NF2 Activates Hippo Signaling and Promotes Ischemia/Reperfusion Injury in Heart

Abstract

Rationale

Neurofibromin 2 (NF2) is an established tumor suppressor that promotes apoptosis and inhibits growth in a variety of cell types, yet its function in cardiomyocytes remains largely unknown.

Objective

We sought to determine the role of NF2 in cardiomyocyte apoptosis and ischemia/reperfusion (I/R) injury in the heart.

Methods and Results

We investigated the function of NF2 in isolated cardiomyocytes and mouse myocardium at baseline and in response to oxidative stress. NF2 was activated in cardiomyocytes subjected to H₂O₂ and in murine hearts subjected to I/R. Increased NF2 expression promoted the activation of Mst1 and the inhibition of Yap, whereas knockdown of NF2 attenuated these responses following oxidative stress. NF2 increased apoptosis of cardiomyocytes that appeared dependent on Mst1 activity. Mice deficient for NF2 in cardiomyocytes, NF2 CKO, were protected against global I/R ex vivo and showed improved cardiac functional recovery. Moreover, NF2 CKO mice were protected against I/R injury in vivo and showed upregulation of Yap target gene expression. Mechanistically, we observed nuclear association between NF2 and its activator MYPT-1 in cardiomyocytes, and a subpopulation of stress-induced nuclear Mst1 was diminished in NF2 CKO hearts. Finally, mice deficient for both NF2 and Yap failed to show protection against I/R indicating that Yap is an important target of NF2 in the adult heart.

Conclusions

NF2 is activated by oxidative stress in cardiomyocytes and mouse myocardium and facilitates apoptosis. NF2 promotes I/R injury through activation of Mst1 and inhibition of Yap, thereby regulating Hippo signaling in the adult heart.

Introduction

Ischemic heart disease is a major component of cardiovascular disease and a leading cause of death worldwide ^{1, 2}. Although much progress has been made in understanding the causes of ischemic heart disease, the molecular underpinnings that occur during ischemia/reperfusion (I/R) injury have not yet been fully elucidated, and the result is current treatments that are only partially effective.

Mammalian sterile 20-like kinase 1 (Mst1) is a ubiquitously expressed and highly conserved serine/threonine kinase ³ that is activated in the heart during stress conditions including I/R, myocardial infarction (MI) and pressure overload ^4-6. Previous work from our group has shown that suppression of Mst1 inhibits I/R injury ^{4, 7} and prevents cardiac remodeling/dysfunction after chronic MI, ⁵ suggesting that Mst1 is a promising target of cardiac therapy for ischemic heart disease. Mst1 is the mammalian homolog of Drosophila hippo and the centerpiece of a signaling cascade that culminates in the phosphorylation and inactivation of the transcriptional co-factor yorkie (mammalian Yap). Our recent work demonstrated Yap to be a critical transcriptional co-factor that mediates cardioprotection and homeostasis of the adult heart ^{8, 9}.

NF2 (also known as neurofibromin 2/schwannomin) is a widely expressed scaffold-like protein that is able to transduce intra- and extracellular signals to modulate various cellular processes ^10-12. NF2 was originally described as a tumor suppressor protein and is linked to several human cancers ^{13, 14}. However, its function in the heart remains largely unexplored. NF2 lacks catalytic function, and conformational changes determine its ability to interact with, and subsequently transduce signals through, effector proteins. NF2 structure is regulated by phosphorylation at Ser518, ¹⁵ a site that can be phosphorylated by PAK2 ^{16, 17} and PKA, ¹⁸ and dephosphorylated by the myosin light chain phosphatase, MYPT-1 ¹⁹. When Ser518 is phosphorylated, NF2 assumes an “open” conformation and can no longer associate with binding partners, effectively inhibiting its function ^{15, 20-22}. Conversely, dephosphorylation of NF2 favors a “closed” conformation that promotes protein interaction and signal transduction. Studies in Drosophila have provided evidence that NF2 can regulate the activity of hippo/yorkie, thereby modulating cell proliferation and survival ^23-25. Yet, to date, evidence linking NF2 and Mst1/Yap in mammalian systems is limited ^26-28.

Herein we demonstrate that NF2 is activated by oxidative stress through dephosphorylation by the protein phosphatase targeting subunit MYPT-1. Active NF2 promotes cardiomyocyte apoptosis through activation of Mst1 and engagement of the Hippo signaling pathway. Interestingly, NF2 is present in both the cytosol and nucleus of cardiomyocytes and promotes phosphorylation and inactivation of Yap, thereby attenuating Yap target gene expression. Mice deficient for NF2 in cardiomyocytes (NF2^flox/flox;Cre^αMHC ; NF2 CKO) show diminished Mst1 activation, increased Yap transcriptional activity, and are protected against I/R injury. These results provide evidence that NF2 modulates Hippo signaling in the mammalian heart to promote acute myocardial injury.

