Djungarian or Siberian hamsters (Phodopus sungorus) acclimated to short photoperiod display episodes of spontaneous daily torpor with metabolic rate depressed by ∼70% and body temperature (Tb) reduced by ∼20°C. To study the cardiovascular adjustment to daily torpor in Phodopus, electrocardiogram (ECG) and Tb were continuously recorded by telemetry during entrance into torpor, in deep torpor, and during arousal from torpor. Minimum Tb during torpor bouts was ∼21°C, and heart rate, ∼349 beats/min at euthermy, displayed marked sinus bradyarrhythmia at ∼70 beats/min. Arousal was typically completed within ∼40 min, followed by a sustained posttorpor inactivity tachycardia (∼540 beats/min). The absence of episodes of conduction block, tachyarrhythmia, or other forms of ectopy throughout the torpor cycle demonstrates a remarkable resistance to arrhythmogenesis. The ECG morphology lacks a distinct isoelectric interval following the QRS complex, and the ST segment resembles the ECG pattern in mice, with a prominent fast transient outward K+ current (Ito,f) determining the early phase of ventricular repolarization. During low-temperature torpor, the amplitudes of the QRS complex substantially increased, suggesting that in the euthermic state the terminal portion of ventricular depolarization is fused with the beginning of repolarization, low Tb acting to decorrelate the superposition between depolarization and repolarization by delaying the repolarization onset. Atrioventricular and ventricular conduction times were prolonged as function of Tb. In contrast, the QT vs. Tb relationship showed marked hysteresis indicating the operation of nonlinear control mechanisms whereby the rapid QT shortening during arousal results from additional mechanisms (probably sympathetic stimulation) other than temperature alone.
to escape from unfavorable climate conditions and/or limited food supply, many small mammals minimize their energy expenditure by entering a state of torpor during certain times of the year or the day, during which periods metabolic rate, body temperature (Tb), and other physiological variables are substantially reduced. Two distinctly different patterns of torpidity have been adopted by most heterothermic mammals: deep hibernation or prolonged torpor in true hibernators and shallow daily torpor in daily heterotherms. Hibernation is characterized by extended torpor bouts over several days or weeks that are interrupted by spontaneous arousals associated with short periods of rewarming to euthermic conditions. In contrast, daily torpor is integrated into the normal circadian cycle of inactivity and is characterized by short-term bouts lasting only a few hours, such that animals return to an active state between episodes of torpor. Daily torpor occurs at distinct times of the day and often occurs during short photoperiods, i.e., in winter. Generally, the terms “daily torpor” and “hibernation” are merely descriptive for the duration of the dormant state and may not be indicative of a difference in fundamental physiological mechanisms between the two categories. Several reviews provide comprehensive coverage of the physiological mechanisms of hibernation and torpor (5, 7, 14, 19, 21, 25, 26, 29, 36, 41).
Remarkably, the Djungarian or Siberian hamster (Phodopus sungorus), which inhabits regions of southwestern Siberia and northeastern Kazakhstan, is not a deep hibernator, although it opportunistically may undergo periods of daily torpor for 5–10 h with Tb as low as 15°C (37). In this species, the expression of daily torpor is part of seasonal acclimation triggered by short photoperiods and involves an array of morphological and physiological adaptations, including weight loss, appearance of winter pelage, reproductive quiescence, and increased capacity for nonshivering thermogenesis. The strictly photic entrained Djungarian hamster displays spontaneous daily torpor during its circadian daytime resting phase, with metabolic rate lowered to ∼30% of the normothermic active state (15, 20, 37) and heart rate reduced by a similar relative scope.
The heart of homeothermic1 mammals tends to develop atrial fibrillation at Tb of ∼30°C, and asystole or ventricular fibrillation ensues with further drop in Tb. In contrast, the heart of hibernators or daily heterotherms continues to beat at substantially lowered Tb and would not experience conduction block or tachyarrhythmia either during entrance into torpor, which is a relatively slow, gradual process (1–3 h), or during arousal, which is relatively rapid (∼40 min). The resistance to cardiac arrhythmias during low-temperature torpor unequivocally presents an adaptive response to short-day exposure because susceptibility to the induction of arrhythmias is maintained in long-day-adapted animals (10). During the torpor cycle, as exemplified by the Djungarian hamster, the animal's Tb drops by ∼20°C, but unlike nonhibernators cardiac contractility is preserved during low Tb and the myocardium of torpid animals is also extremely resistant to the induction of ventricular tachyarrhythmias.
