stretched NRVMs compared with non-stretched NRVMs (Fig. 3b). Next, recombinant GST-fusion proteins that covered the N-terminal FERM (GST-FERM), the kinase (GSTKinase) and the C-terminal (GST-cTERM) domains of FAK were expressed in E. coli and puried to homogeneity (Supplementary Fig. 3b). These constructions were coupled to glutathione sepharose beads and assayed for binding to His-CryAB. The GST-cTERM was the only construct that interacted with His-CryAB, indicating that this domain is sufcient to confer FAK interaction with aB-crystallin (Fig. 3c). Likewise, FAK GST-cTERM was the only construct that precipitated aB-crystallin from NRVM extracts (Fig. 3d). Consistent with the co-immunoprecipitation data, the amount of aB-crystallin precipitated by the GST-cTERM construct increased when samples of stretched (rather than non-stretched) myocytes were used.

To gain information about the stoichiometry of FAK-cTERM and aB-crystallin assembly, we used recombinant His-CryAB and

FAK His-cTERM (Supplementary Fig. 3c) and SEC coupled to a multi-angle light-scattering (MALS) detector. First, we investigated the mass distribution of the His-CryAB oligomer at room temperature. The protein eluted from the column as a single broad, but slightly asymmetric, peak centred on B22.7 min (Fig. 3e). The MALS data shows that at the front end of the peak, the average mass was B750 kDa, at the tail end 550 kDa and at the maximum 580 kDa. This corresponds to an approximate average-oligomer distribution range of 2432 subunits. HiscTERM was also characterized by SEC-MALS, and it was shown to be a monomer in solution (Fig. 3f). The protein eluted from the column as a single peak with average mass of 321 kDa centred on B33.5 min. The SEC-MALS approach of the samples containing the complex His-cTERM-His-CryAB showed three

NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159

distinct peaks. The rst one corresponds to the His-CryAB elution, the late one corresponds to His-cTERM elution and the middle peak (centred on B29 min) corresponds to the complex

His-cTERM-His-CryAB elution (Fig. 3g; Supplementary Fig. 3d). The MALS data from the maximum of the middle peak correspond to an average mass of B1141 kDa. Theoretically, this mass corresponds to a 2:2 stoichiometry of His-CryAB (B23 kDa) and FAK-cTERM (B34 kDa). This stoichiometry was conrmed by experiments where the His-cTERM/His-CryAB complex peak was isolated by SEC and re-examined by SECMALS (Supplementary Fig. 3e). Collectively, these results indicate that FAK binds to the chaperone aB-crystallin through a direct interaction that is mediated by its C-terminal domain.

The ability of His-CryAB to act as a molecular chaperone was assessed via its capacity to prevent the aggregation of FAK-cTERM denatured by heat. FAK His-cTERM was incubated at 42 C in the presence or absence of His-CryAB at molar ratio (His-CryAB:FAK-cTERM) of 1:1. The precipitated and soluble proteins were separated by centrifugation and resolved by SDS polyacrylamide gel electrophoresis (SDS-PAGE). FAK HiscTERM was precipitated in the absence, but not in the presence of His-CryAB, indicating that this construct of aB-crystallin retains its chaperone activity (Supplementary Fig. 4).

FAK-cTERM-aB-crystallin interaction in living cells. To determine whether FAK-cTERM interacts with aB-crystallin in living cells, we employed an assay to detect FAK-cTERM-aB-crystallin proximity by Frster resonance energy transfer (FRET) in HEK 293T cells. In this assay, the cyan uorescent protein (CFP) coupled to the N terminus of aB-crystallin (CFP-CryAB)

served as the FRET donor, while the yellow uorescent protein (YFP) coupled to the N terminus of FAK-cTERM (YFP-cTERM) served as the FRET acceptor (Supplementary Fig. 5). In our system, only the constructs that have the uorescent proteins (CFP and YFP) fused to the N terminus of aB-crystallin and

FAK-cTERM resulted in efcient FRET. FRET was measured here using uorescence lifetime imaging microscopy (FLIM). If energy transfer occurs, the lifetime of the donor-emitted uorescence will decrease because of the new decay path. Plasmid vectors encoding unfused CFP and YFP were co-transfected in cells as negative controls. Despite the co-localization of the donoracceptor uorophores in the confocal images, the uorescence lifetime-weighted mean component (tm) of the donor was 2.5 ns, indicating the absence of FRET (Fig. 4a). A highly efcient positive FRET control was obtained with the CFP uorophore that was coupled directly to YFP through a 15-amino-acid (aa)-long linker (Fig. 4b). The lifetime of CFP decreased to 1 ns, indicating the proximity of CFP to YFP. The tm of CFP-CryAB co-expressed with YFP was 2.4 ns (Fig. 4c), while the co-expression of the pair CFP-CryAB and YFP-cTERM decreased the donor lifetime to 1.8 ns (Fig. 4d). No major difference was observed in the distribution of CFP-CryAB and YFP-cTERM in cells co-expressing both constructs compared with cells expressing each of the constructs alone (Fig. 4d; Supplementary Fig. 5). The reduction in the lifetime of the CFPCryAB conrmed a positive FRET between CFP-CryAB and YFP-cTERM. Thus, the FRET assay data support a direct interaction between FAK-cTERM and aB-crystallin in living cells.

Y925 phosphorylation enhances FAK-aB-crystallin interaction. We next asked whether the phosphorylation of FAK-cTERM at Y925 could inuence the interaction with aB-crystallin. To test this hypothesis, we performed pull-down assays with a recombinant GST-tagged aB-crystallin (GST-CryAB) coupled to glutathione sepharose beads, and His-FAK-cTERM, Y925-

phosphorylated His-FAK-cTERM and a Y925F mutant His-FAK-cTERM. The Y925-phosphorylated FAK-cTERM was obtained by performing in vitro phosphorylation with Src (Fig. 5a; Supplementary Fig. 6a). This phosphorylation was conrmed by immunoblot with a phosphospecic antibody and by tandem mass spectrometry (MS/MS) analysis (Supplementary Fig. 6b,c). Y925 phosphorylation markedly enhanced the binding of FAK-cTERM to GST-CryAB, whereas the cTERM mutation Y925F abolished the interaction (Fig. 5b). The MALS data of cTERMY925F showed no change in its elution pattern with respect to cTERM-WT (Fig. 5c), but the co-elution with His-CryAB (peak centred on B29 min) was disrupted (Fig. 5d). These results indicate that Y925 is required for the interaction of FAK-cTERM with aB-crystallin and that its phosphorylation by Src enhances this interaction.

We next substantiated the inuence of Y925 on the association of FAK with aB-crystallin in cells. For that, we used FAK expression constructs, one of which harboured myc-tagged FAK (myc-FAK) and the other an Y925F mutant version (myc-FAKY925F). These constructs and the empty plasmid (pRc/CMV-myc) were expressed in C2C12 cells (Fig. 5e). Precipitates of a myc-specic antibody from C2C12 cells expressing the myc-FAK-Y925F construct contained markedly less aB-crystallin than those obtained from cells expressing myc-FAK (Fig. 5f). We also used FLIM-FRET approach to test whether the pharmacological inhibition of Src (PP2) could decrease the association of FAK with aB-crystallin. Single-cell FLIM quantication was carried out before and after (30 min) adding PP2 to the culture medium. PP2 treatment induced a shift in the uorescence lifetime from1.8 to 2.4 ns in HEK 293T cells co-expressing the pair CFPCryAB and YFP-cTERM, but not in cells co-expressing the pair CFP-CryAB and YFP (tm 2.4 ns), used as control

(Supplementary Fig. 6d). In line with this data, we demonstrated an increase in the uorescence lifetime to 2.3 ns when the pair CFP-CryAB and YFP-cTERM-Y925F was expressed in HEK 293T cells (Supplementary Fig. 6e). These results support the notion that Y925 is critical for the interaction between FAK and aB-crystallin in cells.

Structural basis of the FAK-cTERM-aB-crystallin interaction. To gather information on the structural features of the FAK-cTERM-CryAB complex, we used crosslinking in combination with mass spectrometry (CXMS) analysis and computational modelling26,27. Given the size and high exibility of the multidomain full-length FAK, we decided to use the truncated FAK-cTERM in the structural and computational studies. FAK His-cTERM and His-CryAB were expressed and co-puried by afnity and SEC. The disuccinimidyl suberate (DSS)-crosslinked FAK His-cTERM-His-CryAB complex was puried by SEC and identied as shifted bands in an SDSPAGE analysis (Fig. 6a; Supplementary Fig. 7a). Liquid chromatographyMS/MS analyses (Supplementary Fig. 7b) of the DSS-crosslinked His-cTERM-His-CryAB complex indicated that the tryptic peptide 920SNDKVYENVTGLVKAVIEMSSK941 is bound by the residue K933 to the K103 of His-CryAB sequence

93VLGDVIEVHGKHEER107 (Fig. 6b,c; Supplementary Fig. 7c). K933 is located on the surface of FAK FAT helix 1, while K103 is located on the b5 strand of aB-crystallin. To model the potential interaction site between FAK-cTERM and aB-crystallin, we performed a molecular docking (GRAMM-X proteinprotein docking) with the solved structures of FAK FAT (PDB code 1K40) and a Mus musculus model of aB-crystallin ACD (derived from PDB code 2KLR). The docking search was run twice with no search space restriction on the possible docking poses and recovered 300 solutions in each submission. Then, we applied

