Piezo1 is the cardiac mechanosensor that initiates the hypertrophic response to pressure overload

Ze-Yan Yu Victor Chang Cardiac Research Institute Hutao Gong Victor Chang Cardiac Research Institute Scott Kesteven Victor Chang Cardiac Research Institute Yang Guo Victor Chang Cardiac Research Institute Jianxin Wu Victor Chang Cardiac Research Institute Jinyuan Li Victor Chang Cardiac Research Institute Siiri Iismaa Victor Chang Cardiac Research Institute Xenia Kaidonis Victor Chang Cardiac Research Institute Robert Graham Victor Chang Cardiac Research Institute Charles Cox Victor Chang Cardiac Research Institute Michael Feneley St Vincent’s Hospital Boris Martinac (  b.martinac@victorchang.edu.au ) Victor Chang Cardiac Research Institute https://orcid.org/0000-0001-8422-7082


Introduction
Despite signi cant advances in cardiovascular medicine over the last 30 years, pathological left ventricular hypertrophy (LVH) secondary to pressure overload resulting from hypertension or aortic stenosis remains a powerful independent predictor of cardiovascular mortality and morbidity 1,2,3,4 . Thus far, the only treatment available for this condition is blood pressure reduction with anti-hypertensive medications or replacement of a stenotic aortic valve. These strategies do not fully reverse the pathological remodelling that occurs once LVH is established. Understanding the molecular mechanisms that drive LVH in response to pressure overload may open avenues to novel anti-hypertrophic therapies.
The two principal stimuli involved in LVH development are neuroendocrine hormones that activate Gq receptors (e.g. angiotensin II) and mechanical forces (e.g. pressure overload) 5 . These stimuli are thought to activate two distinct Ca 2+dependent signaling cascades that ultimately result in LVH: the calcineurin-nuclear factor of activated T-cells (NFAT)-GATA4 pathway and the calmodulin-dependent protein kinase II (CaMKII)-histone deacetylase 4 (HDAC4)-myocyte enhancer factor 2 (MEF2) pathway. Our previous experimental work has demonstrated that the development of LVH in response to transverse aortic constriction (TAC), the most common surgical model of pressure overload, is dependent on activation of the CaMKII-HDAC4-MEF2 pathway but does not require Gq receptor or calcineurin activation 6 .
Moreover, we have shown recently that the Ca 2+ -activated transient receptor potential melastatin 4 (TRPM4) ion channel acts as a positive regulator of pressure overload-induced cardiac hypertrophy, playing a key role in the activation of the CaMKII-HDAC4-MEF2 pathway 7 . Given that TRPM4 is not activated by membrane stretch 8 and that Gq receptor activation is not required, the question remains as to the identity of the molecule at the start of the hypertrophic signaling cascade that senses changes in mechanical load within the myocardium and transduces that mechanical signal into a chemical signal that activates TRPM4 and in turn the CaMKII-HDAC4-MEF2 pathway.
A prime candidate to act upstream of TRPM4 within this mechanosensory signaling cascade that drives LVH is the Ca 2+permeable mechanosensitive ion channel, Piezo1. Since the discovery and cloning of Piezo channels in 2010 9 , this family of structurally distinct ion channels has emerged as key sensors of biomechanical forces in the cardiovascular system 10 .
Despite the signi cant evidence for a key role of Piezo1 channels in vascular physiology and pathophysiology 11,12,13 , little is known about the role of Piezo1 in cardiac biology. A recent study has identi ed that Piezo1 is expressed at low levels in adult murine cardiomyocytes and is calcium permeable, increasing intracellular Ca 2+14 . Genetic ablation of Piezo1 using a cardiomyocyte-speci c Cre driver during early cardiac development resulted in a mild dilated cardiomyopathy that worsened as the mice aged 14 . However, whether Piezo1 expressed in cardiomyocytes is the mechanosensor responsible for initiating the CaMKII-HDAC4-MEF2 hypertrophic signaling cascade to induce LVH in response to pressure overload remains to be determined.
Here, using mice expressing a Piezo1 fusion protein (Piezo1-TdTomato) and a cardiomyocyte-speci c inducible knockout (Piezo1 KO) mouse model, we investigate the role of Piezo1 in LVH induced by pressure overload. We demonstrate that in response to TAC-induced pressure overload, cardiomyocyte expression of Piezo1 increases, and that deletion of cardiomyocyte Piezo1 completely prevents activation of the CaMKII-HDAC4-MEF2 pathway and inhibits the LVH observed in response to TAC. Moreover, we show that the loss of Piezo1 prevents the altered expression of a number of critical Ca 2+ handling proteins, including TRPM4 and the sodium-calcium exchanger (NCX1), that are associated with pressure overloadinduced LVH. Taken together, our new ndings demonstrate that Piezo1 is the mechanosensor that transduces the increased myocardial forces caused by pressure overload into a chemical signal that activates the hypertrophic signaling cascade resulting in pathological LVH.