The key elements preparing the myocardium for the rigor of low-temperature torpor have not been identified. Specifically, the properties of currents underlying the morphology and duration of the cardiac action potential (AP), gap junction channel function, and reprogramming of the cardiac sarcoplasmic reticulum over the short time frame of torpor exhibited by the Djungarian hamster have not been elucidated. Even given today's advanced molecular and electrophysiological techniques, research along these lines is largely restricted to in vitro studies. Alternatively, recent advances in miniature radio transmitter technology have facilitated the continuous monitoring of physiological variables such as electrocardiogram (ECG) and/or Tb in the unrestrained animal maintained in the laboratory under seasonal acclimation to short photoperiod.
In the present study, the temporal cardiovascular adjustment during daily torpor was studied in Djungarian hamsters adapted to short photoperiod with continuous high-resolution recordings of ECG and Tb obtained by telemetry. The time course of alterations of ECG morphology, heart rate dynamics, and Tb was measured during the various stages of the torpor bout: entrance into torpor, deep torpor, and arousal. To evaluate cardiac pulse generation and propagation, the impact of temperature on atrial and ventricular conduction properties was inferred from an analysis of ECG subintervals. The objective of the study was to identify the pattern of the cardiac rhythm along with alterations of ECG morphology and ECG indicators of ventricular conduction and repolarization that ultimately help protect heterothermic animals from cardiac malfunction during sojourn to episodes of low-temperature torpor.
MATERIALS AND METHODS
Animals and Surgical Procedures
Djungarian hamsters (Phodopus sungorus, n = 6) from the breeding stock of the University of Veterinary Medicine, Hannover, Germany were raised in summer under natural photoperiodic conditions. From October to early spring they were individually housed indoors and maintained at ∼17°C ambient temperature (Ta) and natural lighting with food and water ad libitum. In response to shortening of the natural photoperiod in late autumn, the summer agouti pelage had assumed its predominantly white winter color, and animals entered heterothermic torpor episodes for ∼5 h, starting 30–60 min before dawn. The animals retired in a small, well-isolated box covered with cotton batting, assumed a sleeplike curled posture with the eyes closed, and remained behaviorally quiescent until ≥2 h after onset of arousal. At the age of ∼5 mo, the animals (mean ± SD body wt 35 ± 4 g) were implanted intraperitoneally with calibrated temperature (model X, Mini Mitter, Bend, OR) and ECG (TA10EA-F20, Data Sciences, St. Paul, MN) transmitters as described previously (39). Transmitter implantation was performed during inhalation anesthesia (1.5% isoflurane in O2), and the lead configuration was oriented approximately along the axis of lead II, providing for a maximally upright QRS complex. The ECG transmitter setup and electronics were checked ex vivo for proper operation over the Tb range encountered by the experimental animals in vivo. All procedures were in accordance with European Council Directive 86/609/EEC and were performed with permission of the District Government of Lower Saxony, Germany, enforcing the Animal Protection Law, which is in full accordance with American Physiological Society ethical guidelines.
ECG and Tb were continuously recorded during short day lengths of mid-February, and ∼12-h epochs were collected starting at 7 AM. The digitized ECG (sampling rate 2 kHz) was automatically analyzed and annotated to obtain discrete time points corresponding to the successive R-wave maxima with an adaptive QRS template pattern-matching algorithm (Fig. 1A). Briefly, the procedure involves first selecting a representative QRS complex as a template (Fig. 1B). The algorithm then scans the ECG record and calculates the cross-correlation between the template and the time series. In a second pass, the algorithm identifies the points with maximum correlation and the R wave of the ECG is detected with a derivative-based QRS detection algorithm. The ECG morphology was determined for 30-min subepochs by ensemble averaging using the peak of the QRS complex as a fiducial point (Fig. 1C). The ECG intervals (PQ, QRS, QT) of the signal-averaged complex were determined by placing cursors at the appropriate positions.
Physiological signals, such as the heartbeat interval (RRi) time series formed by the time increments between consecutive heartbeats (Fig. 1D), are typically generated by complex, integrated, self-regulatory multicomponent systems ultimately determining the “noisy” and “erratic” appearance of the cardiac rhythm (27, 28). The highly irregular RRi time series may be visualized as a merely regular original signal corrupted by noise. To recover the “baseline” trend (RRB) or unknown function of the signal that is embedded in the “noise”, the technique of empirical mode decomposition (EMD) was employed. EMD was recently devised for unconstrained decomposition of natural source data by adaptively decomposing nonstationary and/or nonlinear multiscale time series into sets of components referred to as intrinsic mode functions (IMFs) (22). Essentially, the signal is considered at the scale of its local oscillations whereby a local low-frequency component is visualized as a local trend supporting a local zero-mean high-frequency component or local detail. EMD is adaptive (signal dependent), since the frequency subbands in which the IMFs live are built up as needed to separate the different oscillating components without using any predetermined filter functions. Although the procedure is basically simple and entirely data driven, the technique lacks any analytical definition and, in principle, represents the output of an iterative algorithm referred to as sifting. A detailed account of EMD and its mathematics is presented elsewhere (28).
