New Approaches to Prevent LEOPARD Syndrome-associated Cardiac Hypertrophy by Specifically Targeting Shp2-dependent Signaling*

Background: In LEOPARD syndrome, hypertrophic cardiomyopathy develops because of hyperactivated prohypertrophic signaling. Results: Pharmacological inhibition of Shp2, focal adhesion kinase, Akt, or mammalian target of rapamycin counteracts the disease mechanism. Conclusion: Interventions at multiple levels of the signaling cascade effectively prevent cardiomyocyte hypertrophy. Significance: Identification of these novel targets will facilitate the development of highly specific therapies for LEOPARD syndrome. In LEOPARD syndrome (LS) patients, mutations in the protein tyrosine phosphatase Shp2 cause hypertrophic cardiomyopathy. The prohypertrophic effects of mutant Shp2 are mediated downstream by hyperactivation of mammalian target of rapamycin. Our goal was to further define the signaling cascade that is essential for the underlying pathomechanism, thus expanding the list of potential future therapeutic targets. Using cultured neonatal rat cardiomyocytes with adenoviral gene delivery and pharmacological inhibitors, we found that hypertrophy induced by a particularly aggressive LS mutation in Shp2 depends on hyperactivation of Akt and focal adhesion kinase as well as mammalian target of rapamycin. Dissecting domain-specific functions of Shp2 using double and truncation mutants, we determined that the hypertrophic effects of mutant Shp2 depend on the two SH2 domains and on an intact catalytic center. The latter finding prompted us to test the efficacy of a Shp2 inhibitor targeted directly at the catalytic pocket. This compound, PHPS1, effectively prevented mutant Shp2-induced hypertrophy. In summary, we identified three novel targets for pharmacological therapy of LS-associated cardiac hypertrophy. Of particular importance is the finding that intervention directly at the mutant Shp2 protein is effective because this would facilitate custom-tailored therapeutic approaches for patients carrying LS mutations in Shp2.

Despite many recent advances, hypertrophic cardiomyopathy (HCM) 2 in pediatric patients poses a particular therapeutic challenge because none of the current pharmacological approaches have been shown to prevent, attenuate, or reverse HCM in humans or impact the prognosis (1). Additionally, childhood-onset HCM is often associated with other congenital malformations, which may further complicate the treatment. We have focused our research efforts on unraveling the disease mechanism of HCM associated with LEOPARD syndrome (LS), which can lead to severe heart failure symptoms in infants already. LEOPARD is an acronym for the manifestations of this syndrome: multiple lentigines, electrocardiographic conduction abnormalities, ocular hypertelorism, pulmonic stenosis, abnormal genitalia, retardation of growth, and deafness (2).
Clinically and histologically, LS-associated HCM resembles severe forms of classic HCM (3). However, sarcomeric protein mutations could not be found in the affected families. Instead, mutations in the protein tyrosine phosphatase Shp2 (encoded by the gene PTPN11) are responsible in 90% of LS patients (4,5). This suggests that the LS-specific disease mechanisms may be very different from those in classic HCM. Therefore, custom-tailored therapeutic approaches for such "phenocopy" forms of HCM as seen in LS could be much more effective than standard treatment protocols.
Recently, substantial progress has been made in unraveling LS-specific signaling mechanisms that promote the associated HCM. Using various in vitro and in vivo models, multiple groups, including ours, discovered in parallel that LS mutations in Shp2 result in hyperactivation of signaling through Akt or mammalian target of rapamycin (mTOR) (6 -9). In two independent mouse models, administration of rapamycin rescued the HCM phenotype. However, the rescue was only complete in the milder model with late-onset HCM (7). In the more severe model with neonatal-onset HCM, cardiac hypertrophy was effectively reversed by rapamycin, but the functional deficit could not be improved (6). Therefore, benefits from rapamycin treatment may be very limited in severely affected pediatric patients with LS. Furthermore, the use of rapamycin or its derivatives may be impeded by the ensuing immunosuppression and other side effects, in particular negative inotropic effects (10). Consequently, the first goal of this study was to test whether other signaling elements upstream of mTOR could be equally effective at inhibiting LSassociated cardiac hypertrophy. Secondly, we wanted to dissect the structure-function relationship of mutant Shp2. Detailed knowledge of the roles of individual domains of the protein would facilitate the future design and development of pharmaceutical compounds to target the mutant Shp2 protein directly and, therefore, potentially have the highest efficacy with the least side effects.
