Research ArticleCardiology

Pacemaker-induced transient asynchrony suppresses heart failure progression

See allHide authors and affiliations

Science Translational Medicine  23 Dec 2015:
Vol. 7, Issue 319, pp. 319ra207
DOI: 10.1126/scitranslmed.aad2899

Disruptive technology

Healthy and the majority of failing hearts beat synchronously. However, some hearts contract with poor coordination and if they are weak, this worsens clinical outcomes. Pacemakers used to reset a heart’s rhythm can also change the synchrony of contraction, making it better or worse, and current therapy called resynchronization makes it better. Perhaps counterintuitively, Kirk et al. demonstrate that using a pacemaker to purposely induce dyssynchrony—but only for part of each day—makes the synchronous failing heart better. In their process, pacemaker-induced transient asynchrony (PITA), the heart’s right ventricle is paced to induce a 6-hour period of dyssynchrony each day, followed by atrial pacing to resynchronize the heart for the remaining 18 hours. In dogs with heart failure, PITA halted chamber dilation and negative remodeling of the heart tissue, improved cellular signaling and force generation, and resulted in normal muscle fiber structure and function, similar to healthy controls. PITA may help the majority of patients with heart failure who have synchronous contraction and thus are not treated with standard resynchronization pacemakers.


Uncoordinated contraction from electromechanical delay worsens heart failure pathophysiology and prognosis, but restoring coordination with biventricular pacing, known as cardiac resynchronization therapy (CRT), improves both. However, not every patient qualifies for CRT. We show that heart failure with synchronous contraction is improved by inducing dyssynchrony for 6 hours daily by right ventricular pacing using an intracardiac pacing device, in a process we call pacemaker-induced transient asynchrony (PITA). In dogs with heart failure induced by 6 weeks of atrial tachypacing, PITA (starting on week 3) suppressed progressive cardiac dilation as well as chamber and myocyte dysfunction. PITA enhanced β-adrenergic responsiveness in vivo and normalized it in myocytes. Myofilament calcium response declined in dogs with synchronous heart failure, which was accompanied by sarcomere disarray and generation of myofibers with severely reduced function, and these changes were absent in PITA-treated hearts. The benefits of PITA were not replicated when the same number of right ventricular paced beats was randomly distributed throughout the day, indicating that continuity of dyssynchrony exposure is necessary to trigger the beneficial biological response upon resynchronization. These results suggest that PITA could bring the benefits of CRT to the many heart failure patients with synchronous contraction who are not CRT candidates.


Congestive heart failure (HF) affects tens of millions of patients worldwide and remains a leading cause of hospitalization and death (1). In about 25% of patients, the disease is worsened by uncoordinated contraction owing to delays in regional electrical activation (2). This major comorbidity can be treated by cardiac resynchronization therapy (CRT), which electrically stimulates both sides of the heart’s left ventricle to restore coordinate contraction, improving both HF pathophysiology and prognosis in humans (3). These salubrious effects were first attributed to enhanced chamber mechano-energetics because CRT reduces wasted cardiac work while augmenting systemic blood flow (4). However, studies have since shown that CRT also profoundly alters myocardial cell and molecular biology to enhance cell survival (5), myofilament function (6), mitochondrial energetics (7), ion channel regulation (8), and β-adrenergic receptor (βAR) signaling (9, 10). Intriguingly, these changes appear to be a consequence of restoring synchrony in a dyssynchronous failing heart, rather than being a generalized response to hemodynamic improvement (58).

In HF patients with dyssynchrony, CRT enhances function and outcomes beyond that observed in failing human hearts that were never dyssynchronous (3). This suggests that the process of transitioning from dyssynchrony to synchrony may itself confer molecular/cellular benefits. If so, then one might also improve synchronous HF by purposely inducing dyssynchrony for a limited period and then reversing it. We first explored this concept in dogs; surprisingly, rather than worsening the outcome, a 2-week midsequence exposure to dyssynchrony (atrial pacing followed by right ventricular (dyssynchronous) pacing, then returning to atrial) improved β-adrenergic and myofibrillar function (6, 9) over HF hearts that were never dyssynchronous. This concept of transiently exposing an organ to a stimulus that could be damaging if sustained, with the goal of gaining benefits upon its removal, is also common to neurostimulation, ischemic pre-conditioning, and immune therapies. In this instance, the therapy is not the stimulus itself but the host’s reactive biology that ensues after its removal.

Applying dyssynchrony for one or more weeks has limitations because the effective duty cycle (that is, how long one waits before repeating exposure) can vary individually, and prolonging exposure to dyssynchrony and thus reduced function may not be well tolerated. To circumvent this, we investigated daily exposure to a period of dyssynchrony followed by resynchronization, which we have termed pacemaker-induced transient asynchrony (PITA), in dogs with synchronous HF from chronic rapid atrial pacing. We then compared PITA biological and physiological outcomes to HF dogs that received only atrial tachypacing and to healthy control animals. Compared to synchronous HF, PITA attenuated progressive chamber dilation and maladaptive remodeling; augmented β-adrenergic responsiveness at chamber and myocyte levels; and yielded normal myofiber structure, contractile function, and cellular force generation.


