Caveolin 3‐dependent loss of t‐tubular I Ca during hypertrophy and heart failure in mice

New Findings What is the central question of this study? Heart failure is associated with redistribution of L‐type Ca2+ current (I Ca) away from the t‐tubule membrane to the surface membrane of cardiac ventricular myocytes. However, the underlying mechanism and its dependence on severity of pathology (hypertrophy versus failure) are unclear. What is the main finding and its importance? Increasing severity of response to transverse aortic constriction, from hypertrophy to failure, was accompanied by graded loss of t‐tubular I Ca and loss of regulation of I Ca by caveolin 3. Thus, the pathological loss of t‐tubular I Ca, which contributes to impaired excitation–contraction coupling and thereby cardiac function in vivo, appears to be attributable to loss of caveolin 3‐dependent stimulation of t‐tubular I Ca. Abstract Previous work has shown redistribution of L‐type Ca2+ current (I Ca) from the t‐tubules to the surface membrane of rat ventricular myocytes after myocardial infarction. However, whether this occurs in all species and in response to other insults, the relationship of this redistribution to the severity of the pathology, and the underlying mechanism, are unknown. We have therefore investigated the response of mouse hearts and myocytes to pressure overload induced by transverse aortic constriction (TAC). Male C57BL/6 mice underwent TAC or equivalent sham operation 8 weeks before use. I Ca and Ca2+ transients were measured in isolated myocytes, and expression of caveolin 3 (Cav3), junctophilin 2 (Jph2) and bridging integrator 1 (Bin1) was determined. C3SD peptide was used to disrupt Cav3 binding to its protein partners. Some animals showed cardiac hypertrophy in response to TAC with little evidence of heart failure, whereas others showed greater hypertrophy and pulmonary congestion. These graded changes were accompanied by graded cellular hypertrophy, t‐tubule disruption, decreased expression of Jph2 and Cav3, and decreased t‐tubular I Ca density, with no change at the cell surface, and graded impairment of Ca2+ release at t‐tubules. C3SD decreased I Ca density in control but not in TAC myocytes. These data suggest that the graded changes in cardiac function and size that occur in response to TAC are paralleled by graded changes in cell structure and function, which will contribute to the impaired function observed in vivo. They also suggest that loss of t‐tubular I Ca is attributable to loss of Cav3‐dependent stimulation of I Ca.


INTRODUCTION
Contraction of cardiac ventricular myocytes is initiated by Ca 2+ influx across the cell membrane during the action potential (AP), via L-type Ca 2+ channels (generating the L-type Ca 2+ current, I Ca ; for review see Bers, 2002). This triggers Ca 2+ -induced Ca 2+ release (CICR) from adjacent sarcoplasmic reticulum (SR), which produces the Ca 2+ transient that activates contraction. Both I Ca and CICR occur predominantly at the t-tubules, which are invaginations of the surface membrane that ensure synchronous Ca 2+ release throughout the cell (Cheng, Cannell, & Lederer, 1994;Kawai, Hussain, & Orchard, 1999) and are therefore central to excitation-contraction (E-C) coupling.
During heart failure, Ca 2+ release from the SR is compromised (Bers, 2006;Gómez et al., 1997a), although most studies show little This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.  (Benitah et al., 2002;Mukherjee & Spinale, 1998;Richard et al., 1998). However, recent work has shown that heart failure induced in rats by coronary artery ligation (CAL) is associated with a decrease in I Ca density at the t-tubules and a compensatory increase at the cell surface, which results in a more uniform spatial distribution of I Ca and little overall change in I Ca density (Bryant et al., 2015); similar changes have been reported in failing human hearts (Sanchez-Alonso et al., 2016). The more uniform distribution might help to explain the paradoxical lack of change in whole-cell I Ca density despite the loss of t-tubules reported in previous studies (Balijepalli et al., 2003;Crossman, Ruygrok, Soeller, & Cannell, 2011;He et al., 2001;Lyon et al., 2009;Song et al., 2006;Wei et al., 2010). Nevertheless, the loss of I Ca at the t-tubules is associated with impaired local Ca 2+ release (Bryant et al., 2015), which will contribute to the smaller slower New Findings

