Early Remodelling of Perinuclear Ca²⁺ Stores and Nucleoplasmic Ca²⁺ Signalling During the Development of Hypertrophy and Heart Failure

Abstract

Background

A hallmark of heart failure is impaired cytoplasmic Ca²⁺ handling of cardiomyocytes. It remains unknown whether specific alterations in nuclear Ca²⁺ handling – via altered excitation-transcription coupling – contribute to the development and progression of heart failure.

Methods and Results

Using tissue and isolated cardiomyocytes from non-failing and failing human hearts, as well as mouse and rabbit models of hypertrophy and heart failure, we provide compelling evidence for structural and functional changes of the nuclear envelope and nuclear Ca²⁺ handling in cardiomyocytes as remodeling progresses. Increased nuclear size and less frequent intrusions of the nuclear envelope into the nuclear lumen indicated altered nuclear structure that could have functional consequences. In the (peri)nuclear compartment there was also reduced expression of Ca²⁺ pumps and ryanodine receptors, and increased expression of inositol-1,4,5-trisphosphate receptors, and differential orientation among these Ca²⁺ transporters. These changes were associated with altered nucleoplasmic Ca²⁺ handling in cardiomyocytes from hypertrophied and failing hearts, reflected as increased diastolic Ca²⁺ levels with diminished and prolonged nuclear Ca²⁺ transients and slowed intranuclear Ca²⁺ diffusion. Altered nucleoplasmic Ca²⁺ levels were translated to higher activation of nuclear Ca²⁺/calmodulin-dependent protein kinase II and nuclear export of histone deacetylases. Importantly, the nuclear Ca²⁺ alterations occurred early during hypertrophy and preceded the cytoplasmic Ca²⁺ changes that are typical of heart failure.

Conclusions

During cardiac remodeling, early changes of cardiomyocyte nuclei cause altered nuclear Ca²⁺ signaling implicated in hypertrophic gene program activation. Normalization of nuclear Ca²⁺ regulation may, therefore, be a novel therapeutic approach for preventing adverse cardiac remodeling.

Heart failure (HF) is characterized by systolic and diastolic dysfunction and abnormalities of intracellular Ca²⁺ handling with disturbed excitation-contraction coupling underlying contractile failure.¹ Current research focuses on better understanding the mechanisms leading to disturbed Ca²⁺ handling during progression from cardiac remodelling (such as hypertrophy) to failure.

The Ca²⁺ cycle in cardiomyocytes that governs contraction and relaxation on a beat-to-beat basis consists of a transient rise in the cytoplasmic free Ca²⁺ concentration ([Ca²⁺]_cyto) and subsequent Ca²⁺ decay. Each cytoplasmic [Ca²⁺] transient (CaT) also elicits a nucleoplasmic CaT.² Much progress has been made towards understanding the role of altered cytoplasmic Ca²⁺ homeostasis in hypertrophy and HF.^3–5 However, nucleoplasmic [Ca²⁺] ([Ca²⁺]_nuc) in HF is understudied and may be critical to cardiac remodeling, because it regulates protein expression through nuclear Ca²⁺-dependent regulation of gene transcription.^{6, 7}

Our previous studies indicated that nucleoplasmic CaTs follow distinct kinetics and may be regulated quite differently from cytoplasmic CaTs.^{2, 8} An important aspect of [Ca²⁺]_nuc is the nuclear envelope (NE), which not only contributes to nuclear structure and insulation from the surrounding cytoplasm, but also controls bidirectional transport of ions (including Ca²⁺) and macromolecular cargo via nuclear pore complexes (NPCs). The NE is also a functional Ca²⁺ store, akin to the sarcoplasmic reticulum (SR), and contains Ca²⁺ pumps (SERCA2) and Ca²⁺ release channels. The regulation of [Ca²⁺]_nuc via Ca²⁺ release from the NE is also important in Ca²⁺-mediated gene expression.^{9, 10} However, it remains unknown whether specific alterations in NE and nuclear Ca²⁺ handling occur in cardiac disease.

Therefore, we analyzed NE structure and function during cardiac remodeling from hypertrophy to HF in animal models of pressure overload and in non-failing and failing human hearts. We demonstrate that NE structure, its molecular composition and nucleoplasmic CaTs undergo significant changes during pressure overload-induced hypertrophy in experimental animal models and in failing human hearts. These nuclear changes precede the changes in cytoplasmic Ca²⁺ dysregulation and, thus, suggest that altered nucleoplasmic [Ca²⁺] is an early event during remodeling and may contribute to the development and progression of cardiac hypertrophy and failure via nuclear Ca²⁺-dependent regulation of gene expression through activation of nuclear Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) and nuclear export of histone deacetylases (HDAC). Normalization of impaired [Ca²⁺]_nuc regulation may, therefore, represent a novel therapeutic target to prevent adverse cardiac remodeling.

Materials and Methods

Detailed description can be found in the Supplemental Material.

Animal models

All experimental procedures involving animals were approved by the local Animal Care and Use Committees according to the Guide for the Care and Use of Laboratory Animals prepared by the U.S. National Academy of Sciences (National Institutes of Health publication No. 85–23, revised 1996). Hypertrophy and HF was induced by transverse aortic constriction (TAC) in C57BL/6 mice or by combined aortic insufficiency and stenosis in New Zealand White rabbits.