To accomplish these goals, we used cultured neonatal rat cardiomyocytes (NRCM) with adenoviral vectors to express various Shp2 mutants and several pharmacological inhibitors. We focused on the LS mutation Q510E-Shp2, which is associated with a particularly aggressive form of biventricular HCM in pediatric patients and, therefore, is ideally suited for proofof-principle studies of disease mechanisms and treatment efficacy (11,12). Biochemically, the Q510E mutation confers dominant-negative effects similar to the effects of other LS mutations (6,13). We previously used the NRCM model system for in vitro assessment of the prohypertrophic effects of the LS mutation Q510E in Shp2 (6). As expected, we found that Q510E-Shp2 overexpression resulted in a robust increase in NRCM size (6). Because this approach increases total Shp2 protein levels, we excluded potential gene dose effects by also overexpressing WT Shp2 (6). WT Shp2 overexpression at equivalent levels did not alter NRCM size, consistent with our previous finding that transgenic WT-Shp2 overexpression in the mouse ventricle also does not induce HCM or any other cardiac abnormalities (14).
For this study, we first validated the NRCM model by confirming that the prohypertrophic effects of Q510E-Shp2 were mediated by mTOR hyperactivation, as seen previously in our mouse model (6). We then tested the roles of signaling proteins upstream of mTOR for promoting hypertrophy. We found that targeting focal adhesion kinase (FAK) or Akt effectively counteracts the prohypertrophic effects of Q510E-Shp2. Furthermore, we found that targeting either the SH2 domains or the catalytic center of Shp2 directly prevents hypertrophy induced by overexpression of Q510E-Shp2. This has not only important implications for the design of future therapies but also advances our insight into the molecular interactions underlying the pathogenesis of LS-associated HCM.
Plasmid and Adenovirus Construction-FVB/N mouse cardiac mRNA was used as template for WT-Shp2 cDNA. Truncation mutants and point mutations were generated using standard polymerase chain reaction methodology. In the truncation mutants, the 211 N-terminal amino acids of Shp2 that comprise the two SH2 domains of the protein were eliminated. After full-length sequencing of the constructs and comparison to the published C57/BL6 Shp2 mRNA sequence (NCBI database accession no. NM_011202.3, Swiss-Prot accession no. P35235.2), the cDNAs were inserted into the pAdTrack-CMV vector and subsequently used for adenovirus generation (AdEasy System, Stratagene). Adenovirus encoding ␤-galactosidase (␤-gal) was obtained from Cell Biolabs. Adenovirus encoding GFP-FAK-related nonkinase (FRNK) was a gift from Dr. Alan Samarel (15). As in our previous study, we confirmed that protein levels of the various Shp2 constructs were equivalent across groups using Western blot analyses (6). GFP-FRNK protein expression was confirmed by the presence of intracellular fluorescent signals as well as by Western blot analyses (data not shown).
mTOR Kinase Activity Assay-The enzymatic activity of mTOR was determined using an ELISA-based assay (K-LISA mTOR activity kit, EMD Chemicals). In short, mTOR protein was immunoprecipitated from NRCM lysates. Purified proteins were then incubated with a p70S6kinase (p70S6K)-GST fusion protein as substrate together with ATP. The phosphorylated substrate was detected with anti-phopspho-p70S6K (Thr-389) antibody, followed by detection with horseradish peroxidase-conjugated antibody and 3,3Ј,5,5Ј-tetramethylbenzidine substrate. Relative activity was determined by reading the absorbance at dual wavelengths of 450/540 nm.
Protein Content, Protein Synthesis Rate, and mRNA Expression-Protein content in NRCM lysates was determined colorimetrically (Bradford method, Bio-Rad) and expressed as total soluble protein yield per 1 million NRCM. Protein synthesis rates were measured using [ 35 S]methionine incorporation as described in Ref. 17. Quantitative RT-PCR was conducted as described previously (6). Primer sequences were as follows: GAPDH, agc ttg tca tca acg gga ag (forward) and ttt gat gtt agt ggg gtc tcg (reverse); ␣-skeletal actin, aat gag cgt ttc cgt tgc (forward) and atc ccc gca gac tcc ata c (reverse).
Data Normalization and Statistical Analysis-Data from mTOR activity assays, cardiomyocyte surface areas, and Western blot signal quantifications were normalized as follows. All measurements obtained from the respective control group were averaged, and then all data (including individual measurements from the control group) were divided by this average. In contrast, activities obtained from the WT group in the PTP activity assays (Fig. 7A) were set as 1.0 for each individual run. Unpaired Student's t tests were used for comparisons of two groups. One-way analysis of variance followed by Scheffé post hoc test was used for comparing multiple groups (StatPlus software). All values are given as mean Ϯ S.E. *, p Ͻ 0.05 versus the respective ␤-gal control group under the same conditions unless specified otherwise in the figure legend.

