Introduction

The transient addition of a single N-acetylglucosamine (GlcNAc) to intracellular serine/threonine residues (O-GlcNAcylation) is a ubiquitous post-translational modification that contributes to the regulation of a variety of proteins and physiologic processes [29]. O-GlcNAcylation occurs through activity of two enzymes: O-linked β-N-acetylglucosamine transferase (OGT) and O-linked β-N-acetyl glucosaminase (OGA), which, under normal physiologic conditions, dynamically add and remove GlcNAc residues respectively [29]. Production of the OGT substrate, UDP-GlcNAc, requires processing of metabolites such as glucose, acetyl-coA and glutamine through the hexosamine biosynthesis pathway [50] and was shown to be increased in diabetic conditions [19, 50]; thus, OGT is also often referred to as a nutrient sensor. In addition to those activities common to all cell types including transcription, translation and cellular metabolism [29], O-GlcNAcylation plays an important role in a variety of physiologic processes specific to cardiomyocyte function including ischemia–reperfusion injury and hypertrophic signaling [4, 20, 39]; however, its role in excitation–contraction (EC) coupling has only recently begun to emerge, and primarily studied only in the context of diabetes/hyperglycemia [9, 13, 19, 31, 52, 53, 55, 68].

L-type voltage-gated Ca2+ channels (Cavs), which are putatively formed by the assembly of α1 (Cav1.2), α2δ1 and β2 subunits, trigger cardiomyocyte Ca2+-activated Ca2+-release and can be directly targeted by signaling pathways that allow the heart to respond to physiologic cues [5, 38]. This modulation of Cav activity is typically transient and beneficial; however, chronic disease states such as hypertrophic cardiomyopathy and diabetes often appropriate these adaptive signaling pathways such that they become pathologic [8, 22, 38, 59]. Several studies demonstrated that cardiac Cav activity is depressed in chronic diabetic models [45, 48, 63]; although, the mechanisms are not well understood. Additionally, utilizing a proteomics approach, it was suggested that several neuronal Cav subunit isoforms bear the O-GlcNAc modification [62]. Despite these findings, a potential role for the regulation of cardiomyocyte Cavs by O-GlcNAcylation is currently unknown.

Here we sought to determine the inherent functional impact of O-GlcNAcylation on cardiac L-type Cav activity by utilizing an inducible cardiomyocyte specific OGT-gene deficient mouse strain (OGTKO). Following induction, OGTKO cardiomyocytes demonstrated significant reductions in Cav current (ICa), with concomitant reductions in Cav α1 and β2 expression, but with increased efficacy of adrenergic stimulation. OGTKO cardiomyocytes also demonstrated depolarizing shifts in Cav voltage-dependent steady-state gating. Consistently, intracellular Ca2+ release and contractility were diminished in OGTKO cardiomyocytes with concurrent reductions in left-ventricular (LV) contractility. We also observed that cardiac Cav α1 and β2 subunits bear the O-GlcNAc modification, while the α2δ1 subunit does not. Thus, results from this study indicate, for the first time, that O-GlcNAc signaling, under basal and non-diseased related conditions, is a crucial and direct regulator of cardiomyocyte L-type Cav activity and therefore EC coupling.

Methods

Ethical approval

Animals were handled in accordance with the NIH Guide for the Care and Use of Laboratory Animals. All protocols involving animals were approved by the Wright State University Institutional Animal Care and Use Committee. Mice were euthanized, under deep anesthesia (5% isoflurane), by thoracotomy and excision of the heart to obtain samples for the biochemical and functional studies.

Generation of the OGTKO and animal use

12–16 week-old male mice were used in all experiments. To create the OGTKO strain, female mice homozygous for a transgene possessing loxP sites flanking the exon encoding amino acids 206–232 of the X-linked OGT gene (004860; The Jackson Laboratory, Bar Harbor, ME, USA) were crossed with male mice possessing one copy of the α-MHC-MerCreMer transgene, which consists of a cardiac-specific α-myosin heavy chain promoter directing expression of a tamoxifen-inducible Cre recombinase (005657; The Jackson Laboratory). Cre expression and OGT gene deletion were accomplished by intraperitoneal injection of tamoxifen (75mgs/kg; T5658; MilliporeSigma, Burlington, MA, USA) dissolved in corn oil (C8267; MilliporeSigma) for four consecutive days. To control for potential spurious effects of tamoxifen injection and α-MHC-MerCreMer expression [6], all control mice were tamoxifen treated and α-MHC-MerCreMer positive but had a normal OGT gene. Animals were used 18–30 days post induction (dpi).

Cardiomyocyte isolation

Cardiomyocytes from the LV wall were isolated using 1 mg/ml collagenase type II (CLS2; Worthington Biochemical, Lakewood, NJ, USA) and 0.65 u/ml protease XIV (P5147; MilliporeSigma) via Langendorff perfusion as previously described [15, 17]. Myocytes were stored in a modified Hank’s balanced salt solution buffered with 10 mmol/liter (mM) (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) and supplemented with 1× essential amino acids (11130051; Thermo Fisher Scientific, Waltham, MA, USA) at room temperature. Myocytes were used 1–6 h following isolation.