Human myocardium

All procedures involving human myocardium were approved by the Ethical Committee of the Medical University of Graz (ref. No. 20-277 ex08/09) and were carried out in accordance with the Declaration of Helsinki. Patient characteristics are summarized in Table S1.

Cardiomyocyte isolation

Murine, rabbit and human ventricular cardiomyocytes were isolated using standard enzymatic dissociation procedures.

Confocal Ca²⁺ imaging of nucleoplasmic and cytoplasmic CaTs

Simultaneous imaging of nucleoplasmic and cytoplasmic CaTs occurred in cardiomyocytes loaded with Fluo-4 (Molecular Probes, Leiden, The Netherlands) using a confocal imaging system (Zeiss LSM 510 Meta or Olympus Fluoview 1000) as described previously.⁸ Cardiomyocytes were field-stimulated via two platinum electrodes. Isoprenaline (30 nM) and angiotensin II (ATII, 100 nM) were used to investigate the effects of β-adrenergic stimulation and IP₃ signaling, respectively. Cytoplasmic and nucleoplasmic fluorescence signals were transformed into calibrated [Ca²⁺] using the previously described method.⁸

Imaging of perinuclear Ca²⁺ stores

Perinuclear Ca²⁺ stores were visualized in cardiomyocytes loaded with the low affinity Ca²⁺ indicator Mag-Fluo-4/AM using a confocal imaging system (Zeiss LSM 510 Meta or Olympus Fluoview 1000). The optical slice thickness was ≤0.76 μm. 2D images at a central depth of the nuclei were collected. The longitudinal axis was drawn through the middle of the nuclei and tubular structures (longer than 1 µm) were counted along the half of the nuclear envelope, which contained more invaginations. Rapid application of caffeine (20 mM) occurred in the presence of 20 mM BDM by wash-in of caffeine for 3 s. Caffeine experiments were conducted in cardiomyocytes isolated from C57BL/6 mice that did not undergo any surgery.

Immunocytochemistry

Immunocytochemistry was performed as previously described¹¹ using the following antibodies: mouse monoclonal anti-nuclear pore complex proteins antibody (ab60080, Abcam, Cambridge, UK), mouse monoclonal anti-SERCA2a and mouse monoclonal anti-RyR antibody (MA3-919 and MA3-916, Thermo Scientific, Rockford, IL, USA), goat polyclonal anti-IP₃R2 antibody (NB100-2466, Novus Biologicals, Littleton, CO, USA), rabbit polyclonal anti-P-CaMKII antibody (ab32678, Abcam, Cambridge, UK) and rabbit polyclonal anti-HDAC4 antibody (sc-11418, Santa Cruz Biotechnology, Santa Cruz, CA, USA). The specificity of the antibodies was confirmed in Western blots.

Electron microscopy and immunogold labeling

Electron microscopic analyses of the NE were performed on human ventricular endocardial trabeculae, prepared as previously described.¹²

For immunogold labeling, ultrathin slices of mouse ventricles were stained with primary antibodies as in Immunocytochemistry, except for the rabbit polyclonal anti-SERCA2a antibody (A010-20, Badrilla, Leeds, UK). Goat anti-rabbit IgG (10 nm) and rabbit anti-goat IgG (5 nm) gold conjugates were from British BioCell International (BBI, Cardiff, UK).

FRET imaging

FRET imaging of CaMKII activation state was performed using the FRET-based biosensor Camui as previously described.¹³

Isolation of cardiac nuclei from human hearts

A detailed protocol for isolation of cardiac nuclei from human myocardium can be found in the Supplemental Material. The final nuclear fraction (N) was tested for expression of Ca²⁺-regulating proteins using standard immunoblot techniques with commercially available antibodies (see section Immunocytochemistry; anti-Nup62 (610497, BD Transduction Laboratories, Oxford, UK) and anti-Nkx2.5 (sc-14033, Santa Cruz Biotechnology, Santa Cruz, CA, USA)). The anti-RyR (1093) antibody used was custom-made.¹⁴ For quantification, signals were normalized to GAPDH (H) or Ponceau staining (H and N).

Drugs and solutions

Unless otherwise indicated, all chemicals were from Sigma-Aldrich (Steinheim, Germany).

Statistics

Data are presented as mean ± SEM. Differences between data sets were evaluated with Wilcoxon's rank sum test for between group comparisons and Wilcoxon's signed rank test for within group comparisons. Correlations were determined using Spearman rank correlation. Significance was accepted at *P<0.05. Statistical analyses were performed with SPSS Version 20.

Results

Hypertrophy vs. HF in mice

Sham-operated mouse hearts showed identical left ventricular (LV) dimensions and systolic function at either 1 or 6 weeks after surgery (not shown). Assessment of LVEDD and LVESD revealed that 1 week post-TAC mice displayed concentric LV hypertrophy, while LV dilation was observed 6 weeks post-TAC (Table S2 and Figure S1a). LV systolic function progressively declined, as indicated by ejection fraction reduction of ~8% one week and ~50% six weeks post-TAC (Table S2). TAC-induced hypertrophy was confirmed by increased heart weight normalized to tibia length or body weight (N=5–10; Figure S1b and S1c). Six weeks post-TAC mice developed pulmonary edema as manifested by an elevated lung weight-to-tibia length ratio (not shown). These data indicate that pressure overload-induced myocardial remodeling was associated with an early onset of compensatory hypertrophy and subsequent progression to overt HF.