6 NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159 ARTICLE

CFP+YFP

CFP-15aa-YFP

[afii9848]m = 2.5 ns

[afii9848]m = 1.0 ns

Ex 800 nm

Ex 488 nm

Ex 800 nm + DIC

Pixel frequency Pixel frequency Pixel frequency Pixel frequency

0.5 ns

CFP-CryAB + YFP

CFP-CryAB + YFP-cTERM

3.0 ns

1 2 3 ns

[afii9848]m = 2.4 ns

[afii9848]m = 1.8 ns

0.5 ns 3.0 ns

Ex 800 nm Ex 488 nm Ex 800 nm + DIC

1 2 3 ns

Figure 4 | Confocal imaging of cells transfected with the donor and acceptor FRET pair. (a,b) Lifetime images of HEK 293T cells transfected with pAmCyan1-C1 and pZsYellow1-C1 vectors (negative control) or with the pAmCyan1-C1 encoding the YFP linked to CFP through a 15-aa-long linker (positive control). In this positive control, the uorescence lifetime, tm, of the donor (2.5 ns) was reduced to 1.0 ns. (c,d) Lifetime images of HEK 293T cells transfected with pAmCyan1-C1-CryAB plus pZsYellow1-C1 vectors or with the pAmCyan1-C1-CryAB plus pZsYellow1-C1-cTERM vectors. The uorescence lifetime of the donor CFP-CryAB (2.4 ns) was reduced to 1.8 ns by FRET. The correspondent images representing the uorescence (excitation at 800 and 488 nm) and differential interference contrast microscopy (DIC) of cells are displayed at the right panels. Scale bar, 10 mm.

distance restrictions between C-alpha of lysine pair of 20 to lter poses into the best energy-ranked solution set. Poses achieved by this approach had lysines 103 and 933 in a position that resembles the DSS-crosslinked lysines. This set of poses was superposed (root mean squared deviation of B0.5 A) and one model was chosen for the analysis. In this model, the face formed by the a1a4 helices of FAK FAT interacts with the b4b8 groove of the ACD (Fig. 6d). The structural model of the FAK FAT-aB-crystallin ACD complex was submitted to energy minimization to relieve the problematic van der Waals contacts and to rectify any potentially incorrect geometry that may have existed in the model (Fig. 6e). Contact surface analyses showed that this interaction buries an area of B1,160 A2. Importantly, this model can explain the 2:2 stoichiometry of FAK-cTERM:aB-crystallin that emerged from SEC-MALS approach, with one FAK-cTERM monomer able to interact with each copy of the binding site on the opposing faces of the aB-crystallin dimer (Supplementary Fig. 7d).

Probing the FAK-cTERM-aB-crystallin complex by mutagenesis. To validate the docking model, we performed mutational analysis guided by the predicted interaction regions between FAK-cTERM and aB-crystallin. In addition to Y925FFAK, the FAK-cTERM mutations H1025NFAK, V1029NFAK and L1035NFAK also attenuated this interaction, whereas the mutations V932NFAK, I936NFAK and L1021NFAK did not affect the interaction with CryAB (Fig. 7a; Supplementary Fig. 8a). The CryAB mutation L89NCryAB decreased the interaction, while the mutations V91NCryAB, V93NCryAB and L137NCryAB did not interfere in the interaction with FAK-cTERM (Fig. 7b; Supplementary Fig. 8b). The CryAB combinatorial mutations V91N/V93NCryAB and V91/V93N/L137NCryAB markedly reduced the interaction with FAK-cTERM (Fig. 7c; Supplementary Fig. 8b).

Far-ultraviolet circular dichroism (CD) was used to examine the effects of mutations on the structure of FAK-cTERM and CryAB. CD spectra for FAK-cTERM exhibited two minima

NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159

Pull-down GST-CryAB

cTERM + Src

cTERM cTERM-Y925FSrcATP

cTERM + Src

+ ATP

cTERM-Y925F

+ Src + ATP

+ +

pY925FAK

FAK

Molar mass

Time (min)

34.4 kDa

Loading control

34.4 kDa

FAK

B-crystallin

Loading control

34.4 kDa

50 kDa

107

105

104

1.0

Normalized LS and RI

Molar mass LS

106

105

104

103

0.80.60.40.20.0

Normalized LS and RI

LS RI

Molar mass (Da)

106

10 15 20 25 30 35 45

Time (min)

IP: myc-constructs

1.00.80.60.40.20.0

myc-FAK-Y925F

125 kDa

B-crystallin myc-tag

pRc/CMV-myc

myc-FAK

FAK

pRc/CMV-myc

myc-FAK

125 kDa

23.5 kDa

pY925FAK 125 kDa

Figure 5 | Tyr925 phosphorylation enhances FAK-aB-crystallin interaction. (a) In vitro Src kinase assay performed with FAK-cTERM, FAK-cTERM-Y925F, recombinant Src kinase and ATP, as indicated. Shown are immunoblots performed with anti-pY925-FAK and anti-FAK antibodies. Immunoblottings performed with anti-FAK were used as loading control. (b) FAK immunoblot from GST-CryAB pull-down assay of FAK His-cTERM, FAK His-cTERM-Y925F, recombinant Src kinase and ATP performed as indicated. aB-crystallin immunoblots were used as loading control (n 4 independent experiments).

(c) SEC-MALS chromatogram of His-FAK-cTERM-Y925F shows its elution as a single peak of 321 kDa. (d) SEC-MALS prole of His-CryAB and His-cTERM-Y925F mixture. His-CryAB and His-cTERM-Y925F elute as isolated proteins. The rst peak corresponds to His-CryAB (6005 kDa) and the second to His-cTERM-Y925F (341 kDa). (e) FAK and pY925-FAK immunoblots of protein extracts from C2C12 cells transfected with pRc/CMV-myc, pRc/CMV-myc-FAK (myc-FAK) or pRc/CMV-myc-FAK-Y925F (myc-FAK-Y925F) vectors. (f) Representative (n 4) anti-aB-crystallin and anti-myc-tag

immunoblots of anti-myc-tag immunoprecipitates (IP) from C2C12 cells treated as indicated.

(208 and 222 nm), characteristic of a-helix-rich proteins. The mutants Y925FFAK, V932NFAK, I936NFAK, L1021NFAK, H1025NFAK, V1029NFAK and L1035NFAK showed no change in the far-ultraviolet CD (Supplementary Fig. 8c). Far-ultraviolet CD spectra for CryAB exhibited single broad ellipticity minima, characteristic of b-sheet and b-turn-rich proteins. The CryAB mutants showed no change in the ellipticity (Supplementary Fig. 8d). These analyses indicate that the mutations did not impact the structure of either FAK-cTERM or CryAB proteins.

Modelling the enhanced binding of phosphoY925 FAK FAT to ACD. The FAK FAT-aB-crystallin ACD complex model revealed that the nonphosphorylated Y925 FAK FAT domain is positioned in close proximity to ACD K92 and is located within a pocket formed by ACD K92 and L94 of the b4 strand and E99 of the b5 strand (Fig. 7d). In this conguration, the phenolic moiety of Y925 makes hydrophobic contacts with the aliphatic groups of K92 and L94, and hydrogen bonds with E99 of the b5 strand.

Molecular dynamics (MD) simulation (YASARA program) of a Y925-phosphorylated FAK FAT model suggested a rearrangement of the phospho-phenolic ring of Y925 and the positively charged side chain of K92 towards each other and a repulsion of

E99 from these contacts (Fig. 7e). The MD simulation estimated that Y925 phosphorylation increases the binding free energy by B122 kcal mol 1 with respect to the nonphosphorylated complex model. Consistent with this model, the CryAB-K92A single mutation markedly decreases the association of CryAB with both the nonphosphorylated and phosphorylated forms of the FAK-cTERM (Fig. 7f; Supplementary Fig. 8b). These results were also validated by SEC-MALS experiments. No change was observed in the elution pattern of CryAB-K92A mutant in comparison with the wild-type CryAB, but the characteristic co-elution peak of the cTERM-CryAB assembly was disrupted (Fig. 7g,h). In addition, FRET data obtained with CFP-CryAB-K92A and YFP-cTERM conrmed the lack of interaction between these proteins in the intracellular environment (Supplementary Fig. 9). Together, these results support the notion that FAK binds to aB-crystallin through a specic interaction that is positively modulated by FAK phosphorylation at Y925.

aB-crystallin depletion prompts FAK degradation by calpain. Given the demonstration that FAK interacts with aB-crystallin in cardiomyocytes subjected to mechanical stress, we asked whether the depletion of aB-crystallin by RNA interference (RNAi;

8 NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159 ARTICLE

b c

His-cTERM + His-CryAB

His-cTERM +

His-CryAB + DSS

Groove

Y925

5 4

K933

K103

6+7

c-TERM

ACD

CryAB 3

FAK FAT

d e

FAK FAT

1 4

K933

K103

K933

K103

ACD

6+7

ACD

Figure 6 | FAK FAT domain and aB-crystallin-interacting surface as determined by CXMS and molecular docking. (a) Coomassie blue-stained SDSPAGE of co-puried His-cTERM-His-CryAB complex and its DSS-crosslinked complex. (b) Ribbon representation of the FAK FAT domain with the crosslinked peptide (SNDKVYENVTGLVKAVIEMSSK) in yellow. The reagent lysine (K933) and the tyrosine 925 are shown in stick representation.(c) Ribbon representation of the ACD of aB-crystallin with the crosslinked peptide (VLGDVIEVHGKHEER) in yellow. The reagent lysine (K103) is arranged near the edge of the hydrophobic binding groove. (d) Ribbon representation of the FAK FAT-a-crystallin (ACD) domain complex. The distance between the lysines crosslinked by the DSS reagent (K933 and K103) is B12 . (e) Surface representation of the FAK FAT-ACD complex (red: negative charges; blue: positive charges; grey: neutral). FAK FAT domainPDB: 1K40 and ACD modelderived fromPDB: 2KLR.