Results
TAC induces identical levels of LVH in WT and Piezo1 P1-tdT/P1-tdT mice We hypothesized that Piezo1 is involved in the signalling cascade that drives pathological LVH in response to pressure overload. To test this hypothesis, we employed Piezo1 P1-tdT/P1-tdT mice that expressed a Piezo1-tdTomato fusion protein from the Piezo1 locus 12 . This reporter mouse allowed us to use Piezo1 fusion proteins to probe Piezo1 expression using a speci c mCherry antibody. We took this approach because there are no speci c mouse anti-Piezo1 antibodies available commercially. We performed transverse aortic constriction (TAC) on Piezo1 P1-tdT/P1-tdT mice and their wild type littermates (WTLs).
TAC increased LV systolic pressure (LVSP) by ~57 mmHg in both WTLs and Piezo1 P1-tdT/P1-tdT mice, when compared with their respective sham-operated controls at 14 days (Table 1, both p < 0.001). This signi cant pressure overload with TAC resulted in enlarged hearts (Table 1, Supplementary Fig. 1A) and signi cant LVH after 14 days in both groups, when compared with their sham-operated controls. The degree of LVH did not differ between WTLs and Piezo1 P1-tdT/P1-tdT mice, whether LVH was assessed by echocardiographically-determined LV mass, wall thickness (h), or wall thickness to chamber radius ratio (h/r), or by postmortem LV weight (LVW), whether normalized to body weight (LVW/BW) or tibial length (LVW/TL) ( Notably, TAC-induced LVH at 14 days was not associated with any evidence of LV dysfunction in either group: there were no signi cant changes in heart rate (HR), echocardiographically-determined LV end-diastolic volume (LVEDV), LV end-systolic volume (LVESV), LV ejection fraction (LVEF), cardiac output (CO), or dP/dt max , dP/dt min , lung weight or lung weight to body weight ratio (Table 1), indicating that our TAC model is a model of pressure overload LVH without ventricular decompensation or heart failure, as reported previously 6,7 .
Early markers of LVH induction in WT and Piezo1 P1-tdT/P1-tdT mice Consistent with the early induction of hypertrophic signalling after TAC 6,7 , gene expression of atrial natriuretic peptide (ANP, Nppa), brain natriuretic peptide (BNP, Nppb) and α-skeletal actin (α-SA, Acta1) was increased signi cantly 48 hours after TAC, preceding the development of signi cant LVH (supplementary Table 1), in both whole LV tissue and isolated LV cardiomyocytes, with no signi cant differences between WTLs and Piezo1 P1-tdT/P1-tdT hearts ( Table 2). The increased expression of these genes in whole LV tissue and isolated LV cardiomyocytes persisted 14 days after TAC in WTLs and Piezo1 P1-tdT/P1-tdT mice ( Table 2).