Methods

An expanded Methods section is available in the online Data Supplement.

Animal models

NF2 floxed mice²⁹ were crossed with α-MHC-Cre transgenic mice³⁰ to generate cardiac-specific knockout (NF2 CKO) mice. Yap floxed mice²⁶ were bred to NF2 CKO mice to generate NF2^flox/flox;Yap^flox/+;Cre^αMHC mice. Mice were housed in a temperature-controlled environment with 12-hour light/dark cycles where they received food and water ad libitum. All protocols concerning the use of animals were approved by the Institutional Animal Care and Use Committee at Rutgers, The State University of New Jersey.

Statistics

All data are reported as mean ± SEM. Evaluation between three or more groups was done using one-way ANOVA. The statistical significance of the differences between groups was calculated using post hoc comparisons. Student's t test was used to evaluate the difference in means between two groups. Statistical analyses were performed using Graph Pad Prism 6.0. A p value less than 0.05 was considered significant.

Results

Regulation of NF2 by oxidative stress

Our previous work demonstrated that Mst1 is activated in response to oxidative stress both in cultured cardiomyocytes and mouse myocardium in vivo ^{7, 31}. Therefore, we first determined the activation status of NF2 using Ser518 phosphorylation as an indicator of the inactive conformation of NF2. Treatment of neonatal rat ventricular myocytes (NRVMs) with H₂O₂, to mimic oxidative stress during reperfusion, elicited activation of both NF2 and Mst1 (Figure 1A-C). To test the activation of NF2 in vivo, we subjected wild-type (WT) C57BL/6 mice to I/R and performed western blots to assay phosphorylation status. Similarly, we found that both NF2 and Mst1 were activated in the myocardium by I/R (Figure 1D-F). We also evaluated samples of failing human hearts and observed increased NF2 and Mst1 activation compared to healthy controls (Online Figure I). Since NF2 phosphorylation is decreased in response to oxidative stress, it is possible that increased phosphatase activity is responsible. To test this, we pretreated NRVMs with the phosphatase inhibitors okadaic acid or calyculin A, followed by stimulation with H₂O₂. The dephosphorylation of NF2 caused by H₂O₂ was partially attenuated in okadaic acid-treated cells at high concentration, but was fully prevented in calyculin A-treated cells indicating likely involvement of PP1 phosphatase (Online Figure II). Previous work identified PP1δ-MYPT-1 as an activator of NF2 through dephosphorylation of Ser518 in mammalian cells ^{19, 32} and in Drosophila ³³. A lack of phosphorylation of MYPT-1 at Ser696 is indicative of its activation. We observed decreased MYPT-1 phosphorylation in response to oxidative stress both in NRVMs and mouse myocardium (Figure 1G-J). To directly test the involvement of MYPT-1 in NF2 regulation, we depleted NRVMs of endogenous MYPT-1 using siRNA. MYPT-1 downregulation was sufficient to attenuate NF2 activation by oxidative stress (Figure 1K). Taken together, these data indicate that MYPT-1 is activated by oxidative stress and mediates the dephosphorylation and activation of NF2 in cardiomyocytes.

Figure 1. Oxidative stress-induced activation of NF2 is mediated by MYPT-1.

(A) Neonatal rat ventricular myocytes (NRVMs) were treated with vehicle or H₂O₂ (100 mM) for the times indicated. Western blot analysis was performed to detect phosphorylated and total NF2 (Ser518) and Mst1 (Thr183). (B and C) Quantification of responses at 30′ time point, n = 3. *, P<0.05. (D) Wild-type C57BL/6 male mice were subjected to ischemia and reperfusion (30′/30′) or sham operation. Left ventricular homogenates were prepared and western blot was performed to detect phosphorylated and total NF2 and Mst1. (E and F) Quantitative analysis of results shown in panel D, n = 4 mice per group. *, P<0.05. (G) NRVMs were treated as described in panel A and westerns were performed to detect phosphorylated and total MYPT-1 (Ser696). (H) Quantification of response at 30′ time point, n = 3. *, P<0.05. (I) Wild-type C57BL/6 male mice were subject to sham or I/R as indicated in panel D and westerns were performed to detect phosphorylated and total MYPT-1. (J) Quantification of results in panel I, n = 3 mice per group. *, P<0.05. (K) NRVMs were transfected with siRNA targeted to MYPT-1 or scrambled control siRNA. 72 hours later, cells were stimulated with H₂O₂ (100 mM) or vehicle for 1 hour followed by western blot analysis. Representative images of 3 independent experiments are shown.