Supplementary Fig. 10ac) could inuence FAK signalling in cardiomyocytes. We examined FAK protein levels in extracts obtained from non-stretched and stretched cardiomyocytes with an antibody specic to FAK C terminus. A marked reduction of FAK protein level was observed in the stretched, but not in the non-stretched NRVMs that were silenced for aB-crystallin (Fig. 8a). This reduction was accompanied by an increase in the cleavage band of B35 KDa in size when compared with their non-stretched counterparts (Fig. 8a).

On the basis of previous data28, we reasoned that the reduction of FAK in the stretched NRVMs depleted of aB-crystallin could be related to calpain-mediated proteolysis of FAK. To examine this more directly, we tested whether calpain 2 could perform proteolysis on puried recombinant full-length FAK (Supplementary Fig. 10d). In fact, in the presence of calcium, calpain 2 displayed signicant proteolytic activity towards FAK, as indicated by the reduction of the 125 kDa band and by the generation of the 35 kDa cleavage band of FAK in the in vitro assay (Fig. 8b). However, when the solution was rst incubated with His-CryAB, we observed a reduction in the FAK cleavage by calpain, indicating that aB-crystallin can function directly to impair the proteolysis of FAK by calpain. To further test this notion, we performed assays with the mutant CryAB-K92A, which is defective in binding FAK. Consistent with the hypothesis that an interaction with aB-crystallin is needed to protect FAK from proteolysis by calpain, the mutant CryAB-K92A was unable to impair FAK degradation by calpain (Fig. 8c). We then examined whether the pharmacological inhibition of calpain

could inuence the FAK protein levels in cardiomyocytes depleted of aB-crystallin. Notably, treatment with the calpain inhibitor E64 abolished the reduction of FAK protein levels in the stretched cardiomyocytes depleted of aB-crystallin (Fig. 8d).

Further evidence for our model was provided by experiments with cardiomyocytes overexpressing myc-FAK or myc-FAKY925F mutant. Western blot performed with anti-myc antibody showed a marked reduction in the amount of myc-FAK-Y925F, but not in myc-FAK, in cardiomyocytes subjected to cyclic stretch. Pretreatment with E64 prevented the reduction of the myc-FAK-Y925F mutant, which is consistent with myc-FAK mutant reduction in stretched cardiomyocytes requiring calpain activity (Fig. 8e). These data support the notion that the lack of interaction with aB-crystallin makes FAK susceptible to proteolysis by calpain in the stretched cardiomyocytes.

FAK depletion by aB-crystallin silencing elicits apoptosis. Since aB-crystallin and FAK may inuence the viability and the hypertrophic response of cardiomyocytes to stress13,19, NRVMs depleted of aB-crystallin were analysed after prolonged (12 h)

mechanical stress. Non-stretched cardiomyocytes depleted of aB-crystallin showed a minor reduction in cell viability, but no change in cell morphology (Fig. 9a). However, aB-crystallin knockdown led to an enhanced apoptosis and an elongated phenotype, instead of the typical hypertrophy, in stretched cardiomyocytes (Fig. 9a,b; Supplementary Fig. 11a). We next explored whether exogenous expression of myc-FAK might

NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159

Pull-down

b c

Pull-down

GST-cTERM

Pull-down

GST-cTERM

GST-CryAB

cTERM

cTERM-H1025N

cTERM-V1029N

cTERM-L1035N

cTERM-L1021N

cTERM-V932N

cTERM-I936N

CryAB

CryAB-L89N

CryAB-V91N

CryAB-V93N

CryAB-L137N

CryAB-V91N/

V93N

CryAB-V91N/

V93N/ L137N

34.4 kDa 50 kDa

6His-tag 6His-tag 6His-tag

CryAB

B-

crystallin FAK

FAK FAT

ACD

23.5 kDa 60 kDa

FAK

K92

Y925 Y925

K92

L94

E99

Pull down

GST-

cTERM

GST-

p-cTERM

GST-p-cTERM

CryAB

pY925

L94

GST-cTERM

CryAB

CryAB-K92A

K92

5 E99

p925-FAK 6His-tagFAK

Normalized LS and RI

60 kDa 23.5 kDa 60 kDa

60 kDa

FAK

g h

108

107

Molar mass LSRI

Molar mass (Da)

1.00.80.60.40.20.0

Molar mass (Da)

104

103

1.00.80.60.40.20.0

107

106

105

Normalized LS and RI

106

105

104

10 15 20 25 30 35 40 45 10 15 20 25 30 35 40 45

Figure 7 | Mutagenesis and computational analysis of the FAK-cTERM and CryAB complex interface. (a) Anti-6His-tag and anti-aB-crystallin immunoblots from GST-CryAB pull-down assay performed with FAK His-cTERM and mutants as indicated (n 4). (b,c) Anti-6His-tag and anti-FAK

immunoblots from GST-cTERM pull-down assay performed with His-CryAB and mutants as indicated (n 4). (d) FAK FAT-ACD complex model

showing the proximity of FAK FAT Y925 (sticks) to the K92 of the ACD. The detailed image shows FAK Y925 docked onto a pocket delimited by K92 and L94 of the b4 strand and E99 of the b5 strand of the ACD. (e) Model showing the rearrangement of a-crystallin K92 towards the phospho-phenolic ring of the Y925-phosphorylated FAK FAT domain. (f) Left panel shows anti-pY925-FAK and anti-FAK immunoblots from GST-cTERM and its

Src-phosphorylated form (GST-p-cTERM). Right panel shows representative (n 4) anti-6His-tag and anti-FAK immunoblots from GST-cTERM or

GST-p-cTERM pull-down assay performed with His-CryAB and His-CryAB-K92A. (g) SEC-MALS chromatogram of His-CryAB-K92A shows that this protein elutes as a single peak of 6402 kDa. (h) SEC-MALS prole of His-CryAB-K92A and His-cTERM mixture. His-CryAB-K92A and His-cTERM elute as isolated proteins. The rst peak corresponds to His-CryAB-K92A (6006 kDa) and the second to His-cTERM (321 kDa).

Time (min) Time (min)

rescue the viability of stretched cardiomyocytes depleted of aB-crystallin. As expected, FAK levels were increased in NRVM cultures transfected with the myc-FAK construct (Fig. 9c). The purity of B97% and the low transfection rate of non-myocytes (o0.02%) in this culture, as estimated by the transfection rates of pEGFP-N1 (Supplementary Table 2), suggest that the

cardiomyocytes contribute substantially to the enhanced expression of FAK in cultures transfected with myc-FAK. Notably, the overexpression of a myc-FAK construct rescued the viability of the stretched NRVMs depleted of aB-crystallin (Fig. 9c; Supplementary Fig. 11a), which is consistent with the notion that the reduction of FAK expression is a critical

10 NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159 ARTICLE

b c

siGFP

siB-crystallin

CryAB

ST NS ST 0 50 50 50 (nM)

Calpain 2:

CryAB K92A

CryAB WT

170 0 50 50 (nM)

FAK FAK

Calpain 2:

170

130

72 55

43 34

130

72 55 43

170

130

72 55 43

FAK

Cleavage

product Cleavage

product

Cleavage

product

1mM CaCI2 1mM CaCI2

E64+SiB-crystallin

SiB-crystallin

E64

Vehicle

d e

SiB-crystallin

E64

Vehicle

E64+SiB-crystallin

Vehicle

myc-FAK

myc-FAK + E64

myc-FAK Y925F

+ E64

NS ST

125 kDa

23.5 kDa

myc-tag

GAPDH

FAK

125 kDa

B-crystallin

37 kDa

Figure 8 | Depletion of aB-crystallin downregulates FAK by calpain-mediated proteolysis. (a) Anti-FAK immunoblots of extracts from NS or ST (2 h) NRVMs transfected with RNAi against aB-crystallin (siaB-crystallin) or GFP (siGFP). Shown are FAK (B125 kDa) and its cleavage band (B35 kDa; n 6).

