Heart rate variability (HRV) is a well-characterized, noninvasive means of assessing cardiac autonomic nervous system activity. This study examines the basic cardiac responses to hypoxic and hypercapnic challenges in seven strains of commonly used inbred mice (A/J, BALB/cJ, C3H/HeJ, C57BL/6J, CBA/J, DBA/2J, and FVB/J). Adult male mice, 8–12 wk of age, were chronically instrumented to a femoral artery catheter for the continuous measurement of systemic arterial blood pressure and heart rate. Mice were exposed to multiple 4-min periods of hypoxia (10% O2), hypercapnia (5% CO2), and combined hypoxia/hypercapnia (10% O2 + 5% CO2). HRV was derived from pulse intervals of the blood pressure tracings. Hypoxia induced increases in high-frequency HRV power and decreased low-frequency (LF) HRV power in most strains. Hypercapnia led to decreased high-frequency HRV power and increased LF HRV power in most strains. Strain differences were most notable in regard to the concomitant exposures of hypoxia and hypercapnia, with FVB/J mice mirroring their own response to hypercapnia alone, whereas CBA/J mice mirrored their own responses to hypoxia. As blood pressure is most likely the driving factor for heart rate changes via the baroreflex pathway, it is interesting that LF, considered to reflect cardiac sympathetic activity, was negatively correlated with heart rate, suggesting that LF changes are driven by baroreflex oscillation and not necessarily by absolute sympathetic or parasympathetic activity to the heart. These findings suggest that genetic background can influence the centrally mediated cardiovascular responses to basic hypoxic and hypercapnic challenges.
- sleep apnea
- blood pressure
- frequency domain
heart rate (hr) variability (HRV) has been used for roughly 20 years as a noninvasive means of investigating cardiac autonomic nervous system (ANS) activity. The frequency domain parameters (those obtained by Fourier-like transformation of tachograms), namely the high-frequency (HF) and low-frequency (LF) ranges, have been correlated with parasympathetic and sympathetic activity, respectively (14). Whereas many studies utilize HRV to prognosticate disease and risk of adverse cardiovascular events (10), the use of HRV parameters to assess the physiological, or even toxicological, effects of acute stimuli has grown over the past few years. The work by Just et al. (8) and Gehrmann et al. (5) established the feasibility of using HRV to assess altered phenotypes in mouse models; given the rapid creation of novel transgenic murine models, HRV has since seen a great deal of use. Our earlier work investigating hemodynamic and HRV differences across sleep states in various mouse strains provided one of few direct comparisons of the genetic influences on HRV (2). The present study expands on that work by investigating in conscious, instrumented mice the HRV responses to hypoxic and hypercapnic stimuli.
The physiological response to hypoxia and hypercapnia can be complex and seems to vary considerably across species (1, 3, 4, 9). Hypoxia and hypercapnia induce locally and centrally mediated changes in HR, cardiac contractility, and peripheral vascular resistance. Hypoxia decreases arteriolar resistance by locally mediated vasodilation (21), but this response is compensated by baroreceptor-mediated constriction of larger vessels, as well as increased HR and contractility, to maintain systemic blood pressure (BP). Hypoxic stimulation of the carotid body chemoreceptors leads to increased sympathetic and parasympathetic output (13), the pressor effects of which may depend greatly on the receptor phenotype of the vasculature. Hypercapnia stimulates central receptors that generally produce sympathetic output, resulting in increased peripheral resistance and elevated BP. Because of the robust and unidirectional (i.e., all sympathetic) effect of elevated arterial Pco2, we predict that the cardiac responses to hypercapnia will be better conserved across strains, whereas the response to hypoxia will show more influence of genetic background.
Data used for HRV analysis were obtained from BP tracings in mice; portions of these BP data have been presented elsewhere (3). The present report describes a post hoc analysis of those data to investigate the characteristics of HR control under the hypoxic and hypercapnic conditions. Portions of the data from Campen et al. (3) are summarized in Table 1.
Adult, male mice (n = 6–12 per strain) were purchased from Jackson Laboratories (Bar Harbor, ME) at 8 wk of age and housed in an Association for Assessment and Accreditation of Laboratory Animal Care-approved microisolator facility. The following strains were examined: A/J, BALB/cJ, C3H/HeJ, C57BL/6J CBA/J, DBA/2J, and FVB/J. Temperature and humidity were continuously regulated to 20–22°C and 40–60% relative humidity, respectively, with a 12:12-h light-dark cycle. Food and water were available ad libitum throughout the study. Protocols were conducted on the assorted strains nonconsecutively to avoid bias. All studies were conducted with the approval of the Johns Hopkins Animal Care and Use Committee.