Characterization of the NRCM in Vitro
Model-NRCM are a well established and frequently used system for hypertrophic signaling studies (18). However, such cultures do represent a reductionist approach. Therefore, we wanted to confirm that increases in cell surface area are indeed accompanied by other classic characteristics of cell hypertrophy. Fig. 1A shows representative images of NRCM immunostained for troponin I expressing Q510E-Shp2 or Y279C-Shp2, a second LS-associated mutation in Shp2. For comparison, NRCM were also exposed to 50 M PE. NRCM cell surface areas were increased significantly by expression of the LS mutants as well as by PE (for quantification, refer to Figs. 2-7). Next, NRCM were scored for sarcomeric organization by a blinded observer. Scores were consistently higher in NRCM expressing LS mutants or exposed to PE (Fig. 1B). Next, we tested whether NRCM expressing LS mutants or treated with PE contained increased amounts of protein. Protein concentrations were increased significantly in lysates obtained from NRCM expressing Q510E-Shp2 or Y279C-Shp2, as well as from PE-treated NRCM (Fig.  1C). Consistently, rates of protein synthesis as reflected in [ 35 S]methionine incorporation were increased by Q510E-Shp2 or Y279C-Shp2 expression and PE treatment (Fig. 1D). Lastly, we determined mRNA expression levels of the ␣-skeletal actin isoform as a marker of the fetal gene program. Similar to PE treatment, both Q510E-Shp2 and Y279C-Shp2 expression increased ␣-skeletal actin expression (Fig. 1E).
mTOR Hyperactivation Is Necessary for Q510E-Shp2-induced Hypertrophy-Next, we confirmed that our in vitro model correctly recapitulated previous findings from in vivo Expression of Q510E-Shp2, Y279C-Shp2, or PE significantly increased the level of organization compared with ␤-gal. n ϭ 10 -12 images/group. *, p Ͻ 0.05 versus the ␤-gal control group under the same conditions. C, NRCM expressing either ␤-gal, Q510E-Shp2, Y279C-Shp2, or treated with PE were lysed, and the total soluble protein was quantified. Protein content/cell number was increased in the three hypertrophied groups compared with the ␤-gal control group. n ϭ 8 -12/group. *, p Ͻ 0.05 versus the ␤-gal control group under the same conditions. D, protein synthesis rates were determined using [ 35 S]methionine. NRCM expressing Q510E-Shp2, Y279C-Shp2, or treated with PE incorporated significantly more radiolabel compared with ␤-gal. n ϭ 10/group. *, p Ͻ 0.05 versus the ␤-gal control group under the same conditions. E, expression of ␣-skeletal actin was quantified using quantitative RT-PCR. NRCM expressing Q510E-Shp2, Y279C-Shp2, or treated with PE demonstrated increased mRNA expression of ␣-skeletal actin. n ϭ 4 per group. *, p Ͻ 0.05 versus the ␤-gal control group under the same conditions. models of LS-associated cardiac hypertrophy. We first determined the degree of mTOR hyperactivation in NRCM expressing Q510E-Shp2 compared with the ␤-gal-expressing control group. mTOR protein was extracted, and kinase activities were determined using an ELISA-based approach. In Q510E-Shp2expressing NRCM, mTOR kinase activity was increased significantly by 40% ( Fig. 2A).
To further support that mTOR is hyperactivated in the presence of Q510E-Shp2, Western blot analyses were used. Representative blots and signal quantifications are shown in Fig. 2, B and C. Phosphorylation levels of mTOR as well as its downstream effectors p70S6K and S6 were increased significantly by Q510E-Shp2 expression. Furthermore, we confirmed that mTOR is a necessary mediator of the effects of Q510E-Shp2 on NRCM size using pharmacological inhibition. Rapamycin prevented Q510E-Shp2-induced hypertrophy in a dose-dependent manner (Fig. 2D). At 0.2 M rapamycin, cell surface areas of Q510E-Shp2-expressing NRCM were completely normalized.