NF2 promotes Mst1 activation and cardiomyocyte apoptosis

NF2 can activate the Hippo pathway in the liver²⁶ and brain²⁷. We sought to determine whether NF2 regulates Hippo signaling in the adult heart. Overexpression of NF2 in NRVMs stimulated Mst1 activation (Figure 2A and B). Mst1 is known to promote cell death; therefore, we evaluated cardiomyocyte apoptosis in response to NF2. We found that increased NF2 expression caused a significant increase in TUNEL-positive cardiomyocytes, and that this response was significantly attenuated by inhibition of Mst1 (Figure 2C and D). Similarly, NF2 elicited caspase-3 activation, which was significantly reduced by inhibition of Mst1 (Figure 2E and F). On the other hand, knockdown of endogenous NF2 using siRNA attenuated both the activation of Mst1 and cardiomyocyte apoptosis driven by H₂O₂ (Figure 2G and H). Taken together, these results indicate that NF2 can engage Hippo signaling at the level of Mst1 to promote cardiomyocyte apoptosis.

Figure 2. NF2 promotes activation of Mst1 and cardiomyocyte apoptosis.

(A) NRVMs were transduced with adenovirus engineered to express FLAG-NF2 or LacZ as a control. After 24 hours, cells were harvested and westerns were performed. (B) Quantification of results obtained from panel A, n = 3. *, P<0.05. (C) NRVMs were transduced with LacZ, Mst1, kinase-inactive (K59) dominant-negative DN-Mst1, or NF2 adenovirus and TUNEL was performed after 48 hours to determine DNA fragmentation as an indicator of apoptosis. (D) Representative images of TUNEL-positive nuclei (arrows), scale bar = 100 mm. (E) NRVMs were treated as described in panel C and westerns were performed to detect caspase-3 (arrow, cleaved 17kd fragment). (F) Quantification of cleaved caspase-3, n = 3. *, P<0.05. (G) NRVMs were transfected with siRNA targeted to Nf2 or scrambled control siRNA for 72 hours. Westerns were performed to detect Mst1 phosphorylation and extent of NF2 knockdown. (H) NRVMs were treated with siRNA against Nf2 or scrambled control and then stimulated with H₂O₂ (100 mM) or vehicle for 6-8 hours. TUNEL was used to measure apoptotic cardiomyocytes. *, P<0.05. n = 3 experiments. Representative images of 3 independent experiments are shown.

Subcellular localization of NF2 in cardiomyocytes

Previous work has demonstrated subcellular localization of NF2 at tight junctions ³⁴, adherens junctions ³⁵, desmosomes ²⁸, the plasma membrane ³⁶ and in the nucleus ³⁷. We sought to investigate the localization of NF2 in NRVMs. Confocal imaging revealed distribution of NF2 in the cytosol, at the cell membrane, and a strong nuclear signal in NRVMs (Figure 3A, Online Figure IIIA). Detection of p-NF2 revealed a diffuse cytosolic and nuclear distribution with a relatively more pronounced presence at the plasma membrane (Online Figure IIIB). We also stained mouse heart sections and observed both nuclear and cytosolic localization of NF2 (Online Figure IVA). As a control, we evaluated HEK293 cells and found predominant plasma membrane distribution of NF2 and no appreciable nuclear signal (Online Figure IIIC). To investigate this biochemically, we separated mouse ventricular lysates into cytosolic and nuclear-enriched fractions. We detected endogenous NF2 and p-NF2 in both fractions (Figure 3B). Similarly, NF2 was observed in cytosolic and nuclear-enriched fractions generated from NRVMs and isolated adult mouse cardiomyocytes (AMCMs)(Figure 3C and Online Figure IVB). Although p-NF2 was detected in cytosolic and nuclear-enriched fractions of NRVMs, we did not observe a change in p-NF2 levels in the cytosol following H₂O₂ treatment, while nuclear p-NF2 levels decreased (Figure 3C). We also probed for MYPT-1 and detected it in the nuclear-enriched, but not in the cytosolic-enriched fractions, of heart homogenates and NRVMs (Figure 3B and C), consistent with a recent report ³⁸. Subcellular fractionation of NRVMs into cytosolic and plasma membrane-enriched fractions revealed faintly detectable levels of NF2 at the plasma membrane, while p-NF2 was more evident (Online Figure IVC).

Figure 3. NF2 associates with Mst1 following oxidative stress.