An example of extraction of all constitutive IMFs in an RRi time series by EMD is shown in Fig. 2A. If all IMFs are added, the resulting summation signal is a near-perfect match of the original signal. The property of EMD to fully decompose a signal into a family of IMFs suggests the use of partial reconstructions to selectively eliminate fast fluctuations in the signal (denoising). The baseline extraction, based on selection of the relevant IMFs according to statistical properties, is shown in Fig. 2C. Notably, the procedure is not based on commonly used linear-stochastic statistics, e.g., averages, and we also emphasize that the elements of the bursty signal that are removed are not noise in the traditional sense but display an intrinsic hidden nonstochastic fine structure (Fig. 2B). The original RRi time series along with its denoised RRB trend is displayed in Fig. 2D.
The time course of RRi, RRB, and Tb during entrance into and arousal from daily torpor is shown in Fig. 3. Individual hamsters displayed a rather steady pattern of daily torpor, but at Ta = 17°C would not take the term “daily” literally, showing episodes of spontaneous daily torpor about every other day. There was some variability in the time of torpor onset, and variability of bout duration ranged from ∼4 to ∼8 h. Arousal typically started at 4 PM, leaving some light for entrainment of the circadian clock located centrally within the suprachiasmatic nucleus of the hypothalamus. In diurnal euthermia, Tb was 35.7 ± 0.5°C, decreasing at a rate of 3.2 ± 0.8°C/h and reaching its minimum of 21.0 ± 2.0°C within 7.8 ± 1.0 h after onset of torpor. During arousal, Tb increased at a maximum rate of 22.5 ± 4.3°C/h and euthermic Tb was usually reached within ∼40 min. Typically arousal from torpor, once under way, was uninterrupted, but recession of arousal and prolonged torpor would occur occasionally (Fig. 3, animal 6). Throughout, the sluggish decline of Tb during entrance into torpor is interrupted by sudden changes of Tb or may display an early stepwise pattern, indicating that Tb is under control around a progressively lowered set point during entrance into torpor. A precise definition of the various stages of the torpor cycle is difficult. Notably, a steady maintenance phase separating entry to and arousal from torpor is generally missing. Hamsters did not become active immediately after arousal but remained in a sleeplike “posttorpor inactivity state” characterized by a lack of locomotion or feeding activity.
RRB at euthermia was ∼175 ms (≈343 beats/min), increasing (decreasing) to maximum (minimum) levels of ∼860 ms (≈70 beats/min) during torpor. During emergence from torpor, animals displayed a substantial tachycardia in the early “posttorpor inactivity state” (RRB ∼105 ms, ≈570 beats/min) that, likely reflecting a state of hypermetabolism, did not revert to pretorpor levels for up to 2 h. The rapid increase of heart rate on arousal suggests that the “chronotropic competence” of the heart is well preserved. The progression of depth of torpor along with substantially lowered Tb was associated with a marked bradyarrhythmia (increasing amplitude of the RRi fluctuations). The appearance of enhanced irregularity of the dynamics appears to signal the onset of the dormant state, although transition is sluggish. The torpor-induced arrhythmicity has previously been described as “skipped beats” and changeover to periods of “asystole” (26), but these terms are merely descriptive and would not comply with arrhythmic classifications generally used in clinical cardiology. Here we emphasize that the electrocardiographic finding of bradyarrhythmia as reflected in the large range of instantaneous RRi (cf. Fig. 1D) was strictly that of sinus bradyarrhythmia, because every P wave was followed by a QRS complex.
There was no evidence of sinoatrial block, junctional rhythm, first- or second-degree atrioventricular (AV) nodal block, or other forms of ectopy. In particular, no episodes of disorganized ventricular tachyarrhythmia, in homeothermic mammals presenting a precursor of lethal ventricular fibrillation, were observed at any Tb. These findings highlight the remarkable resistance of excitation/conduction patterns to evocation of arrhythmogenesis in the excitable medium of the myocardium at Tb changing by ∼20°C within ≤1 h. In contrast, the RRi amplitude rapidly declines as the animal arouses, and variability of RRi is at its minimum in the early phase of the “posttorpor inactivity state.” The time course of RRB, unlike the contour of Tb, is far from being smooth contiguously during the torpor bout but is modulated continually by interspersion of short-term episodes of cardioacceleration as the animal sinks into lethargy. These findings suggest that the highly irregular dynamics of heart rate in daily torpor are not strictly random but remain under deterministic control during periods of depressed metabolism.