PITA blunts progressive chamber dilation and improves β-adrenergic response in HF

Dogs with synchronous HF from chronic rapid atrial pacing (11) received PITA, consisting of right ventricular rapid pacing (dyssynchrony) from 0000 to 0600 each day and atrial rapid pacing (synchrony) for the remaining 18 hours. PITA was initiated after 2 weeks of 100% atrial tachypacing to preestablish HF. At the end of the 6-week protocol, PITA was compared to both HF controls that received only atrial tachypacing throughout and healthy controls. Both PITA and HF dogs were paced at the same rapid rate (200 min−1), with the only variable being which chamber was paced, whereas control dogs were unpaced. We subjected PITA dogs to nighttime dyssynchrony because this is when they are least active. Both HF and PITA dogs were monitored for 24-hour periods to confirm continuous pacing capture, and in PITA dogs, ventricular pacing occurred from 0000 to 0600 with atrial pacing for the rest of the day (fig. S1).

Figure 1A shows example two-dimensional echocardiograms for HF and PITA-treated hearts at initial baseline and 6 weeks (end of study), demonstrating smaller end-diastolic volume with PITA treatment. Summary data for left ventricular volume [end-systolic volume (ESV)] and ejection fraction (EF) (Fig. 1, B and C) and corresponding end-diastolic pressure (EDP) (Fig. 1D) reveal significant increases in ESV and EDP and decreases in EF after 2 weeks of atrial tachypacing. By 6 weeks, both HF and PITA-treated dogs had dilated further, but dilation was less marked with PITA therapy (Fig. 1, A and B). Similarly, EF declined twofold and EDP increased fourfold in HF animals as compared to baseline values, and both EF and EDP changes were partially ameliorated by PITA (Fig. 1, C and D). Although PITA introduced dyssynchrony each night, there was no apparent “memory effect” once atrial pacing was restored during the day. The normal synchronous heart has a dyssynchrony index of 26.9 ± 2.6 versus 75 with dyssynchrony (10). Both HF and PITA models displayed normal synchrony during daytime atrial pacing (HF, 22.5 ± 1.7; PITA, 22.4 ± 3.5).

Fig. 1. PITA improves in vivo cardiac function.

(A) Example end-diastolic echocardiographic images at baseline (BL) and after 6 weeks (end of study) of dog hearts in the HF and PITA-treated groups. The left ventricle (LV) is outlined in yellow. (B and C) Left ventricular ESV (B) and EF (C) assessed by echocardiography at baseline (n = 10), after 2 weeks of atrial pacing (n = 18), at the end of the 6-week atrial pacing protocol (HF, n = 8), and at the end of the 6-week PITA protocol (n = 9). (D) Left ventricular EDP (LVEDP) from hemodynamic studies (baseline, n = 8; 2 weeks, n = 12; HF, n = 13; PITA, n = 10 dogs). EDP data were non-normal and are displayed as box plots. Data were log10-transformed before one-way analysis of variance (ANOVA). Direct comparison to the 2-week group was made by t test. (E) Example left ventricular pressure waveforms in HF and PITA hearts at baseline and with dobutamine (15 μg kg−1 min−1). (F) Dobutamine dose effect on left ventricular contractility (dP/dtmax/IP) (control, n = 6 dogs; HF, n = 9; PITA, n = 8 dogs; some doses missing for individual dogs). Data points indicate individual animals at each dose; symbols are means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 versus control; P < 0.05, P < 0.001 versus HF by one-way ANOVA and Holm-Sidak post hoc test.

Intact heart βAR responsiveness was measured by exposure to intravenous dobutamine. Figure 1E shows example pressure tracings before and after maximal dobutamine stimulation. PITA treatment enhanced maximal rates of pressure rise and decline compared with HF (fig. S2). At baseline, dP/dtmax/IP (maximal rate of pressure rise normalized to instantaneous pressure; a contractility index with less load sensitivity) similarly declined in both the HF and PITA groups compared to healthy controls (Fig. 1F). However, PITA-treated hearts displayed an enhanced dobutamine dose response compared to HF dogs at all doses, resulting in a 38% relative increase in the maximum dobutamine response at 15 μg kg−1 min−1. At this maximum dose, systolic function in PITA-treated hearts was similar to healthy controls, whereas it remained depressed in HF dogs.

PITA enhances myocyte β-adrenergic responsiveness and receptor density

In cardiac myocytes from dogs with HF, baseline sarcomere shortening was depressed compared to both healthy control cells and cells from PITA-treated hearts (Fig. 2A). Corresponding disparities were observed in peak calcium transients (Fig. 2A). HF myocytes also displayed a blunted β-adrenergic response to nonselective (isoproterenol) and selective (β1AR, norepinephrine + prazosin) stimulation. However, cells from PITA-treated hearts displayed essentially normal responses with both forms of β-adrenergic stimulation (Fig. 2A). Although we did not measure isolated myocyte behavior in a separate group of dogs at the 2-week (pre-PITA) time point, prior studies report that substantial β-adrenergic down-regulation occurs by 1 week of tachypacing (12). Thus, these normal responses with PITA suggest functional improvement after initiation of treatment, rather than inhibition of deterioration.

Fig. 2. Myocyte function is depressed in HF after βAR stimulation but near-normal with PITA.