• What is the central question of this study?
Heart failure is associated with redistribution of L-type Ca 2+ current (I Ca ) away from the t-tubule membrane to the surface membrane of cardiac ventricular myocytes.
However, the underlying mechanism and its dependence on severity of pathology (hypertrophy versus failure) are unclear.
• What is the main finding and its importance?
Increasing severity of response to transverse aortic constriction, from hypertrophy to failure, was accompanied by graded loss of t-tubular I Ca and loss of regulation of I Ca by caveolin 3. Thus, the pathological loss of t-tubular I Ca , which contributes to impaired excitation-contraction coupling and thereby cardiac function in vivo, appears to be attributable to loss of caveolin 3-dependent stimulation of t-tubular I Ca .
systolic Ca 2+ transient, and thus contraction, observed in heart failure (Frisk et al., 2016;Louch et al., 2004Louch et al., , 2006. However, it is unclear how these changes are related to the degree of pathology, in particular the transition between hypertrophy and overt failure, although it has recently been suggested that classification of failing hearts might, in future, be based 'less on measures of overall organ function such as ejection fraction, and more on the molecular details of myocyte, and in particular t-tubule, pathophysiology' (Hong & Shaw, 2017). The cause of the redistribution of I Ca is also unknown, although the scaffolding protein caveolin 3 (Cav3) might play a role (Bryant et al., 2015;Bryant, Kong, Cannell, Orchard, & James, 2018), because it appears to localize I Ca to the t-tubules (Bryant et al., 2014) and undergoes downregulation in heart failure (Feiner et al., 2011;Fridolfsson & Patel, 2013).
We have, therefore, performed a post hoc analysis of the response to transverse aortic constriction (TAC) in mice, to investigate the relationship between the whole-heart and cellular responses to TAC, and investigated the effect of decreasing protein binding to Cav3 on I Ca in ventricular myocytes from normal and TAC hearts.

Ethical approval
All animal procedures were approved by the Animal Welfare and Ethics Data from 34 male wild-type (WT) C57Bl/6 mice, specifically bred at the University of Bristol, are reported in this study; surgery was performed at 12 weeks of age and myocyte isolation at 20 weeks of age. Mice were kept in a temperature-controlled, enriched environment with ad libitum access to food and water.
Transverse aortic constriction was used to produce pressure overload, which results in cardiac hypertrophy and heart failure (Rockman et al., 1991;Tachibana, Naga Prasad, Lefkowitz, Koch, & Rockman, 2005). Briefly, animals were anaesthetized with ketamine (75 mg kg −1 I.P., Zoetis UK Ltd., London, UK) and medetomidine (1 mg kg −1 I.P., Orion Corporation, Espoo, Finland) and given buprenorphine (0.05 mg kg −1 S.C., Reckitt Benckiser Health Care (UK) Ltd., Hull, North Humberside, UK) for pain relief; the surgical plane of anaesthesia was monitored using the limb withdrawal reflex. The aortic arch was exposed via a medial sternal thoracotomy and a silk ligature (6-0) placed between the innominate and left carotid arteries and tied round a 27-gauge needle (0.41 mm o.d.). Sham animals underwent the same operation but without placement of the banding suture. Animals were maintained postoperatively for 8 weeks before use. There was no significant difference between sham-operated and age-matched control animals for any of the variables measured, so these data were pooled to form the control group. The ratio of lung weight to tibia length (LW:TL) was taken as an index of heart failure and used to separate the TAC animals into two groups: hypertrophy (HTY) and heart failure (HF; see Results).
In vivo cardiac function was monitored using echocardiography.
Animals were anaesthetized (isoflurane 1-3%, Merial Animal Health Ltd., Harlow, Essex, UK), heart rate was monitored, and measurements of contraction were made from M-mode images acquired from the parasternal short-axis view using a Vevo 3100 (FUJIFILM VisualSonics Inc., Toronto, Ontario, Canada) and an MX550D transducer.