Nuclear envelope remodeling in hypertrophy and failure

In cardiomyocytes from sham-operated mice and rabbits, loading of perinuclear Ca²⁺ stores with the low affinity Ca²⁺ indicator Mag-Fluo-4 revealed a NE and tubular invaginations traversing the nucleus (Figure 1a, left). Rapid application of caffeine (20 mM) reversibly abolished Mag-Fluo-4 fluorescence (Figure 1c) both in the NE and its tubular structures. Fluorescence recovery after depletion was identical in both regions, returning to ~90% of the pre-caffeine level (n=7; Figure 1d). This implies that the NE and its tubular invaginations are functional Ca²⁺ stores capable of releasing and re-accumulating Ca²⁺, both in the nuclear periphery and the regions within the nucleus otherwise remote from the NE.

Figure 1.

Nuclear envelope remodeling in hypertrophy and heart failure. (a) Original 2D confocal images of Mag-fluo-4 fluorescence of nuclei from sham- and TAC-operated mice isolated 1 and 7 weeks after the intervention (top), non-failing and failing rabbit hearts (middle), and non-failing and failing human hearts (bottom). (b) NE invagination density, calculated as number of invaginations per NE circumference, in cardiomyocytes from sham- and TAC-operated mice, isolated 1 and 7 weeks after the intervention (left, n=90/group), non-failing and failing rabbit hearts (middle, n=30/group) and non-failing and failing human hearts (right, n=20/group). * P<0.05 versus 1 week post-sham mice or non-failing rabbit and human controls; # P<0.05 versus 1 week post-TAC mice. (c) Original 2D images of Mag-fluo-4 fluorescence of a nucleus from control mice before, during and after caffeine application. (d) Average values of fluorescence recovery after depletion in different regions of the NE and its invaginations from 7 ventricular cardiomyocytes.

A significant increase in the density of tubular invaginations, calculated as the number of invaginations per NE circumference in the central depth of the nucleus, was observed during physiological aging in young (2–5 months) sham-operated mice (n=90 nuclei/group; Figure 1b). In contrast, TAC-operated mice exhibited a progressive decrease in NE tubular invagination density 1 and 7 weeks post TAC. Similar reductions were observed in cardiomyocytes from non-failing vs. failing rabbit hearts (n=30 nuclei/group) and from non-failing vs. failing human hearts (n=20 nuclei/group), suggesting a misrelationship between the growth of the nuclei and the NE compartment as a general feature of HF, independent of species and etiology of HF. Average nuclear dimensions and number of NE tubular invaginations per nucleus are summarized in Figure S2.

In order to confirm the presence of NE invaginations in cardiac tissue (vs. isolated myocytes), and to investigate their detailed structure, electron microscopy imaging of nuclei from sections of human ventricular trabeculae was performed. The observed invaginations were lined by the inner and outer nuclear membranes, interrupted by numerous NPCs (Figure 2a, right, red arrows and inset), and filled with cytoplasm (Figure 2a, left and middle).

Figure 2.

Nuclear envelope invaginations and localization of Ca²⁺-regulating proteins in non-failing and failing cardiomyocytes. (a) Transmission electron micrograph of a nucleus from human left ventricular trabeculae (left), zoom on the NE invaginations (middle and right) and NPC (red arrows and inset). (b) Original 2D images of cardiomyocytes isolated from sham- and TAC-operated mice 7 weeks after the intervention (left) and non-failing and failing human cardiomyocytes (right) following immunostaining for NPC (top), IP₃R (middle), RyR (middle) and SERCA2a (bottom). Scale bars indicate 5 µm. Transmission electron micrograph of control mouse myocardium stained for (c) RyR, NE and its surrounding (top and middle), and T-tubules (TT) and their surroundings (bottom); (d) IP₃R, zoom on NE (top) and perinuclear region (bottom) and (e) SERCA2a, zoom on NE (top and middle), distribution between inner and outer nuclear membrane (n=11; bottom, left) and expression level on NE in sham- vs. TAC-operated mouse myocardium (n=6; bottom, right). * P<0.05 versus inner nuclear membrane or 7 week post-sham mice. (c)–(e) Red arrows and insets indicate gold particles. Scale bars indicate 0.2 µm.

In summary, these data indicate that the NE of cardiomyocytes contains a network of tubular structures (i.e. NE invaginations) that undergoes significant changes during hypertrophy and HF. With the progression of cardiac remodeling the size of nuclei increases, whereas, at the same time, the number of NE invaginations decreases.

Remodeling of Ca²⁺-regulating proteins in and around the nucleus

We also investigated (peri)nuclear expression of Ca²⁺ release channels, ryanodine receptor (RyR) and inositol-1,4,5-trisphosphate receptor (IP₃R), nuclear pore complexes (NPCs) and sarcoplasmic reticulum Ca²⁺-ATPase (SERCA) that could have functional consequences for [Ca²⁺]_nuc and HF progression.