(A) Confocal microscopy images of NRVMs. Endogenous NF2 (green) and troponin-T (red) were detected by immunofluorescence. Nuclei were stained with DAPI. Scale bar, 10 mm. (B) Cytosolic and nuclear-enriched fractions were prepared from ventricular homogenates of C57BL/6 wild-type mice and subjected to western blot analysis. (C) NRVMs were transduced with siRNA targeted to MYPT-1 (siMY), NF2 or scrambled control (siCTRL). 72 hours later, cells were treated with H₂O₂ (100 mM) or vehicle for 1 hour. Cytosolic and nuclear-enriched fractions were prepared and subjected to western analysis. (D) Control NF2 floxed (NF2^F/F) and NF2 CKO mice were subjected to I/R (30′/30′) or sham operation and ventricular homogenates were separated into cytosolic and nuclear-enriched fractions. GAPDH and Histone H3 were used as markers of cytosolic and nuclear fractions respectively. (E) C57BL/6 wild-type mice were subjected to sham operation (S) or I/R (30′/30”). Homogenates were used for co-IP and subsequent western blotting. (F) NRVMs were treated with H₂O₂ (100 μM) or vehicle for 1 hour. Cytosolic and nuclear-enriched fractions were prepared and subjected to immunoprecipitation using anti-NF2 antibody or control IgG. Immunocomplexes and inputs were analyzed by western blot. Representative images of 3-4 independent experiments are shown.

NF2-dependent Mst1 localization

Because NF2 modified Mst1 activity, we investigated the subcellular localization of the Hippo kinases Lats2 and Mst1. Analysis of Lats2 distribution in cardiomyocytes demonstrated a largely nuclear presence (Figure 3C and Online Figure IVA and B). Mst1 showed a predominant cytosolic presence in NRVMs, AMCMs and heart sections; however, following oxidative stress we observed a nuclear subpopulation of Mst1 (Figure 3C and Online Figure IVB). Knockdown of either MYPT-1 or NF2 attenuated this response in NRVMs (Figure 3C). Furthermore, we observed Mst1 levels increase in nuclear-enriched fractions prepared from WT mouse myocardium subjected to I/R (Figure 3D). Nuclear localization of Mst1 was attenuated in hearts deficient for NF2, suggesting a causal role for NF2 in this process in vivo.

Association of NF2 with Hippo pathway components

Based on previous work demonstrating complexes comprised of NF2 and Hippo in Drosophila ³⁹ and NF2 and Lats in mammalian cells ³⁶, we examined whether NF2, Mst1 and Lats2 associate in cardiomyocytes. Co-IP studies showed an interaction between NF2 and Mst1 in hearts subjected to I/R (Figure 3E), and NRVMs subjected to oxidative stress (Online Figure VA). We performed additional co-IP experiments, this time stimulating NRVMs with H₂O₂ and using cytosolic and nuclear-enriched fractions to determine NF2 interactions. As demonstrated in Figure 3F, association between NF2 and MYPT-1, Mst1 and Lats2 was observed in nuclear-enriched fractions during oxidative stress. Phosphorylation status is important for NF2 conformation change and protein interactions. Therefore, we tested the ability of a phospho-resistant NF2 S518A mutant to associate with Mst1 and Lats2. Co-IP studies showed that NF2 S518A associated with Mst1 and Lats2, and to a greater extent than wild-type NF2, suggesting the importance of phosphorylation of Ser518 in modulating these protein associations (Online Figure VB).

NF2 deletion attenuates cardiac injury ex vivo

To test the physiological role of NF2, we crossed NF2 floxed mice²⁹ with α-MHC-Cre transgenic mice³⁰ to generate cardiac-specific knockout (NF2^flox/flox;Cre^αMHC (NF2 CKO)) mice. These mice had depleted levels of NF2 protein in the myocardium but not in other tissues tested (Figure 4A). Hearts were isolated from NF2 CKO and control NF2^flox/flox mice and subjected to global ischemia and reperfusion ex vivo using the Langendorff preparation. We observed that NF2 CKO hearts had significantly smaller infarct regions compared to control NF2^flox/flox hearts (Figure 4B and C). Cardiac function was assessed at baseline prior to ischemia and serially every 10 minutes through 1 hour of reperfusion. While no differences were observed at baseline, we found that NF2 CKO hearts had significantly improved cardiac function during reperfusion (Figure 4D-G), while no significant difference in heart rate was observed (Figure 4H). Taken together, these data suggest that disruption of NF2 in cardiomyocytes confers protection against global I/R and preserves heart function.

Figure 4. NF2 CKO mouse hearts are protected against global I/R injury and have increased function during reperfusion.