The relationship between RRB and Tb throughout the torpor cycles is shown in Fig. 4 for all individuals (Fig. 4A) along with their group means (Fig. 4B). Group means were obtained by spline interpolation for adjustment of the RRB/Tb database of individual animals to equal data length and bidirectional averaging. Until the onset of arousal, RRB decreased with Tb in an almost linear fashion (ΔRRB ∼55 ms/°C), indicating that RRB was a close function of downregulation of metabolic rate and the ensuing fall of Tb consequent upon a lack of heat production. By contrast, Tb lagged RRB during arousal, yielding the well-established hysteresis between RRB and Tb that is characteristic for all mammals resorting to hibernation or daily torpor as a metabolic saving. Here we emphasize that the hysteresis of the RRB/Tb relationship presents a reflection of a nonlinear phenomenon in the mathematical sense, which strictly implies that the mechanisms underlying entry into and arousal from torpor are substantially different from each other. We also note that the final part of the RRB/Tb loop does not quite match its origin, indicating that the process of arousal was not completed within the window of observation. Hence the duration of the “posttorpor inactivity state” substantially exceeds that anticipated from recovery of Tb to euthermia.
The characteristic ECG morphology in Djungarian hamsters during active-state euthermia is displayed in Fig. 1, A and D. A remarkable feature of the ECG is the absence of a distinct isoelectric interval between the QRS complex and the T wave and a relatively short QT interval. Rather, a sloping ST elevation with an early-peaking dome or hump configuration of the T wave is observed. While an isoelectric ST segment is a common feature of ECG recordings in higher mammals, the ECG of heterothermic hamsters shares some features with the murine ECG morphology, which is attributable to the presence of large K+ currents that dominate the early phase of myocardial repolarization (see discussion). The ECG morphology was unchanged when recordings from early October, i.e., the date of transmitter implantation, were compared with that of mid-February. Strikingly, there was no evidence of photoperiodic adaptation of ECG morphology during progressive shortening of day length over ∼5 mo.
The characteristic changes of ECG morphology across a full-length torpor bout are shown in Fig. 5. The ECGs were ensemble-averaged over 30 min, aligned to the onset of the Q wave, color-coded according to Tb, and displayed by 0.1-mV vertical offset (Fig. 5A). Because some features of the QRS complex remain obscured by superposition of R waves, ECGs were also displayed by 20-ms horizontal offset (Fig. 5B). Progressive diminution of Tb during torpor causes biphasic alteration of P-wave contour, prolongation of P-wave duration, increase of PQ interval, broadening of the QRS complex with deepening of the Q and S waves, enhancement of the dome morphology of the ST segment, and prolongation of the QT interval. Interestingly, all amplitudes of the QRS complex increased substantially with decreasing Tb and the amplitudes of R and S waves were about doubled at minimum Tb. As Tb increased again during arousal, progressive diminution in amplitude of all waves occurred, along with shortening of PQ, QRS, and QT intervals.
A late delta wave or notch following the QRS complex or a small secondary R wave referred to as the Osborn wave, J wave, or elevated J point denotes the approximate end of myocardial depolarization and the beginning of repolarization. It appears that the J point in the surface ECG of the Djungarian hamster is generally masked by the QRS complex, but low Tb during torpor may serve to unmask a latent J wave by moving it out of the QRS complex. The appearance of a transient J wave that disappeared on rewarming was consistently observed in one of six animals.
Cardiac Pulse Propagation
Atrial conduction properties can be inferred from the surface ECG by P-wave contour and duration. In response to daily low-temperature torpor, P-wave patterns assumed a biphasic shape and P-wave duration was increased (Fig. 5A). Prolongation of P-wave duration may result from 1) decrease of intra-atrial conduction velocity in terms of a delayed depolarization without changing the vector of pulse propagation or 2) prolongation of conduction time caused by conduction block associated with consecutive prolongation of the length of the activation pathway.
According to its definition, the PQ interval reflects conduction through the AV node and His bundle to the proximal Purkinje system. Prolongation of the PQ interval was inversely related to Tb in a curvilinear fashion, and hysteresis was virtually absent in the course of entry to and emergence from torpor (Fig. 6A). Because the prolongation of the PQ interval exceeds that of P-wave duration, the PQ prolongation is only accounted for in part by intra-atrial conduction delay and is largely due to decreased AV nodal conduction velocity.
Conduction in His-Purkinje system or ventricles.