Electrophysiology

Cav activity was recorded at room temperature using the whole-cell patch clamp technique as previously described [14, 15]. Briefly, myocytes were added to a recording chamber and bathed in a solution consisting of the following in mM: 136 NaCl, 1 MgCl2, 1 CaCl2, 10 glucose, 10 HEPES, 4 CsCl, 0.02 tetrodotoxin; pH 7.4. Cells were ruptured following seal formation and dialyzed (5 min) with an intracellular recording solution (in mM): 110 Cs+ methanesulfonate, 30 CsCl, 10 HEPES, 0.5 CaCl2, 2 MgCl2, 5 Na2ATP, 5 EGTA; pH 7.2. Cells were voltage clamped at − 50 mV and cell size was determined by integrating the capacitance of the cell following a 25 ms, 10 mV step in voltage. Series resistance was compensated to a minimum of 80%. Analog signals were low-pass filtered (5 kHz) then digitized (50 kHz) and voltage clamp protocols were written and executed using Clampex 10.6 (Molecular Devices, Sunnyvale, CA, USA). Leak currents were always fewer than 50 pA and were ignored.

To record ICa, cells were clamped at − 50 mV then depolarized by a series of voltage steps beginning at − 40 mV and ending at 60 mV for 1.2 s in 10-mV increments. The voltage steps were separated by 15 s. The maximum negative current at each test pulse was divided by capacitance for each cell and averaged to report the current–density/voltage relationships. The rate of fast inactivation was determined by fitting the decay portion of a current trace elicited from a test pulse to 20 mV with a bi-exponential function. To calculate conductance, the maximum negative current was divided by the driving force as described [14, 15]. The reversal potential was determined empirically for each cell and was not different between groups. Conductance/voltage relationships were fit with a single Boltzmann function as described to determine activation gating parameters [14, 15].

To measure Cav voltage-dependent inactivation, cells were held at − 50 mV and subjected to conditioning voltage pulses ranging from − 60 to 10 mV in 10 mV increments for 400 ms. Following the conditioning pulses, the cells were depolarized to 20 mV for 400 ms. Each sweep was separated by 15 s. The maximum negative amplitude from each current trace was normalized to the maximum negative amplitude from the first step to 20 mV. The data were fit with a single Boltzmann function as described [14, 15].

To determine the efficacy of β adrenergic activation on Cav activity, myocytes were clamped at -50 mV and depolarized to 10 mV every 30 s as 1 µM of the β adrenergic agonist isoproterenol was perfused through the recording chamber. Isoproterenol was shown to induce a hyperpolarizing shift in Cav activation [44]; therefore, a 10 mV test pulse was used since this was close to the saturating voltage of the GV curve (Fig. 2d) yet still elicited a measurable current from OGTKO myocytes. Current from each test pulse was normalized to the current elicited from the initial test pulse at the start of the perfusion.

Western blotting and antibodies

Total protein was isolated from OGTKO and control ventricles by homogenization in HEPES buffered saline supplemented with 1% amidosulfobentaine, 1× Complete Ultra protease inhibitor cocktail (5892970001; Roche, Mannheim, DE), 0.01 calpain inhibitors I and II (A2602 and A2603; Apex Bio, Houston, TX, USA), 0.001 thiamet G (13237; Caymen Chemical, Ann Arbor, Michigan, USA) and 0.02% Na+ azide. In the studies investigating ubiquitination, 100 mM N-ethylmaleimide was included in the lysis buffer [18]. Protein lysates were Western blotted as previously described using a bis–tris based buffer system and gels containing an acrylamide/bis-acrylamide ratio of 49:1 to facilitate transfer of the large Cav α1 and α2δ1 proteins [14,15,16,17]. Immunodetection was performed using the following commercially-available antibodies: anti-Cav1.2, anti-Cavβ2, anti-Cavα2δ1 (ACC-003, ACC-105 and ACC-015 respectively; Alomone, Jerusalem, IL), anti-OGT, anti-mono/poly ubiquitin (24083 and 3936 respectively; Cell Signaling, Danvers, MA, USA), RL-2 (MA1-072; Thermo Fisher Scientific), anti-NEDD4-1 (611480; BD, San Jose, CA, USA), goat anti-rabbit IgG-horseradish peroxidase (HRP; AP307P; MilliporeSigma) and goat anti-mouse IgG-HRP (AP308P; MilliporeSigma). HRP signals were normalized to total protein staining by addition of 2,2,2-trichloroethanol to the resolving gel to calculate relative protein expression between OGTKO and controls as described [15, 17]. Image analysis was performed in ImageLab (Bio-Rad Laboratories, Hercules, CA, USA) and Western figures are presented as merged files consisting of the chemiluminescent and colorimetric data to show the molecular weight markers.

Whole-cell Ca2+ signaling and myocyte contractility measurements

LV myocytes were stained with fura-2AM and the dye was allowed to de-esterify. Stained myocytes were stimulated with 20 volts at 1 Hz (MyoPacer; IonOptix, Westwood, MA, USA) and Ca2+ transients were recorded ratiometrically all as previously described [15, 17]. Myocyte contraction was recorded independently by edge detection of unloaded cardiomyocytes at 250 Hz [15]; contraction was induced by stimulation as described above. Ca2+ signals and myocyte contraction were analyzed with FelixGX (PTI; Horiba Scientific, Edison, NJ, USA) and IonWizard 7.4 (IonOptix, Westwood, MA, USA) to compare standard parameters including max slope, time-to-peak, Ca2+ transient duration, activation/inactivation kinetics and fractional shortening; these parameters were calculated as the mean of ten consecutive stimulation events that followed ten unrecorded events to ensure complete SR loading. Whole-cell Ca2+ signals and contractility measurements were made at room temperature.