Akt1/2 Inhibition Ameliorates Q510E-Shp2-induced Hypertrophy-Next, we tested to which extent signaling proteins known to be upstream activators of mTOR were involved in the prohypertrophic effects of Q510E-Shp2. In particular, we tested the role of Akt activation. As shown in Fig. 3, A and B, the level of Akt phosphorylation was increased significantly in Q510E-Shp2-expressing NRCM. To test whether Akt hyperactivation is required for the prohypertrophic effects of Q510E-Shp2, we used Akt inhibitor VIII, which targets both Akt1 and Akt2. Akt1 and Akt2 may have partially redundant functions in the heart (19 -21). Therefore, we did not use an isoform-specific Akt inhibitor. At 2 and 4 M, Akt inhibitor VIII prevented Q510E-Shp2-induced hypertrophy (Fig. 3C).

FAK Hyperactivation Is Essential for Q510E-Shp2-induced
Hypertrophy-Our next step was to test the role of protein kinases further upstream of Akt, in particular FAK. First, we used immunohistochemistry to assess the level of FAK activation in NRCM. FAK phosphorylation was increased in Q510E-Shp2-expressing NRCM (Fig. 4, A and B), whereas total FAK and paxillin remained unchanged (E, F, I, and J). To confirm that the punctate increase in signal intensity correctly reflected FAK hyperactivation, we used a pharmacological approach. In the presence of the FAK/src inhibitor PP2, phospho-FAK signal intensities in Q510E-Shp2-expressing NRCM were similar to those of ␤-gal-expressing controls (Fig. 4, C and D). Expression of total FAK or paxillin was not altered by PP2 (Fig. 4, G, H, K, and L). Secondly, we used Western blot analyses to quantify the level of FAK activation (Fig. 4M). Q510E-Shp2 expression increased the ratio of phospho to total FAK signal more than 2-fold (Fig. 4N).
Subsequently, we measured NRCM cell sizes in the presence of PP2. As seen with rapamycin and Akt inhibitor VIII, we observed dose-dependent effects. At doses Ն 0.5 M, PP2 prevented the Q510E-Shp2-induced cell size increase (Fig. 4O). As an additional negative control, we used PP3, a very similar compound that does not inhibit Src family protein tyrosine kinases. PP3 did not affect cell sizes in ␤-gal or Q510E-Shp2 expressing NRCM (Fig. 4O).
For a more specific test, the endogenous FAK inhibitor FRNK was used in a separate set of experiments. GFP-tagged FRNK was coexpressed with Q510E-Shp2, which resulted in normal NRCM surface areas (Fig. 4P). Together with the PP2 data, this indicates that FAK is indeed a prohypertrophic signaling mediator activated by Q510E-Shp2 expression.
Furthermore, we tested whether the presence of PP2 normalized signaling further downstream using Western blot analyses. As shown in Fig. 4Q, PP2 decreased the levels of Akt, mTOR, and p70S6K phosphorylation to base-line levels or below, indicating that Akt and mTOR are likely to be mediators of the prohypertrophic signal downstream of FAK.

Fibroblast-specific Signals or Calcineurin Activation Do Not Play a Role for Q510E-Shp2-induced Hypertrophy-Because
NRCM cultures may be contaminated with cardiac fibroblasts, we also wanted to exclude that the hypertrophy induced was due to secondary effects. Therefore, we exposed normal NRCM to CM harvested from cardiac fibroblasts that had been infected with ␤-gal, WT-Shp2, or Q510E-Shp2 adenoviruses. In all groups, the presence of CM increased NRCM size, but there were no quantitative differences between groups (Fig.  5A). This indicated that fibroblast-specific effects were unlikely to play a role for the prohypertrophic effects of Q510E-Shp2.
Furthermore, we tested to which extent calcineurin inhibition could ameliorate Q510E-Shp2-induced NRCM hypertrophy. The calcineurin signaling pathway is not directly linked to FAK/Akt/mTOR signaling and, therefore, less likely to be involved in the underlying pathomechanism. However, a recent study has shown that calcineurin does play a role in hypertrophy caused by mutations in RAF1 that were identified in a subset of LS patients (22). In our model, the presence of the calcineurin inhibitor cyclosporin A did not affect Q510E-Shp2induced hypertrophy (Fig. 5B), indicating that calcineurin signaling is unlikely to play a major role.
