Cardiac Memory Is Associated With Decreased Levels of the Transcriptional Factor CREB Modulated by Angiotensin II and Calcium
Cardiac memory (CM) has short- (STCM) and long-term (LTCM) components modulated by calcium and angiotensin II. LTCM is associated with reduced Ito and Kv4.3 mRNA levels. Because the cAMP response element binding protein, CREB, contributes to CNS memory transcription, we hypothesized that it might be a transcriptional factor in CM, influenced by calcium and angiotensin II. We studied STCM in dogs that were AV sequentially paced (AVP) for 2 hours or sham-operated. STCM was evaluated with ECG and vectorcardiogram (VCG), and subepicardial biopsies were taken at 5 to 120 minutes and investigated for CREB. LTCM was studied in dogs paced for 3 weeks and in sham controls. At 3 weeks the heart was excised, biopsies obtained, and CRE binding tested. STCM induction occurred in AVP dogs but not in sham or AVP dogs treated with saralasin or nifedipine. Nuclear CREB was significantly decreased at 2 hours in the AVP no-drug group only. LTCM dogs manifested reduced binding of nuclear proteins to CRE, and CRE binding activity in the promoter region of Kv4.3. In conclusion, there is an association between STCM induction and decreased nuclear CREB that is angiotensin-modulated and calcium-dependent. Moreover, the decreased CRE binding after 3 weeks of AVP combined with CRE binding activity in the Kv4.3 promoter can explain the Kv4.3 mRNA and Ito downregulation that characterize LTCM.
Memory is a form of remodeling common to many organ and cell systems and is widely studied in neural tissues.1–5 Classically, neuronal memory has been considered as short- or long-term.3,4 Long-term memory requires new protein synthesis, whereas short-term memory is thought to result from protein phosphorylation.3 Changes in transcription are associated with long-term neural memory, for which the cAMP response element (CRE) binding protein, CREB, is a major factor.3
Cardiac memory (CM) shares a number of characteristics with that in CNS5 and has been studied similarly in terms of short-6,7 and long-term7,8 processes. Paralleling CNS, CM can be induced by electrical shocks. In heart, these are initiated by ventricular pacing.7,8 Expression of short-term CM (STCM) and long-term CM (LTCM) is related to the transient outward potassium current, Ito. In the setting of LTCM, specifically, Ito is decreased in magnitude and altered in kinetics. Moreover, both STCM and LTCM are prevented by pharmacological blockade of ICa,L, while STCM induction is prevented by AT1 receptor blockade as well. This suggests roles for angiotensin II–modulated,9 calcium-dependent processes in CM evolution.10
CM not only represents a nonpathological form of cardiac remodeling, but it has been implicated in the modulation of repolarization in ventricles8 and atria11 and in modulation of drug actions on the heart.7,12 Both our own earlier data8 and the neurophysiological literature3,4 suggest that a continuum of mechanisms link STCM and LTCM. In heart, this would imply a set of interrelated and contemporaneous processes initiated by ventricular pacing: one modifying the existing channel proteins that determine repolarization to induce the electrocardiographic changes of STCM, another modulating transcriptional activity to alter the channel protein expression required for LTCM.
In this study, we used a canine model of ventricular pacing to induce STCM and LTCM and tested the hypothesis that CREB and phosphorylated (pCREB) are altered by ventricular pacing. Specifically, given the observation that a downstream target in CM, Kv4.3 (one of the molecular correlates of Ito), which has a potential CREB binding site in its promotor region, is downregulated,13 we hypothesized that CREB activity would be downregulated. Moreover, since CM induction is modulated by ICa,L– and angiotensin II–blocking drugs,9,10 we hypothesized that any CREB changes initiated by 2 hours of ventricular pacing would be attenuated by nifedipine or saralasin.
Materials and Methods
Protocols were approved by the Columbia University Institutional Animal Care and Use Committee.