In response to torpor-associated low Tb, QR, QRS, and QT intervals were prolonged, indicating prolongation of conduction times in the bundle branches of the His-Purkinje system. The width of the QRS interval is generally used as a measure of ventricular conduction velocity. Similar to the PQ/Tb function, the QRS/Tb relationship was curvilinear and QRS duration was inversely related to Tb, showing absence of hysteresis in the course of the torpor cycle (Fig. 6B). Thus prolongation of PQ and QRS intervals was an exclusive function of Tb (Fig. 6C). In contrast, the QT interval, reflecting AP duration, was prolonged at decreasing Tb but showed marked hysteresis on arousal (Fig. 6D). Hence, prolongation of AP duration is rapidly reduced during early arousal and almost independent of Tb. Hysteresis, i.e., pronounced complex nonlinearity, is also evident in the RRi/QT relationship (Fig. 6E). The fractional QT interval (QT/RRi) in the Djungarian hamster is relatively short, only ∼18% of the cycle length during low-temperature torpor, although it is ∼60% at euthermia and increases to ∼85% during early posttorpor inactivity state tachycardia (Fig. 6F). These findings suggest that animals expressing daily torpor are capable of preventing excessive increase in refractoriness at low temperature and low heart rate.
The present study is the first demonstrating the time course of high-resolution ECGs in a species undergoing episodes of daily torpor under unrestrained conditions. In the Djungarian hamster, which is strictly photoperiodic, spontaneous daily torpor occurs opportunistically because of short photoperiod exposure at predictable times of the day and does not require other stimulations such as cold and/or food restriction. Under constant conditions of Ta, this species exhibits a robust rhythm typical of a nocturnal species and Tb drops by ∼20°C during torpor episodes. However, unlike in nonhibernating or nontorpid species, the myocardium retains the ability to contract over an approximately eightfold dynamic range of heart rate and is resistant to the development of ventricular dysfunction and life-threatening arrhythmias.
Specifically, the important results and impact in Phodopus may be listed as follows. 1) In the heterothermic state of torpor (Tb ∼21°C, RRi ∼70 beats/min) heart rate is lowered to ∼20% of its euthermic level (RRi ∼350 beats/min) and displays marked sinus bradyarrhythmia. 2) Cardiac dynamics throughout the course of entry to and arousal from torpor are characterized by an absence of conduction block, tachyarrhythmia, or other forms of ectopy. 3) The ECG morphology lacks a distinct isoelectric ST segment suggestive of a narrow triangle-shaped cardiac AP and a prominent fast transient outward K+ current (Ito,f) leading to early rapid repolarization of the excitable myocardium. 4) The inverse relationship of QRS amplitudes and Tb suggests spatiotemporal overlap of ventricular depolarization and repolarization. 5) The QT vs. Tb relationship is characterized by marked hysteresis indicating asymmetric nonlinear system control of cardiac dynamics during entry to and emergence from low-temperature torpor.
ECG Morphology: Mice vs. Djungarian Hamsters
The ECG of Djungarian hamsters at euthermic Tb has peculiar characteristics showing differences from that of larger mammals, particularly the absence of a well-defined T wave separated from the QRS complex by an isoelectric ST segment, accompanied by a relatively short QT interval (cf. Fig. 1, A and C). However, an important feature of the ECG morphology of Djungarian hamsters is that it shows some common features with that of mice and rats but not guinea pigs. The unusual ST characteristics of the murine ECG were recognized as early as 1929 by Agduhr and Stenstrom (1), but the molecular physiology underlying the monophasic AP in mice was identified quite recently (6, 30, 31).
The contour of the murine transmembrane AP is characterized by a rapid depolarization spike (phase 0) and an initial rapid phase of repolarization (phase 1) that is merged with terminal repolarization (phase 3) with little or no plateau (phase 2). When the plateau phase of the epicardial AP is depressed, abbreviated, or even absent, the ECG pattern is characterized by a downsloping ST segment elevation and/or presence of a prominent J wave or elevated J deflection with a triangular dome or hump configuration in the region following the QRS complex. A prominent J wave in the ECG designates the approximate end of ventricular depolarization and the beginning of ventricular repolarization that ends with T-wave offset, while the triangular pattern of the ST segment is presumed to result from dispersion of ventricular repolarization (16, 17, 43). During euthermia, a latent J wave appears to be masked within the QRS complex, suggesting that ventricular depolarization and repolarization overlap, i.e., repolarization is starting in some parts of the heart before depolarization is complete in other parts. Indeed, the conduction delay associated with low Tb during torpor may serve to unmask a J wave that appears as a late delta wave following the QRS complex or as a small secondary R′ wave generally referred to as the Osborn wave since Osborn's observation in the early 1950s (32). However, the appearance of a transient Osborn wave during low Tb was observed only in one of six animals.