(b) Anti-FAK immunoblot from in vitro calpain cleavage assay performed with FL-FAK (full-length FAK) and CryAB-WT. The assay was performed by incubating FAK in the cleavage buffer; FAK with calpain 2 and 1 mM CaCl2; and FAK with His-CryAB-WT, calpain 2 and 1 mM CaCl2 (n 6). (c) FAK

immunoblot from in vitro calpain assay performed with FAK and CryAB-K92A. Shown are FAK; FAK with calpain 2 and 1 mM CaCl2; FAK with His-CryABK92A, calpain 2 and 1 mM CaCl2; and FAK with His-CryAB-WT, calpain 2 and 1 mM CaCl2 (n 6). (d) Attenuation of FAK proteolysis by E64 in

NRVMs depleted of aB-crystallin by RNAi. Shown are representative (n 6) anti-FAK and anti-aB-crystallin immunoblots of NS and ST (2 h) NRVMs

depleted of aB-crystallin or not, treated with vehicle (Lipofectamine 2000, Invitrogen), or E64. (e) Anti-myc-tag and anti-GAPDH immunoblots of extracts from stretched (2 h) NRVMs transfected with vehicle (Lipofectamine 2000, Invitrogen), pRc/CMV-myc-FAK (myc-FAK), pRc/CMV-myc-FAK E64,

pRc/CMV-myc-FAK-Y925F (myc-FAK-Y925F) or with pRc/CMV-myc-FAK-Y925F E64 (n 4). GAPDH served as loading control.

determinant of the reduced viability of stretched cardiomyocyte depleted of aB-crystallin. Additional experiments performed with non-stretched and stretched NRVMs showed that treatment with the calpain inhibitor E64 attenuated the enhanced apoptosis induced by aB-crystallin depletion in stretched NRVMs (Fig. 9d;

Supplementary Fig. 11a). These data suggest an important role for the calpain-induced FAK proteolysis in the reduction of the viability of stretched cardiomyocytes depleted of aB-crystallin.

We next examined whether the depletion of myocardial aB-crystallin potentiate the load-induced apoptosis in the left ventricle of TAC mice. To test this, we used RNAi to induce the depletion of myocardial aB-crystallin in mice subjected to a 7-day

TAC protocol. aB-crystallin silencing was conrmed by western blot in samples of mice left ventricle (Supplementary Fig. 11b). TAC mice depleted of aB-crystallin displayed increased apoptosis of the cardiomyocytes from the left ventricle, as compared with the TAC controls treated with an irrelevant siRNA (siGFP). The basal left ventricular apoptosis was similar between SO mice treated with siRNA targeted to aB-crystallin or with the irrelevant siRNA (Supplementary Fig. 11c). We next sought to examine whether FAK expressed in excess would attenuate the enhanced apoptosis in TAC mice depleted of aB-crystallin. To test this, we used a mice model with cardiac-specic transgenic FAK over-expression17. aB-crystallin silencing in the heart of transgenic

FAK mice was conrmed by western blot (Supplementary Fig. 11b). The apoptosis induced by TAC in the transgenic FAK mice silenced for aB-crystallin was reduced by B50% in comparison with that of the wild-type controls (Supplementary Fig. 11c), likely reecting the protection conferred by FAK.

DiscussionIn this study, we revealed a novel interaction between FAK and aB-crystallin mediated by FAK FAT domain that renders FAK less susceptible to proteolytic degradation by calpain. The experiments we report in isolated cardiomyocytes and in the left ventricle of rodents demonstrate that mechanical stress induces a rapid association of active FAK with aB-crystallin. aB-crystallin knockdown sensitizes mechanically stressed cardiomyocytes to calpain-dependent degradation of FAK and to apoptosis. Furthermore, FAK overexpression was shown to be capable of reversing the stretch-dependent apoptosis in cardiomyocytes depleted of aB-crystallin. Thus, the binding of active FAK to aB-crystallin represents a functionally relevant mechanism by which cells such as cardiomyocytes are enabled to cope with the burden of stress and maintain homeostasis.aB-crystallin is proposed to interact with exposed hydrophobic patches of partially unfolded proteins, preventing their aggregation4. The results of the in vitro and in vivo experiments presented here provide evidence that aB-crystallin can bind FAK in its folded state. This observation is of interest as emerging data suggest a role for aB-crystallin in binding and regulating folded signalling proteins14,15. A complete understanding of how an interplay with aB-crystallin results in regulation of FAK function requires knowledge of structural features of the aB-crystallin:FAK complex. In this study, we used protein chemical crosslinking combined with mass spectrometry and mutational analyses26,27 to show that a surface-exposed patch sited between a1 and a4 helices of the FAK FAT domain binds to the b4b8 hydrophobic groove of ACD. The hydrophobic nature of the FAK FAT a1a4

NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159

100

NS ST

siGFP siGFP

siB-crystallin

100

% Viable cells

75 50

% Apoptosis

75 50

siB-crystallin

siGFP

siB-crystallin

siGFP

siB-crystallin

siGFP

ST12h

c d

Vehicle

siB-crystallin

myc-FAK

siB-crystallin +

myc-FAK

25 20 25 20

5 0

% Apoptosis

* *

% Apoptosis

ST2h

* *

FAK

B-crystallin GAPDH

125 kDa23.5 kDa

37 kDa

Vehicle

myc-FAK

siB-crystallin

+ myc-FAK

Vehicle

siB-crystallin

E64+

siB-crystallin

E64

Vehicle

siB-crystallin

E64+

siB-crystallin

E64

ST2 h

ST12 h

Figure 9 | Depletion of aB-crystallin prompts mechanically stressed cardiomyocytes to apoptosis. (a) Bar graph represents the quantitative analyses of cell viability (trypan blue viability assay) from NS and ST NRVMs transfected with siGFP or siaB-crystallin. *Po0.0001 versus NS NRVMs transfected with siaB-crystallin. #Po0.0001 versus ST NRVMs transfected with siGFP. Data are presented as means.e.m. (n 6). P values were

determined by analysis of variance (ANOVA) and Bonferronis post hoc comparisons. Right panel shows immunouorescence imaging of NS and ST (12 h) NRVMs transfected with siaB-crystallin or siGFP and stained with rhodamine-conjugated phalloidin. Scale bar, 10 mm. (b) Bar graph represents the quantitative analyses of apoptosis (TUNEL assay) performed with NS and ST NRVMs transfected with siGFP or siaB-crystallin. *Po0.0001 versus NS

NRVMs transfected with siGFP. #Po0.0001 versus NS NRVMs transfected with siaB-crystallin. wPo0.0001 versus ST NRVMs transfected with siGFP. Data are presented as means.e.m. (n 6). P values were determined by ANOVA and Bonferronis post hoc comparisons. (c) Representative (from six

independent experiments) anti-FAK, -aB-crystallin and -GAPDH immunoblots of ST (2 h) NRVMs treated with vehicle, transfected with siaB-crystallin, pRc/CMV-myc-FAK (myc-FAK) or pRc/CMV-myc-FAK siaB-crystallin. Bar graph represents the quantitative analyses of apoptosis from stretched

NRVMs treated as indicated. *Po0.0001 versus NRVMs treated with vehicle. #Po0.0001 versus NRVMs transfected with siaB-crystallin. Data are presented as means.e.m. P values were determined by ANOVA and Bonferronis post hoc comparisons. (d) Bar graphs represent the quantitative analyses of apoptosis from NS and ST NRVMs treated as indicated (from four independent experiments). *Po0.0001 versus NRVMs treated with vehicle.

#Po0.0001 versus ST NRVMs transfected with siaB-crystallin. Data are presented as means.e.m. P values were determined by ANOVA and Bonferronis post hoc comparisons.

patch has been reported to favour its sequestration within intermolecular interfaces of FAK protein partners. For example, this patch binds to paxillin, targeting FAK to focal adhesions in an interaction that has to be disrupted to allow downstream signalling by FAK29. Concerning the b4b8 groove of aB-crystallin, it is key to the cross-binding of the aB-crystallin C-terminal tail in the homodimer interface of the aB-crystallin oligomers6,7. This suggests that aB-crystallin will not bind FAK efciently until it is dissociated in suboligomers, when the b4b8 groove is exposed. An interesting point to this observation is that our SEC-MALS data indicate that the aB-crystallin:FAK FAT domain complex assumes a conguration of a dimer of aB-crystallin bound to two FAK FAT domain monomers. This is consistent with the available structural information regarding aB-crystallin homodimer, which displays the equivalent b4b8 grooves of each ACD protomer exposed and free to bind to protein partners.

An important aspect of our biochemical data and computational model is the crucial contribution of FAK Y925 to the core interaction between the FAK C-terminal domain and a previously unrecognized pocket formed by K92 and L94 at the edge of the aB-crystallin b4 strand. This model provides a mechanistic

explanation for the enhanced binding of Y925-phosphorylated FAK C-terminal domain to aB-crystallin. Phosphorylation is predicted to strengthen the binding of the FAK C-terminal domain by repositioning the phenolic side chain of Y925 to within a salt-bridging distance of the positively charged side chain of aB-crystallin K92. These considerations suggest that aB-crystallin recognizes a specic module in the native FAK FAT domain that includes Y925 and other hydrophobic residues. Thus, these data support the concept that aB-crystallin functions not only as a component of the chaperone system but may also bind signalling molecules through specic interactions, clearly illustrating the utility of aB-crystallin as a scaffolding protein. In this context, it is interesting to note that the phosphorylation of FAK Y925 is known to favour an interaction with the adaptor protein Grb2, which has been implicated in the regulation of Erk1/2 pathway22. Therefore, it is likely that the interaction with aB-crystallin results in an attenuation of the Erk1/2 activation by FAK, unless the interactions of FAK with Grb2 and aB-crystallin are conned to distinct subcellular compartments. Consistent with this hypothesis, imaging data of the present study indicate that the stress-induced interaction between FAK and aB-crystallin might be restricted to sarcomeres and the perinuclear cytosol.