Immunostaining of cardiomyocytes confirmed the presence of NPCs both in the NE and its invaginations (n=15; Figure 2b, top). In line with Mag-fluo-4 stainings (Figure 1a), NPC labeling showed increased nuclear dimensions with reduced number of invaginations in 7 weeks post-TAC mice and in cardiomyocytes from failing human hearts (quantitative analyses not shown).

In sham-operated mice and in non-failing human cardiomyocytes, IP₃R2 was found in a striated pattern throughout the cell, in close proximity to the sarcolemma (most prominent) and on the NE. In HF, accumulation of IP₃R2 in the perinuclear region was observed (n=15; Figure 2b, middle). Immunogold labeling revealed that IP₃R2 is localized more prominently on the inner NE surface (Figure 2d, red arrows) but is also seen on the outer NE, as well as in the perinuclear region (Figure 2d, bottom).

A punctate pattern with a striated organization was observed for RyR2, and combined with nuclei staining revealed (Figure S3) that RyR2 did not penetrate into the nuclei but rather formed a "cage" around them, as described previously.¹⁵ A large reduction of RyR2 staining in perinuclear regions, in particular in the longitudinal direction, was observed in 7 weeks post-TAC mice and in cardiomyocytes from failing human hearts (n=15; Figure 2b, middle and Figure S3). Immunogold labeling confirmed that RyR was not expressed on, but only near the NE (Figure 2c, top and middle). As expected, RyR2 was seen in SR-like structures surrounding T-tubules (Figure 2c, bottom).

SERCA2a expression exhibited a characteristic network-like pattern reflecting the SR throughout the cytoplasm. In addition, SERCA2a staining was strongly positive on the NE and its invaginations in sham mouse and non-failing human cardiomyocytes. Consistent with our earlier observation of blunted invagination density in HF, such structures were absent with SERCA2a immunostaining in either murine or human failing myocytes (n=15; Figure 2b, bottom). Immunogold labeling indicated that SERCA2a was mostly expressed on the outer nuclear membrane (n=11; Figure 2e; Figure S4), and that its expression on the NE was significantly decreased in 7 weeks post-TAC mice (n=6; Figure 3c, bottom right).

Figure 3.

Characterization of cytoplasmic, nucleoplasmic and subnucleoplasmic CaTs after pressure overload-induced hypertrophy and heart failure in mouse cardiomyocytes. (a) Linescan imaging of cytoplasmic and nucleoplasmic CaTs in a cardiomyocyte. (b) Averaged original recordings of distinct subcellular regions, as indicated in the scheme in (a): nucleus (red) versus cytoplasm (black) of cardiomyocytes from sham- (left) and from TAC-operated mice isolated 1 (middle) and 7 (right) weeks after intervention. (c) Diastolic [Ca²⁺], (d) amplitude and (e) kinetic parameters (time to peak (left) and DT₅₀ (right)) of nucleoplasmic (red) and cytoplasmic (black) CaTs. (b–e) n=15 myocytes/group. * P<0.05 versus 1 week post-sham. (f) Linescan imaging of subnucleoplasmic CaTs in a cardiomyocyte. (g) Averaged original recordings of CaTs from central (green) versus subnucleolemmal (black) regions of ventricular nuclei from sham- (left) and TAC-operated animals isolated 1 (middle) and 7 (right) weeks after intervention. (h) Diastolic [Ca²⁺], (i) kinetic parameters (time to peak (left), DT₅₀ (middle) and velocity of spread (right)) of the central (green) and subnucleolemmal (black) CaTs. (g)–i) n=10 myocytes/group. * P<0.05 versus 1 week post-sham; # P<.05 versus subnucleolemmal.

Altered expression levels of Ca²⁺-regulating proteins were confirmed by Western Blot analysis of isolated human cardiac nuclei (Figure S5). Nuclear fractions from failing human myocardium exhibited a significant decrease in SERCA2a and RyR2 expression (which was detectable most likely due to remnants of perinuclear regions sticking to the nuclei), while IP₃R2 levels were increased as compared to non-failing hearts.

[Ca²⁺]_cyto and [Ca²⁺]_nuc during electrical stimulation

Fluo-4 fluorescence signals recorded during electrically stimulated CaTs (Figure 3a) were transformed into [Ca²⁺]_nuc and [Ca²⁺]_cyto using our compartment-specific calibration methods (Figure 3b).⁸ No differences in CaTs were observed between the two groups (1 and 7 weeks) of sham-operated mice (not shown). One week post-TAC, diastolic [Ca²⁺]_nuc and [Ca²⁺]_cyto were significantly increased and increased further 7 weeks post-TAC (Figure 3c). The increase of diastolic [Ca²⁺]_nuc was, much more pronounced than [Ca²⁺]_cyto (Figure S6). Peak systolic [Ca²⁺] remained unchanged in either compartment (not shown). Early in hypertrophy CaTs were slowed and CaT amplitudes were reduced only in the nucleoplasm. At seven weeks post-TAC, similar changes of CaTs also occurred in the cytoplasm (Figure 3d–e).