Short-Term Pacing Protocol
Twenty-one dogs (Team Assoc, Inc, Bayville, Conn) were anesthetized with Pentothal (17 mg/kg IV), intubated, and ventilated with isoflurane (1.0% to 1.5%). A percutaneous catheter was inserted in a femoral artery and connected to a pressure transducer for continuous blood pressure monitoring via a computerized recording system (Gould Instrument Systems, Inc, PO-NE-MAH acquisition software). The heart was approached through a left-sided thoracotomy and suspended in a pericardial cradle. Bipolar pacing electrodes were attached to the left atrium (LA) and the anterior wall of the left ventricle (LV) (Figure 1A).
In 7 dogs, LA pacing (AP) was initiated at 20% >sinus rate and the heart was allowed to stabilize for 45 minutes (Figure 1B). Then, AV sequential pacing (AVP), with P-R interval=50 ms to avoid fusion beats, was commenced at 20% >sinus rate in 6 dogs and maintained for 2 hours.
In 5 dogs, we infused saralasin (0.5 μg · kg−1 · min−1) and in 5 others nifedipine (3 μg · kg−1 · min−1, after a 30 μg/kg bolus) for the last 30 minutes of the equilibration period and throughout the AVP protocol (Figure 1B). Four instrumented sham control dogs were paced from the right atrium at 20% >sinus rate throughout the protocol.
Long-Term Pacing Protocol
The pacing protocol for long-term memory has been described previously by Shvilkin et al.8 Briefly, we used sterile surgical techniques to implant bipolar pacing electrodes (model 4965, Medtronic) in the LV inferior wall and the left atrial appendage of 10 dogs. The electrodes were attached to an AV sequential pacemaker (Prodigy DR, MEdtronic) placed in a subcutaneous pocket in the dorsal thorax. After 2 to 3 weeks of recovery, dogs were paced for 21 days with lower and upper tracking rates of 120 and 150 bpm.
Six ECG leads were monitored continuously throughout each experiment. In the STCM protocol, reference ECG and vectorcardiogram (VCG) recordings were made at 15 minutes of AP, just before and just after the initial myocardial biopsy (Figure 1B). Additional ECGs and VCGs were recorded at 60 and 120 minutes of AVP. AVP was briefly interrupted by exclusive AP to record the T-wave changes of CM. Changes in T-wave vector amplitude were analyzed to quantify evolution of CM.8
Quantification of LTCM was performed via ECG and VCG at biweekly intervals as described previously.8
Tissue Acquisition in Short-Term Protocol
Myocardial samples, 6 mm in diameter and 2 mm deep to the epicardium, were taken using a bore, avoiding major blood vessels. Tissues were immediately frozen in liquid nitrogen. An LV reference biopsy was taken at 15 minutes of AP (Figure 1A), with additional biopsies at 5, 20, 60, and 120 minutes of AVP in a circular fashion around the pacing electrode. Distances were 10 to 20 mm from the electrode (Figures 1A and 1B).
To investigate if there was a spatial gradient for pacing-induced changes in CREB, in 3 LV-paced dogs, we took three additional biopsies in a linear direction away from the pacing electrode. By way of comparison, to determine whether dispersion of CREB occurs in the absence of AVP, hearts of 4 additional anesthetized dogs were excised and multiple LV tissue samples were taken spanning the region of interest in our experiments.
Based on findings in the initial protocol, the biopsy procedure was modified in saralasin- or nifedipine-treated dogs. Here, only a reference biopsy and a 2-hour AVP biopsy were taken. The protocol was otherwise unaltered.
Tissue Acquisition in Long-Term Protocol
In the long-term protocol, dogs were anesthetized and their hearts removed at 3 weeks of pacing. Epicardial samples were acquired 10 to 20 mm from the LV pacing electrode.