All surgical procedures were performed under isoflurane anesthesia using aseptic techniques. To ensure exposures were conducted in conscious animals, electrodes for polysomnography were implanted as previously described (18). Briefly, three Teflon-coated wires were inserted into predrilled holes in the left frontal and left and right parietal regions for EEG leads. Two electromyographic electrodes were stitched flat onto the surface of the muscle immediately posterior to the dorsal area of the mouse skull.
To obtain a pulsatile BP trace, the femoral artery was exposed by a 1.5-cm cutaneous incision and blunt dissection of the fascia and surrounding connective tissue and tied with 6-0 suture distal to the point of catheter insertion. A 60-cm Micro-Renathane catheter (MRE025 Braintree Scientific, Braintree, MA), heat-stretched and formed into a J-shape, was inserted with the aid of a 26-gauge needle and advanced ∼0.5–1.0 cm toward the iliac bifurcation. The catheter was secured by suture and cyanoacrylate glue (Quicktite Super Glue, Manco, Avon, OH), then exteriorized at the base of the skull, and secured to the EEG/electromyographic electrodes. The catheter was attached to a single-channel fluid swivel (375/25 Instech Laboratories, Plymouth Meeting, PA) and perfused slowly by an infusion pump (0.5 ml/day) with a sterile saline solution containing heparin (80 U/ml). BP measurements were facilitated by a flow-through pressure transducer. Animals received a minimum of a 5-day recovery period before any exposures.
Monitoring and data analysis.
Both the polysomnography leads and the flow-through transducer were connected to a pen recorder during exposures (Grass Instruments, Quincy, MA). Data from the pen recorder were sampled at 300 Hz, converted to digital format (DI-200 data-acquisition board; Dataq Instruments, Akron, OH), and acquired to optical disks for storage with Windaq/200 acquisition software (Dataq Instruments). BP and HR were determined from signals averaged over 24 h.
Gas exposure protocol.
All exposures were conducted at ∼30 m above sea level; thus ambient pressure approximated 760 mmHg. Mice were placed in a small cylindrical exposure chamber (0.7 liter) with standard bedding. This chamber allowed free movement of the mouse, but was small enough to enable rapid changes of atmospheres. Room air (inspired O2 fraction = 0.209) was forced through the chamber at 2 l/min. The hypoxic atmosphere was produced by adding pure N2 to the chamber at a flow rate equal to that of room air (net flow rate = 4 l/min; inspired O2 fraction = 0.10). Hypercapnia was produced by a standard mixture of 5% CO2, 40% O2, and 55% N2. The hypoxic and hypercapnic atmosphere was produced by a standard mixture of 5% CO2, 10% O2, and 85% N2. The flow rate for the latter two exposures was ∼4 l/min. Oxygen levels in the chamber were continuously recorded (SensorMedics Oxygen Analyzer, OM-11, Yorba Linda, CA).
The exposures began after a 1-h period of acclimation to the exposure chamber. All mice were monitored during exposures to ascertain wakefulness. No exposures were conducted while mice were sleeping. Each exposure lasted 4 min, with 8 min of recovery time following different exposures (i.e., between hypercapnia and hypoxia) and 4 min of recovery between repetitions of the same exposures. Each mouse was exposed to a minimum of two repetitions of each atmosphere.
Beat-to-beat intervals were obtained by identifying systolic peaks of the arterial pressure waveform (Windaq Pro; DATAQ Instruments, Akron, OH). Pulse intervals were sorted to remove any rhythm that was >50% of the preceding interval. The short periods of investigation (4-min exposures) enabled the visual investigation of all abnormal pulse intervals to confirm that movement, and not pulse deficit related to arrhythmia (as occurred often in DBA mice), was the culprit. Missed beats were interpolated (by dividing prolonged intervals), and noise intervals (waveforms interrupting the true BP signal) were summated, to maintain the overall temporal integrity and continuity of the data sets. Arrhythmias were characterized in a separate cohort of mice using ECG leads. Typically, hypoxia-induced arrhythmias were Type II AV-node block, as indicated by a P-wave with no subsequent ventricular activity. Premature atrial or ventricular contractions were not observed with any consistency. DBA/2J was the only strain that consistently demonstrated arrhythmias during hypoxia (3).