(A) Top: Example tracings of sarcomere shortening and intracellular calcium transients from left ventricular lateral wall myocytes isolated from healthy control (Con), HF, and PITA dogs at baseline and after isoproterenol stimulation (0.1 μM) or after norepinephrine (NE; 0.1 μM) and prazosin (Prz; 1 μM) stimulation. Bottom: Sarcomere shortening and intracellular calcium are quantified as means ± SEM [n = 4 to 7 dogs in each group; number of cells per dog, 5.8 ± 0.4 (mean ± SEM)]. For all three groups, isoproterenol and NE + Prz sarcomere shortening and peak Ca2+ transient data are P < 0.05 versus respective baseline; ***P < 0.001 versus control by two-way ANOVA and Holm-Sidak post hoc test. (B) cAMP activity after isoproterenol or forskolin stimulation. Data are individual dogs; means ± SEM (n = 8). (C) Plasma membrane βAR, β1AR, and β2AR density. Data are individual dogs; means ± SEM (n = 8 per group). In (B) and (C), *P < 0.05, **P < 0.01, ***P < 0.001 versus control by one-way ANOVA and Holm-Sidak post hoc test.

CRT globally up-regulates β2AR responsiveness by blunting inhibitory G protein (heterotrimeric guanine nucleotide–binding protein) subunit αi (Gαi) coupling—a change linked to enhanced expression of regulators of G protein signaling, RGS2 and RGS3 (9). We tested if this mechanism applies to PITA by exposing myocytes to a selective β2AR agonist, zinterol, with or without pretreatment with the Gαi inhibitor pertussis toxin. Unlike CRT, β2AR-stimulated sarcomere shortening remained depressed after PITA and in HF, and pertussis toxin increased this zinterol response similarly in both groups (fig. S3), indicating that the β2AR response contributed less to the overall β-adrenergic response observed with isoproterenol stimulation in the myocyte and dobutamine stimulation in vivo. RGS3 and Gαi protein expression rose similarly in both groups, whereas RGS2 and Gαs were not significantly different from healthy controls (fig. S4). Thus, mechanisms for PITA-related βAR responsiveness differed from those previously identified from CRT (9, 10).

Isoproterenol-stimulated cyclic adenosine monophosphate (cAMP) declined in HF myocytes versus controls, and PITA enhanced this response (Fig. 2B). Cells exposed to the direct adenylyl cyclase activator forskolin showed a similar response pattern (Fig. 2B), but differences were relatively small as compared to βAR stimulation. Collectively, these data suggest that PITA modulates membrane receptor density more than G protein coupling or cyclase activity. This hypothesis was tested by a radiolabeled ligand binding assay, which revealed that both β1AR and β2AR membrane densities were reduced in HF and to a lesser extent with PITA (Fig. 2C). This change in receptor density was not transcriptionally regulated because mRNA expression was similar for both receptors (fig. S5A). Because receptor density may decline from phosphorylation by the G protein–coupled receptor kinases GRK2 or GRK5, we assessed the protein expression of each kinase. Both rose similarly in the HF and PITA groups (fig. S5B), and so unlikely explained the disparity.

PITA-treated myocytes have normal myofilament calcium activation

Systolic reserve also depends on the myofilament response to calcium, which is globally depressed in both synchronous and dyssynchronous HF and normalized by CRT (6). To test the impact of PITA on this behavior, myocytes were isolated from the left ventricle (lateral wall endocardium), chemically skinned, and then exposed to steady-state calcium. Maximal activated force (Fmax) normalized to cellular cross-sectional area declined in HF and was fully restored by PITA (Fig. 3A). However, cross-sectional area was increased in HF myocytes (Fig. 3, B to D), and this change principally explained the decline in normalized Fmax because peak force alone was similar among the three groups (Fig. 3E).

Fig. 3. Myofilament function is depressed in HF and recovered with PITA.

(A) Mean force as a function of calcium concentration (± SEM) and fitted curves for skinned myocytes from the left ventricular lateral wall for control, HF, and PITA. Inset shows force data normalized to Fmax. Summary results for Fmax and EC50 from these curve fits are shown as individual myocytes and means ± SEM [control, n = 15 from 6 dogs; HF, n = 26 myocytes from 8 dogs; PITA, n = 14 myocytes from 4 dogs; number of cells per dog, 3 ± 0.2 (mean ± SEM)]. (B) Representative images (×40) of control, HF, and PITA skinned myocytes stretched to a sarcomere length of 2.1 μm. (C) Myocyte cross-sectional area (CSA) of all myocytes examined in (A). Data are means ± SEM. (D) Mean skinned myocyte width ± SEM (n = 150 cells per group). (E) Fmax from (A) without normalization to CSA. Data are individual myocytes (means ± SEM). (F) Expression of phospho-TnI S22/S23 normalized to total TnI. Data are individual dogs and means ± SEM (n = 4 per group). (G) Expression of phospho–GSK-3β S9 normalized to total GSK-3β (n = 4 per group). For all panels, **P < 0.01, ***P < 0.001 versus control, unless otherwise indicated, by one-way ANOVA with Holm-Sidak post hoc test.

HF also shifted the force-calcium relationship leftward [calcium sensitization or decline in calcium concentration resulting in half-maximal force (EC50)] (Fig. 3A), which PITA reversed. The sensitization with HF was consistent with reduced troponin I (TnI) phosphorylation at S22/S23 (13) (Fig. 3F) and a rise in glycogen synthase kinase 3β (GSK-3β) activity (6), reflected by a decline in its phosphorylation at S9 (Fig. 3G). Reduced TnI S22/S23 phosphorylation was also observed with PITA, but GSK-3β S9 was more phosphorylated (reducing activity), which desensitizes myofilaments. This reduced GSK-3β activity and calcium desensitization with PITA is consistent with the changes observed with dyssynchrony (6).