ECG Wave Amplitudes
A consistent finding from continuous ECG monitoring in Djungarian hamsters undergoing daily torpor was the progressive increase of the amplitudes of all ECG waves during low-Tb episodes, which diminished on Tb returning to euthermic levels. Despite intensive cardiovascular research in hibernators over the past 60 years, this phenomenon escaped the attention of earlier studies. Credit for its first recognition goes to the work of Richards et al. (35), who showed in mice that the amplitude of all waves increased during exposure to cold. These authors also proposed that depolarization and repolarization overlap to a greater extent in mice than in larger animals. Along these arguments for the murine heart, we argue from analogy that, in heterothermic hamsters, the terminal portion of ventricular depolarization is fused by partial superpositioning with the beginning of repolarization at euthermia, low Tb acting to decorrelate and unmask the separation between depolarization and repolarization by delaying the onset of repolarization.
Action Potential Waveforms and Ionic Currents
A number of voltage-gated transient outward (Ito) and delayed rectifier (IK) K+ current channels as well as inward rectifier (IKI) and ligand-gated (IKATP and IKACh) K+ current channels are expressed in most cardiac cells (30). The specific murine ECG morphology that, by the argument of pattern analogy, is extended here to Djungarian hamsters is attributable to the dominance of the rapidly activating and inactivating 4-aminopyridine-sensitive transient outward K+ current, Ito,f, that determines the early phase of ventricular repolarization (phase 1) while the delayed, slowly inactivating 4-aminopyridine-insensitive K+ current, IKs, is responsible for the later phase of repolarization (phases 2 and 3) back to the resting membrane potential (4, 17). Much like in mice and rats, the high density of Ito,f appears to underlie the shortening of the plateau phase of the AP and the expression of a prominent J wave and the hump or dome configuration of the R-ST junction observed in Djungarian hamsters.
The myocardial AP has recently been recorded in euthermic Djungarian hamsters acclimated to long and short photoperiods (8). Remarkably, no photoperiod-dependent differences in AP duration or shape were detected between these groups, which is in line with the present findings of absence of changes of ECG morphology during seasonal shortening of day length over ∼5 mo. However, the AP pattern in acutely torpid hamsters was not determined. Because the AP contour is strongly determined by the repolarizing properties of the Ito,f, the impact of low Tb on adaptation of the molecular determinants of Ito,f within the short time frame of daily torpor exhibited by Djungarian hamsters remains to be elucidated. Notably, the triangular shape of the AP in Djungarian hamsters is characterized by the absence of a distinct plateau (phase 2) and resembles that of the murine ventricular transmembrane AP, showing an early rapid repolarization phase (1) followed by a slow terminal phase (3). Hence, the voltage time course of the AP suggests the presence of a prominent Ito,f in this species, which yields the pattern of the early ST segment observed in the ECGs of animals in the present study. The absence of a pronounced plateau phase has also been observed in the hedgehog (Erinaceus europaeus), which is a true hibernator (10).
Although mice generally do not undergo episodes of deep torpor, they use torpor as a major adaptive mechanism when deprived of food (33). Remarkably, transgenic leptin-deficient A-ZIP/F-1 mice that have virtually no white adipose tissue display fasting-induced deep torpor with Tb as low as ∼24°C (13), the time course approximating that of the Djungarian hamsters in the present study. It is tempting to hypothesize that all mammalian species that have APs without a pronounced plateau as evidenced by its specific ECG pattern may experience the ability to resort to torpor as a metabolic saving and safely resist the impact of low Tb during periods of reversible metabolic depression. A comparative ECG analysis across hibernators and daily heterotherms, however, is lacking. Specifically, a prominent role of Ito,f in defining the unique features of ventricular repolarization and its contribution to electrical stability at low temperature has not been delineated.
The APs of epicardial and midmyocardial cells generally display a prominent Ito-mediated phase 1 that is absent in endocardial cells (2). Spatiotemporal heterogeneity of the time course of repolarization of phases 2 and 3 in the three cardiac cell types provides for the formation of opposing voltage gradients on either side of the midmyocardial region, which for a large part generate the inscription of the T wave in the ECG. As a result, full repolarization of the epicardial AP coincides with the peak of the T wave (Tpeak), full repolarization of the midmyocardial AP being coincident with the end of the T wave (Tend). Hence, the Tend–Tpeak interval may serve as an index of transmural dispersion of repolarization (3). Inspection of Fig. 5A reveals that the position of the Tpeak wave remains fairly stable and the QTpeak interval displays little, if any, delay in the course of transition through a heterothermic torpor bout. Thus the duration of the epicardial AP appears to exhibit a remarkable stability despite the marked Tb-related depression of heart rate during torpor. In contrast, Tend is progressively prolonged as Tb is decreased. Hence, the prolongation of the Tend–Tpeak interval difference suggests that the spatiotemporal dispersion of transmural voltage gradients during repolarization in hearts of Djungarian hamsters is markedly increased with lowering Tb. Notably, the amplitude of the T waves appears to undergo little change during the torpor cycle (Fig. 5B).