The NE invaginations and SERCA2/RyR2 distribution suggest the following working model of nuclear CaTs. The rise in [Ca²⁺]_nuc is driven mainly by RyR2-dependent rise in perinuclear [Ca²⁺]_cyto, and the primarily cytosolic facing SERCA2 distribution suggests that Ca²⁺ may largely have to diffuse out of the nucleus to be pumped back into the NE and SR. These aspects would slow the rise and fall of nuclear vs. cytosolic CaTs (as observed) and could be exacerbated by the reduction in invaginations during the progression to HF. We measured this diffusional delay during development of HF in our TAC mice by measuring CaTs at the surface and center of both the nucleus and cytosol (Figure 3f–j and Figure S7). Kinetics of cytosolic CaTs were spatially uniform and unaltered 1 week post-TAC (n=10; Figure S7). However, at seven weeks post-TAC there was some slowing of the central CaT vs. the sub-sarcolemmal CaT (Figure S7), presumably caused by the reported reduction in T-tubular density that occurs in HF.¹⁶ In contrast, nuclear CaT propagation velocity (difference in time to peak between sub-nucleolemmal and central nucleoplasmic region divided by the distance) was already slowed significantly at 1 week post-TAC (Figure 3j, right). Selective slowing of the central nucleoplasmic CaTs 1 week post-TAC (Figure 3i–j) led also to a more pronounced increase in diastolic central [Ca²⁺]_nuc already 1 week post-TAC (Figure 3h). This observation might be particularly relevant at high stimulation frequencies, when diastolic [Ca²⁺]_nuc is more affected than [Ca²⁺]_cyto due to its slower kinetics.⁸ In the rabbit HF model we also observed elevated diastolic [Ca²⁺]_nuc and slowed kinetics of [Ca²⁺]_nuc rise (Figure S8).

Importantly, in human myocytes (both for moderate and severe HF) we saw quite similar effects as in the 1- and 7-week post-TAC mice (Figure 4). Although a low number of moderately failing hearts is a limitation of the study, our results suggests that early alterations in [Ca²⁺] transients that occur selectively in the nucleus might be clinically relevant for the progression of cardiac remodeling in HF patients.

Figure 4.

Characterization of cytoplasmic and nucleoplasmic CaTs of cardiomyocytes from non-failing, moderately failing and severely failing human hearts. (a) Averaged original recordings of CaTs in nucleus (red) versus cytoplasm (black) of cardiomyocytes from healthy controls (N=6, left), moderately failing (N=2, middle) and end-stage failing human hearts (N=5, right). (b) Diastolic [Ca²⁺], (c) amplitude and (d) kinetic parameters (time to peak (left) and DT₅₀ (right)) of nucleoplasmic (red) and cytoplasmic (black) CaTs. (a–d) n= 6–10 myocytes/group. * P<0.05 versus non-failing.

Correlation of NE invaginations with nuclear [Ca²⁺] transient kinetics

In mouse and rabbit cardiomyocytes, we simultaneously quantified NE invaginations (with Mag-fluo-4) and spatiotemporal aspects of cytoplasmic and nucleoplasmic CaTs (with rhod-2). Figure 5a shows that nuclear CaT propagation is slowed in cardiomyocytes with lower NE invagination density. Correlation analysis indicates a strong inverse correlation between the density of NE invaginations and Ca²⁺ propagation to and from the nucleus (time to peak and DT₅₀) in both sham and HF conditions (Figure 5b). This means that there is indeed a functional link between NE structure and nuclear CaTs kinetics.

Figure 5.

Correlation between nuclear envelope invaginations and nuclear CaT kinetics in cardiomyocytes from non-failing and failing hearts. (a) Rabbit: Original confocal images of 3 nuclei stained with Mag-fluo-4 (left) and corresponding Rhod-2 recordings of CaTs in the same nuclei (right). (b) Kinetic parameters (time to peak (left) and DT₅₀ (right)) of the nucleoplasmic (corrected for cytoplasmic) CaTs as a function of NE invagination density in cardiomyocytes from sham- and TAC-operated mice (top) and non-failing and failing rabbit hearts (bottom). In all groups, there was a significant correlation (with the correlation coefficient r as given in the figure). n=12–14 myocytes/group.

Frequency-dependent changes of nucleoplasmic [Ca²⁺] transients

Figure 6a–b show recordings in which stimulation frequency was gradually increased from 0.5 to 5 Hz (mouse cells) or from 0.2 to 1 Hz (rabbit cells). In all 3 groups of mice diastolic [Ca²⁺] in the nucleus and cytoplasm rose with an increase in stimulation frequency (Table S3). However, during early remodeling (TAC 1 week) diastolic [Ca²⁺]_nuc was already elevated at very low stimulation rate (in contrast to diastolic [Ca²⁺]_cyto). At faster stimulation rates, diastolic [Ca²⁺]_nuc also rose much more prominently than [Ca²⁺]_cyto. In failing cardiomyocytes (TAC 7 weeks), these changes in nucleoplasmic and cytoplasmic diastolic [Ca²⁺] were exacerbated. Systolic peak [Ca²⁺] at 1 and 7-weeks post-TAC was comparable to the sham group in both compartments (at most frequencies). The increase in diastolic [Ca²⁺] resulted in a progressive decline in the CaT amplitude with faster stimulation rates. The reduction of CaT amplitude started at lower frequency in the nucleoplasmic vs. cytoplasmic compartment. Similar changes were observed in cardiomyocytes from non-failing vs. failing rabbit hearts (Figure 6b).