O-GlcNAc affinity precipitation

Protein was extracted as described above and preabsorbed with streptavidin coated magnetic beads (88816; Thermo Fisher Scientific) for ~ 2 h at 4 ℃. To capture proteins with the O-GlcNAc modification, biotinylated succinylated wheat germ agglutinin (sWGA; B-1025S; Vector Laboratories, Burlingame, CA, USA) was coupled to the streptavidin magnetic beads. Following preabsorption, protein was added to the lectin coupled beads and mixed overnight at 4 ℃, washed extensively with lysis buffer and eluted in 1xLDS-PAGE sample buffer at 60 ℉ for 5 min. To ensure specificity, parallel experiments were always performed where sWGA was preabsorbed with 500 mM GlcNAc during the bead/lectin coupling and 50 mM during protein/bead incubations. Following elution, eluates were Western blotted as described above.

Echocardiography

Transthoracic echocardiography was performed using a Vevo 3100 (Fujifilm, Japan) equipped with a digital transducer centered on 30 MHz (20–46 MHz). Mice were anesthetized with ~ 2.0–2.5% isoflurane/O2. B and M mode images were acquired digitally and analyzed using VevoLab 3.1 (VisualSonics) to compare standard hemodynamic parameters as described by us previously [12, 15]. Briefly, 2-D mode in parasternal long axis and the parasternal short axis at the mid-papillary muscle level were imaged. From this parasternal short axis view, the 2-D guided M mode across the anterior and posterior wall were recorded and LV anterior wall thickness at diastole and systole (LVAWd and LVAWs, respectively) and LV posterior wall thickness at diastole and systole (LVPWd and LVPWs, respectively) were measured. Using the leading edge method, all other parameters including ejection fraction (EF) and fractional shortening (FS) were measured from three consecutive cardiac cycles according to the guidelines of the American Society of Echocardiography [40] and as described by us previously [12, 15].

Statistical analysis

Data are presented as mean ± SEM and significance was determined using unpaired student’s t-tests and Mann–Whitney rank sum tests where appropriate with a p criteria of < 0.05. Animal numbers are listed as “N” and cell numbers as “n”. All data are available upon request.

Results

Cardiomyocyte deletion of the OGT gene reduces OGT and O-GlcNAcylation levels by at least 90%

To assess the amount to which induced cardiomyocyte deletion of the OGT gene diminishes OGT expression, Western analysis was performed on ventricular protein lysates and indicated a nearly 90% reduction in OGTKO OGT (Fig. 1a and c; N = 3; p = 0.001). To determine the levels to which O-GlcNAcylation was reduced in OGTKO cardiomyocytes, the monoclonal antibody RL-2, which recognizes O-GlcNAcylated serine/threonine residues with only some other peptide dependence [61], was used and indicated an ~ 97% reduction in OGTKO cardiomyocyte O-GlcNAcylation (Fig. 1b and c; N = 3; p = 0.001). These results demonstrated that, following induction, OGT gene deletion markedly reduces OGT expression and cardiomyocyte O-GlcNAcylation.

Fig. 1
figure 1

Cardiomyocyte OGT-gene deletion significantly reduces OGT and O-GlcNAc levels. a, b Western and immunodetection analysis (upper panels) indicated that, following tamoxifen induction, OGTKO (KO; gray squares) ventricular OGT expression and LV myocyte O-GlcNAcylation are markedly reduced compared to controls (c; black circles). Signals were normalized to total protein levels (lower panels). Molecular weights in kilodaltons are listed on the right. c Mean ± SEM normalized OGT and O-GlcNAcylation signals (N = 3; *p = 0.001; t-test)

Full size image

Cav activity is diminished with chronic reductions in O-GlcNAcylation, while adrenergic efficacy is increased

To investigate whether reductions in O-GlcNAcylation affect Cav activity, we compared whole-cell LV cardiomyocyte ICa from OGTKO and control animals (Fig. 2a). As discussed in the Methods, here and throughout, all control animals were α-MHC-MerCreMer positive, were administered tamoxifen, but possessed a normal OGT gene; thereby, controlling for any effects of tamoxifen-treatment and Cre expression. OGTKO cardiomyocytes demonstrated significant reductions in ICa density at all channel-activating test potentials with no effect on reversal potential (Fig. 2b). For example, at a test potential of 10 mV, OGTKO ICa density was reduced by 38% compared to controls [Fig. 2b and supplemental Table 1 (ST1); p < 0.0001; for all ICa parameters: N = 3–4; n = 18–20]. Cell capacitances were statistically the same between groups (Fig. 2c and ST1). The voltage-dependence of Cav steady-state gating was determined and indicated nearly uniform depolarizing shifts of 5–6 mV for voltage-dependent inactivation and activation respectively, including rightward shifts in the activation and inactivation midpoints but with no effect on the slope of the Boltzmann fits (Fig. 2d and ST1; p ≤ 0.001). To determine the rate of fast inactivation, current traces elicited from a 20 mV test pulse were fit with a bi-exponential function. A 20 mV test potential was utilized because this voltage occurred at a saturating point of the GV curve (Fig. 2d) so that the inherent effect of a rightward shift in voltage-dependent activation, as observed in OGTKO myocytes, would not influence the measurement of inactivation rate. Both the slow and fast components of Cav inactivation were significantly slower in OGTKO cardiomyocytes by 14 and 36% respectively (Fig. 2e and ST1; p ≤ 0.03; N = 3–4; n = 18–20).