Pressure overload induces Piezo1 upregulation
Under baseline conditions, Piezo1 mRNA levels were low in isolated LV cardiomyocytes when compared to whole heart tissue, brain, aorta and lung (Fig. 1A). This is consistent with previous reports 9,11 . We used the Piezo1-tdTomato fusion protein to measure Piezo1 protein levels. Western blots demonstrated a clear band at ~320 kDa, indicative of the Piezo1-tdTomato fusion protein, in both whole heart tissue and isolated cardiomyocytes (Fig. 1B). Under baseline conditions, Piezo1 protein levels were low in isolated cardiomyocytes, particularly when compared with the aorta and lung (Fig. 1C).
Two days after TAC, however, Piezo1 mRNA expression increased signi cantly, ~2-fold in LV tissue (p < 0.01) and ~6-fold in isolated LV cardiomyocytes (p < 0.001, Fig. 1D), when compared with sham-operated hearts. The upregulation of Piezo1 mRNA expression was not maintained 14 days after TAC, by which time mRNA expression was similar to that in shamoperated controls in both LV tissue and isolated cardiomyocytes (Fig. 1D). In contrast, Piezo1 protein levels did not increase signi cantly in either LV tissue or isolated cardiomyocytes 2 days after TAC ( Fig. 1E and F), but did increase signi cantly 14 days after TAC in both LV tissue (~1.6-fold, p < 0.001) and isolated cardiomyocytes (~1.4-fold, p < 0.01) ( Fig. 1E and F).
Targeted deletion of Piezo1 from cardiomyocytes in the adult heart To further investigate whether Piezo1 plays an important role in the induction of LVH secondary to pressure overload, we generated an inducible cardiomyocyte-speci c Piezo1 knockout mouse (Piezo1 KO) that permitted targeted deletion of Piezo1 from cardiomyocytes in adult mice (8 weeks old) ( Fig. 2A, B). Tamoxifen inducible α-MHC-MerCreMer transgenic mice (Cre transgene under control of the α-myosin heavy-chain (Myh6) promoter 15 ) were crossed with Piezo1 / mice, and offspring backcrossed until P1 / MCM +/were generated (see Methods and Fig. 2B, C). To account for the potential of nonspeci c Cre-recombinase mediated cardiotoxicity 16 , P1 wt/wt MCM +/mice were designated as controls for Cre. P1 / MCM +/male mice aged 8-10 weeks were indistinguishable in cardiac function and anatomical parameters from ageand sex-matched P1 wt/wt MCM +/and P1 wt/wt MCM -/control mice ( Supplementary Fig. 2).
To establish a dosing regimen of tamoxifen that maximized Cre-recombinase activity at the Piezo1 locus but minimized tamoxifen-induced cardiotoxicity 16,17 , we injected different concentrations of tamoxifen into 8-week-old male P1 / MCM +/and P1 wt/wt MCM +/mice on 3 consecutive days, and allowed 10 days for them to recover. We injected the same volume of peanut oil into P1 / MCM +/mice as the treatment control. We observed that tamoxifen induced Cre recombinase activity in a dose-dependent manner (Supplementary Fig. 3A -C). Piezo 1 deletion was observed in the cardiomyocytes of mice injected with tamoxifen but not peanut oil. P1 / MCM +/mice treated with tamoxifen at 30 mg/kg/d exhibited normal heart structure and function when compared with mice treated with peanut oil (Supplementary Table 2). Higher doses of tamoxifen, 50 mg/kg/d or 100 mg/kg/d, caused LV dilatation with impaired contraction (Supplementary Table 2).
Consequently, a tamoxifen dose of 30 mg/kg/d was used in subsequent experiments.
As treatment for three consecutive days did not result in a satisfactory level of Piezo1 excision we increased the number of consecutive days of tamoxifen dosing. Ultimately, tamoxifen at 30 mg/kg/d for six consecutive days produced e cient inducible deletion of Piezo1 in P1 / MCM +/mice when measured ten days after the last injection with no impact on cardiac function (Fig 2). We con rmed successful Cre-recombination using PCR ampli cation of cardiac genomic DNA 18 . As shown in Figure 2C, Cre-mediated recombination of Piezo1 was detected only in the heart, with no excision identi ed in liver or lung tissues. A stronger signal was identi ed in isolated cardiomyocytes when compared to whole heart tissue from tamoxifentreated P1 / MCM +/mice, and no excision was evident in cardiomyocytes isolated from control mice treated with peanut oil (Fig.2 C).
Congruent with these results, and consistent with previous ndings of the relative e ciency for the αMHC-MerCreMer construct 15,19 , expression of Piezo1 mRNA was reduced by approximately 78% in isolated cardiomyocytes from tamoxifen treated P1 / MCM +/mice (Piezo1 KO mice, p < 0.001) when compared with mice injected with peanut oil (Fig. 2C, D). These ndings con rm that tamoxifen 30 mg/kg/d for 6 days induced speci c deletion of Piezo1 from cardiomyocytes.
Next, we assessed whether tamoxifen-induced deletion of Piezo1 in cardiomyocytes had any impact on baseline cardiac function and structure. To do this, we treated P1 / MCM +/mice with the same tamoxifen dose (30 mg/kg/d) as a control for Cre expression (termed α-MHC-MCM +/mice). We performed echocardiographic measurements 10 days after mice received their last injection of tamoxifen for Piezo1 KO mice, α-MHC-MCM +/mice and P1 / MCM +/control mice (peanut oil injected). The results from this analysis showed that there were no signi cant differences in body weight (Fig. 2E), heart rate ( Fig. 2F), cardiac function (Fig2. G-J) or left ventricular morphology between any of the three groups ( Fig. 2 K, L).
Cardiomyocyte-speci c deletion of Piezo1 inhibits the hypertrophic response to pressure overload TAC or sham surgery was performed on Piezo1 KO mice, α-MHC-MCM +/mice and P1 / MCM +/control mice. After 14 days, body weight, heart rate, LV end-diastolic and end-systolic volumes, ejection fraction and dP/dt were not signi cantly different between TAC or sham-operated animals, or between the three genotypes ( Fig.3, Supplementary Table 3).