PITA improves sarcomere structure and myofiber depression with HF

Unaltered maximal force despite greater cell cross-sectional area with HF suggests the presence of either a subpopulation of very weakened myofibrils or a widespread moderate weakening of all myofibrils, which are normalized in PITA-treated hearts. We therefore examined skinned myocytes by electron microscopy (EM). Many sarcomeres from HF cells exhibited structural defects (Fig. 4A), including a wider and wavier Z-band along with wavy/misaligned actin-myosin filaments (Fig. 4B, arrows). Transverse sections showed more space between myofibrils and a reduced myofibril diameter, an appearance also seen in longitudinal sections (Fig. 4C). In control myofibrils, a regular hexagonal arrangement of myosin filaments, each surrounded by six symmetrically placed actin filaments, was observed and confirmed by Fourier analysis (Fig. 4C, inset). In many HF myofibrils, this arrangement was less ordered, with a more diffuse fast Fourier transform (FFT). PITA-treated cells displayed normal myofibrillar structural pattern, fiber size, and intermyofibrillar space, although a wider, wavier Z-band remained. In both control and PITA cells, there was normal sarcomere structure, whereas in HF, among 3727 imaged sarcomeres, 1525 (40%) had disarrayed myofilaments (Fig. 4D). These disarrayed sarcomeres clustered in regions, with <20% of imaged fields exhibiting both normal and disarrayed sarcomeres (Fig. 4E).

Fig. 4. Myofilament structure and function is disrupted in HF but restored in PITA.

(A) Longitudinal sections of skinned myocytes imaged by EM. Normal structures were observed in control and PITA, but HF sarcomeres showed myofilament disarray. Scale bar, 1 μm. (B) Higher magnification highlighting the disrupted myofilament structures: weak and wavy Z-band and M-band and bent/curved filaments with irregular spaces in between that were observed in some HF myofibrils, as indicated by arrows. Scale bar, 200 nm. (C) Transverse sections of myofilaments showing smaller-diameter myofibrils with increased spaces in between and loss of regular filament lattice structure in HF compared with control and PITA-treated animals. Inset: FFTs of boxed areas confirmed loss of normal lattice structure in the HF group. Scale bar, 200 nm. (D) Percentage of normal and disarrayed sarcomeres identified by EM (n = 3727 sarcomeres examined from n = 3 HF dogs). (E) Disarrayed sarcomeres cluster together in HF. The percent of EM fields containing normal, disarrayed, or a mixture was quantified in the HF group (n = 28 fields). (F) Isolated myofibril force. A subpopulation of fibrils generating extremely low force is indicated by the dashed line. Data are individual myofibrils and shown as box plots because HF data were non-normally distributed (control, n = 15; HF, n = 18; PITA, n = 18 myofibrils). P values determined by Mann-Whitney rank sum nonparametric test. (G) Fmax compared after removing the weak myofibrils from (F) (n removed from the analysis: control, 0; HF, 7; PITA, 1 myofibril). n.s., not significant by one-way ANOVA.

To test if heterogeneity of sarcomere structure reflected functional disparities in subpopulations of myofibrils, myocytes were disrupted, and maximum force in individual myofibrils was assessed. Similar to whole, chemically skinned myocytes, mean myofibrillar Fmax declined in HF but was normal with PITA (Fig. 4F). However, this was not due to a general population behavior, but rather to the presence of a subgroup of very low force–generating fibrils in HF [7 of 18 of those tested (~40%)], defined by Fmax <150% of noise levels (Fig. 4F, dashed line). Only one such myofibril (out of 18) was found in PITA-treated hearts, and none (out of 15) in healthy controls. By excluding low force–generating myofibrils, we noted similar myofibrillar Fmax among the groups (Fig. 4G).

To test if disrupted myofibrils in HF resulted from deficiencies in myofilament protein expression, we performed nonbiased quantitative proteomics using sequential window acquisition of all theoretical fragment-ion spectra mass spectrometry (SWATH-MS) to determine protein concentrations in myofilament-enriched samples. There were 490 unambiguous proteins quantified (table S1). Proteins involved with motor force generation, such as TnI, troponin T, troponin C, tropomyosin, actin, obscurin, titin, α-actinin, myosin binding protein C, and myosin light chains (MLC1 and MLC2), were similarly expressed in each group (Fig. 5A). Four proteins showed differential expression [false discovery rate (FDR), one-way ANOVA/q < 0.05, P < 0.0017]; of these, two differed between HF and PITA (Holm-Sidak post hoc test): Bcl2-associated athanogene 3 (BAG3) and nebulin-related anchoring protein (NRAP). Both BAG3 and NRAP increased with HF, but rather than reversing with PITA, they increased even more (Fig. 5B), indicating that these proteins, involved in forming and maintaining sarcomere structure, may be elements of a compensatory mechanism.

Fig. 5.

Proteomic analysis of myofilament-enriched samples revealed changes in sarcomere assembly chaperones. (A) Ratio of the core myofilament proteins in both HF and PITA normalized to control samples (n = 4). Proteins that lay along the identity line (dashed line, slope = 1) were similarly expressed in HF and PITA. The two red dots (BAG3 and NRAP) were significantly different between HF and PITA: P < 0.0017 (FDR, 0.05) by one-way ANOVA. Tm, tropomyosin; MyBPC, myosin binding protein C; TnC, troponin C; TnT, troponin T. (B) SWATH-MS analysis of BAG3 and NRAP expression. Data are individual animals; means ± SEM (n = 4). **P < 0.01, ***P < 0.001 versus control, unless otherwise indicated, by one-way ANOVA and Holm-Sidak post hoc test. βMHC, β–myosin heavy chain.