Fig. 2
figure 2

Cav activity is diminished, gating depolarized and inactivation slowed in OGTKO LV myocytes. a Representative whole-cell voltage-sensitive Ca2+ current (ICa) density traces recorded from control (black) and OGTKO (gray) LV myocytes. b Current density–voltage relationships indicated significant reductions in OGTKO current densities. c Cell size measured as capacitance was not different between groups. d Cav voltage-dependent gating and window current were significantly depolarized by 5–6 mV in OGTKO LV myocytes. e The rates of the fast and slow components of Cav inactivation were significantly slower in OGTKO myocytes. a–e N = 3–4; n = 18–20; *p ≤ 0.04; t-test

Full size image

A requisite function of cardiac L-type Cavs is response to increased sympathetic drive through activation of cardiomyocyte adrenergic receptors resulting in increased ICa [38]. To test whether OGTKO Cavs respond normally to adrenergic activation, cardiomyocyte Cav activity was recorded before and after perfusion with 1 µM of the β-adrenergic receptor agonist isoproterenol. Myocytes were voltage-clamped at − 50 mV and depolarized to a near-saturating 10 mV test pulse every 30 s during the perfusion. Results from these experiments indicated that isoproterenol induced a robust increase in ICa in both control and OGTKO myocytes (Fig. 3e and f); however, ICa from OGTKO myocytes demonstrated a 51% larger increase following isoproterenol treatment compared to controls (Fig. 3c and ST1; ~ 1.8 versus 2.7-fold difference; N = 3–4; n = 7–8; p = 0.02). Despite the increased efficacy of isoproterenol on OGTKO Cavs, following isoproterenol treatment, OGTKO ICa was still ~ 20% lower than control ICa under basal conditions suggesting that the dysregulation of adrenergic signaling was not solely responsible for the decrease in basal OGTKO ICa.

Fig. 3
figure 3

The efficacy of isoproterenol on ICa is heightened in OGTKO LV myocytes. a Representative ICa density traces following control perfusion (basal) and perfusion with 1 µM isoproterenol (ISO). b Scatter plots of perfusion time with 1 µM ISO versus normalized current indicating increased efficacy of adrenergic stimulation on OGTKO Cav. c Mean ± SEM maximum fold change of ICa following perfusion with1 µM ISO. b and c: N = 3–4; n = 7–8; *p ≤ 0.04; t-test

Full size image

Cav α1 and β2 protein levels are reduced in OGTKO hearts with α2δ1 levels unaffected

To test whether the observed reduction in OGTKO ICa density was caused by a decrease in the number of functional Cav, Western analysis was performed on cardiac protein lysates using antibodies that recognize the three putative cardiac Cav subunits. Cav α1 and β2 expression, but not α2δ1, were significantly reduced in OGTKO ventricles by 52 and 42% respectively (Fig. 4a–c and f; N = 6; p ≤ 0.0004). Under these experimental conditions, only the truncated α1 form was detected, which was shown to represent at least 80% of the total α1 protein pool in cardiac tissue [1, 36] and can migrate in SDS-PAGE conditions through a range of apparent molecular weights (~ 175 kDa to ~ 210 kDa) largely depending on the extent of processing and the acrylamide concentration and acrylamide/bis-acrylamide ratio [11, 24, 28]. The specificities of all Cav subunit antibodies were tested using their respective immunizing peptides provided by the manufacturer of the antibodies.

Fig. 4
figure 4

Cav α1 and β2 expression are decreased, NEDD4-1 expression and protein ubiquitination are increased, and Cavα2δ1 levels are similar in OGTKO ventricles. a–e Western and immunodetection results (top panels) from OGTKO (KO; gray squares) and control (C; black circles) ventricular protein lysates. Molecular weights in kilodaltons are listed on the right. Optical densities were normalized to total protein levels (bottom panels). f Mean ± SEM normalized expression levels. N = 5–6; *p < 0.05; t-test

Full size image

Expression system studies previously showed that the ubiquitin ligase, neuronal precursor cell-expressed developmentally down-regulated 4 isoform 1 (NEDD4-1), down-regulates expression and activity of the L-type Cav [56]. It was also shown that O-GlcNAcylation negatively regulates NEDD4-1 expression [34]. We therefore rationalized that NEDD4-1 would be upregulated in OGTKO cardiomyocytes and this might contribute to the observed Cav downregulation. Western analysis indicated that, indeed, NEDD4-1 expression levels were 41% higher in OGTKO hearts (Fig. 4d and f; N = 6; p < 0.05). A major function of NEDD4-1 is to ubiquitinate proteins which targets them for degradation [32]. While the study by Rougier et al. suggested that the regulation of L-type Cavs by NEDD4-1 overexpression did not involve increased Cav1.2 ubiquitination [56], interest remained in determining whether ubiquitination levels were altered in OGTKO hearts, which would be consistent with increased NEDD4-1 expression. To test this, blotted protein was subjected to immunodetection using an antibody that recognizes mono- and poly-ubiquitinated proteins; the data indicated that protein ubiquitination was increased by 60% in OGTKO hearts (Fig. 4e and f; N = 5; p = 0.008).