Importantly, TAC induced the same increase in LV systolic pressure in all three genotypes (Fig. 3C). Echocardiographic indices of LVH -LV wall thickness to chamber radius ratio (h/r) and LV mass -increased signi cantly (both p < 0.001) after Cardiomyocyte-speci c deletion of Piezo1 inhibits myocardial brosis in response to pressure overload Pathological LVH is associated with cardiac brosis, with upregulated collagen expression and deposition. We evaluated interstitial cardiac brosis in response to pressure overload 14 days after TAC in Piezo1 KO hearts and α-MHC-MCM +/hearts using Masson's trichrome staining (Fig. 3L). There was increased interstitial cardiac brosis in α-MHC-MCM +/--TAC hearts when compared with their sham controls (p < 0.001), but there was no increase in cardiac brosis in Piezo1 KO hearts after TAC (Fig. 3L, M). Consistent with these ndings, Collagen III (Col3a1) mRNA expression increased ~6-fold in α-MHC-MCM +/--TAC hearts when compared with their sham controls (p < 0.001), but a more attenuated response was observed in Piezo1 KO hearts after TAC when compared with sham-operated hearts (~2-fold; p < 0.05, Fig. 3N) and with αMHC-MCM +/-TAC-operated hearts (p < 0.001, Fig. 3N).
Cardiomyocyte-speci c deletion of Piezo 1 inhibits early markers of induction of hypertrophy in response to pressure overload Although LVH had not yet developed 2 days after TAC ( Supplementary Fig. 4), induction of hypertrophic signalling in α-MHC-MCM +/hearts was already evident at this time, as was apparent from 5 to10-fold increases in gene expression of ANP (Nppa), BNP (Nppb) and α-SA (Acta1) (all p < 0.001). However, this early response of markers of hypertrophic signalling was completely absent in Piezo1 KO hearts 2 days after TAC (Fig. 4A). The increased expression of these hypertrophy-associated genes was maintained 14 days after TAC in α-MHC-MCM +/hearts (~4 to 10-fold, all p < 0.001), while an attenuated late response was observed in Piezo 1 KO mice 14 days after TAC (~2 to 5-fold, p < 0.001 for both Nppa and Nppb, not signi cant for Acta1, Fig. 4B).
Cardiomyocyte-speci c deletion of Piezo1 prevents activation of the CaMKII-HDAC4-MEF2 hypertrophic signalling pathway in response to pressure overload The cytoplasmic and the nuclear fractions of LV tissue were separated as described in the Materials and Methods. Purity of the isolated fractions was con rmed by western blot analysis using antibodies against marker proteins speci c for cytoplasmic (glyceraldehyde 3-phosphate dehydrogenase, GAPDH) and nuclear (Histone H2B) fractions ( Supplementary   Fig. 5).
As expected, α-MHC-MCM +/mice exhibited strong activation of the CaMKII-HDAC4-MEF2 hypertrophic signalling pathway 2 days after TAC (Fig. 5), consistent with our previous ndings in TAC-induced hypertrophy 6,7 . The hallmark of this activation is increased levels of both total and activated CaMKII, which is auto-phosphorylated at threonine 287 (p-CaMKII) 20,21 , and the resultant increase in the cytoplasmic to nuclear ratio of HDAC4 (p < 0.01), indicating nuclear export of HDAC4, with consequent de-repression of MEF2A (p < 0.001, Fig. 5).
Remarkably, Piezo1 KO mice failed to exhibit any evidence of activation of the CaMKII-HDAC4-MEF2 hypertrophic signalling pathway 2 days after TAC: the ndings in Piezo1 KO mice 2 days after TAC were indistinguishable from those in their shamoperated controls (Fig. 5). These data indicate that Piezo1 is essential for activation of the CaMKII-HDAC4-MEF2 hypertrophic signalling pathway in response to pressure overload-induced by TAC.
Cardiomyocyte-speci c deletion of Piezo1 is associated with activation of the calcineurin-NFAT hypertrophic signaling pathway in response to pressure overload As expected, α-MHC-MCM +/mice exhibited no evidence of activation of the calcineurin-NFAT hypertrophic signalling pathway 2 days after TAC (Fig. 6), consistent with our previous ndings in TAC-induced hypertrophy 6,7 . One explanation for this nding is that pressure overload activates CaMKII and activated CaMKII inhibits calcineurin activation 22 . Because Piezo1 KO mice exhibited no evidence of CaMKII activation in response to pressure overload, yet they exhibited incomplete inhibition of LVH, we postulated that the residual LVH was driven by calcineurin activation in the absence of its inhibition by CaMKII. The results obtained in Piezo1 KO mice 2 days after TAC supported this hypothesis: the hallmark of calcineurin activation, an increase in the nuclear to cytoplasmic NFAT ratio, indicating translocation of NFAT to the nucleus due to dephosphorylation by activated calcineurin, was clearly evident in Piezo1 KO mice 2 days after TAC (p < 0.01, Fig. 6A 7A, E and F, both p < 0.01), replicating our previous ndings in wild type mice 7 , but Piezo1 KO mice exhibited no change in TRPM4 mRNA 2 days after TAC (Fig. 7A, E and F), indicating that Piezo1 is upstream of TRPM4, and mediates the response of TRPM4 to pressure overload.
Conversely, the upregulation of Piezo1 mRNA expression 2 days after TAC in wild type hearts (p < 0.001, Fig. 7B) was not abolished in TRPM4 KO hearts (p < 0.01) 2 days after TAC, con rming that TRPM4 is downstream of Piezo1 in the response to pressure overload. Nevertheless, the magnitude of upregulation of Piezo1 mRNA 2 days after TAC was diminished signi cantly in TRPM4 KO mice (p < 0.001, Fig. 7B), suggesting that TRPM4 plays a signi cant role in the feedback regulation of Piezo1 in response to pressure overload.
α-MHC-MCM +/hearts exhibited signi cant increases in both NCX1 mRNA (Slc8a1) (Fig. 7C) and protein expression 2 days after TAC ( Fig. 7E and G, both p < 0.01). In contrast, Piezo1 KO hearts exhibited no change in NCX1 mRNA or protein levels 2 days after TAC (Fig. 7C, E and G), indicating that Piezo1 is upstream of NCX1, and mediates the response of NCX1 to pressure overload.
α-MHC-MCM +/hearts exhibited signi cant decreases in the gene expression of the T-type calcium channel (Cacna1h) 2 days after TAC (p<0.01), which was abolished in Piezo1 KO hearts (Fig. 7D), but the absence of any change at the protein level (Ca v3.2 ) 2 days after TAC in α-MHC-MCM +/hearts (Fig. 7H) casts some doubt on the functional signi cance of the Ttype calcium channel in mediating the response to pressure overload governed by Piezo1.