Randomly distributed dyssynchrony does not replicate benefits from PITA

A key concept underlying PITA is that dyssynchrony is applied for a contiguous period sufficient to induce biological effects that lead to benefits following resynchronization. In this key sense, the therapy should not be equivalent to random dyssynchrony such as occurs in patients who develop frequent premature ventricular contractions or lose CRT pacing in a dyssynchronous failing heart, both thought to worsen disease (1416). To test this, we altered the pacemaker program to flip a four-sided coin every 2 min, with right ventricular pacing instituted only if one specific side was obtained. Thus, the right ventricle was paced for 25% of the day, similar to PITA, but this was distributed randomly, as confirmed by 24-hour monitoring (fig. S6). The random protocol was initiated at week 3 (Rand) in six dogs and at week 1 (Rand6) in three dogs.

In contrast to PITA, in vivo cardiac function was unaltered over HF in hearts receiving randomly distributed right ventricular pacing (Fig. 6, A and B). Intravenous dobutamine induced similar augmentation of dP/dtmax/IP in both groups (Fig. 6C). Isolated myocytes also showed similar reduced rest and isoproterenol-stimulated contraction and peak calcium transient responses (Fig. 6D). Sarcomere function, as measured by Fmax, and calcium sensitivity, EC50, were unchanged in myocytes from the dogs receiving random pacing initiated at the beginning of the protocol compared to those from HF dogs (Fig. 6E). Further, cells from randomly paced hearts displayed similar myofilament disarray as in HF (Fig. 6, F and G), where 22.5% of 3067 sarcomeres displayed the disarrayed phenotype observed in HF. Thus, randomly inducing dyssynchrony by right ventricular pacing throughout the day did not replicate the benefits from PITA.

Fig. 6.

Randomly distributed right ventricular pacing (dyssynchrony) does not confer beneficial effects compared to HF. (A) Changes in ESV (ΔESV) and EF (ΔEF) assessed by serial echocardiography at 6 weeks versus baseline. (B) Absolute left ventricular EDP in random right ventricular pacing in HF dogs. (C) Influence of dobutamine infusion on contraction (dP/dtmax/IP). In (A) to (C), data are means ± SEM (Rand, n = 6; HF, n = 5 dogs). (D) Peak sarcomere shortening and peak calcium transient at baseline and with isoproterenol stimulation in HF and Rand. Data are means ± SEM (baseline HF, n = 13 cells from 4 dogs; baseline Rand, n = 34 cells from 7 dogs; isoproterenol HF, n = 30 cells from 4 dogs; isoproterenol Rand, n = 71 cells from 7 dogs). Black dashed line, baseline control mean; gray line, isoproterenol-stimulated control mean. #P < 0.05 versus respective baseline by two-way ANOVA and Holm-Sidak post hoc test. (E) Mean force as a function of calcium concentration (±SEM) and fitted curves for myocytes from the left ventricular lateral wall for HF and Rand6 (6 weeks of 6 hours/day randomly distributed right ventricular pacing). Summary data for Fmax and calcium sensitivity (EC50) are shown as means ± SEM (HF, n = 26 myocytes from 8 dogs; Rand6, n = 10 myocytes from 3 dogs). (F) EM structural imaging of sarcomeres. Image is representative of n = 3 sarcomeres. Scale bar, 500 nm. (G) Transverse EM image section, showing filament lattice structure. Scale bar, 200 nm. Inset: FFTs confirmed loss of normal lattice structure.


Here, we show that PITA, a daily circumscribed period of pacemaker-induced cardiac dyssynchrony that is then reversed, suppresses progressive heart dilation, and enhances in vivo and cellular β-adrenergic response, intact myocyte function, calcium handling, and myofilament function and ultrastructure in a canine model of tachycardia-induced HF. Whereas PITA diminished global dilation and dysfunction, it did not prevent it entirely; however, this is not surprising given that tachypacing stress was applied throughout. By comparison with a tachypaced CRT model, PITA reduced chamber volumes even more (5, 10). Myocyte functional improvements from PITA were similar to those reported from CRT in dogs (6, 9), but we found that their molecular mechanisms differed, in that PITA did not involve RGS proteins and had opposing effects on GSK-3β activity, revealing PITA as a new and distinct form of HF therapy.

PITA may at first seem counterintuitive because dyssynchrony depresses function when sustained, yet here we find benefits when it is applied for several hours daily and then removed. One analogy is myocardial preconditioning, where cardiac protection is generated after first applying several brief periods of intense ischemia followed by reperfusion (17). More sustained or randomly induced ischemia depresses function. Very short-term dyssynchrony pacing (three 5-min episodes separated by synchronous pacing) may also induce preconditioning, as reported in a rabbit model (1820); however, unlike PITA, short-term dyssynchrony would not be applicable to chronic disease.

PITA affected two primary behaviors underlying systolic reserve, βAR responsiveness and myofilament function, thus blocking defects that otherwise develop with synchronous HF. The primary cause of improved βAR responsiveness was increased receptor density, and although several possible explanations were tested, the mechanism remains unclear. The change in calcium sensitivity with PITA is opposite to that from CRT (6), and this difference could relate to disparities in GSK-3β phosphorylation, which declines in synchronous HF and CRT but increases in dyssynchronous HF (6). The rise in GSK-3β phosphorylation with PITA is consistent with the effects of its associated dyssynchrony and may explain its restoration of calcium sensitivity to normal.