Cav α1 and β2 subunits are O-GlcNAcylated but the α2δ1 subunit is not

To investigate whether Cav subunits are directly O-GlcNAcylated, we performed affinity precipitation with biotinylated sWGA that was coupled to streptavidin-coated magnetic beads. sWGA binds terminal GlcNAc residues and can be used to precipitate O-GlcNAcylated proteins [46]. In some previous cases, affinity reagents that recognize the O-GlcNAc modification were shown to cross-react with extracellular N-glycans that possess terminal GlcNAcs [54]. Unlike here, this is often not a limitation because most proteins studied in the context of O-GlcNAcylation are nucleocytoplasmic and are typically not N-glycosylated. The α2δ1 subunit is heavily N-glycosylated and, while it does not appear to occur at high levels, it is possible that the cardiac α1 subunit is N-glycosylated as well [15]; β2 is cytosolic. To overcome the potential spurious reactivity of α1 and/or α2δ1 terminal N-linked GlcNAcs with sWGA, we utilized a mouse model developed by our lab where hybrid/complex N-glycosylation, and therefore the addition of terminal N-linked GlcNAcs, cannot occur (MGAT1KO) [15, 17]. Detergent soluble protein lysates from MGAT1KO hearts were subjected to sWGA precipitation and indicated that both the α1 and β2 subunits were precipitated by sWGA while α2δ1 was not (Fig. 5). Similar experiments were performed using protein lysates from OGTKO and control ventricles (not shown). Precipitation of Cav α1 and β2 by sWGA in control lysates was identical to that which was precipitated from MGAT1KO lysates with only marginal pull-down observed in OGTKO lysates, which was likely the result of spurious reactivity as described above, non-specific binding, and/or residual OGT expression. Together, these data, along with the functional data (Fig. 2), offer strong evidence that Cav α1 and β2 are directly O-GlcNAcylated under basal conditions and that this modification is an important regulator of Cav function.

Fig. 5
figure 5

Cav α1 and β2 are O-GlcNAcylated but Cavα2δ1 is not. Affinity purification (AP) using the lectin sWGA, which recognizes terminal GlcNAc residues, was performed in the presence (+) or absence (−) of 500 mM N-acetylglucosamine (GlcNAc). GlcNAc was used to demonstrate specificity as free excess GlcNAc would saturate binding of the lectin and prevent precipitation of O-GlcNAcylated proteins by the bead-lectin sorbent. Precipitated protein eluates were subjected to Western and immunoblot (IB) analysis using Cav subunit specific antibodies and results indicated that sWGA precipitates Cav α1 and β2 but not Cav α2δ1. Images are representative of 5–6 independent experiments

Full size image

Intracellular Ca2+ release and contractility are diminished and more variable in OGTKO myocytes

In cardiomyocytes, ICa is the primary trigger for release of Ca2+ from intracellular stores [60]. Therefore, to test whether the reductions in OGTKO Cav activity impact intracellular Ca2+ release, we performed fura-2 ratiometric Ca2+ photometry in field-stimulated (1 Hz/20 volts) OGTKO and control LV myocytes (Fig. 6a). OGTKO myocytes demonstrated an ~ 27% decrease in the magnitude and rate of intracellular Ca2+ release and a 12% slowing in the time-to-peak (TTP) of the Ca2+ transient (Fig. 6b–d). Contrastingly, Ca2+ transient duration and rate of decay, which are indicators of the time required for intracellular Ca2+ to be extruded or re-sequestered, were not different between groups (Fig. 6d; for all Ca2+ transient parameters: N = 3; n = 43–66; p ≤ 0.003 where significant). The robust effects on Ca2+ transient activation with no observed effects on Ca2+ transient duration or relaxation suggest that the reduction in OGTKO ICa density contributes to the diminished Ca2+-activated Ca2+-release observed in OGTKO LV myocytes although this does not rule out an effect of reductions in O-GlcNAc levels on ryanodine receptor activity. These data also suggest that, while phospholamban was shown to be O-GlcNAcylated [68], prevention of O-GlcNAcylation under basal conditions does not measurably impact the role of phospholamban in Ca2+ re-sequestration.

Fig. 6
figure 6

The magnitude and kinetics of intracellular Ca2+ release are reduced and more variable in OGTKO LV myocytes with no measurable effect on Ca2+ re-sequestration or efflux. a Representative fura-2 ratiometric Ca2+ transients recorded from control (black) and OGTKO (gray) LV myocytes. b–d Mean ± SEM Ca2+ transient parameters for control (black circles) and OGTKO (gray squares) LV myocytes. e The relative standard deviation (RSD) of Ca2+ handling parameters were calculated by normalizing each parameter’s standard deviation from ten consecutive Ca2+ transients by their respective means and indicated greater variability in OGTKO myocytes. For all Ca2+ handling parameters: N = 3; n = 43–66; *p ≤ 0.03, t-test; #P ≤ 0.009, Mann–Whitney rank sum test. TTP: time-to-peak; CaD50-85: Ca2+ transient duration at 50 and 85% return-to-baseline values; Tau: single exponential Ca2+ transient inactivation time constant

Full size image

To determine whether the decrease in OGTKO Ca2+-activated Ca2+-release could impact contractility, we assessed sarcomere shortening of unloaded and field stimulated (1 Hz/20 volts) LV myocytes from OGTKO and control hearts (Fig. 7a and b). Fractional shortening was significantly reduced, and the rate of shortening was slowed, each by ~ 40% in OGTKO myocytes, with no significant effect on relaxation kinetics (Fig. 7c and d; N = 3; n = 18–29; p ≤ 0.04). In addition to the reductions in myocyte contraction, it was apparent that OGTKO fractional shortening demonstrated significantly greater beat-to-beat variability compared to controls (Fig. 7a and b). To quantify this, the relative standard deviations of ten consecutive contractions were compared between groups. Strikingly, the variability in contractile parameters from rhythmic contractions was ~ 78–177% greater in OGTKO myocytes compared to controls (Fig. 7e; N = 3; n = 18–29; p ≤ 0.02). A similar analysis was performed on the recorded Ca2+ transients (Fig. 6) and while the magnitude of variability was not as large as what was observed for contractility, all OGTKO Ca2+ transient parameters demonstrated significantly greater variability than controls (18–33%; Fig. 6e; N = 3; n = 43–66; p ≤ 0.03).