Myocyte isolation and detubulation
Animals were injected with heparin (500 I.U., I.P.) and killed by cervical dislocation. The heart was rapidly excised and Langendorff perfused at 37 • C, and myocytes were isolated using a standard enzymatic method (Bryant et al., 2014). Myocyte detubulation (DT; physical and functional uncoupling of the t-tubules from the surface membrane) was achieved using formamide-induced osmotic shock, as described previously (Brette & Orchard, 2003;Bryant et al., 2014Bryant et al., , 2015Kawai et al., 1999); this procedure has no effect on I Ca , the action potential or the systolic Ca 2+ transient in rat atrial cells, which lack t-tubules (Brette, Komukai, & Orchard, 2002).
C3SD peptide (Pepceuticals Ltd., Enderby, Leicestershire, UK) was used to disrupt binding of Cav3 to its protein partners (Bryant et al., 2014;Couet, Li, Okamoto, Ikezu, & Lisanti, 1997;Feron et al., 1998;MacDougall et al., 2012). Myocytes were incubated in 1 mol l −1 TAT-C3SD for 45 min at room temperature before use; we have previously shown that a scrambled version of this peptide has no effect on I Ca (Bryant et al., 2014). C3SD corresponds to the scaffolding domain of Cav3; it was designed to compete with endogenous Cav3 for its protein partners (Couet et al., 1997) and has previously been shown to inhibit Cav3-dependent signalling (Feron et al., 1998;MacDougall et al., 2012).

Recording and analysis of I Ca
Myocytes were placed in a chamber mounted on a Nikon Diaphot inverted microscope. Membrane currents and cell capacitance were recorded using the whole-cell patch-clamp technique, using an Axopatch 200B, Digidata 1322A A/D converter and pClamp 10 (Molecular Devices, LLC. San Jose, CA, United States). Pipette resistance was typically 1.5-3 MΩ when filled with pipette solution, and pipette capacitance and series resistance were compensated by ∼70%. Currents were activated from a holding potential of −80 mV by a step depolarization to −40 mV for 200 ms (to inactivate the sodium current, I Na ) followed by steps to potentials between −50 and +80 mV for 500 ms, before repolarization to the holding potential, at a frequency of 0.2 Hz. The amplitude of I Ca (in picoamperes) was measured as the difference between the peak inward current and the current at the end of the depolarizing pulse, and was normalized to cell capacitance (in picofarads; a function of membrane area) to calculate I Ca density (in picoamperes per picofarad). The inactivation phase of I Ca , during the step depolarization, was fitted to a double exponential function to give fast ( f ) and slow ( s ) time constants. Surface membrane current density was obtained from currents measured in DT myocytes after correction for incomplete DT, as described previously (Bryant et al., 2014(Bryant et al., , 2015Horiuchi-Hirose et al., 2011;Kawai et al., 1999); t-tubular membrane current density was calculated from the loss of membrane current and capacitance after DT. Detubulation efficiency, measured from images of intact and DT cells stained with di-8-ANEPPS, was ∼91%, and was not different between the groups.