The Role of ERK Signaling in Q510E-Shp2-induced Hypertrophy-Because ERK hyperactivation has been shown recently to play a role in other LS models (23,24), we quantified the levels of ERK1/2 activation in NRCM expressing Q510E-Shp2 or Y279C-Shp2. In our serum-free cultures, no significant differences were detected, but the presence of Q510E-Shp2 or Y279C-Shp2 did result in non-significant reductions in ERK1/2 activation (Fig. 5, C and D). We also wanted to include a gainof-function mutation in Shp2 for comparison and chose the Noonan syndrome-associated Q79R-Shp2 mutation. This mutation renders the Shp2 protein constitutively active and thereby stimulates ERK signaling (14,16). As shown in Fig. 5, C and D, NRCM expressing Q79R-Shp2 displayed a clear trend toward ERK1/2 hyperactivation. This is consistent with a number of previous studies, including our own (6,13,16,25), demonstrating that under conditions of serum or specific growth factor stimulation, cultured cells expressing LS mutants of Shp2 exhibit reduced ERK activation, whereas cells expressing gain-of-function mutants of Shp2 display increased ERK activation.
We also tested whether ERK inhibition affected NRCM size in our model. Fig. 5E shows that the mitogen-activated protein kinase/ERK kinase 1 (MEK1) inhibitor U0126 partially reduced the size of Q510E-Shp2-expressing NRCM. However, Q510E-Shp2-expressing NRCM remained significantly larger in size compared with the ␤-gal group.
Taken together, our data therefore indicate that ERK signaling is unlikely to play a major role in the hypertrophic response to Q510E-Shp or Y279C-Shp2 expression. We suspect that the partial effect of U0126 on NRCM size may be due to cross-talk between the ERK and Akt pathways.
The Role of SH2 Domains for the Prohypertrophic Effects of Q510E-Shp2-Because we found multiple feasible therapeutic targets upstream of mTOR, we considered the possibility that intervention directly at the level of Shp2 could also be effective. How this could be accomplished has not yet been explored. To better define the structure-function relationship of Shp2, we examined the relative contributions of the different Shp2 protein domains to the prohypertrophic effects. First, we tested the necessity of the presence of the two SH2 domains for the path- ogenic effects. New cDNA constructs were generated, encoding only the PTP domain and the linker region between the C-terminal SH2 domain and the PTP domain but lacking both SH2 domains. One construct had no other mutations introduced (termed WT-trunc), whereas the other construct also contained the Q510E mutation (termed Q510E-trunc).
The truncated proteins were expressed in NRCM at levels equivalent to the levels of all other Shp2 constructs. The truncated Shp2 proteins were extracted by immunoprecipitation using an antibody against the C-terminal FLAG tag. The purified proteins were incubated with a tyrosine-phosphorylated src peptide to determine in vitro phosphatase activity. PTP activity of WT-trunc was close to normal, and the activity of Q510E-trunc was reduced significantly and not different from that of full-length Q510E-Shp2 (Fig. 6A).
Subsequently, NRCM surface areas were determined in a new set of cultures. Expression of WT-trunc had no effect on NRCM size, similar to full-length WT-Shp2 overexpression. However, deletion of the two SH2 domains of the Q510E mutant abolished the prohypertrophic effect (Fig. 6B). Lastly, we tested the level of Akt activation in the presence of the two truncation mutants. Neither expression of WT-trunc nor of Q510E-trunc affected the degree of Akt phosphorylation, consistent with the NRCM size data (Fig. 6, C and D).
The Role of the Catalytic Center for the Prohypertrophic Effects of Q510E-Shp2-Next, we evaluated the role of the catalytic center for the prohypertrophic effects of mutant Shp2. To completely destroy any possibly remaining enzymatic activity of the Q510E-Shp2 protein, we generated a double mutant containing both the C459S and the Q510E mutations. The C459S mutation completely abolishes Shp2 phosphatase activity by replacing the cysteine residue essential for nucleophilic attack of the phosphate group of the substrate. To ensure that the C459S substitution itself did not have any hypertrophic effects, we also introduced the C459S mutation into the full-length WT sequence.
As before, we first assessed PTPase activities of the mutant proteins in vitro (Fig. 6A). As expected, the C459S mutation significantly reduced enzymatic activity of the Shp2 protein.
The double mutant C459S-Q510E displayed similarly low phosphatase activity. Next, NRCM surface areas after infection with the respective constructs at equivalent levels were determined. Expression of C459S-Shp2 did not affect NRCM size (Fig. 6B). Notably, combination of the C459S substitution with the Q510E mutation completely abolished the prohypertrophic effects of Q510E-Shp2 (Fig. 6B).