12 NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159 ARTICLE

Our results also help underscore the functional relevance of the interaction between FAK and aB-crystallin in the heart. Using cultured cardiomyocytes as a model system, we have demonstrated that the depletion of aB-crystallin results in a signicant reduction in FAK protein levels in mechanically stretched cells. Likewise, in mice hearts depleted of aB-crystallin, mechanical stress induced by pressure overload was accompanied by a reduction in FAK protein levels. In addition, our data indicate that the decline in FAK protein levels that accompanies the depletion of aB-crystallin is related to calpain-mediated proteolysis. Therefore, we reason that on the interaction with aB-crystallin, FAK can escape from degradation by calpain, a proteolytic enzyme that is activated and plays an important role in preserving cardiac and skeletal muscle homeostasis in stress conditions30. However, the method by which the interaction with aB-crystallin impedes calpain-mediated FAK proteolysis remains unclear. It is notable that calpain cleaves FAK after Ser-745, between the two C-terminal proline-rich regions28, which are located in an unstructured region not far from the FAK FAT domain. Thus, we speculate that the interaction of the FAK C-terminal domain with aB-crystallin may somehow impede the access of calpain to the FAK cleavage site.

FAK regulates many distinct cellular functions, including migration, proliferation, survival, cellcell signalling, mechanosensing, cell cycle and gene transcription31. Particularly in cardiomyocytes, FAK is activated by stimuli such as mechanical forces, neurohormonal factors and hypoxia to promote hypertrophic growth and survival32. Here we present evidence that the interaction with aB-crystallin and the consequent protection of FAK against calpain-induced proteolysis represents a critical event in the survival of cardiomyocytes under mechanical stress. As such, pharmacological inhibition of calpain attenuated FAK degradation and apoptosis in stretched cardiomyocytes that were also depleted of aB-crystallin.

Moreover, overexpression of FAK either in cultured cardiomyocytes or in the left ventricle of cardiac-specic FAK transgenic mice blunted the stress-induced apoptosis of cardiomyocytes depleted of aB-crystallin. It is noteworthy that the enhanced interaction between FAK and aB-crystallin may persist unabated up to the advanced stages of pathologic hypertrophy in mice. The consequence of this to the pathophysiology of maladaptive hypertrophy was not explored in the present study, but it could possibly slow the disease progression. In this context, one would expect that the depletion of cardiac aB-crystallin or FAK might predispose to the apoptosis of cardiomyocytes and further deterioration of load-induced pathologic hypertrophy. In fact, the ablation of aB-crystallin in the mice heart has been shown to enhance the maladaptive hypertrophy induced by pressure overload33. Similarly, FAK depletion seems to predispose to eccentric hypertrophy and heart failure34 in response to pressure overload in mice, although it has also been shown by an independent group18 that depletion of FAK attenuates the load-induced hypertrophy in the mice heart. Furthermore, by identifying the interaction of FAK with aB-crystallin through the FAT domain, our study also provides a rationale for the understanding of the regulation of FAK activity by FRNK (FAK-related nonkinase), a separate protein comprising the C terminus of FAK, which is expressed under certain pathophysiological conditions to act as a dominant negative regulator of FAK activity35. We anticipate that an enhanced FRNK expression during stress conditions might compete with FAK for the binding with aB-crystallin, therefore prompting

FAK to degradation by calpain, which might negatively impact the viability of cardiomyocytes subjected to mechanical stress. This hypothesis is supported by evidence36 indicating that overexpression of FRNK results in cardiomyocytes apoptosis.

Moreover, mice with cardiac-restricted expression of FRNK exhibit remarkably similar developmental and hypertrophic defects as mice with similar temporal deletion of FAK34.

Thus, the protection of FAK from calpain-induced cleavage provides a mechanism that enables the cardiomyocytes to survive under stress conditions. Importantly, in addition to mechanical stress, FAK interaction with aB-crystallin was demonstrated to occur in response to oxidative stress, indicating that this mechanism may be relevant to the protection of cardiomyocytes to distinct forms of stress. Relevant to this, calpain is involved in the degradation of unfolded sarcomeric and cytoskeletal proteins of cardiomyocytes in stress conditions and disease states30. However, calpain activation may also target signalling molecules, including FAK. It is therefore reasonable to posit that the interaction between FAK and aB-crystallin represents an essential strategy to protect FAK and potentially other signalling molecules during stress conditions in which the activation of calpain is necessary for cardiomyocyte homeostasis.

Methods

Cultured NRVMs. Neonatal male rats (Wistar; 12-day old) were handled in compliance with the principles of laboratory animal care formulated by the Institutional Committee for Ethics in Animal Research (State University of Campinas). Each primary culture of NRVMs was prepared using about 70 neonatal rat hearts. After the decapitation, the hearts were rapidly excised, rinsed in cold saline solution and minced coarsely. The myocytes were isolated by digestion with pancreatin and type-2 collagenase and then puried on a discontinuous Percoll gradient, suspended in plating medium containing 10% horse serum, 5% fetal bovine serum and 0.5% penicillinstreptomycin, and plated in type-I collagen Bioex plates or in Petri plates (25 mm) and incubated for 44 h under 95% air5% CO2. The medium was then replaced by serum-free DMEM and the cells were incubated for 24 h before use16. The purity of cardiomyocyte cultures was estimated using 40,6-diamidino-2-phenylindole (DAPI) and anti-a-actinin antibody (1:100) for staining of nuclei and the cardiomyocyte sarcomeres, respectively. The percentage of cardiomyocytes was B97%. NRVMs cultured in

Bioex plates were stretched in a Flexercell FX-3000 strain unit to 10% of their resting length at a frequency of 1 Hz (0.5 s stretch/0.5 s relaxation) for variable periods of time. At the conclusion of the experimental protocols, cells were scraped from membranes and lysed for immunoblottings or xed for analysis by uorescence microscopy.

NRVM treatment. To obtain myc-FAK or myc-FAK-Y925F overexpression, NRVMs were transfected with 500 ng ml 1 of pRc/CMV-myc-FAK or pRc/CMV-myc-FAK-Y925F constructs. Given that myc and FAK are both expressed as endogenous proteins, we used EGFP construct, DAPI and anti-a-actinin staining to access the efciency of gene transfer in neonatal cardiomyocytes. We obtained

B20% of transfected NRVMs. To determine whether interaction with aB-crystallin inhibits the cleavage of FAK by calpain, NVMRs were treated with 10 mM of the calpain inhibitor E64 (N-[N-(L-3-trans-carboxyirane-2-carbonyl)-L-leucyl]-agmatine; [1-[N-[(L-3-trans-carboxyoxirane-2-carbonyl)-L-leucyl]amino]-4-guanidino-butane] (Sigma-Aldrich). To deplete aB-crystallin, NRVMs were transfected with 200 ng ml 1 of a specic siRNA using the Lipofectamine 2000 reagent (2 ml ml 1;

Invitrogen)37. NRVMs transfected with siRNA against GFP was used as control. The primers used for the in vitro synthesis of the siRNAs are displayed in Supplementary Table 3.

Isolation of adult rat ventricular myocytes. Adult cardiomyocytes were isolated from male Wistar rats (78-week old) after SO or TAC (2 h), anaesthetized with a mixture of ketamine (30 mg kg 1) and xylazine (5 mg kg 1). Cardiomyocytes were isolated by enzymatic dissociation of the whole heart on a Langendorff apparatus. Hearts were excised in ice-cold Tyrodes solution and perfused retrogradely with the oxygenated solutions: (1) modied Tyrodes solution containing1.0 mM Ca2 (2 min), (2) Ca2 -free Tyrodes solution (23 min), (3) Ca2 -free Tyrodes solution containing 0.2 Wnsch U ml 1 of collagenase (Sigma, USA) for 20 min and (4) followed by washing in Kruftbrhe (KB) solution for 5 min.

Digested atrium and ventricles were cut off and minced in KB solution, pipetting to dissociate the cells, then ltered through a nylon mesh (200 mm pore size) to remove big pieces of undigested tissues. Isolated cells were rinsed in KB solution and allowed to settle by gravity three times to remove debris and non-cardiomyocytes. Resuspended cells in KB solution were loaded above the top layer of a Percoll gradient formed by 20%, 40% and 70% Percoll and centrifuged at 100g for 20 min to further purify cardiomyocytes. Puried myocytes after Percoll gradient separation were homogenized in lysis buffer and used in western blotting and immunoprecipitation protocols. Modied Tyrodes solution contained: 10 mM HEPES, 105 mM NaCl, 0.6 mM KH2PO4, 5.4 mM KCl, 6 mM NaHCO3, 0.6 mM

NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159

NaH2PO4, 1 mM MgCl2, 5 mM KHCO3, 1 mM CaCl2, 20 mM taurine and5 mM glucose (pH 7.35NaOH), and KB solution: 10 mM HEPES, 20 mM KCl, 1 mM MgCl2, 70 mM K -glutamate, 10 mM KH2PO4, 25 mM glucose, 20 mM taurine, 10 mM b-hydroxybutyric acid, 0.1% albumin and 0.5 mM EGTA(pH 7.25KOH)38.