Confocal imaging and analysis of cell membrane
The surface and t-tubular cell membranes were stained by incubating cells with 5 mol l −1 di-8-ANEPPS for 10 min. Volume imaging of labelled cells was carried out using an LSM 880 confocal microscope (Zeiss, Carl Zeiss AG, Oberkochen, Germany) in Airyscan 'superresolution' mode, with a 1.4 numerical aperture (NA) ×63 oil immersion objective, with voxel size set to 40 nm in plane and 180 nm along the optical axis. Image analysis was performed using MATLAB R2015a (Mathworks Inc., Natick, MA, USA) and ImageJ (v1.50; National Institutes of Health, Bethesda, MD, USA). The regularity of t-tubule staining was quantified by applying a two-dimensional (2D) fast Fourier transformation to an offset-subtracted square region of the interior of the cell, and the power of the first harmonic normalized to that of the average image intensity (P 1 /P 0 ). The t-tubule density and orientation were obtained by first processing the volumetric data with a tubuleenhancing three-dimensional (3D) filter, segmenting using an Otsu threshold [graythresh(), Matlab, Mathworks Inc.] and skeletonizing using 'Skeletonize (2D/3D)' in ImageJ. The skeleton was then used to calculate t-tubule density (length per cell volume, in micrometres per cubic micrometre) and as a mask to locate Eigenvectors for extracting t-tubule angles. Tubule orientation is expressed relative to the z-disc plane, so that 0 deg corresponds to a transverse tubule, whereas 90 deg corresponds to a tubule that is orthogonal to the z-disc.

Confocal measurement of intracellular Ca 2+ and membrane potential
Cells were imaged using a water immersion objective with 1.2 NA and ×40 magnification, with the confocal pinhole set to 1 Airy unit. Linescans were recorded at 0.5-1.0 ms per line. Cells were field stimulated at 0.2 Hz at 1.5 × threshold using parallel Pt wires.
Intracellular Ca 2+ and membrane potential were recorded simultaneously at a single t-tubule in myocytes loaded with the Ca 2+ indicator fluo-4 AM (5 mol l −1 for 25 min; Thermo Fisher Scientific, Waltham, MA, USA) and the voltage-sensitive dye di-4-AN(F)EPPTEA (0.5-1 g ml −1 for 15 min; supplied by Dr Leslie Loew; Yan et al., 2012), as described previously (Bryant et al., 2015). Line scans were recorded along a t-tubule, with excitation at 514 nm and emitted fluorescence collected between 518 and 560 nm for Ca 2+ , and between 590 and 700 nm for voltage. Release of Ca 2+ at the t-tubule was determined as described previously (Bryant et al., 2015). In brief, the di-4-AN(F)EPPTEA fluorescence was used to determine the upstroke of the AP at the t-tubule. The latency of Ca 2+ release at each point along the scan was determined from the time between the upstroke of the AP and the time when the Ca 2+ signal became >5 SD above the prestimulus value. The latency to the time of maximal rate of rise of Ca 2+ was also determined, and the SD of latencies for each cell was used as a measure of the heterogeneity of release.
Spatially averaged (whole-cell) Ca 2+ transients were recorded via 1 ms line scans along the length of cells loaded with fluo-4 AM only, using 488 nm excitation and recording fluorescence between 495 and 650 nm.

Western blot analysis
Samples of myocyte lysates (10 g protein) were run on 4-15% gradient SDS-PAGE gels and transferred onto an Immobilon-P membrane. The blot was probed with antibodies to Cav3 (BD Transduction Laboratories, San Jose, CA, USA; 610420, dilution 1:5000),

Statistics
Data are expressed as means ± SD, unless stated, of N animals for in vivo data and n/N cells/animals for in vitro data, as shown in the figure legends. Data normality was assessed using the Shapiro-Wilk test, and subsequent testing was performed using Student's unpaired t test or the Mann-Whitney U test; one-way ANOVA (post hoc tests used Bonferroni correction) or the Kruskal-Wallis test.
The I Ca density-voltage relationship curves were analysed using a repeated-measures (RM) ANOVA, with voltage and intervention (i.e. DT and C3SD) as factors (Figures 6 and 7); I Ca properties elicited by a step depolarization to a single voltage were analysed with twoway ANOVA; post hoc tests used the Bonferroni correction ( Figures 6   and 7). The errors in derived variables (specifically, I Ca density at the t-tubule membrane; Figure 6) and the subsequent statistical analysis (Student's unpaired t test), were calculated using propagation of errors from the source measurements and thus, to indicate the precision with which the mean value has been estimated, are expressed as SEM. The limit of statistical confidence was taken as P < 0.05.

