Recent studies have suggested a genetic component to heart rate (HR) and HR variability (HRV). However, a systematic examination of the genetic contribution to the variation in HR and HRV has not been performed. This study investigated the genetic contribution to HR and HRV using a wide range of inbred and recombinant inbred (RI) mouse strains. Electrocardiogram data were recorded from 30 strains of inbred mice and 29 RI strains. Significant differences in mean HR and total power (TP) HRV were identified between inbred strains and RI strains. Multiple significant differences within the strain sets in mean low-frequency (LF) and high-frequency (HF) power were also found. No statistically significant concordance was found between strain distribution patterns for HR and HRV phenotypes. Genomewide interval mapping identified a significant quantitative trait locus (QTL) for HR [LOD (likelihood of the odds) score = 3.763] on chromosome 6 [peak at 53.69 megabases (Mb); designated HR 1 (Hr1)]. Suggestive QTLs for TP were found on chromosomes 2, 4, 5, 6, and 14. A suggestive QTL for LF was found on chromosome 16; for HF, we found one significant QTL on chromosome 5 (LOD score = 3.107) [peak at 53.56 Mb; designated HRV-high-frequency 1 (Hrvhf1)] and three suggestive QTLs on chromosomes 2, 11 and 15. In conclusion, the results demonstrate a strong genetic component in the regulation of resting HR and HRV evidenced by the significant differences between strains. A lack of correlation between HR and HRV phenotypes in some inbred strains suggests that different sets of genes control the phenotypes. Furthermore, QTLs were found that will provide important insight to the genetic regulation of HR and HRV at rest.
- autonomic nervous system
cardiovascular disorders are prominent public health concerns, and their complexity can be difficult to dissect. However, measurements of heart rate (HR) and heart rate variability (HRV) have become increasingly useful clinical tools for assessment of autonomic nervous system function and cardiovascular health (1). Reduced HRV and increased resting HR are well-established indicators of increased cardiovascular risk (5, 33). A number of factors have been identified as determinants of HRV (29) that do not fully explain the reported variance in HRV (24, 41). Importantly, genetic factors may significantly influence HR and HRV but have not been studied in sufficient detail.
HRV, in the frequency domain, is divided into a number of frequency ranges. In this study, the variation in HR in the low- (LF) and high-frequency (HF) ranges were of principal interest. LF HRV can be attributed to delays on baroreceptor feedback as a consequence of blood pressure fluctuations. HF HRV can be ascribed to respiratory sinus arrhythmia, where the HRV peak within the HF HRV range matches the breathing frequency (26).
Epidemiological studies have begun to address the genetic influence on HRV in human subjects, and heritability was estimated to account for 13–22% (31) and 28–34% (32) of the interindividual variance in HRV. Unfortunately, these studies can be difficult to interpret since the effect of respiration on HRV was not considered. Nevertheless, they do provide an excellent platform from which a more detailed study of the genetic contribution to HR and HRV can be performed while respiration is monitored.
Studies that have measured HR and HRV from conscious, unrestrained inbred rodents have used a limited number of strains, and the effect of respiration on HRV was not considered (21, 36). Initial evidence for a genetic contribution to HR was provided by Kreutz et al. (22) who reported an association between Hr-sp1 locus on rat chromosome 3 and HR regulation. Furthermore, differences in the HR of inbred mouse strains have been reported (35). However, Campen et al. (7) found no significant between-strain differences in HRV. Interestingly, the ECG was recorded during periods of sleep, and therefore the physical activity of these mice was presumably similar and between-strain differences in respiration may have been minimized, making the ECG record more appropriate for HRV analysis. Hoit et al. (19) investigated cardiovascular variation and reported a 91 and a 79 beats/min difference in HR between A/J and C57BL/6J mice using echocardiography and tail-cuff measurements, respectively. Tankersley et al. (36, 39) have reported an ∼80 beats/min difference between C3H/HeJ and C57BL/6J mice using radiotelemetry.
The current ambiguity in the literature regarding the genetic regulation of HR and HRV in humans and rodents warrants further and more detailed investigation. In the present study, HR and HRV were calculated from ECG recorded from inbred strains of mice using telemetric methods while monitoring minute ventilation (V̇e). The purpose of this study was to assess the within- and between-strain variation in HR and HRV and to identify quantitative trait loci (QTLs) that account for significant genetic variance in HR and HRV phenotypes.