Discussion
Recently, we reported that the TRPM4 ion channel is a positive regulator of LVH induced by pressure overload, playing a key role in the activation of the CaMKII-HDAC4-MEF2 hypertrophic signaling pathway 7 . Activation of CaMKII and downstream hypertrophic signaling were inhibited after TAC in a mouse model in which TRPM4 was deleted in cardiomyocytes (TRPM4 KO), and this was associated with signi cant inhibition of the hypertrophic response to pressure overload. Since TRPM4 is not a stretch-activated 8  These new observations support the hypertrophic signalling pathway outlined schematically in Figure 8.  30 , and it is known that aortic constriction provides this type of calcium signal 31 . In contrast, calcineurin activation requires a sustained increase in the resting intracellular calcium level 32 . There is evidence that calmodulin can distinguish high-amplitude versus low-amplitude calcium signals via the differential calcium sensitivity of its N-and C-lobes 33,34 , and may thus provide a gatekeeper role in the differential activation of the CaMKII-dependent and calcineurin-dependent hypertrophic signalling pathways (Fig. 8). In addition, once activated, CaMKII inhibits calcineurin activation 22 .
We reported previously that calcineurin activation was not required for the hypertrophic response to pressure overload with TAC, which was associated with activation of the CaMKII-HDAC4-MEF2 pathway 6 . In the present study, we found that cardiomyocyte-speci c deletion of Piezo1 prevented any activation of the CaMKII-HDAC4-MEF2 signalling pathway secondary to TAC, and reduced the amount of LVH developed after TAC by approximately 60%. While this observation strengthens the causal link between activation of the CaMKII-HDAC4-MEF2 signalling pathway and the induction of LVH secondary to pressure overload 22 , it might reasonably have been expected that the complete absence of CaMKII activation secondary to pressure overload in Piezo1 KO hearts would prevent any hypertrophic response. Our ndings support the conclusion, however, that the residual LVH in Piezo1 KO hearts after TAC is due to the absence of the usual inhibition of calcineurin activation by activated CaMKII. We observed no evidence of calcineurin activation after TAC in the α-MHC-MCM +/-(control) hearts, consistent with our previous report 6 , but in Piezo1 KO hearts in which CaMKII activation after TAC was abolished, calcineurin activation was clearly evident. It is notable, however, that calcineurin activation in Piezo1 KO hearts after TAC produced signi cantly less LVH than observed with CaMKII activation when Piezo1 was present and calcineurin activation was absent, emphasising the importance of the CaMKII-HDAC4-MEF2 signalling pathway in pressure overload LVH.
In the same previous report 6 , we also demonstrated that Gq receptor activation was not required for the hypertrophic response to pressure overload with TAC since we found the same amount of LVH after TAC in the presence and absence of functioning Gq receptors. This raises the question: how is calcineurin activated after TAC in Piezo1 KO mice? It is important to note that TAC increases diacylglycerol (DAG), a downstream marker of Gq activation 35 . The implication of the present results, therefore, is that Gq is indeed activated directly by TAC, but that activation has no impact on the hypertrophic response in the presence of Piezo1 activation because Piezo1 activation results in CaMKII activation, which inhibits Gqdependent calcineurin activation. When Piezo1 is deleted, CaMKII is not activated and Gq-dependent calcineurin activation in response to TAC is no longer inhibited. We observed the same phenomenon of calcineurin activation associated with residual LVH after TAC in the TRPM4 KO mouse, in which CaMKII activation was strongly inhibited 7 .
Although pressure overload resulted in a signi cant increase in Piezo1 mRNA 2 days after TAC, Piezo1 protein was not increased at this early time point when evidence of hypertrophic signalling is already abundant. Since we have shown that Piezo1 is the instigator of the hypertrophic signalling cascade after TAC, it would seem that the normal very low expression of Piezo1 protein is su cient for this purpose due to the ampli cation of the calcium signal by TRPM4. This raises an interesting question regarding the late increase in Piezo1 protein observed 14 days after TAC, a time when the hypertrophic response to TAC has already plateaued 6 . We speculate that stretch-activated Ca 2+ entry via Piezo1 may be maximal when Piezo1 protein expression is low, and that increased Piezo1 protein production may act to reduce stretch activation and Ca 2+ entry by cooperative load sharing between the increased Piezo1 molecules, and this is the subject of our ongoing research.
An independent research group has very recently reported inhibition of hypertrophy after TAC in a Piezo1 KO mouse model based on the same Cre recombinase strategy used in our experiments 36 . While the inhibition of pressure overload hypertrophy with Piezo1 deletion is con rmatory, the authors did not examine the mechanism of the inhibition of hypertrophy in vivo, and did not examine the CaMKII-HDAC4-MEF2 hypertrophic signalling pathway or the central role of produces a tonic increase in the calcium concentration 37,38 , which is well known to activate calcineurin 32 , but this is not relevant to the high-amplitude calcium signal produced by aortic constriction and necessary for CaMKII activation 30,31 .
Given the signi cant differences also in maturity and terminal differentiation between NRVCMs and adult cardiomyocytes and the absence of data on CaMKII activation in their study of NRVCMs, it is di cult to determine the relevance of the NRVCM data to our ndings in adult mice. We did demonstrate in adult mice in vivo, however, that due to inhibition of calcineurin activation by activated CaMKII, calcineurin activation is absent in the hypertrophic response to TAC initiated by Piezo1.
In summary, our study not only identi es Piezo1 as the primary instigator of hypertrophic signalling in response to pressure overload, it provides a mechanistic explanation for the central role of TRPM4 in amplifying the initial Ca 2+ signal via NCX1 to activate CaMKII via calmodulin, and thus activate the CaMKII-HDAC4-MEF2 hypertrophic signalling pathway. In addition, our new ndings reveal the potency of the inhibition of calcineurin activation by activated CaMKII as the main explanation for the apparent segregation of the CaMKII-mediated and calcineurin-mediated hypertrophic signalling pathways, despite the fact that activation of both CaMKII and calcineurin is Ca 2+ -calmodulin dependent. The closely cooperative role we have demonstrated between Piezo1 and TRPM4 in providing a stretch-activated biological signal seems likely to provide a universal paradigm for Piezo1's broader role in all biological tissues. These new ndings may also provide targets for the development of novel therapies to prevent or reverse pathological hypertrophy and its harmful clinical sequelae.