(A) Ventricular (ht), lung (lu), liver (lv) and skeletal muscle (sk) homogenates were prepared from control NF2 floxed (NF2^F/F) and NF2 CKO mice and subjected to western analysis. (B and C) Hearts from control and NF2 CKO mice were isolated and subjected to global I/R (30′/60′) using the Langendorff method. Infarct size was determined by TTC staining. Scale bar, 1 mm. (D-G) Parameters of ventricular function were determined during the reperfusion phase, including left ventricular developed pressure (LVDP), left ventricular end diastolic pressure (LVEDP), dP/dT max, and dP/dT min. Time 0 min represents the start of reperfusion. (H) Heart rate was determined at baseline and after 60 min of reperfusion. *, P<0.05 versus control NF2 floxed (NF2^F/F). N.S. = not significant. N = 8 mice per group.

Deletion of NF2 affords cardioprotection against I/R injury

To determine whether disruption of NF2 also conferred protection of the myocardium in vivo, we subjected NF2 CKO and control NF2^flox/flox mice to I/R and determined infarct size 24 hours later. We found no difference in the myocardial area at risk (AAR) following I/R; however, infarct size was significantly reduced in NF2 CKO mice indicating that NF2 deletion is cardioprotective (Figure 5A-C). We also subjected mice to I/R and examined heart function by echocardiography 2 weeks post-infarction. We observed that left ventricular ejection fraction (LVEF%) was significantly greater in NF2 CKO mice compared to controls (Figure 5D). Since NF2 regulates Hippo signaling in the mouse liver²⁶ and brain²⁷, we examined the activation of the conserved pathway components Mst1 and Yap. The phosphorylation of both Mst1 and Yap were increased in NF2^flox/flox hearts in response to I/R, indicating Hippo pathway activation. However, this response was significantly attenuated in NF2 CKO hearts (Figure 5E-G). The activation of known cardioprotective targets AKT and ERK1/2 showed negligible differences between NF2^flox/flox and NF2 CKO hearts (Figure 5E, H and I). Consistent with these findings, we also observed Yap downregulation in NRVMs overexpressing NF2, and increased Yap levels in NRVMs treated with siRNA targeting NF2 (Online Figure VI).

Figure 5. Myocardial Yap activity is increased and I/R injury is attenuated in NF2 CKO mice.

(A-C) Control NF2 floxed (NF2^F/F) and NF2 CKO mice were subjected to I/R (30′/24hr) and area at risk (AAR) and infarct size were determined by Alcian blue and TTC staining respectively. Scale bar, 1 mm. *, P<0.05. N.S. = not significant. N = 5 mice per group. (D) Left ventricular ejection fraction (LVEF%) was determined by echocardiography prior to intervention and 2 weeks after I/R. *, P<0.05. N = 4-5 mice per group. (E) Ventricular homogenates from control NF2 floxed (NF2^F/F) and NF2 CKO mice were prepared following sham operation and I/R (30′/30′) and western blot analysis performed. (F-I) Quantitative results from westerns in panel E. *, P<0.05. N.S. = not significant. (J) Real time quantitative PCR was performed using cDNA prepared from control NF2 floxed (NF2^F/F) and NF2 CKO mice subjected to either sham or I/R (30′/120′). *, P<0.05 versus NF2^F/F sham. #, P<0.05 versus NF2^F/F I/R. N = 3-4 mice per group. Representative images for each experiment are shown.

Yap activation is upregulated in NF2 CKO myocardium

Yap is a transcriptional co-factor and pro-survival signaling molecule in cardiomyocytes⁹; therefore, we hypothesized that altered gene expression due to Yap modulation might explain the protective effect observed in NF2 CKO mice. We first tested whether established Yap target genes were altered following Yap modification in NRVMs. Overexpression of Yap in NRVMs caused a significant increase in mRNA levels of ctgf, cyr61, fgf2 and birc5 compared to LacZ control (Online Figure VIIA). Conversely, shRNA-mediated knockdown of endogenous Yap significantly reduced Yap target gene expression (Online Figure VIIB). To test whether NF2 affects Yap transcriptional activity in NRVMs, we used a TEAD luciferase reporter system and either overexpressed or silenced NF2, which resulted in a significant decrease and increase in reporter gene expression, respectively (Online Figure VIIC and D). We also examined expression of these Yap targets by qRT-PCR and observed a significant upregulation of the aforementioned genes in NF2 CKO hearts compared to controls (Figure 5J). Taken together these results suggest that Yap activates a genetic program in the heart similar to that observed in the liver²⁶ and brain²⁷, and indicate that Yap activity is upregulated in hearts deficient for NF2.