Isolation of Nuclear Extracts
Frozen samples were rinsed in ice-cold hypotonic buffer A (10 mmol/L HEPES, 1.5 mmol/L MgCl2, 10 mmol/L KCl, 0.5 mmol/L DTT, 0.5 mmol/L Na3VO4, and protease inhibitors). Tissues were minced in 1 mL of buffer A with 20 strokes in a glass Dounce homogenizer and centrifuged at 10 000 rpm for 6 minutes. The supernatant containing the non-nuclear fraction was removed. The pellet was washed in 1 mL of buffer A and resuspended and incubated on ice for 1 hour in buffer B (20 mmol/L HEPES, 1.5 mmol/L MgCl2, 0.2 mmol/L EDTA, 0.5 mmol/L DTT, 25% glycerol, 500 mmol/L NaCl, 0.1% Triton X-100, 0.5 mmol/L Na3VO4, and protease inhibitors). Samples were centrifuged at 14 000 rpm and the supernatants, containing the nuclear fraction, were saved. Protein contents of the nuclear fraction was measured according to Bradford.14 Two times SDS-PAGE loading buffer was added to the supernatants in a 1:1 ratio.
Western Blot Analysis
Thirty micrograms of nuclear protein or protein from the crude fraction (containing cytosolic and membranous fractions) was separated on a 4% to 20% Tris-glycine gradient gel (Novex) and transferred to a PVDF membrane (BioRad). Membranes were blocked in phosphate-buffered saline (PBS) containing 0.1% Tween 20 (PBST) and 2% BSA for 2 hours. Membranes were incubated in one of the following primary antibodies (CREB, Cell Signaling 1:2000; pCREB, Cell Signaling 1:2000; histone, Santa Cruz 1:1000) at 4°C in PBST, 2% BSA overnight.
Membranes were washed in PBST 3×5 minutes and exposed to the appropriate HRP-linked secondary antibody (rabbit IgG, Cell Signaling 1:2000; mouse IgG, Santa Cruz 1:2000; goat IgG, Santa Cruz 1:5000) in PBST, 2% BSA for 1 hour at room temperature. Membranes were washed (6×5 minutes) with PBST, and immunodetection was performed using an enhanced chemiluminescence method (Amersham Pharmacia Biotech).
To control for loading conditions, membranes were stripped (2×1 hour) at 80°C in 20 mmol/L glycine, 0.1% Tween (pH 2.5). Membranes were blocked and incubated again with anti-histone 1 as a primary antibody.
Results were scanned and quantified using densitometry and ImageQuant software.
Electrophoretic Mobility Shift Assay (EMSA)
Nuclear extracts were isolated using a nuclear extracting kit (Panomics) according to the manufacturer’s instructions. EMSAs were performed with the Panomics kit according to the manufacturer’s instructions. Briefly, 5 μg of nuclear extract was incubated with 10 ng of 5′ biotin-labeled probe, 2 μL of 5× binding buffer (Panomics), and 1 μg of poly d(I-C) in a total volume of 10 μL for 30 minutes at 15°C to 20°C. Samples were separated on a 6% polyacrylamide gel in 0.5× TBE at 4°C and transferred to a Pall Biodyne B membrane with standard electroblotting procedures for 30 minutes at 300 mA at 4°C. Bound oligonucleotides were fixed to the membrane in a UV crosslinker for 3 minutes. Biotinylated oligonucleotides were recognized by streptavidin-HRP conjugate and detected by enhanced chemiluminescence. Oligonucleotide sequences of the probes used were as follows: CREB, 5′-biotin-AGAGATTGCCTGACGTCAGAGAGCTAG-3′ (Panomics) and biotin-5′-ATTTGGGGACTGCAGAGTGATTT-3′ for a probe based on a potential binding motif for CREB in the promotor region of human Kv4.3 (probe CRE-1).
Data are expressed as mean±SEM. ANOVA was used to analyze effects of pacing, sham, or drug treatment on T-wave vector amplitude, CREB, and pCREB. Statistics were done on raw data. Tukey’s test was used to evaluate the presence of a spatial CREB gradient. P<0.05 was considered significant.
Short-Term Cardiac Memory
During STCM experiments without drugs, arterial pressure was stable at 128±8/72±7 kN/m2 after performing the thoracotomy. Pressure was lower but remained stable in the nifedipine- (111±4/67±7 kN/m2) and saralasin- (117±3/68±5 kN/m2) treated dogs. No change in QRS complex configuration or duration was noted and there were no arrhythmias in any of the dogs (data not shown).