The more prominent change with PITA was the recovery of maximal activated force normalized to cell cross-sectional area. With HF, this decline in Fmax was primarily due to a lack of increase in force despite myocyte enlargement, which has been previously observed in experimental and human HF (2124). Therapies that reduce cell size, such as ventricular assist devices (21), CRT (6, 25), and now PITA, also leave absolute Fmax unaltered, but the normalized value rises. To our knowledge, the data presented here are the first to test if uncoupling of Fmax and cell size results from generalized myofibrillar depression or the presence of a subpopulation of weak myofibrils. Our results support the latter, with a remarkable quantitative similarity in the decline in normalized Fmax (38%) and the amounts of disrupted sarcomeres (40.3%) and weak myofibrils (39%). We cannot prove that the functionally defective myofibrils had damaged sarcomeres because EM cannot be performed on an isolated myofibril. However, the clustering of structural defects in the EM analysis supports their presence within a given fiber.

The mechanism by which PITA prevents myofibrillar defects remains unanswered. Overall global chamber dilation and stress were less in PITA-treated hearts compared with HF hearts, and this might affect signaling and/or force to stimulate myofibrillogenesis and enhance fiber quality control. Compromised energy availability and reduced protein turnover in HF have been linked to depressed proteasome function and autophagy (26, 27). This might hinder the cell’s ability to eliminate poorly formed proteins and structures, compromising sarcomere architecture and function. In this regard, the relative PITA-induced increase in BAG3 and NRAP is intriguing because both are critically involved with sarcomere structure. BAG3 maintains myofibrillar integrity during mechanical stress through its interaction with heat shock cognate 70 (28), whereas NRAP acts as a scaffold for actin–α-actinin assembly at the start of myofibrillogenesis (29). That both rose in HF may reflect a compensatory yet inadequate response if cell size is increasing, yet when cells are again smaller as with PITA, their further rise might prove more effective in sarcomere restructuring. Proof of their role cannot be established in the current dog model but will require development of a chronic pacing mouse model combined with genetic manipulation.

The application of dyssynchronous pacing by PITA differs importantly from conditions where the heart is prematurely activated by abnormal excitable tissue (for example, extrasystoles) or where CRT is used in dyssynchronous HF yet intermittently fails to capture both ventricles. Both are considered as contributors to worsening human HF (15, 16). However, dyssynchrony from these clinical settings is not sustained for hours and absent otherwise, but rather randomly distributed throughout the day. We found that random dyssynchrony did not replicate PITA benefits, indicating that continuity of dyssynchrony exposure at least long enough to trigger a biological response once it is resynchronized is key to its efficacy. How long this period needs to be and whether there is a better time of day to deliver it than at night remain to be determined. These are difficult questions to answer in a large animal model given the expense and number of animals required, but may be tested in rodents. The simplicity of the concept suggests that one might translate this to humans directly, conducting a proof-of-concept study first before assessing alternative dosing options.

Our study has some limitations. First, we used a tachypacing model to induce HF. Although tachycardia is not a very common cause of the disease in humans, it is a well-established and accepted large animal HF model reproducing many organ and cellular/molecular aspects of the human disease (30). Unlike myocardial infarction, which itself leads to uncoordinated wall motion and regional heterogeneity of cellular properties, the current model allowed us to test the effect of a single change: the chamber being paced. Global dilation without regional infarction is a common HF presentation, so the results certainly have implications to a substantial population even if excluding this subgroup. Tachypacing also leads to early global dysfunction (31) and depression of excitation-contraction coupling (32) and βAR signaling (12), all within days to 1 week, which is why we waited until week 3 to initiate PITA. Our 2-week global data mirror these prior studies, but we did not elect to replicate earlier myocyte findings because this would require sacrifice of an entirely separate set of dogs at the earlier time point.

We also focused our cellular and molecular analysis to a single site (lateral wall) rather than duplicate these experiments in the anterior wall as well. This is because prior studies in always dyssynchronous hearts have shown that myofibrillar (6) and βAR regulatory (10) dysfunction are similar in both territories. This also holds for survival signaling and other factors (3). Because PITA induced dyssynchrony for only 6 hours (versus 24 hours), regional disparities would be even less likely. Our study design did not test if PITA could rescue more advanced HF. Doing so would have required extending the pacing protocol for several months, raising mortality and necessitating a different protocol. However, we did delay implementation until demonstrable disease had been induced. Last, canine models do not permit molecular cause-effect analyses.

In summary, transient exposure to dyssynchrony by PITA therapy ameliorates HF pathobiology and improves cardiac rest and reserve function. We identified a novel mechanism for the decrease in maximal activated force in HF—the addition of structurally and functionally incompetent myofibers and sarcomeres—and showed that PITA reverses and/or avoids this defect, improving cellular and global function in the process. These results suggest a new therapy for the sizable proportion of HF patients who are not candidates for CRT, many of whom already receive implantable programmable stimulation devices to prevent sudden cardiac death or to provide rate-responsive pacing.