Figure 6.

Frequency-dependent changes of nucleoplasmic versus cytoplasmic CaTs, phosphorylation of CaMKII and HDAC4 nuclear export in non-failing and failing cardiomyocytes. (a) Original recordings of nucleoplasmic (red) versus cytoplasmic (black) CaTs of cardiomyocytes from sham- (top) and TAC-operated mice (1 (middle) and 7 weeks (bottom) after intervention) during increases of stimulation frequency from 0.5 to 5 Hz. (b) Original recordings of nucleoplasmic (red) versus cytoplasmic (black) CaTs of cardiomyocytes from non-failing (top) and failing rabbits (bottom) during increases of stimulation frequency from 0.2 to 1 Hz. (c) Original confocal images (top) of cardiomyocytes from sham- and TAC-operated mice (1 and 7 weeks after intervention) immunostained for phospho-CaMKII following 10 min stimulation at 0.5 (left) or 5 Hz (right). Average phospho-CaMKII fluorescence values (bottom left) and increases in phospho-CaMKII levels calculated as difference to sham (DF, bottom right) from 25 cardiomyocytes. *P<0.05 versus 0.5 Hz (left) or 1 week post-TAC group (right); # P<0.05 versus cytoplasm. (d) Original confocal images of CFP and YFP emission signals from non-failing and failing rabbit cardiomyocytes expressing Camui (left) and average data of Camui activation (F_CFP/F_YFP) in nucleus and cytoplasm of 32 non-failing and 90 failing cardiomyoytes at baseline conditions (right). * P<0.05 versus non-failing. (e) Original confocal images (left) of non-failing rabbit cardiomyocytes immunostained for HDAC4 following 15 min stimulation at 0.2 (top) or 1.5 Hz (bottom) and average values of nucleoplasmic/cytoplasmic HDAC4 fluorescence ratio in 28 non-failing and 20 failing cardiomyocytes. * P<0.05 versus 0.2 Hz non-failing; # P<0.05 versus 0.2 Hz failing. (c)–(e) Scale bars indicate 20 µm.

In further experiments, we investigated whether β-adrenergic stimulation can diminish or prevent the disproportionate rise in [Ca²⁺]_nuc in hypertrophied and failing cardiomyocytes to increased stimulation frequencies (Figure 6a, right). β-adrenergic stimulation greatly accelerates CaT decay and, thereby, might blunt the frequency-dependent increases in diastolic [Ca²⁺].¹⁷ Application of isoprenaline (30 nM) resulted in a robust increase of both nucleoplasmic and cytoplasmic CaTs in control cardiomyocytes. However, the preferential rise in diastolic [Ca²⁺]_nuc was still observed, and again was more pronounced as HF progressed (1- vs. 7-week post-TAC). Average data and statistical analysis related to Fig 6a are summarized in Table S3.

Collectively, these data revealed three important points: 1) the frequency-dependent increase in diastolic [Ca²⁺]_nuc is much larger than the increase in diastolic [Ca²⁺]_cyto; 2) it persists with β-adrenergic stimulation; and 3) this effect is even more pronounced in failing myocytes.

Frequency-dependent activation of nuclear CaMKII and nuclear export of HDAC

In cardiomyocytes from sham-operated mice stimulated at low frequency (0.5 Hz), there was only weak staining of the autonomously active form of CaMKII, phospho-CaMKII, in the cytoplasm and within the nucleus (Figure 6c). Higher stimulation frequency (5 Hz) increased CaMKII phosphorylation, corresponding to the increase in [Ca²⁺]_nuc and [Ca²⁺]_cyto under the same pacing conditions, with much larger increases in the nucleus (particularly in the NE), and especially markedly in HF (n=25; Figure 6c). The phospho-CaMKII signals were prevented by KN-93 (1µM) preincubation, confirming the functional basis of the observed signals.

We also expressed the FRET-based biosensor CaMKII activity reporter Camui in control and HF rabbit cardiomyocytes. Camui displayed similar cellular distribution as the endogenous phospho-CaMKII detected by immunocytochemistry, consistent with known CaMKII localization. Camui signals indicated significantly higher CaMKII activation in the nucleus and cytoplasm of failing cardiomyocytes at rest (n≥32; Figure 6d).

Histone-deacetylase 4 (HDAC4) is a transcriptional regulator that is a downstream effector of CaMKII in the nucleus.¹⁰ That is, CaMKII binds to and phosphorylates HDAC4 to drive HDAC4 nuclear export and derepression of hypertrophic transcription. We found that in HF vs. control myocytes, HDAC4 was less nuclear at low frequency stimulation and was more readily driven out by increasing pacing frequency (n≥20; Figure 6e) as expected from the frequency-dependent increase in nuclear CaTs and CaMKII activation (Figure 6a–c).