Fig. 7
figure 7

OGTKO LV myocyte contractility is diminished and more variable. a, b Representative cell shortening from control (black) and OGTKO (grey) field-stimulated (20 volts/1HZ) LV myocytes. c, d Mean ± SEM fractional shortening (FS) and contraction (contract)/relaxation (relax) kinetics (change in length/time; ΔL/ΔT) for control (black circles) and OGTKO (gray squares) LV myocytes. e Relative standard deviation (RSD) of FS and contractile kinetics. N = 3; n = 18–29; *p ≤ 0.04; t-test

Full size image

The deficits in OGTKO cardiomyocyte contractility are conserved at the whole-heart level

To determine whether the reductions and increased variability of myocyte Ca2+ handling and contractility were conserved at the whole-heart level, echocardiography of the LV was performed on OGTKO and control animals pre-induction (baseline) then again at 19 dpi (the time at which the myocyte Ca2+ transient and contractility measurements were made), and at 29 dpi. At 19 dpi, EF and FS were significantly reduced by ~ 37 and 45% respectively in OGTKO animals compared to baseline levels with consistent increases in end systolic volumes (ESV) and internal LV dimension at systole [(LVIDs) Fig. 8 and ST2; N = 7; p ≤ 0.04]. There were, however, no changes in wall thickness, end-diastolic volume (EDV) or internal dimensions (LVIDd) that would indicate hypertrophy (ST2); this is consistent with the similar OGTKO and control cell sizes as indicated through membrane capacitance measurements (Fig. 2c). The loss of contractility was progressive in that at 29 dpi, OGTKO EF and FS were further reduced by ~ 30 and 33% respectively compared to 19 dpi (Fig. 8b and ST2; N = 7; p = 0.01). At 29 dpi, compared to baseline values, an ~ 23% reduction in OGTKO LVPWs was observed with no other changes in wall thicknesses (ST2; N = 7; p = 0.04) suggesting that OGTKO animals forego any hypertrophy and enter directly into heart failure. Control animals showed no significant change in any recorded parameter at any time point tested, and at baseline, all measured parameters showed no difference in systolic function between OGTKO and control hearts (Fig. 8 and ST2; N = 7). Together, these data strongly suggest that the EC coupling and contractile dysfunctions conferred by OGT deletion in cardiomyocytes and described in Figs. 6 and 7 translate directly to aberrant LV shortening and contractility with no indication of hypertrophy at the cellular or LV levels.

Fig. 8
figure 8

OGTKO LV contractility is reduced concurrent with diminished myocyte EC coupling and worsens over time. a Representative M mode images taken from control (left) and OGTKO (right) left ventricles at pre-induction (baseline; top) and 19 days post induction (dpi; bottom). b Mean ± SEM scatter plots of LV ejection fraction (EF) and fractional shortening (FS) parameters measured from echocardiographs of control (black circles) and OGTKO (gray squares) hearts at baseline (B), 19 and 29 dpi. N = 7; *p = 0.01: 19 dpi versus B; #p = 0.01: 29 dpi versus 19 dpi

Full size image

Discussion

Here we show that chronic reductions in cardiomyocyte O-GlcNAcylation significantly impact Cav function. We observed reductions in OGTKO ICa density, with no impact on cell size, and consistent decreases in Cav α1 and β2 expression. Further, the data showed increased efficacy of adrenergic stimulation on OGTKO Cav. We also observed rightward shifts in OGTKO Cav steady-state gating. Together, the data show that reduced O-GlcNAc levels impact Cav through at least three previously undescribed mechanisms: (1) reduced Cav subunit expression, (2) depolarizing shifts in Cav steady-state gating and window current, and (3) increased adrenergic stimulation efficacy. Furthermore, while it is not possible at this time to rule out a contribution of aberrant contractile protein function, some of which were shown to be dynamically O-GlcNAcylated [52, 53], or dysregulation of intracellular Ca2+ handling proteins, the observed reductions and increased variability in Ca2+ handling and contractility are consistent with reduced and altered Cav function including the observed rightward shift in OGTKO Cav window current and slowing of inactivation (Fig. 2d and e) suggesting a critical role for O-GlcNAcylation in these processes. Finally, the severe impact of OGT gene deletion on cardiomyocyte EC coupling is observed at the whole-heart level as evidenced by the concurrent and progressive loss of OGTKO LV contractility (Fig. 8) in the absence of hypertrophy.