Genetic inhibition of Yap abolishes the protection observed in NF2 CKO mice

Finally, we tested whether Yap activity is required for the protective effect of NF2 depletion in vivo. NF2 CKO, Yap^flox/+;Cre^αMHC , NF2-Yap double deficient (NF2^flox/flox;Yap^flox/+;Cre^αMHC ) and control (NF2^flox/flox;Yap^flox/+ ) mice were generated. The expected reduction in NF2 and Yap protein was confirmed by western blot (Figure 6A) and these mice were subjected to I/R. Consistent with our findings above, the NF2 CKO mice had significantly smaller infarcts and reduced TUNEL⁺ staining compared to control mice (Figure 6B, D and E). There were no differences in AAR/LV between the groups (Figure 6C). Interestingly, we observed a significant increase in infarct size in the Yap^flox/+;Cre^αMHC and NF2^flox/flox;Yap^flox/+;Cre^αMHC myocardium versus controls (Figure 6B and D). Furthermore, we observed a significant increase in TUNEL⁺ cardiomyocyte nuclei in Yap^flox/+;Cre^αMHC hearts compared to controls as well as in NF2^flox/flox;Yap^flox/+;Cre^αMHC hearts compared to NF2 CKO hearts following I/R (Figure 6E and F). Taken together, these results indicate a loss of cardioprotection in the NF2^flox/flox;Yap^flox/+;Cre^αMHC mice that is likely due to normalization of Yap activity.

Figure 6. Hemizygous deletion of myocardial Yap abrogates the cardioprotection observed in NF2 CKO hearts following I/R.

Control (NF2^flox/flox;Yap^flox/+) NF2 CKO (NF2^flox/flox;Cre^aMHC), Yap^flox/+;Cre^aMHC, and NF2-Yap double deficient mice (NF2^flox/flox;Yap^flox/+;Cre^aMHC) were generated. (A) Ventricular homogenates from all groups were subjected to western analysis to determine levels of NF2 and Yap protein. (B-D) All groups were subjected to I/R (30′/24hr) and infarct size and area at risk (AAR) determined by TTC and Alcian blue staining respectively. Scale bar, 1 mm. (E) Representative images from TUNEL staining of heart sections following I/R (30′/24hr) are shown. Sections were counterstained with DAPI (blue) and troponin-T (green) to identify nuclei and cardiomyocytes, respectively (Merged). Scale bar, 50 μm. (F) Average data as determined by TUNEL. *, P<0.05. N.S. = not significant. N = 5-8 mice per group.

Discussion

NF2 is a recognized tumor suppressor that has been shown to promote Hippo signaling in Drosophila ^23-25. Subsequent work has demonstrated that NF2 can modulate mammalian Hippo signaling in the murine liver²⁶ and brain²⁷. In a recent report from Marian and colleagues, NF2 was shown to be activated in a mouse model of arrhythmogenic cardiomyopathy (AC) and in heart samples from AC patients²⁸. Murine models of AC displayed increased activation of Hippo signaling and reduced Yap activation, which was shown to contribute to the enhanced adipogenesis observed. However, prior to our study, a loss-of-function mouse had not been used to test whether NF2 regulates Hippo signaling and contributes to myocardial injury caused by I/R. We demonstrate here that NF2 is activated in cardiomyocytes and mouse myocardium in response to oxidative stress and I/R, and contributes to cardiac injury through engagement of the Hippo pathway and subsequent inactivation of the transcriptional co-factor Yap.

Regulation of NF2 through multiple posttranslational modifications has been demonstrated previously. NF2 is negatively regulated via Ser518 phosphorylation by PKA¹⁸ and PAK2^{16, 17}, which promotes its inactive conformation, and thereby attenuates downstream signaling^{15, 22}. Alternatively, dephosphorylation of Ser518 mediated by MYPT-1 activates NF2¹⁹. NF2 function can also be modulated through Akt-mediated phosphorylation at Thr230 and Ser315⁴⁰. These Akt-dependent modifications both inhibited the association of NF2 with effectors while also eliciting increased ubiquitin-mediated NF2 degradation. Our study demonstrates that MYPT-1 associates with NF2 and promotes its dephosphorylation, indicating that MYPT-1 is an important promoter of NF2 activity in cardiomyocytes. MYPT-1 was also activated (as determined by regulatory phosphorylation at Thr696) following I/R; however, establishing evidence that MYPT-1 is required for NF2 activation in vivo is a limitation of this study and warrants further investigation. More recent work has demonstrated that NF2 is also a substrate of sumoylation. It was reported that SUMO-1 modification at Lys76 led to decreased active conformation status, altered subcellular localization, and impaired tumor suppressor function suggesting that sumoylation is important for proper NF2 signaling⁴¹. Whether or not NF2 is sumoylated in cardiomyocytes, and if this modification modulates NF2 activity in the heart, remains to be determined.