Mice. Swiss mice (68-week old) were handled in compliance with the principles of laboratory animal care formulated by the Animal Care and Use Committee of the State University of Campinas. Procedures such as jugular vein catheterization, aortic banding, echocardiographic examination and arterial vessel catheterization for blood pressure monitoring were performed under anaesthesia with a mixture of ketamine (100 mg kg 1) and xylazine (5 mg kg 1). Aortic banding was performed by TAC by tying a 7-0 nylon (Dalon; B.Braun Melsungen AG,

Germany) suture ligature against a 27-gauge needle. SO animals were prepared in the same way, except for the aortic constriction. For the blood pressure monitoring, the right carotid and the right femoral arteries were cannulated with ame-stretched PE-50 polyethylene tube. Blood pressure in the carotid and femoral arteries was simultaneously recorded for a 10-min period to determine the transconstriction systolic gradient. The recordings were obtained from a COBE transducer (Arvada, USA), connected to a GP4A Stemtech amplier (Stemtech, USA). The amplier output was connected to an analogue-to-digital board and this to a computer loaded with WINDAQ-PRO Data Acquisition software (DATAQ Instruments, USA), for continuous blood pressure monitoring. Pulsatile blood pressure from carotid and femoral catheters was recorded in individual channels and sampled at 100-Hz39,40.

Immunoprecipitation. NRVMs or tissues homogenized in lysis buffer were normalized (500 mg) and incubated with anti-aB-crystallin polyclonal (2 mg; Santa Cruz Biotechnology), anti-myc-tag monoclonal (2 mg; Sigma-Aldrich) or with anti-FAK-c20 (2 mg; Santa Cruz Biotechnology) polyclonal antibodies. The proteins were collected after addition of 20 ml of protein A-sepharose beads (GE Health-care). Western blots of the immunoprecipitates were performed with anti-FAK-c20 (1:1,000; Santa Cruz Biotechnology), anti-p397FAK (1:1,000; Invitrogen), anti-p925FAK (1:1,000; Santa Cruz Biotechnology), anti-aB-crystallin (1:1,000; Santa

Cruz Biotechnology) or anti-myc (1:2,500; Sigma-Aldrich) antibodies. Quantication of bands was performed using Image J (Image Processing and Analysis in JAVA-program/National Institutes of Health NIH). The uncropped blots are displayed in Supplementary Fig. 12.

Microscopy analysis. Fixed cells or tissues were blocked with 1% bovine serum albumin, 0.1% Triton-X and 50 nM glycine in 0.1 M phosphate-buffered saline (PBS) on ice. Then, samples were incubated with anti-FAK c-20 (1:100) and subsequently with anti-aB-crystallin (1:100) antibody overnight at 4 C.

Next, Alexa Fluor-488-conjugated goat anti-rabbit (1:200) and Alexa Fluor-568-conjugated goat anti-mouse (1:200) secondary were used at room temperature, followed by Alexa Fluor-647 phalloidin (1:50; Invitrogen). Alternatively, NRVMs were incubated with rhodamine-conjugated phalloidin (1:50; Molecular Probes) at room temperature for 2 h. Slides were then mounted with VectaShield with DAPI. Confocal images were obtained with a LEICA TCS SP8 microscope. Fluorescence images were obtained with a LEICA DM4000 uorescence microscope.

Protein cloning and purication. The pGEX2T constructs encoding the GST FERM (GST-FERM; B70 kDa), GST-Kinase (B55 kDa) and GST-FAK C terminus (GST-cTERM; B60 kDa) fusion proteins were described previously26. The GSTCryAB construct (1175 aa; B49.5 kDa) encoding the full-length protein aB-crystallin was cloned in the BamHI site of pGEX2T. 6xHis-tagged aB-crystallin (His-CryAB) was obtained by cloning the aB-crystallin sequence between the NdeI and XhoI sites of pET-15b (1175 aa; B23 kDa), while the His-cTERM construct was obtained by cloning its sequence into the NdeI and BamHI sites of pET-15b (7651,052 aa; B34 kDa). The cloning primers are listed in Supplementary Table 3.

Plasmid encoding the c-Src kinase domain containing a YopH phosphatase was provided by Dr John Kuriyan. Plasmid encoding the wild-type myc-FAK construction (pRc/CMV-myc-FAK) was provided by Dr Steven K. Hanks and its mutated form was obtained by site-directed mutagenesis. The 6xHis-tagged constructions were expressed in E. coli Bl21 (DE3) C41 cells and puried by afnity chromatography (HisTrap HP 5 ml, GE Healthcare Bio-Sciences, Piscataway, NJ, USA). Next, an ion exchange chromatography was performed using buffer A 50 mM Tris, pH 7.5, 50 mM NaCland buffer B50 mM Tris, pH 7.5, 1 M NaCl (HiTrap Q HP and HiTrap SP HP, GE Healthcare)followed by a SEC in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl and 10% glycerol (HiLoad 16/60 Superdex 200 pg, GE Healthcare). The His-cTERM-His-CryAB complex was obtained by incubation of the previously puried His-cTERM and His-CryAB (1:1 molar; overnight; 4 C) followed by SEC in 50 mM phosphate, pH 7.5, and 150 mM NaCl buffer (HiLoad 16/60 Superdex 200 pg, GE Healthcare). The elution peak containing the copuried complex was isolated and used to the cross-linking experiments. See also Supplementary Information for full-length FAK purication.

Pull-down assay. His-CryAB coupled to Ni2 sepharose beads was incubated with full-length recombinant human GST-FAK (FL-FAK; Invitrogen, USA) and total extracts from non-stretched (NS) or stretched (ST) NRVMs, as indicated (see Results). GST-FERM, GST-KINASE, GST-cTERM and GST coupled to glutathione sepharose beads were incubated with His-CryAB or total extracts from NS or ST NRVMs. GST-CryAB coupled to glutathione sepharose beads was incubated with His-cTERM plus Src, His-cTERM plus Src and ATP or with His-cTERM-Y925F plus Src and ATP. GST-cTERM coupled to glutathione sepharose beads was incubated with His-CryAB or His-CryAB mutants. The precipitated pellets from the various assays were resolved by SDSPAGE and the membranes immuno-blotted with anti-aB-crystallin (1:1,000; Santa Cruz Biotechnology), anti-FAKc20 (1:1,000; Santa Cruz Biotechnology) or anti-6his-tag (1:2,500; Abcam) antibodies.

SEC-MALS. The oligomeric state of His-CryAB and His-FAK-cTERM in solution was analysed by SEC-MALS. The measurements were performed onKTA FPLC system (GE Healthcare) connected to a triple-angle static light-scattering detector miniDAWN TREOS and a refractive index detector Optlab-TrEX (Wyatt Technology, Santa Barbara, CA, USA). Puried proteins (1.5 mg ml 1; 500 ml) were applied to a Superdex200 HR 10/30 column (GE Healthcare) with a ow rate of0.5 ml min 1. Refractive index and light scattering were measured and the data were analysed using ASTRA 6 software after calibration with a bovine serum albumin standard.

FRET-FLIM. The pAmCyan1-C1 and pZsYellow1-C1 vectors were acquired from Clontech Laboratories. CFP was linked to YFP through a 15-aa-long linker (CFP-15aa-YFP) by inserting the YFP sequence between the SalI and BamHI sites of the pAmCyan1-C1 vector. The CFP-CryAB (wild-type or K92A) and YFP-cTERM (wild-type or Y925F) constructs were generated by subcloning their sequences into the BamHI site of the pAmCyan1-C1 and the pZsYellow1-C1 vectors, respectively. The primers used in cloning reactions are listed in Supplementary Table 3. The confocal and FLIM images of HEK 293T cells expressing diverse combinations of constructs were acquired with a Zeiss LSM780-NLO microscope (Carl Zeiss AG, Germany). The uorescence signal was excited with a Mai-Tai laser (Newport, USA) or with an argon laser (Lasos, Germany). The laser was focused by an EC Plan-Neouar 40/1.30 Oil DIC, and the same objective collected the signal. The

confocal images were acquired with the internal detectors, while the FLIM images were detected with a Hybrid-PMT in the non-descanned port. The FLIM system is composed of a hybrid-PMT, TCSPC cards and a controller computer (all obtained from Becker&Hickl, Germany). CFP was excited at 800 nm and YFP was excited at 488 nm. A short-pass lter at 470 nm was used to reject any YFP uorescence originating from the excitation at 800 nm.

In vitro Src-mediated phosphorylation of FAK C terminus. FAK-cTERM and FAK-cTERM-Y925F were dialyzed into phosphorylation buffer (50 mM HEPES, pH 7.0, 30 mM MgCl2, 1 mM dithiothreitol, 10% glycerol, 6 mM EGTA, 10 mM sodium orthovanadate and 10 mM sodium tartrate) and incubated with Src and ATP (1 mM) at 30 C for 90 min. The samples were analysed by MS/MS or incubated with SDS Laemmli sample buffer for western blotting with a phosphospecic antibody.

Site-directed mutagenesis. The Stratagene QuikChange II XL Site-Directed Mutagenesis Kit was used to produce the myc-FAK, FAK-cTERM domain and the CryAB mutants using the mutagenic primers listed in Supplementary Table 3. DNA samples from the plasmids pRc/CMV-Myc-FAK, pET15b-cTERM and pET15b-CryAB were used as templates. The mutations were conrmed by DNA sequencing.