It has been suggested that C459S-Shp2 may adopt a closed conformation (13). Therefore, it could be possible that introducing the C459S mutation abolished the prohypertrophic effects of Q510E-Shp2 simply by preventing any interactions with other proteins. To test this, we combined the Q79R and C459S mutations to promote an open conformation. The Q79R-Shp2 mutation is thought to activate Shp2 by disrupting interactions between the PTP and SH2 domains, thereby promoting unfolding of the protein (26). We found previously, in vitro as well as in mouse models, that Q79R-Shp2 does not alter cardiomyocyte cell size (6,14). Consistent with our previous findings, PTPase activity of the Q79R-Shp2 mutant was increased significantly (Fig. 6A). As expected, combination of the C459S and Q79R mutations lowered the resulting enzymatic activity of the protein to close to zero (Fig. 6A). Surprisingly, after infection with the Q79R-C459S double mutant, NRCM surface areas remained normal (Fig. 6B).
We also determined the levels of Akt activation in NRCM expressing C459S-Shp2, C459S-Q510E-Shp2, and Q79R-C459S-Shp2. Consistent with the cell size data, the degree of Akt phosphorylation was unchanged compared with the control, whereas only expression of Q510E-Shp2 resulted in upregulation of Akt activity (Fig. 6, C and D).
Pharmacological PTP Inhibition Prevents Q510E-Shp2-induced Hypertrophy-Because the C459S mutation abolished the prohypertrophic effects of Q510E-Shp2, we tested whether FIGURE 5. Fibroblast-specific effects or calcineurin-dependent signaling do not contribute to the hypertrophic mechanism, whereas MEK1 inhibition results in a partial effect on NRCM size. A, non-infected NRCM were exposed to CM harvested from independently cultured cardiac fibroblasts. These fibroblasts had been infected previously with ␤-gal, WT-Shp2, or Q510E-Shp2. 50 M PE in fresh serum-free medium served as a positive control. n ϭ 7-8/. *, p Ͻ 0.05 versus control; #, p Ͻ 0.05 versus vehicle. B, the presence of the calcineurin inhibitor cyclosporin A (CsA) did not reduce cell surface areas of Q510E-Shp2-expressing NRCM. n ϭ 4/group. *, p Ͻ 0.05 versus the respective ␤-gal control group under the same conditions. C and D, quantification of ERK1/2 phosphorylation using Western blot analyses. Because NRCM were cultured in serum-free conditions, no significant differences compared with WT were found in the groups expressing Shp2 mutants. n ϭ 4 -6/group. E, effects of MEK1 inhibition on NRCM size. U0126 partially reduced NRCM size in the Q510E-Shp2expressing groups. n ϭ 8 -12/group. *, p Ͻ 0.05 versus the respective ␤-gal control group under the same conditions. a pharmacological inhibitor that is specific for the catalytic center of Shp2 would affect Q510E-Shp2-induced hypertrophy. NRCM were infected with Q510E-Shp2 or ␤-gal control viruses and incubated with either vehicle or 10 M PHPS1 (Shp2 inhibitor V). PHPS1 was designed to bind at the catalytic site and, thereby, disrupt interactions between Shp2 and all of its substrates (27). As shown in Fig. 7A, the presence of PHPS1 completely abolished the prohypertrophic effects of Q510E-Shp2. To confirm that this effect was not limited to the Q510E mutation, we also tested the combination of PHPS1 with Y279C-Shp2 expression. As expected, incubation with PHPS1 normalized the size of NRCM expressing Y279C-Shp2 (Fig. 7B). To further establish that the signaling mechanism in Y279C-Shp2expressing NRCM closely resembles that of Q510E-Shp2 expression at other levels of the pathway as well, we extended our studies to FAK and mTOR. Both PP2 and rapamycin effectively reversed the Y279C-Shp2-induced increase in NRCM size (Fig. 7, C and D). This indicates that the two LS mutations Q510E-Shp2 and Y279C-Shp2 both induce NRCM hypertrophy through similar mechanisms.
Lastly, we assessed the effects of PHPS1 on downstream signaling. In NRCM expressing Q510E-Shp2 and incubated with PHPS1, the levels of Akt and mTOR activation were reduced close to normal levels (Fig. 7E).

DISCUSSION
In this study, Q510E-Shp2-induced cardiomyocyte hypertrophy depended on hyperactivation of FAK, Akt, and mTOR signaling. Hypertrophy was counteracted effectively not only by inhibition of mTOR, as shown before in mice (6), but also by the FAK/src inhibitor PP2, the endogenous FAK inhibitor FRNK, and the Akt1/2 inhibitor VIII. Dissecting domain-specific functions of Shp2 using double and truncation mutants, we found that the prohypertrophic effects of Q510E-Shp2 require the presence of the two SH2 domains and a functional catalytic center. Consistent with the cell size data, Akt hyperactivation was also only noted when the SH2 domains or the catalytic center, respectively, remained intact. The latter finding suggests that eliminating catalytic activity by preventing access to the pocket could also abolish the prohypertrophic effects of Q510E-Shp2. We therefore tested the Shp2-specific phosphatase inhibitor PHPS1, which had been specifically designed to bind the catalytic site of Shp2. Importantly, PHPS1 effectively prevented hypertrophy induced by Q510E-Shp2 as well as by another LS mutant.