The resulting tachograms (Fig. 1A) were then transformed using a Lomb-type periodogram to determine the frequency spectrum (Fig. 1B; Ref. 16), using a computer program developed in association with the US Environmental Protection Agency and Dr. William P. Watkinson. This method was preferred over traditional Fourier analysis, as the Lomb-type transform is better suited for analyzing discrete data series as opposed to continuous, evenly sampled waveforms. The LF range was calculated as the area under the curve from 0.2 to 1.5 Hz, and the HF range was calculated between 1.5 Hz and the Nyquist frequency (HR frequency divided by 2, typically around 5 Hz). Determination of the HF peaks confirmed that this HF range was appropriate to all strains.
Phenotypic measurements of systemic arterial blood pressure (PSA), HR, and HRV for control and exposed periods were averaged for each strain. The statistical difference in cardiovascular parameters between control and exposed periods was determined by paired t-test within each strain. One-way ANOVA was used to identify significant change differences in cardiovascular parameters between the seven different strains. If the ANOVA was significant, a post hoc test (Scheffé’s method) was used to identify which strains were significantly different. Data are reported as means ± SE, and differences were considered significant if P < 0.05.
Baseline cardiovascular values.
Table 1 shows the baseline comparisons among the seven mouse strains. In general, the FVB/J mouse stood out as having elevated PSA, HR, HF, and lower LF/HF and standard deviation of normal intervals (SDNN) than other strains. A/J mice displayed significantly lower PSA and LF/HF values than most strains. These PSA and HR values are similar to, albeit higher than, reported 24-h averages from earlier work (2).
Hypoxia (10% O2) induced varying changes in HF values across the strains of mice (Fig. 2). C3H mice demonstrated the greatest HF increase, whereas FVB/J mice showed decreases in HF. In general, LF (Fig. 3) and LF/HF (Fig. 4) behaved opposite to HF, decreasing in BALB/cJ, C3H/HeJ, C57BL/6J, and CBA/J mice. SDNN measures tended to follow the behavior of the LF values (Fig. 5). Again, FVB/J mice demonstrated opposite responses to most other mice, with an increasing trend in LF over the 4-min exposure; such opposite behavior was also seen in PSA (Fig. 6) and HR (Fig. 7) alterations. Genetic differences were apparent in HF and SDNN measures, with FVB/J and DBA/2J mice demonstrating significantly different patterns than CBA/J and C3H/HeJ mice (ANOVA; P < 0.05). Interestingly, DBA/2J mice displayed marked arrhythmogenesis during hypoxia, displaying over 20 skipped beats/min during the 4th min of the challenge. These arrhythmias were characterized by ECG in a separate cohort of mice, and it was confirmed that the patterns typically resembled Type II AV block, with little incidence of premature contraction (3). Unfortunately, such arrhythmias may invalidate strong conclusions on the resultant HRV parameters during hypoxia. No significant arrhythmias were detected in other strains. The arrhythmias of the DBA/2J mice may be related to the severe hypotension and bradycardia observed; alternatively, such arrhythmias may impair the maintenance of normal BP.
During exposure to 5% CO2, all strains demonstrated significant decreases in HF power (Fig. 2), which were, for the most part, mirrored by increases in LF (Fig. 3), LF/HF (Fig. 4), and SDNN (Fig. 5) parameters. PSA consistently increased during CO2 exposure (Fig. 6), whereas HR values decreased concomitantly (Fig. 7), presumably due to baroreflex modulation. No hypercapnia-related arrhythmogenesis was observed in any strain. While all strains behaved similarly in the direction of the response to CO2, DBA/2J mice did show a lower magnitude of HF power response (0.71 ± 0.41 ms2) compared with FVB/J mice (2.85 ± 1.08 ms2; P < 0.01).
Combined hypoxia and hypercapnia.
When administered concomitantly, hypoxia and hypercapnia elicited varying responses among strains. A/J and FVB/J mice demonstrated decreases in HF power, whereas C3H/HeJ and C57BL/6J mice demonstrated increases (Fig. 2). Similarly, LF power and LF/HF were elevated in A/J and FVB/J mice, whereas C3H/HeJ, C57BL/6J, and CBA/J mice demonstrated significant decreases in LF (Fig. 3). Significant differences among the strains were primarily driven by the responses of the FVB/J mice compared with the C57BL/6J and C3H/HeJ mice for HF, LF, and LF/HF measures. SDNN measures were either increased (A/J, BALB/cJ, and FVB/J) or showed no significant trend; no strain-related effect was observed for SDNN (Fig. 5).