Severity of response to TAC
Transverse aortic constriction resulted in cardiac hypertrophy, indicated by a significant increase in the ratio of heart weight to tibia length (HW:TL ratio; in grams per millimetre × 10 3 ) from 10.5 ± 1.9 (N = 15) in control animals to 17.0 ± 4.8 (N = 19) in TAC animals (P < 0.001; Figure 1a). Pulmonary congestion, a symptom of congestive heart failure, also increased in TAC animals, indicated by an increase in LW:TL (in grams per millimetre × 10 3 ) from 7.8 ± 1.2 (N = 15) in control animals to 12.9 ± 6.6 (N = 19) after TAC We therefore used LW:TL as an index of heart failure to separate the TAC cohort into two groups: hypertrophy (HTY) and failing (HF), defined as LW:TL < 2 SD and LW:TL > 2 SD above the mean of the control group, respectively (indicated by the dashed line in Figure 1b).

Cell size and structure
Analysis of cell morphology revealed that hypertrophy in response to TAC was also graded at the cellular level. In intact myocytes, cell capacitance increased by 30% in HTY and by 91% in HF (Figure 3a).
Cell volume, calculated from cell width and length, also showed a graded increase (Figure 3b). In DT myocytes, cell capacitance and cell volume increased with TAC (P < 0.001 for both). However, neither cell capacitance nor cell volume increased in HTY cells, but both increased significantly in HF (Figure 3c,d). These data are consistent with cellular hypertrophy increasing with the severity of the cardiac response to pressure overload.
Heart failure has previously been shown to be associated with disrupted t-tubule structure, and representative images of t-tubules from the control, HTY and HF groups (Figure 4a) appear to show an increasing propensity for discontinuities in the t-tubule network and an increase in longitudinal or oblique elements (Louch et al., 2006).
Fourier analysis confirmed this impression. The amplitude of the first harmonic of t-tubule labelling (P 1 /P 0 ) was reduced in TAC myocytes (P < 0.001) and showed a significant graded decrease (Figure 4b),

Distribution and regulation of I Ca
To determine the distribution of I Ca between the surface and t-tubular membranes, I Ca was measured in intact and DT myocytes. Figure 6 shows In intact cells, I Ca density (at 0 mV) showed a graded decrease with TAC ( Figure 6d). In contrast, I Ca density in DT myocytes was not significantly different between the three groups (Figure 6e), showing that I Ca density at the surface membrane is not changed in HTY or HF, and thus that the reduction in I Ca density in intact myocytes must be at the t-tubule membrane. Calculation of t-tubular I Ca density supports this idea. Figure 6f shows that I Ca density at the t-tubule membrane decreases in a graded manner with severity of response to TAC; I Ca density was reduced in HTY (P < 0.001 versus control) and further reduced in HF (P < 0.001 versus control). (d) T-tubule orientation (angles in degrees from the z-disc; data binned between the angles shown on the x-axis) for control (n/N = 40/8), HTY (n/N = 26/5) and HF (n/N = 21/5) myocytes. * P < 0.05 or # P < 0.05, ** P < 0.01 or ## P < 0.01, *** P < 0.001 or ### P < 0.001; two-way ANOVA, Bonferroni-corrected post hoc test. In (d), asterisks indicate a significant difference between control and HF, and hash symbols indicate a significant difference between HTY and HF, at the same angle; there were no significant differences between control and HTY at any angle or had been pretreated with C3SD. In control myocytes, C3SD significantly decreased I Ca density (P < 0.01), suggesting significant Gapdh 37 kDa * * * * * * F I G U R E 5 Changes in protein expression after TAC. (a-c) Representative Western blots for Jph2 (a), Bin1 (b) and Cav3 (c) and Gapdh expression in myocyte lysates taken from two control (C1 and C2, replicated on the three blots), two HTY (HTY1 and HTY2) and two HF (HF1 and HF2) hearts.
(d-f) Mean data for Jph2 (d), Bin1 (e) and Cav3 (f) protein expression normalized to Gapdh in control, HTY and HF hearts (N = 5, 4 and 4 hearts, respectively, each analysed in duplicate). ** P < 0.001, one-way ANOVA, Bonferroni-corrected post hoc test that the decrease in t-tubular I Ca density in HF is attributable, at least in part, to loss of regulation by Cav3.