ECG and VCG T-wave changes typical of CM were readily induced. Figure 2A demonstrates an atrially paced QRS and T wave on ECG before (Figure 2A) and after (Figure 2C) AVP (Figure 2B). The deeper T wave in Figure 2C tracks the inverted QRS during AVP, consistent with CM. Figure 2D overlays pre- (Figure 2A) and post- (Figure 2C) AVP T-wave vectors, demonstrating a clear increase in amplitude. Figure 2E summarizes T-vector amplitude changes at 60 and 120 minutes. CM is evident in the paced group (P<0.05), but not the sham or the paced dogs treated with nifedipine or saralasin.
CREB Expression and Activation in the Short-Term Protocol
Figure 3 demonstrates representative Western blots for nuclear CREB (Figure 3A) and pCREB (Figure 3B) from paced and sham control dogs. Figures 3C and 3D summarize all experiments. A decrease in total nuclear CREB was seen over the duration of the experiment in paced (P<0.05) but not sham dogs. No significant changes occurred in pCREB.
The effects of saralasin and nifedipine on CREB at the time of peak change (2 hours) are shown in Figure 4 and summarized in Figure 5. The pacing-induced decrease in CREB is attenuated by saralasin and by nifedipine, while no significant changes occur in pCREB in any setting.
Spatial Gradient for Change in CREB
That CREB dispersion does not occur in instrumented, nonpaced dogs is shown in Figure 6A. To investigate whether decreases in nuclear CREB are uniform throughout the ventricle or occur in a gradient with distance from the pacing site, we analyzed samples taken in a linear fashion, away from the LV electrode (Figure 1A) at 2 hours of pacing. CREB density was reduced significantly at the site closest to the pacing electrode (P<0.05) and returned toward the reference value with distance from the electrode (P>0.05 vs Ref) (Figure 6B).
Long-Term Cardiac Memory
Electrophysiology and CRE Binding
CM was demonstrable in the animals paced for 3 weeks but not the shams (T-wave vector displacements were of 0.61±0.16 and 0.11±0.04, respectively [P<0.05]). EMSAs demonstrated less nuclear protein binding to the CRE in extracts isolated from LTCM dogs than those from shams (Figure 7A). In analyzing the promoter region of Kv4.3, we found a sequence closely resembling the CRE sequence (TGACGTCT versus TGACGTCA) 1209 bp upstream to the transcription start site. To determine whether these CRE binding nuclear proteins may bind also to this Kv4.3 CRE-1 sequence, we performed EMSAs using this sequence as a probe. Nuclear proteins bound to CRE-1, and competition with this binding by the consensus CRE oligonucleotide, demonstrated it to be specific (Figure 7B).
CM has been studied extensively over the past two decades.5,8,11,15–19 Its induction is attenuated by AT1 receptor blockers, ACE inhibitors,9,20 and ICa,L blockade.10 Although little is known about the signaling pathways involved, downstream changes in ion channels that regulate cardiac electrical activity have been described. As one example, induction of CM by ventricular pacing is associated with reduction of the phase 1 action potential notch, Ito, and mRNA for Kv4.3.8,13 Moreover, Ito-blocking drugs suppress CM induction.6 Interestingly, in independent experiments, angiotensin II has been implicated in the modulation of Ito.20
Assuming there is linkage between angiotensin II, calcium, and CM, what we have lacked is an understanding of pathways connecting the upstream factors (eg, AT1 receptors, ICa,L) and the downstream changes in ion channels. In this study, we focused on a portion of a possible signaling pathway, by testing the hypothesis that an important transcription factor in neuronal memory might be modulated in a fashion consistent with cardiac memory. In selecting CREB for study, we did so understanding that four associations would have to be considered: a reduction in transcriptional activity, consistent with the reduction seen in putative downstream targets,13,15 blockade of any change in CREB by two interventions known to suppress CM, angiotensin II receptor blockade9 and ICa,L blockade,10 a gradient for change with distance from the pacing site, and an association between CREB and the downstream target.