Role of IP₃ in regulation of nucleoplasmic [Ca²⁺] transients

IP₃ is an important regulator of nucleoplasmic CaTs.^{2, 18} Inhibition of IP₃Rs (using 2-APB) causes selective decreases in diastolic nucleoplasmic [Ca²⁺].⁸ Furthermore, perinuclear IP₃R expression is augmented in HF (Figures 2 and S4). Figure 7a shows linescan images of CaTs from electrically stimulated cardiomyocytes in the absence or presence of ATII, which is known to cause IP₃ production in cardiomyocytes.¹⁹ In sham-operated mice, application of ATII increased diastolic [Ca²⁺] in both the cytoplasm and nucleus (n=15; Figure 7b and 7c). The increase in [Ca²⁺]_nuc was much larger than the increase in diastolic [Ca²⁺]_cyto, and this effect was augmented in 7 weeks post-TAC cardiomyocytes. Large increases in diastolic [Ca²⁺] led to significant increases in systolic [Ca²⁺] in both compartments (Figure 7d), while CaT amplitude remained unaltered. ATII raised systolic [Ca²⁺]_cyto to similar levels in sham vs. TAC cardiomyocytes, while upon ATII application systolic [Ca²⁺]_nuc was significantly higher in 7 weeks post-TAC cardiomyocytes compared to controls. Application of 2-APB (3 µM) completely blocked the effect of ATII (not shown), consistent with the ATII effects being IP₃-mediated.

Figure 7.

IP₃-dependent changes of nucleoplasmic vs. cytoplasmic CaTs in non-failing and failing mouse cardiomyocytes. (a) Linescan images of cytoplasmic and nucleoplasmic CaTs in a non-failing (top) and failing (bottom) mouse cardiomyocyte without (Normal Tyrode, NT) or with angiotensin II (+ATII; 100 nM). (b) Averaged original recordings of nucleoplasmic (red) versus cytoplasmic (black) CaTs of cardiomyocytes from sham- (top) and TAC-operated (bottom) mice in the absence (left) or presence (right) of ATII. Average diastolic (c) and systolic (d) [Ca²⁺] (n=15 cardiomyocytes/group). * P<0.05 versus NT; # P<0.05 versus sham ATII.

Discussion

The present study is the first to provide direct compelling evidence for structural and functional changes of the NE and nucleoplasmic Ca²⁺ handling during cardiac remodeling and HF in mouse and rabbit models of pressure overload and in human hearts. The progressive decrease in NE invagination density and changes in the Ca²⁺-regulatory protein expression patterns (schematically summarized in Figure 8) were associated with alterations of nucleoplasmic [Ca²⁺] handling and consequent activation of gene transcription via the nuclear CaMKII-HDAC4 axis in electrically stimulated cardiomyocytes isolated from hypertrophied and failing hearts. Changes in nuclear CaTs occurred before cytoplasmic CaTs were affected, with an onset so early that they may well be involved in the development and progression of hypertrophy and HF.

Figure 8.

Scheme of perinuclear Ca²⁺ stores and nucleoplasmic Ca²⁺ signaling alterations during cardiac remodeling. Ca²⁺ influx (red arrows) and removal (green arrows) pathways in the nucleus are highly balanced to assure specific regulation of Ca²⁺-dependent signaling and transcription (top). During development of hypertrophy, nuclear structure is altered, leading to a bigger nucleus size and less frequent NE invaginations (bottom). Expression levels of receptors and channels involved in Ca²⁺ homeostasis are altered, with increased perinuclear IP₃R expression and decreased perinuclear expression of RyR and SERCA. These structural and functional changes in NE and perinuclear regions contribute to the alterations of nucleoplasmic Ca²⁺ handling, especially to the increase in diastolic [Ca²⁺]_nuc. ET-1, endothelin-1; ATII, angiotensin II; PE, phenylepinephrine; ECM, extracellular matrix; PLC, phospholipase C; IP₃, inositol-1,4,5-trisphosphate; IP₃R, IP₃ receptor; RyR, ryanodine receptor; SERCA, sarcoplasmic reticulum-Ca²⁺-ATPase.

The NE of numerous cell types (including cardiomyocytes) contains invaginations including deep, branching tubular structures (for review see⁹). As caffeine experiments (Figure 1c) demonstrated, the NE and its invaginations represent functional Ca²⁺ stores capable of releasing and re-accumulating Ca²⁺. It is tempting to speculate that – similar to T-tubular sarcolemmal invaginations that are critical for coordinated Ca²⁺ cycling throughout the myocyte – nuclear tubular invaginations may be critical for minute control of nucleoplasmic Ca²⁺ release and removal. Recent work suggested that the NE invaginations might be an artifact caused by NE folding due to the shorter sarcomere length in isolated cells.¹⁵ However, we also observed NE invaginations in multicellular cardiac tissue (Figure 2a). Furthermore, the fact that sarcomeres in cardiomyocytes rhythmically shorten and lengthen during the contraction-relaxation cycle suggests that NE invaginations might change in number and depth in the beating heart.