To our knowledge, this is the first study to test these various pharmacological inhibitors in the context of LS-associated HCM. Importantly, our novel findings demonstrate that pharmacological inhibition of particular signaling steps upstream of mTOR could become feasible therapeutic approaches in the clinical arena. Our results complement the recent findings of two other groups showing that inhibition of phosphoinositide 3 kinase with LY294002 or wortmannin is also effective against the prohypertrophic effects of LS mutations in Shp2 (8,9). This is not surprising because PI3K is an important activator of Akt and, in turn, is regulated by FAK (28,29). Therefore, it appears that targeting any step of this signaling pathway could be considered for new treatment strategies for LS-associated HCM.
However, inhibitors acting upstream of mTOR are also likely to cause substantial side effects. In particular, FAK and Akt are essential regulators of normal growth and development. Mouse models with genetic manipulations to specifically disrupt FAK activity revealed severe developmental phenotypes (30 -32). In contrast, deletion of one Akt isoform does not affect base-line cardiac function in rodent models (33,34). However, lack of Akt1 diminishes somatic growth, and deletion of Akt2 causes insulin resistance and diabetes (35,36). Furthermore, Akt1 deletion exacerbates cardiac hypertrophy after pressure overload, indicating that Akt1 is essential for cardiac adaptation to such stresses (34). Therefore, long-term administration of Akt inhibitors may not be advisable, and this may limit the clinical applicability of our findings. However, in the case of LS patients, the therapeutic goal would be normalization, not a complete blockade of Akt signaling. Therefore, low-dose administration of Akt inhibitors could be of great benefit for affected patients.
In recent years, specific Shp2 inhibitors have become available for research applications. The fact that a Shp2-specific phosphatase inhibitor prevents LS-associated hypertrophy in our model is particularly exciting. This inhibitor was originally developed with the goal to treat an entirely different class of Shp2 mutations, i.e. gain-of-function mutations as found in patients with Noonan syndrome or juvenile myelomonocytic leukemia (27). Our data suggest that such inhibitors might also be beneficial for LS patients, thus increasing the number of patients that could benefit from the use of Shp2-specific inhibitors. In terms of side effects, Shp2-specific inhibitors may have additional advantages compared with FAK and Akt inhibitors. In the case of the Shp2 inhibitor NSC-87877, it has been shown that the inhibitor preferentially binds Shp2 protein if the gainof-function mutation E76K is present, although the reasons for this are still unclear (37). If PHPS1 also demonstrated preferential binding to mutant Shp2, the possibility of off-target effects would be reduced greatly, but this has not yet been tested.
At first sight, our finding that a pharmacological inhibitor reverses the effects of two different dominant-negative mutations in Shp2 appears to be a paradox. However, our data are consistent with two other studies in which inhibition of enzymatic activity of the respective LS-associated Shp2 mutant by replacing an essential arginine in the catalytic pocket with a methionine also abolished the phenotypic effects of mutant protein expression (23,24).
To explain this apparent paradox, we need to consider the possibility that the C459S alteration simply interferes with the dominant-negative effects of the resulting protein, thereby allowing the endogenous intact Shp2 protein to function normally. However, studies in other tissues, such as endothelial cell cultures or 293T cells, have found that C459S does indeed exert dominant-negative effects in those systems (13,38,39). Therefore, loss of dominance in the C459S-Q510E double mutant is, in our opinion, an unlikely explanation. On the other hand, our truncation mutants lack both SH2 domains and, therefore, most likely also lost their ability to interact with the regular binding partners of Shp2, such as receptor tyrosine kinases. Consequently, loss of dominance could very well explain why the Q510E-trunc mutant failed to promote NRCM hypertrophy in our studies.