We also note that the QTend interval and hence the overall length of ventricular repolarization is not a simple function of Tb alone. The hysteretic loops in the Tb/QT and RRi/QT relationships (cf. Fig. 6, D and E) indicate that prolongation of the QT interval during entrance into torpor and rapid QT shortening during arousal are a reflection of a nonlinear phenomenon. Hence, the time course of rapid QT interval shortening during the rewarming process designates a very different physiological state that presumably involves neuroautonomic cues other than temperature alone, e.g., autonomic sympathetic stimulation. Moreover, the fractional QT interval length (QT/RRi), ∼60% of the cycle length in the animal's euthermic state, is within the range reported for eutherian mammals and is reduced to ∼18% during low-temperature torpor episodes (cf. Fig. 6F). Hence, Djungarian hamsters cannot produce long QT intervals, which affords the prevention of an excessive increase in refractoriness and helps resist arrhythmogenesis. In the hedgehog (E. europaeus), which is a typical seasonal hibernator, the QT interval in winter is only 7% of the cycle length at Tb = 16°C, although it is 47% during summer when the animal is active (10).
Transmural Conduction at Low Temperature
In the mammalian heart, conduction of the cardiac AP depends on AP upstroke velocity and synchronization of ventricular repolarization by cell-to-cell communications via gap junctions (24). Gap junctions are composed of transmembrane proteins belonging to the connexin (Cx) family, each having a unique expression pattern in the working myocardium and conduction system of mammalian cardiomyocytes (11, 18). Because Cx43 is the principal ventricular gap junction protein, it has been hypothesized that transmural AP gradients are in part produced by heterogeneous Cx43 expression in epicardial, midmyocardial, and endocardial layers (34).
Immunoreactivity for Cx43 is elevated in cardiac myocytes of cold-acclimatized hibernating hamsters (Mesocricetus auratus) and reverses to euthermic control levels within 2 h after arousal from torpor (38). However, because Cx43 immunostaining was also elevated in cold-acclimated control hamsters that did not hibernate, this change may not be strictly associated with the hibernating phenotype that is protected against conduction block at low Tb and ventricular fibrillation during rewarming. In a similar study in hibernating Siberian ground squirrels (Citellus undulatus), a marked 45% upregulation of Cx43 in working myocardium and appearance of Cx45 in working ventricular myocardium was observed, the latter normally being restricted to the conduction system (12). At low Tb (3°C), conduction velocity was markedly decreased but conduction patterns remained normal and arrhythmias due to conduction block did not develop. However, even at normal Tb the relationship between gap junction density and conductance velocity is not firmly established and is clearly nonlinear (23). The functional significance of a coordinated compensatory response by gap junction overexpression remains unclear in the absence of functional assessment of gap junction behavior at low Tb. Comparative screening of protein sequences in 12 different mammals including true hibernators, the species of the present study, which undergoes daily torpor (P. sungorus), and several nonhibernating species revealed no evidence for specific Cx43 amino acid differences in hibernators vs. nonhibernators (42).
AV conduction as inferred from prolongation of the PQ interval (∼65 to ∼190 ms, ΔPQ ≈ 192%) was reduced in the hamsters in the present study (cf. Fig. 6A) and appears to be exclusively due to temperature, although the PQ/Tb relationship is curvilinear. The approximate Q10 of 2.3 is close to that expected for temperature-dependent biochemical reactions. Similarly, ventricular conduction determined from broadening of the QRS interval (∼32 to ∼52 ms, ΔQRS ≈63%) was dependent on temperature (cf. Fig. 6B), but the relative decrease of ventricular conduction capacity was substantially smaller than decreased AV conduction capacity because the PQ/QRS relationship displayed a slope of ≈6.7 (cf. Fig. 6C). Hence, it appears that decreased conduction velocity per se and spatiotemporal heterogeneity of pulse propagation were both responsible for low-temperature-induced lengthening of atrial conduction time.