Study design

The study was designed to identify differences in dogs that were synchronously tachypaced for the entire pacing protocol (6 weeks) and those that were treated with PITA. A subgroup of dogs was randomly assigned to the control group and received no pacing. The HF and PITA groups were paced at the right atria 100% for the first 2 weeks. All paced dogs received echocardiographic and hemodynamic measurements at baseline (before pacing) and at the 2-week time point. At the beginning of week 3, dogs were randomly assigned to either continued atrial pacing (HF) or PITA treatment. At the end of the 6-week protocol, echocardiographic, hemodynamic, and dobutamine challenge experiments were conducted. The animals were then euthanized, heart tissues were harvested, and additional experiments were conducted as described below. The number of dogs in each group was determined by power analysis (for in vivo results) and matched our previous work with this dog model of HF (5, 6, 9, 10), although n values for each assay done on tissue samples varied for each particular experiment and are indicated in the figure legends.

Canine models

All studies were approved by the Johns Hopkins Animal Care and Use Committee and performed by trained personnel. Details of the canine model have been reported previously (5, 10). For the current study, an Anthem CRT-P pacemaker (PM3212, St. Jude Medical) was modified through a material transfer agreement between Johns Hopkins University and St. Jude Medical. The proprietary device communicated with a custom laptop-based programmer/emulator and pacemaker firmware to pace at 200 beats per minute (bpm) (above the device’s normal upper limit of 180) and automatically implement time-of-day changes in pacing mode (right ventricle versus right atrium). With the exception of healthy controls, animals were tachypaced at 200 bpm, 24 hours/day, in VVI mode for 6 weeks to induce HF, with one group (HF) being atrially paced the entire time and the other receiving PITA (atrial pacing from 0600 to 2359, right ventricular pacing from 0000 to 0600) starting on week 3. Right ventricular free-wall pacing induces left ventricular dyssynchrony similar to that with a left bundle branch block as indexed by variance in wall motion timing by tissue Doppler (10). To date, a fully implantable device with the time-of-day switch to alter between synchrony and dyssynchrony pacing has not been commercially developed, but recommended methods for reproducing this program are provided in Supplementary Methods.

A third group received right ventricular pacing starting on week 3, but distributed randomly during the 24 hours. The algorithm followed a four-sided coin flip assigning right ventricular/atrial pacing site at a 1:3 ratio. Last, three dogs received random pacing starting on week 1. At the end of study, animals were anesthetized with pentobarbital and intubated, and the heart was excised under ice-cold cardioplegia.

Intact myocyte isolation and functional studies

Primary isolated left ventricular myocytes were obtained as previously described (10). Cells from the lateral endocardium were studied because cell function and adrenergic signaling have been shown to be similar from this and anterior/septal walls in the pacing models (10). Sarcomere shortening and whole-cell calcium transients were assessed with an inverted microscope (Ellipse TE2001, Nikon) equipped with an image fluorescence system (MyoCam, IonOptix). Studies were performed at 37°C, and cells were field-stimulated at 1 Hz, as described (10). Drug doses were 100 nM isoproterenol, 100 nM norepinephrine, 1 μM prazosin, 1 μM zinterol, and pertussis toxin at 1.5 μg/ml, as used previously (9, 10).

Isolated myofibrils

Tissue from the endocardium of the left ventricular lateral wall was flash-frozen in liquid nitrogen and stored at −80°C. Isolated myofibril functional experiments were carried out as described (33). Briefly, thin strips of tissue were dissected from the tissue and were “skinned” in relaxing solution plus 1% Triton X-100. Muscle strips were then mechanically homogenized in the presence of relaxing buffer but without Triton. The myofibrils were lifted between two glass microtools: a very stiff holding probe and an L-shaped force probe. Sarcomere length was calculated from the FFT of the image. Sarcomere length was set at 2.1 μm. The deflection of the force probe in response to contraction of the myofibril was detected by a split photodiode placed at the camera port of the inverted microscope. The myofibril was maintained at 15°C during this experiment.

Electron microscopy

Membrane-permeabilized myocytes were prepared as described in Supplementary Methods. Isolated cardiac myocytes were fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) at room temperature for 2 hours. Fixed myocytes were postfixed with 1% OsO4 in 0.1 M sodium cacodylate buffer, dehydrated in a graded alcohol series, and embedded in Epon. Sections (70 nm thick) were cut with a diamond knife and then stained on copper grids with uranyl acetate followed by lead citrate (34). Images were obtained at 80 kV on a Tecnai Spirit transmission electron microscope (FEI) using a Gatan Erlangshen charge-coupled device camera. FFTs of selected regions were obtained using the ImageJ [National Institutes of Health (NIH)] FFT function (process/FFT/FFT).

βAR density

Myocardial sarcolemmal membranes were prepared by homogenizing whole hearts in ice-cold buffer containing 50 mM tris, 2 mM EDTA, 12.5 mM MgCl2, 10 mM sucrose, aprotinin (2 μg/ml), and leupeptin (2 μg/ml). Membrane protein concentration was determined by the Bio-Rad protein assay. Total specific binding was determined by incubating 25 μg of membrane homogenate with a saturating concentration of 125I-cyanopindolol (500 pM; PerkinElmer) in binding buffer.

Total βAR density was determined using 20 mM alprenolol, whereas 300 nM CGP 20712A was used to saturate all β1ARs. β2AR density was obtained by subtracting β1AR density from total βAR density. Binding assays were conducted at 37°C for 90 min and terminated by rapid vacuum filtration over glass fiber filters (Whatman GF/B catalog no. FPD-196), which were subsequently washed and analyzed in a gamma counter. Specific binding was reported as femtomoles of receptor per milligram of membrane protein.