Calcium release after TAC
To determine whether the observed changes in I Ca were associated with altered Ca 2+ release at the t-tubules, the latency between membrane depolarization and subsequent Ca 2+ release at a t-tubule was examined as described previously (Bryant et al., 2015). Figure 8a shows transverse line-scan images of electrically stimulated Ca 2+ transients and the corresponding spatially averaged di-4-AN(F)EPPTEA (membrane potential) signal recorded at a single t-tubule in representative control, HTY and HF myocytes. Figure 8b shows that the latency of Ca 2+ release, measured from the start of the AP upstroke to the start of the rise of Ca 2+ (filled bars) and to the maximal rate of rise of Ca 2+ (grey bars), became significantly longer in HF, but not HTY, indicating an increase in latency with severity of response to TAC. Figure 8c shows that the heterogeneity of the maximal rate of rise of Ca 2+ also increased significantly in HF but not in HTY. These changes could not be ascribed to changes in SR Ca 2+ content, measured using 10 mM caffeine, which did not change significantly, even in HF [F/F 0 ; control 4.2 ± 2.1 (n/N = 12/4), HTY 6.2 ± 3.6 (n/N = 7/4) and HF 3.5 ± 1.0 (n/N = 10/5), P = 0.06].
To investigate whether the loss of Cav3-dependent stimulation of I Ca observed after TAC could account for these changes in Ca 2+ release, we investigated the effect of inhibiting this stimulation using C3SD on Ca 2+ release in control myocytes. Figure 8a  Thus, in the mouse, pressure overload is associated with a decrease in t-tubular I Ca density, which is graded with the severity of the response to TAC. This graded decrease in t-tubular I Ca density is associated with graded impairment of Ca 2+ release at the t-tubule and accompanied by physical disruption of the t-tubule network. These changes will decrease the synchronization of local Ca 2+ releases within the cell, which will contribute to the graded decrease in ejection fraction observed in vivo, although it has little effect on the rise of the spatially averaged Ca 2+ transient.

DISCUSSION
The present post hoc analysis of the response to TAC shows graded changes in cardiac growth and function in response to TAC-induced pressure overload, which are paralleled by graded disruption of t-tubule network structure and t-tubular I Ca density, and impairment of CICR and SR Ca 2+ release. These data provide the first demonstration of loss of t-tubular I Ca after TAC in mice and support the suggestion (Hong & Shaw, 2017) that graded changes in cardiac function reflect graded changes in cell, particularly t-tubule, structure and function. These changes are accompanied by a loss of Cav3dependent regulation of I Ca and a decrease in Cav3 and Jph2 protein expression in HF. As there is no change in I Ca density at the cell surface after TAC, and Cav3-dependent stimulation of I Ca occurs predominantly at t-tubules (Bryant et al., 2014), loss of such stimulation might underlie the decrease in t-tubular I Ca density.