The mechanisms underlying the formation of pronounced sinus bradyarrhythmia during torpor are essentially unclear. It is unlikely that lability of myocardial conduction properties and refractoriness was accounting for the marked sinus bradyarrhythmia during low-temperature torpor. Thermally induced instability of spontaneous sinus node autorhythmicity, potentially elicited by floating resting membrane potentials of pacemaker cell assemblies, and attenuation of neuroautonomic sympathovagal output to the heart are presumably involved in the generation of randomlike beat-by-beat dynamics. Further analysis reveals that heartbeat interval fluctuations during torpor were close to but not strictly random dynamics in the mathematical sense (M. Meyer, unpublished observations).
Adaptations to Low-Temperature Torpor
The molecular and cellular bases of the adaptive mechanisms employed by animals exhibiting low-metabolism and low-temperature torpor episodes are not well known and controversial (40). Strictly, distinction must be made between large (time)-scale, i.e., circannual, adaptations and those acting on short scales, i.e., adaptations acting over the time frame of a torpor bout that may last from hours in daily heterotherms to weeks in true hibernators. Large-scale winter adaptations include changes in pelage, reproductive quiescence, and increased capacity for nonshivering thermogenesis and are primarily mediated by reduced photoperiod rather than temperature per se. Clearly, preserved cardiac contractility during the exposure to low temperatures and the resistance to ventricular fibrillation in torpid animals presents an adaptive response because it occurs in hedgehogs (E. europaeus) after exposure to short day length, whereas in long-day length-adapted animals the induction of arrhythmias is still possible (10).
Because animals undergoing torpor epochs arouse periodically, the ability to function in both euthermic and heterothermic states must be maintained continuously. Hibernators and daily heterotherms appear to employ mechanisms to preserve mRNA pools that could facilitate the resumption of gene expression during arousal for the replenishment of protein pools. In this regard, the observed sleeplike “posttorpor inactivity state,” during which heart rate and presumably metabolic rate are enhanced beyond normal levels, is particularly noticeable and may reflect rapid restoration of protein pools. Alternatively, the intense posttorpor sleep has recently been associated with the replacement of dendritic and synaptic plasticity (21). A widespread adaptation on short timescales at the molecular level during low-temperature torpor is unlikely to occur; rather, it seems that animals exploit the temperature-dependent reduced rates of biochemical reactions and use spontaneous interbout rewarmings to recover from the biochemical suppression. Only a small number of gene transcripts (∼5%) show alteration over a daily hypometabolic torpor bout in Phodopus, and a large part of these transcripts are involved in regulation of protein turnover and degradation (8). Clearly, any biochemical adjustment involved in maintaining function during torpor must not compromise function during euthermia. Torpor is used by species of the three evolutionary branches of mammals: placentals, marsupials, and monotremes. Phylogenetic distribution of torpor among mammalian species provides a strong argument for many mammals experiencing resistance to cold and reversible hypometabolism. The widespread distribution of torpor in mammals may result from reactivation or retention of gene expression patterns that are normally required for fetal or neonatal life.
Perspectives and Significance
The heart of heterothermic animals, as exemplified by the Djungarian or Siberian hamster (P. sungorus), shows a remarkable integrity of normal excitation patterns during highly irregular sinus rhythm at Tb ∼21°C, despite an approximately eightfold slowing of heart rate. Cardiac dynamics revealed no manifestations of conduction block, tachyarrhythmia, or other forms of ectopy. The ECG is characterized by the absence of a distinct isoelectric ST segment and an early-peaking dome or hump configuration of the T wave. These findings suggest the presence of a prominent repolarizing Ito,f and a narrow triangular ventricular AP with no plateau phase in Phodopus and (all?) other heterothermic species undergoing hibernation or daily shallow torpor. The present findings derived from the intact heart warrant detailed follow-up studies of sinus node autorhythmicity and AP properties across myocardial layers at low temperatures. Specifically, the evolutionary or adaptive role of Ito,f and other membrane currents relevant for ventricular repolarization including synchronization of this process mediated by gap junctions and their impact on electrical stability and arrhythmia protection require further evaluation, both on large (short vs. long day adaptation) and short (time frame of torpor episode) timescales. Identification of the molecular and cellular adaptations employed by animals exhibiting low-temperature torpor provides an important lead for future directions of research that will have profound implications for the susceptibility of humans and other homeothermic species to cardiac arrhythmias and hence be of therapeutic benefit.
This work was supported by the Center for Neurogenomics and Cognitive Research and the Max Planck Society.
We acknowledge the excellent technical support by Jens Brinkmann and Anja Ronnenberg. Major parts of the data analysis were performed by Alexander Mertens during an internship in the Fractal Physiology Group.
↵1 In homeothermy, animals maintain a constant (37°C) Tb; in heterothermy, there are regulated fluctuations of Tb between high (37°C) and low (0–37°C); euthermy is a state of warm (∼37°C) Tb.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society