We found that during the evolution of STCM on ECG and VCG (Figure 2), CREB levels decreased (Figures 3 and 4⇑). That there was no parallel change in pCREB has several possible explanations: (1) unphosphorylated CREB could be targeted specifically; (2) changes in pCREB might have evolved and dissipated over a time frame not incorporated in the study (note, our early data points were at 5 and 20 minutes); (3) an initial rise in pCREB might be masked by a general degradation of CREB, not allowing us to register an increase in pCREB but preventing its parallel downregulation of CREB. Regardless of the cause, CRE binding was clearly reduced when pacing was extended to 3 weeks (Figure 7A), consistent with diminished CREB activity in LTCM. If as postulated CREB activity is downregulated in LTCM, this may provide an explanation for the downregulation of genes containing CRE or CRE-like sequences in their promotor region as is seen during LTCM. Our demonstration of both CRE binding activity in the Kv4.3 promotor region (Figure 7B) combined with our previously shown downregulation of Kv4.3 mRNA and of Ito13 provides the first example of such a target and provides further support for our hypothesis.
The gradient of change in nuclear CREB (Figure 6B) is consistent with regulation by a paracrine system. We9 and others21 have shown that ventricular pacing alters stress-strain relationships in the myocardium. This makes angiotensin II a likely candidate as an initiator of the cascade since its release is altered by changes in stretch on myocytes.22,23 Furthermore, it is known that angiotensin II increases ICa,L.24 Taken together, it seems plausible to assume that the mechanism involved is activated by a pacing-induced local change in stress-strain relationship leading to a change in local angiotensin II release with associated effects on calcium homeostasis. Exactly how these changes in calcium regulation translate into a decreased amount of CREB will be a subject of future experiments.
Whereas ICa,L blockade inhibits the induction of STCM and LTCM, angiotensin II blockade prevents STCM only. A possible explanation for the loss of effect of receptor blockade might be compensatory upregulation of the AT1 receptor, allowing induction of LTCM. We have recently shown in rat pups that the administration of losartan delays the normal developmental decrease of the AT1 receptor, indicating that blockade of the AT1 receptor can lead to its upregulation.25
Very importantly, our study focuses attention on the role of pacing, per se, as a determinant of transcription in the heart. Although generally considered as an uneventful—and usually beneficial—intervention, pacing has long been known to have effects on both electrophysiological remodeling, seen as CM8,16 and structural remodeling.26 The latter is most frequently expressed as a thinning of myocardium in the region of the pacing electrode.27
Our demonstration of the effect of pacing on an important transcriptional factor opens a range of possibilities for further exploration. We state this not only because Kv4.3 has the potential to be regulated by CREB but because a variety of cardiac genes have a CRE element in their promoter region, including phospholambam,28,29 the β1 and β2-adrenoreceptors,30,31 the potassium channel pore-forming unit Kv1.5,32 and protein phosphatase 2Aα.33 The effects of modulation by ventricular pacing on the expression of these and other CRE element-regulated genes have not been reported, yet may be important in determining outcome in both healthy and diseased hearts. For example, pacing might have therapeutic effects beyond rhythm control alone via induction of transcriptional activity leading to synthesis of proteins beneficial to the diseased heart. Alternatively, it may contribute to an increased incidence of death (eg, Wilkoff et al34). In either event, location of the pacing lead would be a major consideration as the effect of pacing is greatest near the electrode and decays with distance.
This work was supported by NIH Grants HL-67101 and DA12962 and a grant from the Dutch Heart Foundation (NHS 2000.021). The authors acknowledge with gratitude the guidance and advice of Dr Aviva J. Symes, the expert technical assistance of Linda Samaniego, and the careful attention to the preparation of the manuscript by Eileen Franey.
Original received March 26, 2003; revision received July 1, 2003; accepted July 17, 2003.
Mills VA, Orbach-Arbouys S, Gershon RK. Relation of DNA synthesis and suppression to development of memory by T cells. J Immunol. 1975; 114: 200–205.
Kandel ER. The molecular biology of memory storage: a dialogue between genes and synapses. Science. 2001; 294: 1030–1038.