MATERIALS AND METHODS
Thirty male inbred strains of mice (20–30 g) were studied (Table 1). All mice, except for CAST (24 wk old), SPRET (37 wk old), and MOLF (49 wk old), were 8–12 wk of age (Table 1). CAST, SPRET, and MOLF mice were studied after additional aging because the body mass of these mice was not sufficient to meet surgery requirements when at an age similar to the other strains. Twenty-nine male recombinant inbred (RI) strains (20–30 g) were studied. All mice were studied at 8–21 wk of age (Table 2).
All mice were purchased from Jackson Laboratory (Bar Harbor, ME) with the exception of the AKR mice, which were purchased from the National Cancer Institute (Bethesda, MD), and the ICR mice, which were purchased from Taconic Farms (Raleigh, NC). Information regarding the known characteristic features of the different strains is publicly available through The Jackson Laboratory Mouse Phenome Project (http://phenome.jax.org/pub-cgi/phenome/mpdcgi?rtn=docs/home). All mice were housed individually with a 12-h:12-h (7:00 am to 7:00 pm) light-dark cycle. Food (NIH-31) and water were provided ad libitum. Animals were handled in accordance with The National Institutes of Health Humane Care and Use of Laboratory Animals guidelines and the American Physiological Society's “Guiding Principles in the Care and Use of Animals.” The study protocol was reviewed and approved by the National Institute of Environmental Health Science Animal Care and Use Committee.
Surgical Implantation of Radiotelemetry Transmitter
Mice were anesthetized using inhaled isoflurane, and buprenorphine (0.1 mg/kg) was given for analgesia. A 3-cm midline dorsal incision was made in the skin, and a subcutaneous tissue pocket was made using a blunt instrument. An ETA-F20 transmitter (DSI; Arden Hills, MN) was placed inside the tissue pocket and sutured to the left latissimus dorsi muscle. The anodal and cathodal leads were tunneled subcutaneously and sutured over the left superficial gluteus and right trapezius muscles, respectively. All incisions were closed using wound clips, and animals were allowed 5 days to recover.
All data were recorded from conscious, unrestrained mice at the same time of day (9:00–11:00 am) to control for circadian variation. Animals were placed individually in whole body plethysmographs (Buxco Electronics, Wilmington, NC) and allowed at least 30 min to acclimate. To monitor the possible confounding effects of breathing on HRV (6), we simultaneously recorded 20 min of radiotelemetry ECG and pulmonary function data from each mouse during periods of quiet rest or sleep when breathing rate and depth were consistent.
With 20 min of ECG from individual mice, R-waves were marked and R-R intervals were extracted using specialist software (Dataquest A.R.T. v. 3.1, DSI). The resultant R-R series was then uploaded to custom software (HRV Webstart edition development) created in collaboration with Drs. Todd Jenkins (East Carolina University, NC) and Matthew J. Campen (The Lovelace Foundation, NM). Any R-R interval that was 150% greater or less than the median R-R intervals for a given series was considered an arrhythmia and removed; HRs were then calculated. HRV was calculated using a Lomb periodogram, which was developed specifically for spectral analysis of unevenly spaced data.
Currently, a standard LF cutoff has not been agreed on for mice. However, it is important to consider statistical as well as physiological rationales when choosing the LF cutoff. We analyzed 120 s of ECG data at a time, and it is well known that the LF cutoff should be limited to 1/6 of the sampling length (i.e., 0.05 Hz). In our case and in that of Campen et al. (7), we used a LF cutoff where spectral power commonly reduced almost to zero (∼0.2 Hz). Since 0.2 Hz is greater than the minimum cutoff of 0.05 Hz, we feel confident that our LF range is appropriate.
The LF range was set at 0.2–1.5 Hz, and HF was 1.5–5.0 Hz. Total power (TP) was determined by the summation of the LF and HF range values.
Breathing frequency (f) tidal volume (Vt), and minute ventilation normalized to body mass (nV̇e) were used to indicate changes in pulmonary function, which are thought to affect HRV assessments. Values for f, Vt, and V̇e were calculated using specialist software (Buxco Electronics).