Materials And Methods
Mice and genotyping Reactions were separated on 2% agarose gels yielding the following band sizes: P1 + : 160 bp, P1 f : 330 bp, P1 − : 230 bp 18 . All mice were on the C57BL/6J genetic backgroud, and male mice aged between 8-13 weeks old were used throughout.

Induction of LVH
As previously described 6,7 , male mice were subjected to TAC to induce pressure overload. Mice were anesthetized with 5% iso urane and ventilated at 120 breaths/min (Harvard Apparatus Rodent Ventilator). The transverse aortic arch was accessed via an incision in the second intercostal space, and constricted with a ligature tied around a 25-gauge needle, which was then removed. The TAC procedure was modi ed from a published paper 39 . Sham mice underwent the same procedure but the ligature was not tied. Simultaneous direct pressure recordings (1.4 F pressure catheter, AD Instruments, P/L) from both the right carotid artery and the aorta distal to the ligature (n=20 mice) indicated a TAC pressure gradient of 60 ± 8 mmHg with this technique. Animals were sacri ced after 2 days or 14 days.

Echocardiographic measurements
As previously described 40 , echocardiography was performed using an MS400 18-38 MHz transducer probe and VEVO 2,100 ultrasound system (VisualSonics Inc., Canada). The mice were anesthetized (3-5% iso urane for induction, 1-2% iso urane for maintenance with adjustment to maintain heart rate at ~500 bpm) and imaged at the endpoint of study to assess cardiac function. The acquisition of images and evaluation of data were performed by an operator blinded to treatment.