Increased NEDD4-1 and protein ubiquitination in the OGTKO are consistent with decreased Cav expression and I Ca

Work by Rougier et al. suggested that NEDD4-1 overexpression regulates L-type Cavs by disrupting the exit of Cavs from the Golgi [56]. Here, studies into a potential role of endogenous NEDD4-1 in reducing OGTKO Cav activity and subunit expression indicated that NEDD4-1 levels and protein ubiquitination are elevated in OGTKO ventricles (Fig. 4d–f) but a direct relationship between these phenomena and the reduction in OGTKO Cavs has yet to be established. O-GlcNAcylation was specifically shown to limit protein degradation by blocking ubiquitination [42]; therefore, it is reasonable to predict increased ubiquitination in the absence of OGT. Studies investigating whether OGTKO Cav are retained in the Golgi or demonstrate differences in ubiquitination or interactions with ubiquitin-related proteins are currently ongoing but are, as of yet, inconclusive. A subsequent study by the same group also identified the de-ubiquitinating (dub) enzyme USP2 as a negative regulator of L-type Cavs [57] and O-GlcNAcylation was shown by other groups to regulate several dub proteins [58]. Dub proteins and ubiquitin ligases such as NEDD4-1 serve antagonizing roles in the ubiquitination process; thus, it is surprising that such proteins would have similar effects on Cav activity/expression. Although it is important to point out that the studies investigating the impact of NEDD4-1 and USP2 on the L-type Cav were performed in over-expression systems that likely do not recapitulate a native cardiomyocyte. The regulation of membrane proteins through ubiquitination and, ultimately, degradation is critically important as well as being exceedingly complex involving thousands of proteins [21]. Whether and how O-GlcNAcylation regulates ventricular myocyte Cav expression, and thereby ICa density, through an interplay with ubiquitination will require additional work; however, it is clear that OGT-gene deletion and, presumably, the subsequent reductions in O-GlcNAcylation, impact ubiquitin signaling as evidenced by the significant increase in ubiquitinated proteins in OGTKO hearts (Fig. 4e and f) suggesting an important link between these processes.

Potential mechanisms by which reduced O-GlcNAcylation impacts Cav gating and results in increased adrenergic stimulation efficacy

While alterations in the OGTKO NEDD4-1/ubiquitin pathway may well explain the reduced ICa density and Cav subunit expression levels, it is unlikely that this pathway can explain the additional observation of a rightward shift in OGTKO Cav steady-state-gating and window current or slowing of inactivation. Thus, we initially hypothesized that the depolarization of steady-state gating was caused through electrostatic mechanisms conferred by an increase in Cav subunit phosphorylation in the OGTKO. This speculation was based on the fact that O-GlcNAcylation often overlaps with phosphorylation where the same serine/threonine residue can be reciprocally modified [30] and cardiac Cav α1 and β2 subunits are substrates of several kinases [23, 25, 51], and, as reported here for the first time, are O-GlcNAcylated under basal conditions in ventricular myocytes (Fig. 5). Thus, it is conceivable to infer conservation of the two modifications among some α1 and β2 serine/threonine residues. Based on this, we rationalized that, in the absence of O-GlcNAcylation in OGTKO cardiomyocytes, the O-GlcNAcylation/phosphorylation equilibrium of Cav α1 and β2 would shift toward increased phosphorylation. The impact of phosphorylation on cardiac Cav function can vary greatly depending on the kinases involved and the Cav subunits and residues that are phosphorylated with both increases and decreases in Cav activity being reported as well as rightward and leftward shifts in gating [7, 35, 37]. As one example, a phosphomimetic mutation in Cav β2 resulted in a decrease in Cav activity and a marked rightward shift in activation and inactivation gating [7]. However, we speculate that if increased Cav phosphorylation in the OGTKO contributes to the effect of reduced O-GlcNAcylation on Cav gating, it may not occur through any one specific phosphorylation pathway, but rather through a non-specific electrostatic mechanism. That is, as suggested by others [3, 27], a general increase in the number of negatively charged phosphate groups attached to the cytosolic portions of OGTKO Cav subunits, or other proteins in sufficiently close proximity, may effectively hyperpolarize the membrane potential sensed by the Cav voltage sensors, thereby requiring a greater depolarization for OGTKO Cav gating to occur as is observed (Fig. 2d). An investigation into the role of increased Cav phosphorylation in conditions of reduced O-GlcNAcylation and how this may impact channel function utilizing proteomic and pharmacologic approaches is necessary and ongoing. To this end, identifying the specific Cav α1 and β2 residues that are O-GlcNAcylated and the phosphorylation pathways that are disrupted in the OGTKO will be of paramount importance in resolving these questions. Alterations in Cav phosphorylation or signaling pathways that utilize phosphorylation may also contribute to the slowing of OGTKO Cav inactivation (Fig. 2e). However, because Ca2+ influx accelerates Cav inactivation [Ca2+-dependent inactivation (CDI) [41]], under these experimental conditions, it is difficult to speculate a cause of the slowing of OGTKO Cav inactivation other than simply due to the reduced current (and thereby, reduced CDI). This is supported by the fact that the fast component of inactivation, which is largely thought to be ascribed to CDI [26], is impacted to a greater degree (~ 36% versus 14%; Fig. 2e and ST1) in OGTKO cardiomyocytes. This does not, however, rule out a direct effect on calmodulin, which is the Ca2+ sensor in CDI [49]. Additional experiments utilizing different intracellular Ca2+ concentrations and chelators or charge carriers (e.g., Ba2+) will be required to explore this phenomenon more rigorously. Irrespective of the precise mechanism, the uniform rightward shifts in OGTKO Cav voltage-dependent steady-state gating and window current and the slowing of inactivation would be expected to confer a complex set of gains and losses of Cav function that would alter intracellular Ca2+ dynamics and, potentially, action potential duration [2]; these effects are all consistent with the significantly increased variability of EC coupling observed in OGTKO LV myocytes (Figs. 6e, 7a, b and e).