Rosen MR, Cohen IS, Danilo P Jr, Steinberg SF. The heart remembers. Cardiovasc Res. 1998; 40: 469–482.
del Balzo U, Rosen MR. T wave changes persisting after ventricular pacing in canine heart are altered by 4-aminopyridine but not by lidocaine: implications with respect to phenomenon of cardiac “memory.” Circulation. 1992; 85: 1464–1472.
Plotnikov AN, Shvilkin A, Xiong W, de Groot JR, Rosenshtraukh L, Feinmark S, Gainullin R, Danilo P, Rosen MR. Interactions between antiarrhythmic drugs and cardiac memory. Cardiovasc Res. 2001; 50: 335–344.
Shvilkin A, Danilo P Jr, Wang J, Burkhoff D, Anyukhovsky EP, Sosunov EA, Hara M, Rosen MR. Evolution and resolution of long-term cardiac memory. Circulation. 1998; 97: 1810–1817.
Plotnikov AN, Yu H, Geller JC, Gainullin RZ, Chandra P, Patberg KW, Friezema S, Danilo P Jr, Cohen IS, Feinmark SJ, Rosen MR. Role of L-type calcium channels in pacing-induced short-term and long-term cardiac memory in canine heart. Circulation. 2003; 107: 2844–2849.
Herweg B, Chang F, Chandra P, Danilo P Jr, Rosen MR. Cardiac memory in canine atrium: identification and implications. Circulation. 2001; 103: 455–461.
Yu H, McKinnon D, Dixon JE, Gao J, Wymore R, Cohen IS, Danilo P Jr, Shvilkin A, Anyukhovsky EP, Sosunov EA, Hara M, Rosen MR. Transient outward current, Ito1, is altered in cardiac memory. Circulation. 1999; 99: 1898–1905.
Costard-Jackle A, Goetsch B, Antz M, Franz MR. Slow and long-lasting modulation of myocardial repolarization produced by ectopic activation in isolated rabbit hearts. Evidence for cardiac “memory.” Circulation. 1989; 80: 1412–1420.
Yu H, Gao J, Wang H, Wymore R, Steinberg S, McKinnon D, Rosen MR, Cohen IS. Effects of the renin-angiotensin system on the current Ito in epicardial and endocardial ventricular myocytes from the canine heart. Circ Res. 2000; 86: 1062–1068.
Sadoshima J, Izumo S. Molecular characterization of angiotensin II-induced hypertrophy of cardiac myocytes and hyperplasia of cardiac fibroblasts: critical role of the AT1 receptor subtype. Circ Res. 1993; 73: 413–423.
Chandra P, Sosunov EA, Anyukhovsky EP, Patberg KW, Sun LS, Rosen MR. AT-1 receptor blockade accelerates developmental changes in ventricular repolarization. PACE. 2003; 26: 1067.Abstract.
van Oosterhout MF, Arts T, Muijtjens AM, Reneman RS, Prinzen FW. Remodeling by ventricular pacing in hypertrophying dog hearts. Cardiovasc Res. 2001; 49: 771–778.
van Oosterhout MFM, Prinzen FW, Arts T, Schreuder JJ, Vanagt WYR, Cleutjens JPM, Reneman RS. Asynchronous electrical activation induces asymmetrical hypertrophy of the left ventricular wall. Circulation. 1998; 98: 588–595.
Toyofuku T, Zak R. Characterization of cDNA and genomic sequences encoding a chicken phospholamban. J Biol Chem. 1991; 266: 5375–5383.
Collins S, Altschmied J, Herbsman O, Caron MG, Mellon PL, Lefkowitz RJ. A cAMP response element in the β2-adrenergic receptor gene confers transcriptional autoregulation by cAMP. J Biol Chem. 1990; 265: 19330–19335.
Mori Y, Matsubara H, Folco E, Siegel A, Koren G. The transcription of a mammalian voltage-gated potassium channel is regulated by cAMP in a cell-specific manner. J Biol Chem. 1993; 268: 26482–26493.