The subcellular localization of NF2 has been investigated in multiple cell types with varied observations^{34-37, 42, 43}. Recent findings from the Pan laboratory demonstrated that plasma membrane association of NF2 and direct binding with Lats1/2 is critical for engagement of Hippo signaling and subsequent inactivation of Yap³⁶. In other work, Giancotti and colleagues have demonstrated that NF2 can shuttle between the cytosol and nucleus, an observation that may be mediated by a nuclear localization motif present in the N-terminal portion of NF2 ⁴⁴. It is likely that observed differences in subcellular localization of NF2 are due to cell type specificity, and that NF2 has important cellular functions in multiple locations. In NRVMs, AMCMs and mouse heart sections we observed NF2 in the cytosol and nucleus, and relatively lower levels at the plasma membrane, using both biochemical fraction enrichment and immunofluorescence-based approaches. In contrast, p-NF2 was detected in cytosolic and nuclear fractions, with relative enrichment at the plasma membrane. Importantly, nuclear p-NF2 levels decreased after oxidative stress indicating this is an important site of NF2 activation. We also observed a subpopulation of Mst1 in nuclear-enriched fractions of stressed cardiomyocytes and hearts, and Lats2 appeared almost exclusively nuclear, similar to our prior work⁸. The AC study mentioned above also examined the distribution of Hippo components by immunostaining heart sections²⁸. In mouse myocardium, p-NF2 was reported to localize to cardiomyocyte desmosomes. This is not inconsistent with our current findings, which demonstrate p-NF2 at the plasma membrane. It should be noted that this prior study did not observe localization of p-Mst1 in the nucleus. It is possible that this apparent discrepancy concerning Mst1 is due to the type of stress imposed, or the potential for heterogeneity between cardiomyocytes in the heart. Indeed, our NF2 staining results suggest that nuclear NF2 may be more pronounced in a select subset of cardiomyocytes. We believe it is likely that NF2 serves important physiological functions in multiple subcellular regions. Based on our localization and immunoprecipitation results, we propose that the nucleus is one of perhaps several important focal points of NF2-Hippo signaling in the cardiomyocyte.

Our prior work demonstrated a deleterious role for Mst1 during I/R^{4, 7}. These studies interrogated both upstream regulation and downstream effectors of Mst1 in this setting. We identified a K-Ras/RASSF1A/Mst1 complex present at mitochondria that mediated activation of Mst1 at this subcellular locale. We also found that active Mst1 phosphorylates Bcl-xL, thereby attenuating Bcl-xL-Bax interaction and increasing Bax activation and cardiomyocyte apoptosis⁷. Importantly, we demonstrated that this non-canonical Hippo signaling (i.e. not through Lats2/Yap) contributes to I/R injury⁴⁵. Our current findings extend this work by demonstrating that NF2 is a regulator of canonical Hippo signaling in the myocardium and implicating the nucleus as a target organelle in which this signaling occurs. Based on our findings we hypothesize that NF2 and K-Ras/RASSF1A function to activate Mst1 in separate subcellular locations (mitochondria vs. nucleus) leading to engagement of different downstream effectors (Bcl-xL vs. Lats2/Yap). Mst1 has also been shown to inhibit cardiomyocyte autophagy and contribute to ischemic injury³¹. Although we found that NF2 promotes Mst1 activation, we did not observe a significant effect of NF2 on cardiomyocyte autophagy (Online Figure VIII), although further study is needed to examine this in greater detail. Taken together, we propose that inhibition of Mst1 likely confers cardioprotection by preventing multiple signaling mechanisms and would be advantageous versus targeting a single pathway in isolation.

Mst1 contains a regulatory phosphorylation site, Thr183, in its N-terminal catalytic domain⁴⁶. Autophosphorylation of Thr183 has been shown to be important for Mst1 activation and homodimerization, the latter mediated by its C-terminal SARAH domain⁴⁷. The SARAH domain also allows for heterodimerization between Mst1 and RASSF1A, NORE1 and Salvador (Sav1), known activators of the kinase. Structural studies of Mst1 demonstrated the importance of h1 and h2 helices for homodimerization of Mst1 monomers⁴⁸; however, the precise molecular mechanism of Mst1 activation remains unclear and it is possible that additional kinases may play an important role^{49, 50}. We observed association between NF2 and MYPT-1, Mst1 and Lats2 and increased phosphorylation of Mst1 following oxidative stress. Our results lead us to speculate the formation of a nuclear complex that could include additional components (e.g. Sav1, MOB1A/B, MAP4K/Happyhour)^{51, 52}. We hypothesize that recruitment of Mst1 to this complex may be facilitated by Lats2, as a direct interaction between NF2 and Lats2 has been demonstrated previously³⁶.