CXMS. The co-puried His-cTERM-His-CryAB complex in 50 mM phosphate buffer, pH 7.5, and 150 mM NaCl was incubated with DSS (disuccinimidyl sube-rate; suberic acid bis; Sigma-Aldrich) dissolved in dry dimethylformamide in a nal 1:100 protein/DSS ratio for 2 h at room temperature26, which was followed by quenching with 50 mM Tris buffer, pH 7.6. The proteins were reduced (5 mM dithiothreitol, 25 min at 56 C), alkylated (14 mM iodoacetamide, 30 min at room temperature in the dark) and digested with trypsin (Promega). The samples were dried in a vacuum concentrator and reconstituted in 100 ml of 0.1% formic acid.

The digested sample was fractionated by using Oasis SPE cartridges (Waters) in elution steps of 5, 10, 15, 20, 25 and 70% acetonitrile aqueous solution. Each fraction was dried in evaporator for a nal volume of 10 ml and analysed by liquid chromatographyMS/MS performed using a Synapt mass spectrometer (Waters). For chromatography, solvent A was 0.1% aqueous formic acid and solvent B acetonitrile with 0.1% of formic acid. The gradient started with 3% solvent B, increasing linearly to 30 over 40 min, going to 70% for 10 min and going back to 3% after 55 min. The MS/MS data were acquired selecting the top three precursors with 1 s MS/MS acquisition time, selecting precursor ions of charge 42 . The

raw data were processed using Mascot Distiller 2.4.3 (Matrix Science) and the processed spectra were used for crosslinked peptide identication using Crux search for xlinks41 and XLinker Identier42 softwares. Crosslinked identication was performed using the following parameters: 0.1 Da tolerance for both MS and

14 NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159 ARTICLE

MS/MS, hydrolysed DSS for lysines and oxidation for methionine as variable modications, iodoacetamide as xed modication on cysteines and three missed cleavages, trypsin digestion. Manual validations of crosslinking assignment were performed on singly charged deconvoluted spectrum.

Molecular modelling, docking and dynamics simulation. The structure of ACD of Homo sapiens (PDB: 2KLR) was used as a template for the molecular modelling of the Mus musculus ACD (YASARA program). Molecular docking was conducted with the FAK FAT domain of Mus musculus (PDB: 1K40) and the a-crystallin model using the GRAMM-X web server for the molecular docking. To select the most appropriate solution from all the resulting docked conformations, a discrimination procedure based on the CXMS data was applied. The representative model for the complex between FAK FAT and ACD was selected, and submitted to energy minimization and MD simulation using YASARA. For that, all hydrogen atoms and other missing atoms from the model were created using force eld parameters obtained from YAMBER3. A simulation box was dened at 15 around all atoms of each complex. Protonation was performed based on the pH7.5. Cell neutralization was reached lling the box with water molecules and Na/Cl counter ions. A short MD was performed for the solvent adjustment, deleting water until the density of 0.997 g ml 1. A short steepest descent energy minimization was carried out until the maximum atom speed dropped below 2,200 m s 1.

Then, 500 steps of simulated annealing were performed with a temperature of 0 K. Finally, we performed an 8 ns simulation at 298 K and a non-bonded cutoff of7.86 A. A snapshot was saved every 25 ps. Simulation time was adjusted to stabilize the contacts between proteins.

In vitro calpain cleavage assay. Recombinant FAK (1 mg) was incubated in cleavage buffer (50 mM Tris-HCl, pH 7.4, 137 mM KCl and 1 mM MgCl2) with 50 nM puried recombinant calpain 2 (Calbiochem) plus 1 mM CaCl2 at 30 C for 30 min in the absence or presence of His-CryAB or His-CryAB-K92A (1:1 molar). The cleavage reactions were stopped by addition of 5 SDS Laemmli sample

buffer.

Cardiomyocyte viability. Non-stretched or stretched NRVMs transfected with siRNA against GFP or aB-crystallin were washed with PBS (pH 7.4), trypsinized for 5 min and centrifuged (12,000 g). The supernatant was aspirated and the cardiomyocytes were resuspended in a suitable volume of PBS (1 ml) with 0.8% trypan blue. Then, the cells were counted in a Neubauer chamber. The number of dead cells (blue cellsTrypan ( )) in a total of 300 cells was counted in each experi

ment. Viable cells were represented as the mean percentage of dead cells per total cells43.

TUNEL assay. Apoptotic nuclei of non-stretched or stretched cardiomyocytes treated with E64, RNAi or overexpression assays, or the myocardium of wild-type or FAK transgenic mice were assessed using In Situ Cell Death Detection Kit (Roche). Fixed cells or deparafnized myocardial sections were washed with PBS and incubated with Proteinase K 20 mg ml 1 for 10 min at 25 C. For the myocardium, the endogenous peroxidase activity was inhibited using 3% hydrogen peroxide. Subsequently, cells or the myocardium were incubated with TUNEL reaction mixture containing terminal deoxy-nucleotidyl transferase and uorescein-dUTP for 1 h at 37 C. For positive control, samples were incubatedwith DNAse I before the addition of the TUNEL reaction mixture. For negative control, samples were incubated only with uorescein-dUTP. Cells were mounted with Vectashield with DAPI and then examined by uorescence microscope43. Heart tissues were incubated with substrate mixture of 0.5 mg ml 1 3,30-diaminobenzidine (Sigma, St Louis, MO, USA) and 0.3% hydrogen peroxide, according to manufacturers instruction. The sections were counterstained with Harris haematoxylin and then examined by light microscope.

Statistical analyses. Analysis of variance and Bonferronis post hoc comparisons were used to compare the actual densitometric data from the immunoblots or the quantication of cell viability or apoptosis in the respective experimental groups. Po0.05 was considered to be statistically signicant. The data are presented as the means.e.m. for the indicated number of experiments. When presenting the data, we normalized the densitometric values and dened the mean for the control group as 100.

References

1. Vabulas, R. M., Raychaudhuri, S., Hayer-Hartl, M. & Hartl, F. U. Protein folding in the cytoplasm and the heat shock response. Cold Spring Harb. Perspect. Biol. 2, a004390 (2010).

2. Chen, B., Retzlaff, M., Roos, T. & Frydman, J. Cellular strategies of protein quality control. Cold Spring Harb. Perspect. Biol. 3, a004374 (2011).

3. Portbury, A. L., Willis, M. S. & Patterson, C. Tearin up my heart: proteolysis in the cardiac sarcomere. J. Biol. Chem. 286, 99299934 (2011).

4. Basha, E., ONeill, H. & Vierling, E. Small heat shock proteins and alpha-crystallins: dynamic proteins with exible functions. Trends Biochem. Sci. 37, 106117 (2012).

5. Bagneris, C. et al. Crystal structures of alpha-crystallin domain dimers of alphaB-crystallin and Hsp20. J. Mol. Biol. 392, 12421252 (2009).

6. Jehle, S. et al. N-terminal domain of alphaB-crystallin provides a conformational switch for multimerization and structural heterogeneity. Proc. Natl Acad. Sci. USA 108, 64096414 (2011).

7. Braun, N. et al. Multiple molecular architectures of the eye lens chaperone alphaB-crystallin elucidated by a triple hybrid approach. Proc. Natl Acad. Sci. USA 108, 2049120496 (2011).

8. Bullard, B. et al. Association of the chaperone alphaB-crystallin with titin in heart muscle. J. Biol. Chem. 279, 79177924 (2004).

9. Golenhofen, N., Arbeiter, A., Koob, R. & Drenckhahn, D. Ischemia-induced association of the stress protein alpha B-crystallin with I-band portion of cardiac titin. J. Mol. Cell. Cardiol. 34, 309319 (2002).

10. Vicart, P. et al. A missense mutation in the alphaB-crystallin chaperone gene causes a desmin-related myopathy. Nat. Genet. 20, 9295 (1998).

11. Rajasekaran, N. S. et al. Human alpha B-crystallin mutation causes oxidoreductive stress and protein aggregation cardiomyopathy in mice. Cell 130, 427439 (2007).

12. Sun, L., Trausch-Azar, J. S., Ciechanover, A. & Schwartz, A. L. Ubiquitinproteasome-mediated degradation, intracellular localization, and protein synthesis of MyoD and Id1 during muscle differentiation. J. Biol. Chem. 280, 2644826456 (2005).

13. Morrison, L. E., Hoover, H. E., Thuerauf, D. J. & Glembotski, C. C. Mimicking phosphorylation of alphaB-crystallin on serine-59 is necessary and sufcient to provide maximal protection of cardiac myocytes from apoptosis. Circ. Res. 92, 203211 (2003).

14. Mao, Y. W., Liu, J. P., Xiang, H. & Li, D. W. Human alphaA- and alphaB-crystallins bind to Bax and Bcl-X(S) to sequester their translocation during staurosporine-induced apoptosis. Cell Death Differ. 11, 512526 (2004).

15. Liu, S., Li, J., Tao, Y. & Xiao, X. Small heat shock protein alphaB-crystallin binds to p53 to sequester its translocation to mitochondria during hydrogen peroxide-induced apoptosis. Biochem. Biophys. Res. Commun. 354, 109114 (2007).

16. Torsoni, A. S., Constancio, S. S., Nadruz, Jr W., Hanks, S. K. & Franchini, K. G. Focal adhesion kinase is activated and mediates the early hypertrophic response to stretch in cardiac myocytes. Circ. Res. 93, 140147 (2003).

17. Clemente, C. F. et al. Focal adhesion kinase governs cardiac concentric hypertrophic growth by activating the AKT and mTOR pathways. J. Mol. Cell. Cardiol. 52, 493501 (2012).