SWATH mass spectrometry

Myofilament-enriched samples were solubilized in 8 M urea, 0.1% SDS. For each sample, 200 μg of protein was digested using the filter-aided sample preparation (FASP) method (35). Samples were digested with a trypsin/Lys-C solution; acidified with trifluoroacetic acid; desalted on Oasis HLB μ-elution plates (Waters); eluted with 80% acetonitrile (ACN), 0.1% formic acid (FA); and dried down. Samples were resuspended in 0.1% FA at a concentration of 4 μg/μl for liquid chromatography–MS/MS analysis.

Samples were analyzed on a 5600 TripleTOF mass spectrometer (Sciex) equipped with an Eksigent ekspert 415 nanoLC, Ekspert cHiPLC, and Ekspert nanoLC 400 autosampler. Pooled samples were analyzed in shotgun mode [data-dependent acquisition (DDA)] to create a spectral ion library. Individual samples were analyzed using SWATH acquisition [also known as DIA (data-independent acquisition)] with 100 variable SWATH windows from 400 to 1250 mass/charge ratio (m/z). For both DDA and SWATH acquisitions, samples were loaded onto a trap column (nano cHiPLC Trap) for 10 min and then separated using a nano cHiPLC ChromXP C18-CL column using a linear AB gradient (where A is composed of 0.1% aqueous FA and B is 0.1% FA in ACN) from 5 to 35% solvent B for 120 min, 35 to 85% B for 2 min, holding at 85% for 5 min, and then reequilibration at 5% B for 17 min. Exogenous retention time peptide standards (iRT, Biognosys) were directly spiked into each sample before DDA and SWATH analyses.

DDA files were searched using the Paragon algorithm with the canine UniProt (July 2014) library, resulting in 529 proteins (which was reduced down to 490 proteins unambiguously identified in the individual samples) and 8803 peptides identified at 1% global FDR. The resulting .group file was loaded into the SWATH 2.0 microapp within PeakView software (Sciex) to construct the peptide ion library. The resulting ion library was uploaded into the BaseSpace OneOmics cloud server (, and protein areas were calculated using the Protein Expression Extractor app within BaseSpace, implementing the most likely ratio normalization. Individual protein areas were then normalized to βMHC protein area for each sample.

Statistical analyses

Comparisons between multiple experimental groups (with no repeated measures) were performed by one-way ANOVA with Holm-Sidak post hoc test. Data involving multiple treatments (Figs. 2A and 6D and figs. S2 and S3) were analyzed by two-ANOVA with Holm-Sidak post hoc test. Dobutamine dose data (Fig. 1F) were analyzed by analysis of covariance (ANACOVA), one-way ANOVA at each dose, and Holm-Sidak post hoc test. Myofibril data (Fig. 4F) were analyzed by Mann-Whitney rank sum nonparametric test, HF versus control and PITA. Before analysis, EDP data were transformed using a standard −log10 transformation because raw data were nonnormal. Repeated measures were treated as independent to avoid issues with missing data points. P < 0.05 was considered significant.



Fig. S1. Holter monitoring confirmed atrial pacing during the day and ventricular pacing at night.

Fig. S2. β-Adrenergic response in HF and PITA.

Fig. S3. β2AR stimulation revealed little improvement in PITA.

Fig. S4. Western blots of βAR signaling.

Fig. S5. Changes in βAR density were not due to mRNA expression of GRK signaling.

Fig. S6. Holter monitoring confirmed randomly distributed 2-min periods of right ventricular pacing.

Table S1. SWATH MS protein areas for each sample, normalized to βMHC.


Acknowledgments: We thank N. Witayavanitkul, A. Dvornikov, and M. Phillip for their technical expertise and assistance. Funding: This work was supported by the NIH [P01-HL077180 to D.A.K. and J.E.V.E.; T32-HL007227, P01-HL059408, and Abraham and Virginia Weiss and Michael and Janet Huff Endowments to D.A.K.; NHLBI-HV-10-05(2) to J.E.V.E.; HHSN268201000032C to J.E.V.E.; R01-AR034711 to R.C.; and P01-HL75443 to H.A.R.], the American Heart Association (14SDG20380148 to J.A.K.), and the Burroughs Wellcome Fund (1012753 to J.A.K.). Author contributions: J.A.K. performed skinned myofilament studies, isolated myofibril studies, Western blots, and data analysis; prepared the manuscript; and provided intellectual input. K.C. performed isolated myocyte functional studies and Western blots. K.H.L. and R.C. conducted EM and image analysis. E.K. and T.G.F. developed the custom pacemaker software. R.J.H. and J.E.V.E. conducted MS studies. G.P. and H.A.R. conducted βAR experiments. R.S.T. performed canine surgeries. I.P. and T.P.A. conducted echocardiographic studies. P.d.T. supervised isolated myofibril studies. D.A.K. orchestrated the study, assisted in data analysis, edited the manuscript, and provided intellectual input. Competing interests: J.A.K. and D.A.K. are listed as inventors on a pending patent on the use of temporary dyssynchrony to improve cardiac function in failing hearts. H.A.R. is a scientific cofounder for Trevena Inc., a company that is developing G protein–coupled receptor–targeted drugs. E.K. and T.G.F. hold stocks in St. Jude Medical Inc., which developed the pacemaker system and software. Data and materials availability: Proteomic data were uploaded to PeptideAtlas (, accession number: PASS0076.
View Abstract

Navigate This Article