Interestingly, the HR of the A/J and FVB/J mice showed significant decreases in response to hypoxia and hypercapnia in combination (Fig. 7), whereas PSA showed no change in A/J mice and a major (>10 mmHg) increase in FVB/J mice (Fig. 6). Both the BALB/cJ and C57BL/6J mice demonstrated PSA increases in response to combined hypoxia and hypercapnia, but showed no significant changes in HR.
As an initial investigation to better understand how genetics may influence the pattern of HRV response to hypoxia and hypercapnia, we assayed various strains of inbred mice and found significant strain-related differences. As HRV parameters reflect cardiac ANS activity, we chose stimuli that are consistently associated with sympathetic and parasympathetic responses. Specifically, CO2 elicits a sympathetic response, whereas hypoxia elicits a combination of ANS responses (13). Our results showed clear patterns of HRV changes during these exposures that are consistent with the traditional interpretations of LF (sympathetic) and HF (parasympathetic) power. Importantly, while most strains of mice showed similar responses, the FVB/J mouse demonstrated clearly opposite HRV changes during hypoxia and combined hypoxia/hypercapnia.
The patterns of HRV changes during hypercapnia reflect the traditional interpretations of a sympathetic response, i.e., increased LF, LF/HF, and SDNN, and decreased HF (14). Patterns for hypoxia generally followed a “parasympathetic” interpretation, although not in all strains. The FVB/J mice displayed a notable decrease in HF, with increases in LF, LF/HF, and SDNN. DBA/2J mice also showed decreases in HF during hypoxia, although the degree of arrhythmogenesis observed in this strain most likely invalidates the HRV calculations. The dramatic difference of the FVB/J strain is consistent with earlier reports of the strain’s hypersympathetic tendencies (17). At baseline, FVB/J mice demonstrated the highest HR and PSA while conscious; on the other hand, baseline sympathetic HRV parameters (LF, LF/HF, SDNN) were lower than the other strains. It does seem apparent that, as HR approaches or surpasses 700 beats/min in mice, the opportunity for variation wanes. Such is often the case at the lower end of the HR spectrum (i.e., <500 beats/min in mice).
Certain findings, however, do not reflect the traditional interpretations of HRV, e.g., that the LF power tends to behave opposite to HR. It seems intuitive that, if LF (or LF/HF) represents cardiac sympathetic activity, then HR would be expected to be positively correlated. A possible explanation is that the sympathetic activity elicited by hypercapnia leads to LF increases, despite HR reduction because of the oscillating nature of baroreflex modulation. Even though the predominant cardiac stimulus is vagal and negatively chronotropic, the baroreflex maintenance of PSA leads to an oscillating ANS output that still modulates the HR at a frequency consistent with the sympathetic output. Stated another way, sympathetic activity causes a pressor response that is attenuated by a baroreceptor feedback loop oscillating at a frequency of ∼0.4 Hz in the rat (7), which is probably a good estimate for mice, as well. Therefore, while the HR decreases, LF HR fluctuation, entrained by baroreflex modulation, increases as a result of heightened sympathetic activity.
It is well known that the ANS plays a significant role in balancing the hemodynamic response to hypoxia and hypercapnia in other species. Hirakawa et al. (6) reported increases in sympathetic activity and withdrawal of vagal activity during hypocapnic hypoxia in the rat, whereas isocapnic and hypercapnic hypoxic exposures led to increased parasympathetic activity, possibly as a result of chemoceptor-related pressor responses and baroreceptor feedback. This study was followed by an analysis of HRV in Wistar rats that demonstrated withdrawal of LF and HF power during hypocapnic hypoxia and considerable increases in the LF and HF power during hypercapnic hypoxia (12). Direct comparisons of these results with those of the present study are not plausible, due to differences in mathematical procedures. Our algorithm for frequency-domain HRV parameters returned relative frequencies; thus the total power (sum of LF and HF; not reported) remained fairly constant across all mice (except for the arrhythmic DBA/2J mice). However, withdrawal of LF power was quite commonly observed in mice during hypoxia, whereas the combination of hypercapnia and hypoxia led to increases in LF power in two strains (A/J and FVB/J), similar to that reported in rats.