Invasive hemodynamic measurements
As previously described 6,7 , after 14 days of sham or TAC, mice were anesthetized by inhalation of iso urane (1.5%) and a 1.4F micro-tip pressure catheter (Millar Instruments Inc, Houston, Texas, USA) was inserted into the left ventricle via the right carotid artery. The heart rate, aortic systolic pressure, LV systolic pressure, +dP/dt, and -dP/dt were recorded (LabChart 6 Reader, AD Instruments, P/L). Animals were sacri ced, and the heart weight (HW) and left ventricle weight (LVW) normalized to body weight (BW) and to tibia length (TL) were measured as indicators of LVH.
Mouse LV cardiomyocyte isolation and puri cation As previously described 7 , the mice were heparinized and euthanized according to the Animal Research Act 1985 No 123 (New South Wales, Australia). Hearts were dissected and perfused through the aorta and the coronary arteries by 10 mL pH 7.2 perfusion buffer containing 135 mM NaCl, 4 mM KCl, 1 mM MgCl 2 , 0.33 mM NaH 2 PO 2 , 10 mM HEPES, 10 mM Glucose, 10 mM 2,3-Butanedione 2-monoxime (BDM), and 5 mM Taurine, with a Langendorff apparatus at 37 degrees for 5 minutes.
Next, 30 mL digestion buffer composed of the above solution and Collagenase B, D (dose by BW: 0.4 mg/g, Roche) and Protease Enzyme Type XIV (dose by BW: 0.07 mg/g, Sigma-Aldrich) was used to perfuse the hearts for 15 minutes. After the perfusion, the heart was removed from the setup and placed into a pH 7.4 transfer buffer containing 135 mM NaCl, 4 mM KCl, 1 mM MgCl 2 , 0.33 mM NaH 2 PO 2 , 10 mM HEPES, 5.5 mM Glucose, 10 mM BDM, and 5 mg/mL BSA. Both atria and the right ventricle were discarded, and the LV muscle was torn into small pieces and gently triturated in transfer buffer with a pipette to isolate cardiomyocytes. The suspension was then ltered through a 200 micro lcon cup lter (BD), and centrifuged at 20xg for 2 minutes. After that, the cardiomyocytes were puri ed by a method described previously 41 . We con rmed that rod-shaped cardiomyocytes accounted for more than 85% of the total puri ed cardiomyocytes. The isolated cardiomyocytes were frozen immediately in liquid nitrogen and stored at -80°C for following experiments.

Quantitative Real-Time Polymerase Chain Reaction (RT-PCR)
Gene expression was determined by quantitative RT-PCR. Total RNA was extracted and puri ed from LV tissue and isolated cardiomyocytes with the RNeasy Fibrous Tissue Mini Kit (QIAGEN), following the manufacturer's protocol. RNA (500 ng) was reverse transcribed into cDNA using the SuperScript III First-Strand Synthesis SuperMix kit (Invitrogen). cDNA was subjected to PCR ampli cation to detect ANP (Nppa), BNP (Nppa), α-SA (Acta1), collagen III (Col3a1), Piezo1,Trpm4, Slc8a1, Cacna1h gene expression, using the PCR master mix LightCycler 480 SYBR Green I Master (Invitrogen) and performed with the CFX384 Touch Real-Time PCR Detection System (Bio-Rad). Samples were run in technical triplicates and the mRNA expression levels were normalized to those of GAPDH to calculate relative gene expression using the delta-delta Ct method.
Three independent experiments yielded similar results. The mouse RT-PCR primers (Sigma-Aldrich) used are shown in Supplementory Table 4.

Western blotting
For total protein extraction, LV tissue and isolated cardiomyocytes were lysed in a pH 7.4 lysis buffer containing 150 mM NaCl, 50 mM Tris-HCL, 1% Triton X-100, 1 mM sodium orthovanadate, 1 mM beta-glycerophosphate, 5 mM dithiothreitol and MiniComplete protease inhibitors (Roche). The cytoplasmic and the nuclear fractions of LV tissue were separated as described previously with con rmed high fraction purity 6,7 . Brie y, LV tissue was lysed using NE-PER nuclear and cytoplasmic extraction reagents (Pierce Biotechnology) and  Table 4) were used at room temperature for one hour.
Immunologic detection was accomplished using SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermo Scienti c). Protein levels were quanti ed by densitometry using ImageJ (NIH) software. Protein levels were normalized to relative changes in Histone H2B for the nuclear fraction and GAPDH for the cytoplasmic fraction and expressed as fold changes relative to those of control animals. Experiments were repeated independently three times with similar results.

Histological analysis
As previously described 7 , Masson's trichrome stain was used to quantify brosis in the LV (collagen bres stain blue). The hearts were excised from iso urane-euthanized mice and washed with phosphate-buffered saline (PBS). Hearts were then cut longitudinally in the coronal plane, embedded into optimal cutting temperature (OCT) compound (Sakura Finetek), and gradually frozen in melting isopentate, precooled in liquid nitrogen to avoid tissue damage, and stored at -80°C for following experiments. Serial 6 mm sections were sliced with a cryostat (Leica) and stained using a Masson's trichrome staining kit (Sigma-Aldrich), following the manufacturer's instructions. Images of the LV were obtained with 4 to 6 elds per section 42 using a bright eld microscope (Leica). Blue-stained areas of brosis within sections were determined using colour-based thresholding 43 and measured with ImageJ software (NIH; http://rsbweb.nih.gov/ij/). The percentage of total brotic area was calculated by taking the sum of the blue-stained areas divided by the total LV area.