Classification of cardiac hypertrophy and failure
Transverse aortic constriction is widely used in mice as an experimental model of pressure overload-induced left ventricular hypertrophy and heart failure. However, the phenotype of this model can vary because even small differences in constriction can result in large changes in resistance, which varies with the fourth power of the radius (Sutera & Skalak, 1993).
We therefore grouped mice according to their LW:TL ratio, and an LW:TL of 2 SD above the mean LW:TL of the control group was taken to indicate congestive heart failure. Mice that showed an LW:TL ratio below the threshold (Figure 1b) were classified as having cardiac hypertrophy but without being in overt heart failure.
Other measures of pathology, including ejection fraction and fractional shortening (Figure 2), HW:BW, LW:BW, cell size and changes in t-tubular organization (Figures 3 and 4), also showed gradation between the control, HTY and HF groups, supporting this classification criterion. Despite these changes, stroke volume and cardiac output were maintained, although the mechanism appeared to be different in HTY and HF animals. The lack of change in HTY mice appeared to be because there was little change in either contraction or heart size (Figures 2b,c,f,g). In contrast, in HF, although fractional shortening and ejection fraction decreased, the heart was larger, so that cardiac output was maintained by ejecting a smaller fraction of a larger diastolic volume (i.e. a dilated cardiomyopathy phenotype).

T-tubule structure after TAC
Examination and analysis of 3D t-tubule structure in myocytes from control, HTY and HF mice showed that t-tubule network disruption is graded with the severity of response to TAC (Figure 4), consistent with 2D Fourier analysis of ventricular tissue from rats after TAC (Wei et al., 2010). However, Fourier analysis does not show the changes in the complex architecture of the t-tubule network that underlie the changes in regularity. To overcome this, t-tubule density and orientation, both of which are strongly linked to cardiac contraction (Crossman et al., 2015), were measured in 3D. The t-tubule density showed a graded reduction with severity of response to TAC (Figure 4c), and there was a change in the orientation of the The mechanisms underlying t-tubule disruption are unknown, although t-tubules are known to be labile (Balijepalli et al., 2003;Brette & Orchard, 2003;Crossman et al., 2011;Forbes, Hawkey, & Sperelakis, 1984;He et al., 2001;Louch et al., 2004;Ziman et al., 2010). A number of proteins have been implicated in the formation and maintenance of t-tubules, including Bin1 (Caldwell et al., 2014;Hong et al., 2010), Jph2 (Reynolds et al., 2013) and Cav3 (Parton, Way, Zorzi, & Stang, 1997;Ziman et al., 2010). In the present study, Jph2 and Cav3 expression

I Ca and Ca 2+ release after TAC
Although most studies suggest that I Ca density is unaltered in HF (Benitah et al., 2002;Kamp & He, 2002), recent work in failing rat heart cells has shown a decrease in t-tubular I Ca that is offset by an increase in I Ca at the cell surface (Bryant et al., 2015) as implied by earlier work on human heart (Schroder et al., 1998). In the present study, there was a graded decrease in t-tubular I Ca density with increased severity of response to TAC (Figure 6e), although I Ca density at the surface membrane was unchanged; therefore, the whole-cell I Ca density decreased. However, the disruption of t-tubule structure and decreased I Ca density in the remaining t-tubules observed after TAC, both of which will impair Ca 2+ release, appear to be consistent between species and aetiologies.
To investigate the consequences of the graded decrease in t-tubular I Ca on E-C coupling, we measured Ca 2+ release at single t-tubules. There was a small but non-significant increase in latency and AP Initial rise Max. rate Heterogeneity ( suggest that the gain of E-C coupling in mice is high; therefore, in normal conditions, I Ca density is more than sufficient to activate CICR maximally (Cannell, Berlin, & Lederer, 1987). Given that the intersection with the surface membrane was excluded from measurements of local Ca 2+ release, these data suggest that latency (and thus, by implication, I Ca ), changes along the t-tubule, as shown in Figure 8a.
We have shown previously that inhibition of I Ca using a Ca 2+ channel blocker causes similar impairment of Ca 2+ release (Bryant et al., 2015).
The use of C3SD in control myocytes to mimic the loss of Cav3dependent stimulation of I Ca observed after TAC resulted in a similar decrease in I Ca (30%) and increase in latency (27%) as observed in HF myocytes (33 and 26%, respectively). Thus, loss of Cav3-dependent stimulation of I Ca is sufficient to account for the decrease in I Ca and increase in latency observed in HF.
The increase in latency and heterogeneity of Ca 2+ release observed in HF will contribute to the increased heterogeneity of Ca 2+ release reported previously and ascribed predominantly to disruption of t-tubule structure (Louch et al., 2004(Louch et al., , 2006. However, TAC had little effect on the amplitude of the spatially averaged Ca 2+ transient, consistent with the lack of effect on SR Ca 2+ content, which has previously been reported to increase, not change or decrease in hypertrophy and HF (Bers, 2006;Gómez et al., 1997b;Ibrahim et al., 2012;Mork et al., 2007). Perhaps surprisingly, given the increase in latency and heterogeneity of local Ca 2+ release, the maximal rate of rise of the whole-cell Ca 2+ transient was not significantly altered by TAC or by pretreatment with C3SD peptide, suggesting that other factors dominate the rising phase of the Ca 2+ transient. The relationship with the in vivo data is complex, both because of the presence of neurohumoral influences in vivo and because, although ejection fraction and fractional shortening decreased, this was in the presence of increased heart size. Given that wall tension will increase with dilatation, the Law of Laplace shows that a reduced ejection fraction does not necessarily imply reduced contractility.
It appears, however, that impaired Ca 2+ release in HF is attributable, in part, to disruption of t-tubule structure and decreased I Ca density in the remaining t-tubules.