Other hypotheses for HRV control should also be considered. Mechanical influences on the heart could be expected to play some role in these situations as well. For one, the LF fluctuation brought on by baroreflex changes would undoubtedly alter venous return at the same frequency, leading to stretch receptor activation in the right atria. Such stimulation may amplify the alterations in HRV, depending on the timing of the reflex arch. Meanwhile, hypoxia and hypercapnia both lead to ventilatory stimulation, which should be reflected in the HF range. Other studies have noted clear entrainment of HRV frequencies with respiratory rate, even with a dissociation with sympathetic nerve activity (22), suggesting that the pleural pressure influence on venous return may also mechanically influence HRV. Tankersley et al. (20) demonstrated the effects of hypoxia and hypercapnia on ventilation, and, while genetic differences were observed, both stimuli consistently increased rate and volume of ventilation. However, given that hypoxia and hypercapnia typically induced opposite patterns of HRV changes, despite the likelihood that both led to increases in respiratory frequency and tidal volume, this suggests that there is more to the HRV behavior than mechanical forces.
HRV has been reported to reflect deficits in ANS input and/or cardiac responsiveness in a variety of diseases and with natural aging processes and is an important predictor of several adverse cardiovascular events (10). HRV has also been studied, as in the present report, in terms of its response to acute physiological changes. Our major interest stems from recent findings of air-pollution influences (as well as other forms of occupational exposure) on HRV (19). The difficulty with interpreting such studies is that acute HRV alterations do not necessarily predict adverse outcomes (i.e., mortality and morbidity). Thus while HRV parameters may decrease as a result of an acute stimulus (e.g., ambient air pollution levels increase; Refs. 11, 15), these changes do not imply an adverse effect but rather a physiological change. The present study certainly emphasizes the point that acute changes do not necessarily import any adverse functional demise, as all signals returned to baseline quickly (within 4 min) and consistently following the 4-min gas challenges.
The most significant drawback of the present study is the absence of supporting ventilatory data. As significant ventilatory response differences are noted between strains exposed to hypoxia and hypercapnia (20), this has a high potential for interacting with HR, especially in the HF range. Tankersley et al. (20) reported resting respiratory frequencies ranging from 1.8 Hz in the C3H/HeJ mouse to 2.7 Hz in the C57BL/6J mouse. Moreover, responses to hypoxia elevated frequencies into the 3.5- to 5-Hz range, and hypercapnia stimulated even greater elevations. As the HF range for HRV was from 1.5 to 5.0 (or the Nyquist frequency, which varied with HR), respiration is the predominant oscillating physiological process that occurs in this HF range. Respiration can affect HR by both neural (stretch receptor reflex arcs) and mechanical (venous return changes with thoracic pressure) means. While the HF HRV range equally encompasses the respiratory rates that we anticipate observing with the hypoxic and hypercapnic exposures, the entrainment of HR oscillations may be more significant and nonlinear over certain ranges. One would expect that particular harmonic frequencies (i.e., respiratory rate equals one-half or one-third of HR) would have a greater impact on entraining the HR and therefore affecting the HRV. Thus the genetic effect on ventilation could certainly drive the effects observed in the present study; however, there is no certainty that the effect of respiration on HF is linear.
Because the implantation of the femoral artery catheter is fairly invasive, this environmental factor cannot be ruled out as a confounder to the interpretation of these data. We have previously noted that certain transgenic strains display a poor recovery from the catheterization surgery (e.g., ob/ob mice; unpublished observations); thus genetic variation may have some effect on the recovery. The mice from the present study were monitored closely to ensure normal recovery and general health throughout the protocols. Mice that did not recover appetite or appeared agitated were excluded from the study. There was no relationship between strain and exclusion from the study; survival was more closely related to the learning curve for performing the surgery. Because strains were investigated nonconsecutively, this source of error does not bias our conclusions. We feel that the robustness of the observed findings implies, at least, a predominance of genetic influence on the hypoxic hemodynamic responses.
In summary, the profile of pacemaking and hemodynamic responses to hypoxia and hypercapnia strongly indicates that genetic differences exist in the manifestation of ANS responses among these strains of mice. Our results are consistent with previous interpretations of HRV in mice (5, 8) and reflect the ANS alterations that accompany hypoxic and hypercapnic exposures. Further investigation into the genetic differences between FVB/J mice and most other commonly used strains might help clarify the observed differences in HRV responses.
This research was funded by National Heart, Lung, and Blood Institute National Research Service Award Grant F32HL-68417 (M. J. Campen).
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 © 2005 the American Physiological Society