Statistics
All experiments and analyses were blinded. Averaged data are presented as mean ± standard error of the mean (SEM). The statistical analyses were performed using GraphPad Prism software, version 7.04 (GraphPad). For comparisons among multiple sets of data with one factor or two factors, one-way or two-way ANOVA was used accordingly, followed by Tukey's post-hoc test. p < 0.05 was considered statistically signi cant.

Source Data
All source data are provided with this paper. 3.81 ± 0.50 *** Table 2. Early markers of LVH induction in response to left ventricular pressure overload in WTLs and Piezo1 P1-tdT/P1-tdT mice. Relative mRNA expression of ANP (Nppa), BNP (Nppb) and α-SA (Acta1) in LV tissue and cardiomyocytes (CMs) 2 days and 14 days after TAC compared to sham (n = 3-5/group). The relative mRNA expression was normalized with GAPDH and calculated as fold change relative to sham in 2 days and 14 days groups, respectively. Results are presented as mean ± SEM, Two-way ANOVA followed with Tukey's post-hoc test, *p < 0.05, **p < 0.01, ***p < 0.001, compared between sham-and TAC-operated groups. quantitative densitometric analysis of 320 kDa Piezo1 protein expression (bottom panel) from isolated cardiomyocytes (CMs), brain, whole heart, aorta and lung tissues collected from Piezo1P1-tdT/P1-tdT mice under baseline conditions. Relative Piezo1 mRNA and protein expression from isolated cardiomyocytes, brain, heart, aorta, and lung were normalized to GAPDH (n = 3/group). (D) Relative Piezo1 mRNA expression in LV tissue and in LV cardiomyocytes (CMs) 2 days and 14 days after TAC or sham surgery (n = 5/group) from Piezo1P1-tdT/P1-tdT mice. (E) Representative western blots of Piezo1 protein expression in LV tissue and in LV cardiomyocytes. (F) Quantitative densitometric analysis of 320 kDa Piezo1 protein expression from LV tissue and LV cardiomyocytes 2 days and 14 days after TAC from Piezo1P1-tdT/P1-tdT mice (n = 4/group). Relative Piezo1 mRNA and protein expression in the LV tissue and cardiomyocytes were normalized to GAPDH and calculated as fold change relative to sham in the 2 day and 14 day of groups, respectively. Results are presented as mean ± SEM, Two-way ANOVA followed with Tukey's post-hoc test, **p < 0.01, ***p < 0.001 vs. sham-operated groups, n.s., no signi cant difference. Figure 1-source data 1. Source data le (Excel) for Figure 3A, C, D and F. T-test and One-way ANOVA followed with Bonferroni post-hoc test, ***p < 0.001 vs. peanut oil treated cardiomyocytes. Figure 2-source data 1. Source data le (Excel) for Figure 4C, D, E, F, G, H, I, J, K and L. were normalized with GAPDH and Histone H2B, respectively. Fold changes and cytoplasmic/nuclear ratios were calculated relative to sham groups for each genotype (n = 6/group). Results are presented as mean ± SEM, Two-way ANOVA followed with Tukey's post-hoc test, **p < 0.01, ***p < 0.001 vs. sham-operated group.  Western blots showing the expression of NFAT and GATA4 in cytoplasm (left) and nucleus (right). (B) Cytoplasmic and nuclear (right) quantitative data were normalised by GAPDH and Histone H2B, respectively (n = 6/group). Fold changes and nuclear/cytoplasmic ratio were calculated relative to sham groups, in each genotype. Results are presented as mean ± SEM, Two-way ANOVA followed with Tukey's post-hoc test, **p < 0.01, ***p < 0.001 vs. sham-operated group. (n = 5/group) and protein expression (n = 6/group) in the LV tissue was normalized to GAPDH and calculated as fold change relative to the 2 days post-sham group, respectively. Results are presented as mean ± SEM, Two-way ANOVA followed with Tukey's post-hoc test, **p < 0.01 vs. sham-operated groups; ### p < 0.001 vs. TAC-operated groups. Figure 7source data 1. Source data le (Excel) for Figure 6A, B, C, D, F, G and H.

Figure 8
Schematic of the Piezo1-dependent signalling pathway that drives LVH secondary to pressure overload. Piezo1 is the cardiomyocyte mechanotransducer that converts the increased mechanical forces secondary to pressure overload into Ca2+ entry into the cell. This increases local [Ca2+], resulting in TRPM4 activation. The Na+-permeable TRPM4 activity then likely induces reverse activity of the Na+/Ca2+ exchanger through local Na+ loading. This leads to a high-amplitude increase in local [Ca2+]. Calmodulin responds to this high-amplitude Ca2+ stimulus through the lower a nity Ca2+ binding site at its N-lobe, which then preferentially activates CaMKII and thus stimulates the CaMKII-HDAC4-MEF2 pathway to induce LVH6. In contrast, calcineurin is activated preferentially by low-amplitude Ca2+ signalling via Gq-coupled receptors and calmodulin, and calcineurin activation is strongly inhibited by activated CaMKII. ECM: extracellular matrix, CaM: calmodulin.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.