There are several important consequences of the NE invaginations penetrating into the nucleus. In general, they facilitate intranuclear regulation of ions and transcription factors that travel between cytoplasm and nucleoplasm by decreasing the diffusion delay and by increasing membrane surface area, which may be critical for the regulation of gene transcription. The presence of NPCs on the NE invaginations ensures effective nucleo-cytoplasmic ion diffusion and cargo transport in regions that would otherwise be remote from the nuclear periphery. Expression of SERCA pumps on the invaginations enables Ca²⁺ removal from deep within the nucleus, but since most SERCA2 pumps face the cytoplasmic side, their position with respect to nuclear pores may be important in shaping nucleoplasmic CaT kinetics.

Our working hypothesis, based on NE invaginations and NPC, SERCA2 and RyR2 distribution is the following (Figure 8): Normally the rise in [Ca²⁺]_nuc is driven mainly by the RyR2-dependent rise in [Ca²⁺]_cyto that occurs during E-C coupling, and Ca²⁺ diffusion via NPCs (including invaginations) is the cause of slower rise times and peak [Ca²⁺]_nuc. NE SERCA2 is mainly facing the cytoplasm (where there is also much more SERCA on SR) indicating that most Ca²⁺may have to diffuse out of the nucleus to be pumped back into the NE and SR (accounting for the much slower [Ca²⁺]_nuc decline). The loss of NE invaginations which occurs early in HF (in mouse, rabbit and human) would reduce NPC and SERCA within the nuclear core and exacerbate the slowing of nuclear CaT transients and elevate diastolic [Ca²⁺]_nuc especially at higher heart rates. Indeed, we have directly measured all those effects here, even in the presence of -adrenergic stimulation. The overall decrease in SERCA functional expression in HF slows [Ca²⁺]_cyto decline, which further increases diastolic [Ca²⁺]_nuc in HF. The profound increase in diastolic [Ca²⁺]_nuc and consequent higher CaMKII activation level that we observed in myocytes from the early TAC group at higher pacing frequencies may be causally involved in the remodeling processes leading to HF, in particular when considering the often elevated heart rates of patients with HF. Indeed, the SHIFT trial data demonstrate a beneficial effect of lowering heart rate by ivabradine in patients with advanced systolic heart failure, both in terms of attenuated LV remodeling²⁰ and less frequent hospitalizations.²¹

The simple working hypothesis above gets complicated somewhat by the IP₃R localization that was preferentially (but not exclusively) facing the nucleoplasm. G_q-coupled receptors (α-adrenergic, endothelin and ATII receptors) that induce IP₃ production may be more activated during HF progression.^{22, 23} Moreover, elevated [Ca²⁺]_nuc would enhance the IP₃ sensitivity of IP₃Rs²⁴ and could well synergize with the other factors which elevate [Ca²⁺]_nuc in HF and at high heart rate, as we have shown for ATII. Indeed, IP₃R-mediated Ca²⁺ release from the NE can elevate local [Ca²⁺]_nuc independently from [Ca²⁺]_cyto.^{2, 10} This pathway may be central in cardiac excitation-transcription coupling^{10, 25} and, hence, increased IP₃R expression was proposed to be important during hypertrophy and HF.^{10, 26, 27} In line with this notion, we observed increased perinuclear IP₃R expression, and higher diastolic [Ca²⁺]_nuc vs. [Ca²⁺]_cyto in HF cardiomyocytes treated with ATII.

Perinuclear RyR expression was reduced here in HF, which would slow further the rise in [Ca²⁺]_nuc but could also shift perinuclear Ca²⁺ signaling in favor of IP₃R-mediated Ca²⁺ release during the progression from hypertrophy to HF. Nuclear factor of activated T-cells (NFAT) is a mediator of calcineurin-dependent nuclear signaling, and ATII and endothelin-1 could activate this system in adult ventricular cardiomyocytes.²⁸ It was speculated that the perinuclear region could represent a local reserve of NFAT that is poised for shuttling in and out of the nucleus when local [Ca²⁺] is elevated via IP₃-mediated Ca²⁺ release.²⁸ Furthermore, nuclear calcineurin is increased in human HF²⁹ and required for full transcriptional effects of NFAT.³⁰ Similarly, in adult ventricular myocytes α-adrenergic receptor activation selectively increased nuclear CaMKIIδ phosphorylation (without altering SR associated CaMKII), an effect attributed to Ca²⁺ mobilized through nuclear IP₃-sensitive stores.³¹ Accumulation of IP₃Rs in this region in cardiomyocytes from failing hearts would support the idea of enhanced NFAT and CaMKII signaling and activation of a hypertrophic gene program.

During the past two decades great advances have been made in understanding alterations in cytoplasmic ion homeostasis in cardiomyocytes during the development of hypertrophy and HF. Here, we show for the first time how nucleoplasmic Ca²⁺ homeostasis is altered during this hypertrophy-HF process. We provide evidence for structural and functional alterations of the nucleus and nuclear Ca²⁺ signaling as remodeling progresses. Importantly, the changes were observed not only in animal models of hypertrophy and HF but also in human HF. Our results implicate [Ca²⁺]_nuc as an important determinant of cardiac remodeling which may contribute to the development and progression of hypertrophy and HF. Normalization of nucleoplasmic Ca²⁺ regulation may, therefore, be a novel therapeutic approach for preventing adverse cardiac remodeling.