In addition to an interplay with ubiquitination and phosphorylation, O-GlcNAcylation was shown to affect protein function by modulating protein-to-protein interactions [67]. Therefore, an additional possible explanation for the reduction in OGTKO Cav activity and the shift in gating may stem from a disruption in the Cav complex, which could impact the stability of the channel. Additionally, evidence suggests an indispensable role for Rad, which is a member of the Rad and Gem/Kir Ras-related GTP-binding protein (RGK) family, in the regulation of cardiomyocyte Cav function. Under basal conditions, Rad inhibits Cav activity and, during adrenergic stimulation, Rad dissociates from Cavs removing its tonic inhibition [44]. Cardiomyocyte-specific deletion of the Rad gene resulted in an increase in Cav activity and a hyperpolarizing shift in activation gating [47]. Several RGK proteins are regulated by O-GlcNAcylation [43, 66], thus, given the data provided here, it is interesting to speculate a potential role for the regulation of Cavs by O-GlcNacylation that is mediated through Rad. If, for example, reductions in O-GlcNAcylation affect the expression of Rad or interaction of Rad with Cavs while also making Rad more susceptible to PKA phosphorylation during adrenergic stimulation, this could explain the effect of OGT gene deletion on Cav activity including the increased efficacy of adrenergic stimulation (Fig. 3). Preliminary studies into whether Rad expression is altered in the OGTKO and whether it is O-GlcNAcylated have been hampered by the lack of a suitable Rad antibody but are ongoing.

Potential pathophysiologic effects of altered O-GlcNAcylation on cardiac function

The reduction in OGTKO Cav activity was surprising in that some of the impetus for these studies was predicated on the fact that experimental models of diabetes/hyperglycemia were shown to result in reduced cardiomyocyte L-type Cav activity, with mechanisms largely unknown [45, 48, 63], while increased O-GlcNAcylation was shown to contribute to the pathologic molecular remodeling associated with diabetes [9, 19, 31, 50, 52, 53, 68]. An initial hypothesis, therefore, was that OGTKO Cav activity would be increased. If increased O-GlcNAcylation is involved in the reduction of Cav activity in diabetic models, there are several possible explanations for the apparent discrepancies between previous disease-related studies and our data here. As one example, there may be a multi-phasic effect of O-GlcNAc signaling on Cav activity in conditions of hyperglycemia. Others have also shown a protective effect of increased O-GlcNAcylation on cardiac function in that acute increases in O-GlcNAcylation, caused by the release of an unidentified humoral factor during a remote ischemic preconditioning intervention (rIPC), contributed to improved restoration of contractility following ischemia reperfusion injury [33]. Interestingly, the group also showed that diabetic patients, for whom O-GlcNAc levels would likely be higher under basal conditions due to the diabetes-related hyperglycemia, offered similar cardioprotection before rIPC but no additional protection after rIPC [33]. More specific models that link hyperglycemia, temporal increases in O-GlcNAcylation, and Cav function will be required to investigate these phenomena more rigorously. Pointedly, however, results from this study indicate that O-GlcNAcylation is a critical and direct regulator of cardiac Cav function; therefore, it is likely that in this model, where cardiomyocyte O-GlcNAcylation is drastically reduced and O-GlcNAc homeostasis is disrupted, a more severe phenotype is observed which is not fully reflective of diabetic/hyperglycemic disease-states where changes in O-GlcNAcylation would likely be more subtle and/or transient. Nevertheless, in addition to providing the first insight into this novel form of cardiac L-type Cav regulation, use of the OGTKO model, along with the development of more refined models that target individual Cav α1 and β2 O-GlcNAcylation sites, should prove invaluable in elucidating the role of O-GlcNAcylation in mediating the effects of various signaling pathways that modulate Cav function in health and disease. Additionally, determining the impact of more transient changes in O-GlcNAc levels on Cav function should help uncover how O-GlcNAcylation regulates Cav activity under basal conditions. Initial studies using glucosamine or OGA inhibitors, both of which can acutely increase O-GlcNAc levels, on isolated cardiomyocytes (1–6 h), indicated that such an approach is ineffective in modulating Cav function although this may be due to an inability to incubate adult mouse cardiomyocytes for a sufficient duration that alters Cav O-GlcNAcylation yet still allows for reliable ICa recordings.

It was previously shown in a similar model that ablation of the OGT-gene in adult mouse cardiomyocytes depressed cardiac function in non-stressed animals as evidenced by significantly reduced EF that onsets at approximately 4 weeks of age [65, 69]. Here we show that the deficits in OGTKO cardiac contractility (EF and to a larger extent FS) onset even earlier (19 dpi) with no indications of hypertrophy at the LV or myocyte levels (Figs. 2c and 8 and ST1 and ST2). Additionally, our data, as well as those from others, indicate that the reductions in ventricular contractility worsen over time along with an increase in systolic volume by 29 dpi (ST2) and an increase in diastolic volume by 2 months [65] suggesting dilation. Inducible cardiomyocyte OGT ablation was also shown to exacerbate the functional deterioration observed in pressure-overload and infarct models [10, 64, 69]. Several lines of questioning were investigated without identifying a causative mechanism. Our data here indicate for the first time that OGTKO LV myocyte Cav activity, intracellular Ca2+ release and contractility are all markedly reduced (Figs. 2, 6 and 7) by as early as 18 dpi (the earliest time-point tested). These data, therefore, suggest that diminished OGTKO cardiomyocyte EC coupling, largely resulting from a direct reduction in L-type Cav activity, plays a major role in the development of heart disease that likely progresses to heart failure, thus highlighting the critical role of O-GlcNAcylation in these processes as well as in the function of healthy and diseased hearts.