An alternative explanation for the paradox could lie in our limited understanding of the molecular interactions between mutant Shp2 and its various substrates. Recent studies have demonstrated that in vitro phosphatase activities greatly depend on the substrate chosen for the assay and, therefore, that individual assays with only one substrate may be misleading (40). Consequently, it is possible that Q510E-Shp2, although unable to dephosphorylate the src peptide, may have retained substantial activity for other substrates. It is feasible that the putative loss-of-function mutations, in fact, induce gain-of-function effects regarding certain signaling pathways, which would explain why both the presence of a PTP inhibitor as well as the introduction of the C459S mutation confer protection against the prohypertrophic effects of Q510E-Shp2. However, whether or not Q510E-Shp2 is able to dephosphorylate certain substrate(s) remains unclear. This issue is currently impossible to address because many substrates of Shp2 are still unknown, as suggested by unbiased screening approaches (41). Substrate trapping mutants have been employed to identify some of the substrates of Shp2, such as EGF receptor (42,43), but have the potential of changing substrate specificities themselves, which might skew the results. Identification of all Shp2 substrates would most likely also lead to further insight into how exactly Q510E-Shp2 increases FAK activation. FAK is thought to be a substrate of Shp2 (44,45), but the effects seen in this study may not be due to direct interaction. To fully understand how disease-causing mutations such as Q510E alter the functionality of Shp2, comprehensive assessments of substrate FIGURE 7. PHPS1 inhibits both Q510E-Shp and Y279C-Shp2 induced NRCM hypertrophy. A, Q510E-Shp2 was expressed in NRCM in the presence of 10 M PHPS1 or vehicle. n ϭ 12/group. PHPS1 reduced NRCM size in the Q510E-Shp2-expressing group to close to normal levels. *, p Ͻ 0.05 versus the respective ␤-gal control group under the same conditions. B, Y279C-Shp2 was expressed in NRCM in the presence of 10 M PHPS1 or vehicle. n ϭ 8/group. Again, PHPS1 prevented Y279C-Shp2-induced hypertrophy. *, p Ͻ 0.05 versus the respective ␤-gal control group under the same conditions. C, the FAK/src inhibitor PP2 normalized cell surface areas of Y279C-Shp2-expressing NRCM. n ϭ 7-8/group. D, rapamycin abolished the Y279C-Shp2induced NRCM size increase in a dose-dependent manner. n ϭ 7-8/group. *, p Ͻ 0.05 versus the respective ␤-gal control group under the same conditions. E, Western blot signal quantification to evaluate downstream mediators. In the presence of PHPS1, the levels of Akt and mTOR phosphorylation were normalized in Q510E-Shp2-expressing NRCM. n ϭ 5/group. *, p Ͻ 0.05 versus the respective vehicle-treated ␤-gal control group. affinities and enzymatic activities for individual substrates will be needed.
Another explanation for the paradox could be that Q510E-Shp2 binds and traps certain substrates, thus reducing their availability. The phosphatase inhibitor orthovanadate competes with the substrates at the binding site of Shp2 and, thereby, interferes with substrate trapping (43). Presumably, PHPS1 would also compete with the regular substrates and prevent trapping in a similar fashion, so this concept would be consistent with our data. However, substrate trapping cannot explain why introduction of the C459S mutation abolishes the pathogenic effects. In fact, substrate trapping effects have been described for C459S-Shp2 itself (43). Importantly, C459S-Shp2, as well as the Q79R-C459S double mutant, did not have any hypertrophic effects in our study, which would, therefore, argue against substrate trapping being the key event of the molecular pathomechanism.
Our finding that the Q79R-C459S double mutant did not have any hypertrophic effects also suggests that the complex molecular underpinnings of the pathogenic effects of LS mutations may be described insufficiently by current concepts. It has been proposed that LS mutations may preferentially adopt "open-inactive" conformations, whereas C459S-Shp2 would be "closed-inactive" (13). The Q79R-C459S double mutant would presumably also be open-inactive and should, therefore, have induced NRCM hypertrophy. Why the double mutant did not have the expected effects is difficult to explain because crystal structures of mutant Shp2 proteins are not yet available. But our findings do indicate that deeper insight into the structure and functionality of normal and mutant Shp2 is needed urgently and will most likely lead to refinement of the current concepts of the three-dimensional structure of various Shp2 mutants.
In summary, we found that pharmacological reduction of signaling through Shp2/FAK/Akt/mTOR at all levels of the signaling cascade effectively counteracts the prohypertrophic response to Q510E-Shp2 expression. Importantly, this greatly broadens the range of tools that could be used to treat patients with such mutations in the future. Our discovery that interventions targeted directly at the mutant Shp2 protein could be highly effective is particularly exciting because already available compounds, such as PHPS1, are likely to be suitable, thus facilitating translation into the clinic. With such new custom-tailored approaches, morbidity and mortality of severely affected LS patients may improve substantially in the future.