Caveolin 3 levels and regulation of I Ca after TAC
The observation that incubation of myocytes with C3SD peptide reduced I Ca density and increased f and s in control cells, but not in HTY or HF myocytes (Figure 7), suggests a constitutive Cav3dependent increase in I Ca density and inactivation in control myocytes, which is lost after TAC. The reason for this is unknown; however, I Ca has previously been shown to be tonically stimulated by protein kinase A (Bracken, ElKadri, Hart, & Hussain, 2006;Bryant et al., 2014;Hussain, Drago, Bhogal, Colyer, & Orchard, 1999). Given that the binding partners of Cav3 include proteins that regulate Ca 2+ channel phosphorylation (Balijepalli, Foell, Hall, Hell, & Kamp, 2006), and disruption of Cav3 binding using C3SD inhibits such stimulation (Bryant et al., 2014), it appears possible that Cav3 binding to its partners is disrupted after TAC. Previous work has also shown that Cav3-dependent stimulation of I Ca occurs predominantly at t-tubules (Bryant et al., 2014).
One possible explanation for a loss of Cav3 binding after TAC is a decrease in Cav3 expression. Jph2, Bin1 and Cav3 have been implicated in the development and maintenance of t-tubules and in the regulation of t-tubular I Ca density (Bryant et al., 2014;Hong et al., 2010). In the present study, there was no change of Bin1 expression in HTY or HF; however, Jph2 and Cav3 expression decreased significantly in HF but not in HTY mice ( Figure 5). Thus, the loss of Cav3-dependent stimulation and significant decrease in t-tubular I Ca density in HTY myocytes occurred with no significant changes in Cav3 expression ( Figure 5), whereas in HF myocytes the decrease in t-tubular I Ca density was accompanied by a significant decrease in Cav3 expression.
Given the role of Jph2, Bin1 and Cav3 in determining t-tubule structure and I Ca localization (Bryant et al., 2014(Bryant et al., , 2015Caldwell et al., 2014;Hong et al., 2010;Parton et al., 1997;Reynolds et al., 2013;Ziman et al., 2010), it appears possible that changes in protein trafficking precede decreased expression, which could explain why a decrease in t-tubular I Ca density and altered t-tubule structure precede marked decreases in protein expression, which appear to be associated with the HF rather than the HTY phenotype (Reynolds et al., 2016), although it remains possible that another mechanism, as yet unknown, inhibits Cav3-dependent stimulation of I Ca .