Based on the literature regarding the function of Yap in the mammalian heart, it is not entirely surprising that ablation of a negative regulator of Yap activity, in this case NF2, would prove cardioprotective against I/R injury. Our previous work demonstrated that inhibition of Lats2 in the myocardium afforded protection against I/R injury through increased Yap-FoxO1 transcriptional activation⁸. Through loss-of-function studies, Yap has been shown to be critical for adult heart homeostasis and protection against cardiomyocyte apoptosis, as well as a promoter of cardiomyocyte growth and proliferation^{9, 53, 54}. Conversely, strategies to increase Yap expression and activity stimulate proliferation of cardiomyocytes and cardioprotection against MI and resection of the mouse heart^{53, 55}. Our current study demonstrates that the cardioprotection observed in NF2 CKO mice is abrogated by Yap heterozygosity in cardiomyocytes. We believe this is strong evidence that Yap functions downstream of NF2; however, we cannot exclude the possibility that Yap acts in a parallel pathway that supersedes the NF2 effect in this setting.

Our results indicate that NF2 is activated by oxidative stress and has a presence in the nuclei of cardiomyocytes where it promotes activation of Mst1 and inhibitory phosphorylation of Yap (Figure 7). This work further supports the notion that increased Yap activity during acute stress/injury serves to protect the myocardium and highlights Yap as an attractive target for potential future therapies against MI.

Figure 7. Schema illustrating a working hypothesis of NF2-related signaling in the cardiomyocyte during oxidative stress.

We propose that oxidative stress stimulates dephosphorylation of NF2 via MYPT-1-PP1δ thereby promoting an active conformation of NF2. Active NF2 associates with Mst1 and Lats2 in the cardiomyocyte nucleus, promotes Mst1 activation, and negatively regulates Yap target gene expression.

Novelty and Significance.

What Is Known?

• Myocardial infarction (MI) results in cardiomyocyte loss and insufficient self-renewal, which contribute to injury and compromised heart function.

• Increasing evidence suggests that Hippo-Yes-associated protein (Yap) signaling can modulate cardiomyocyte death and proliferation in the postnatal heart.

• Neurofibromin 2 (NF2) has been linked to the Hippo-Yap pathway in mouse liver and brain, and is associated with arrhythmogenic cardiomyopathy (AC) in human and murine hearts.

What New Information Does This Article Contribute?

• NF2 interaction with mammalian sterile 20-like kinase 1 (Mst1) and large tumor suppressor kinase 2 (Lats2) in the cardiomyocyte nucleus is associated with Mst1 activation following oxidative stress.

• Phosphorylated (inactive) NF2 is decreased in failing human hearts and after ischemia/reperfusion in mouse hearts.

• Cardiomyocyte-specific deletion of NF2 confers protection against ischemia/reperfusion injury in the mouse heart.

• Yap transcriptional activity is upregulated in NF2 cardiomyocyte-specific knockout hearts, and evidence implicates heightened Yap function as a mediator of the observed cardioprotection.

Cardiomyocyte loss contributes to injury after MI. The Hippo-Yap pathway has emerged as an important modulator of cardiomyocyte survival and proliferation, yet mechanisms that underlie signaling initiation remain incompletely understood. These findings indicate that NF2 facilitates activation of the core Hippo pathway kinase Mst1, and negatively regulates the transcription co-factor Yap, to influence cardiomyocyte apoptosis and myocardial injury following ischemia/reperfusion.

Nonstandard Abbreviations and Acronyms

αMHC

alpha myosin heavy chain

AAR

area at risk

AC

arrhythmogenic cardiomyopathy

AMCMs

adult mouse cardiomyocytes

CKO

cardiomyocyte-specific knockout

IP

immunoprecipitation

I/R

ischemia/reperfusion

Lats2

large tumor suppressor kinase 2

LV

left ventricle

MI

myocardial infarction

Mst1

mammalian sterile 20-like kinase 1

MYPT-1

myosin phosphatase target subunit 1

NF2

neurofibromin 2

NRVMs

neonatal rat ventricular myocytes

PP1δ

serine/threonine-protein phosphatase 1 delta

RASSF1A

Ras association domain family protein 1A

Sav1

Salvador

TTC

triphenyltetrazolium chloride

TUNEL

terminal deoxynucleotidyl transferase

dUTP

nick end labeling

WT

wild-type

Yap

Yes-associated protein