18. DiMichele, L. A. et al. Myocyte-restricted focal adhesion kinase deletion attenuates pressure overload-induced hypertrophy. Circ. Res. 99, 636645 (2006).

19. Cheng, Z. et al. Targeted focal adhesion kinase activation in cardiomyocytes protects the heart from ischemia/reperfusion injury. Arterioscler. Thromb. Vasc. Biol. 32, 924933 (2012).

20. Schaller, M. D. et al. Autophosphorylation of the focal adhesion kinase, pp125FAK, directs SH2-dependent binding of pp60src. Mol. Cell. Biol. 14, 16801688 (1994).

21. Calalb, M. B., Polte, T. R. & Hanks, S. K. Tyrosine phosphorylation of focal adhesion kinase at sites in the catalytic domain regulates kinase activity: a role for Src family kinases. Mol. Cell. Biol. 15, 954963 (1995).

22. Schlaepfer, D. D. & Hunter, T. Evidence for in vivo phosphorylation of the Grb2 SH2-domain binding site on focal adhesion kinase by Src-family protein-tyrosine kinases. Mol. Cell. Biol. 16, 56235633 (1996).

23. Deramaudt, T. B. et al. FAK phosphorylation at Tyr-925 regulates cross-talk between focal adhesion turnover and cell protrusion. Mol. Biol. Cell 22, 964975 (2011).

24. Carragher, N. O., Westhoff, M. A., Fincham, V. J., Schaller, M. D. & Frame, M. C. A novel role for FAK as a protease-targeting adaptor protein: regulation by p42 ERK and Src. Curr. Biol. 13, 14421450 (2003).

25. Jin, J. K. et al. Localization of phosphorylated alphaB-crystallin to heart mitochondria during ischemia-reperfusion. Am. J. Physiol. Heart Circ. Physiol. 294, H337H344 (2008).

26. Santos, A. M. et al. FERM domain interaction with myosin negatively regulates FAK in cardiomyocyte hypertrophy. Nat. Chem. Biol. 8, 102110 (2012).

27. Fioramonte, M. et al. Analysis of secondary structure in proteins by chemical cross-linking coupled to MS. Proteomics 12, 27462752 (2012).

28. Chan, K. T., Bennin, D. A. & Huttenlocher, A. Regulation of adhesion dynamics by calpain-mediated proteolysis of focal adhesion kinase (FAK).J. Biol. Chem. 285, 1141811426 (2010).29. Gao, G. et al. NMR solution structure of the focal adhesion targeting domain of focal adhesion kinase in complex with a paxillin LD peptide: evidence for a two-site binding model. J. Biol. Chem. 279, 84418451 (2004).

NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 15

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6159

30. Patterson, C., Portbury, A. L., Schisler, J. C. & Willis, M. S. Tear me down: role of calpain in the development of cardiac ventricular hypertrophy. Circ. Res. 109, 453462 (2011).

31. Frame, M. C., Patel, H., Serrels, B., Lietha, D. & Eck, M. J. The FERM domain: organizing the structure and function of FAK. Nat. Rev. Mol. Cell Biol. 11, 802814 (2010).

32. Franchini, K. G. Focal adhesion kinasethe basis of local hypertrophic signaling domains. J. Mol. Cell. Cardiol. 52, 485492 (2012).

33. Kumarapeli, A. R. et al. Alpha B-crystallin suppresses pressure overload cardiac hypertrophy. Circ. Res. 103, 14731482 (2008).

34. Peng, X. et al. Inactivation of focal adhesion kinase in cardiomyocytes promotes eccentric cardiac hypertrophy and brosis in mice. J. Clin. Invest. 116, 217227 (2006).

35. DiMichele, L. A. et al. Transient expression of FRNK reveals stage-specic requirement for focal adhesion kinase activity in cardiac growth. Circ. Res. 104, 12011208 (2009).

36. Heidkamp, M. C., Bayer, A. L., Kalina, J. A., Eble, D. M. & Samarel, A. M. GFPFRNK disrupts focal adhesions and induces anoikis in neonatal rat ventricular myocytes. Circ. Res. 90, 12821289 (2002).

37. Marin, T. M. et al. Shp2 negatively regulates growth in cardiomyocytes by controlling focal adhesion kinase/Src and mTOR pathways. Circ. Res. 103, 813824 (2008).

38. Zhang, Y. et al. Dedifferentiation and proliferation of mammalian cardiomyocytes. PLoS ONE 5, e12559 (2010).

39. Clemente, C.F. et al. Targeting focal adhesion kinase with small interfering RNA prevents and reverses load-induced cardiac hypertrophy in mice. Circ. Res. 101, 13391348 (2007).

40. Franchini, K. G., Torsoni, A. S., Soares, P. H. A. & Saad, M. J. A. Early activation of the multicomponent signaling complex associated with focal adhesion kinase induced by pressure overload in the rat heart. Circ. Res. 87, 558565 (2000).

41. McIlwain, S., Draghicescu, P., Singh, P., Goodlett, D. R. & Noble, W. S. Detecting cross-linked peptides by searching against a database of cross-linked peptide pairs. J. Proteome Res. 9, 24882495 (2010).

42. Du, X. et al. Xlink-identier: an automated data analysis platform for condent identications of chemically cross-linked peptides using tandem mass spectrometry. J. Proteome Res. 10, 923931 (2011).

43. Brar, B. K. et al. Cardiotrophin-1 can protect cardiac myocytes from injury when added both prior to simulated ischaemia and at reoxygenation. Cardiovasc. Res. 51, 265274 (2001).

Acknowledgements

This work was funded by Fundaao de Amparo Pesquisa do Estado de Sao Paulo

(FAPESP; Grants 2006/54878-3, 2007/55930-1, 2007/59442-1, 2008/53519-5, 2008/

57805-2, 2010/02628-9) and Conselho Nacional de Pesquisa (CNPq: Grants 304366/

2009-9, 475158/2010-5). We thank the access to equipment and assistance provided by

the National Institute of Science and Technology on Photonics Applied to Cell Biology

(INFABIC) at the State University of Campinas.

Author contributions

M.B.M.P., A.M.S. and K.G.F. conceived the research and designed the experiments.

M.B.M.P., A.M.S., A.C.C., S.R.C. and A.H.M.P. conducted the experiments. D.C.G.,

A.O.T.-C. and C.H.I.R. performed SEC-MALS experiments. A.M.S., A.A.d.T. and C.L.C.

performed FRET-FLIM experiments. M.B.M.P., A.M.S., A.R.F. and F.C.G. performed

crosslinking and MS experiments and analysis. A.M.S. and P.S.O. performed molecular

modelling and docking. M.B.M.P., A.M.S. and K.G.F. wrote and edited the manuscript.

M.B.M.P. and A.M.S. contributed equally to the work.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/

http://www.nature.com/naturecommunications

Web End =naturecommunications Competing nancial interests: The authors declare no competing nancial interests. Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/

http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Pereira, M. B. M. et al. aB-crystallin interacts with and prevents

stress-activated proteolysis of focal adhesion kinase by calpain in cardiomyocytes.

Nat. Commun. 5:5159 doi: 10.1038/ncomms6159 (2014).

16 NATURE COMMUNICATIONS | 5:5159 | DOI: 10.1038/ncomms6159 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

Word count: 13622

Show less

Abstract

Translate

Focal adhesion kinase (FAK) contributes to cellular homeostasis under stress conditions. Here we show that αB-crystallin interacts with and confers protection to FAK against calpain-mediated proteolysis in cardiomyocytes. A hydrophobic patch mapped between helices 1 and 4 of the FAK FAT domain was found to bind to the β4-β8 groove of αB-crystallin. Such an interaction requires FAK tyrosine 925 and is enhanced following its phosphorylation by Src, which occurs upon FAK stimulation. αB-crystallin silencing results in calpain-dependent FAK depletion and in the increased apoptosis of cardiomyocytes in response to mechanical stress. FAK overexpression protects cardiomyocytes depleted of αB-crystallin against the stretch-induced apoptosis. Consistently, load-induced apoptosis is blunted in the hearts from cardiac-specific FAK transgenic mice transiently depleted of αB-crystallin by RNA interference. These studies define a role for αB-crystallin in controlling FAK function and cardiomyocyte survival through the prevention of calpain-mediated degradation of FAK.

Details

Title

[alpha]B-crystallin interacts with and prevents stress-activated proteolysis of focal adhesion kinase by calpain in cardiomyocytes

Author

Pereira, Michelle B M; Santos, Aline M; Gonçalves, Danieli C; Cardoso, Alisson C; Consonni, Sílvio R; Gozzo, Fabio C; Oliveira, Paulo S; Pereira, Ana Helena M; Figueiredo, Alana R; Tiroli-cepeda, Ana O; Ramos, Carlos H I; De Thomaz, André A; Cesar, Carlos L; Franchini, Kleber G

Pages

5159

Publication year

2014

Publication date

Oct 2014

Publisher

Nature Publishing Group

e-ISSN

20411723

Source type

Scholarly Journal

Language of publication

English

DOI

https://doi.org/10.1038/ncomms6159

ProQuest document ID

1612284517

[alpha]B-crystallin interacts with and prevents stress-activated proteolysis of focal adhesion kinase by calpain in cardiomyocytes

Jump to:

Full Text

Abstract

Details

Suggested sources