Genomewide scans for QTLs were done with the entire RI data set and parental phenotypes. We used Web QTL (www.genenetwork.org) (43) for interval analyses by fitting a regression equation for the effect of a hypothetical QTL at the position of each marker and at intervals between the markers. Over 8,500 informative markers and 2,400 unique strain distribution patterns (SDPs) exist for the AXB/BXA RI set. Likelihood of the odds (LOD) score curves on the interval map are provided for each chromosome and graphically represent the approximate position of the QTL(s). The dominance and additive properties of each putative QTL were also evaluated. The significance of each association (the base-10 LOD score) was determined and plotted against the linear position on the chromosome. Permutation tests were used to establish empirically the significance thresholds for the genomewide QTL mapping results following Churchill and Doerge (10) and Doerge and Churchill (12). For each genome scan, 1,000 permutations were done to establish significant and suggestive linkage threshold values as indicated in each figure. These values correspond to the genomewide probabilities proposed by Lander and Kruglyak (23) and are as have been published previously for mouse and rat RI data sets (9, 11, 16, 42). The significant threshold level approximately corresponds to a LOD score of 3.0 (i.e., P = 10−4) or higher, whereas the suggestive threshold level approximately corresponds to a LOD score of 2.1 (i.e., P = 10−3).
Data from individual mice were used to calculate the strain average (mean). Strain averaged data were then used to make comparisons between strains. Means ± SE are presented. Differences between strains in nV̇e, f, Vt, HR, and HRV were assessed independently using a one-way ANOVA with Tukey's post hoc test. However, the LF and HF HRV phenotypes failed the normality test, and therefore between strains differences were assessed using a Kruskal-Wallis ANOVA on ranks with Dunn's post hoc test. The relationships between HR and TP, body mass or age, and nV̇e were assessed using a linear regression analysis. Statistical significance was accepted at <0.05.
Inbred Strain Comparison
Multiple statistically significant differences in nV̇e (in ml·min−1·g−1) and f (in ml·min−1·g−1) were found between inbred strains (Table 1). Interestingly, the SDPs for nV̇e, Vt (in ml), and f were discordant (Table 1). No statistically significant relationship was found between age (in wk) or body mass and nV̇e, Vt, or f (data not shown), although all strains (except the wild-derived CAST, SPRET, and MOLF strains) were age matched (Table 1).
HR and HRV.
Statistically significant differences were found in baseline HR and HRV between the inbred strains (Figs. 1A and 2A). Mean (±SE) HR (in beats/min) ranged from 480.5 ± 11.4 (129Svlm) to 776.9 ± 11.5 (CAST). Whereas mean HR for C57L mice was significantly lower than CAST HR, the mean HRs of both strains were significantly higher than all other strains (P < 0.05; Fig. 1A). The range in mean (±SE) TP (in ms2/Hz) was 1.39 ± 0.2 (PL) to 3.35 ± 0.1 (HeJ; Fig. 2A) for the inbred strains. Significant between-strain differences in LF and HF power, the two components of TP, were also found (Fig. 2B), and the SDPs were not the same as TP. Indeed, the relationship between mean LF and HF was different and in some cases reversed (e.g., NZB vs. DBA2; Fig. 2). Furthermore, no correlation between baseline HRV and nV̇e, Vt, or f was found between these inbred strains.
Mean HR and TP did not cosegregate (R = 0.18) among the strains investigated (Fig. 3A); that is, HR did not predict TP, and in the majority of strains, a wide range of values for TP was observed at similar HRs. The most notable exceptions were the CAST and C57L strains. Despite the higher HRs, TP remained within the range of the other strains, suggesting that HR and TP are independent from one another in these strains.
RI Strain Comparison
nV̇e, f, and Vt were not significantly different between the RI strains (Table 2). The SDPs for each phenotype were different among these strains. It was not possible to predict nV̇e based on f or Vt alone. The strains used different Vt to f ratios to achieve the required nV̇e.
HR and HRV.
Mean (±SE) HR (in beats/min) ranged from 530.3 ± 38.1 (AXB8) to 719.5 ± 53.2 (BXA4; Fig. 1B). This range in RI strain HRs was observed despite no significant difference between the A and B6 parental strains in the present study. The range in mean (±SE) TP (in ms2/Hz) was 1.47 ± 0.5 (AXB24) to 3.18 ± 0.2 (BXA4) for the RI strains (Fig. 2C). As observed for the inbred strains, the SDPs for the LF and HF were different from each other (e.g., AXB13 vs. BXA4; Fig. 3B). No overall concordance was found (R = 0.42; Fig. 3B) between HR and TP, although a wide range of HRs was observed at similar TP values.
QTL mapping with AXB/BXA RI strains.
A genomewide search for QTLs for each of the HR and HRV phenotypes was performed with the entire RI data set. Interval mapping identified a significant QTL for HR (Fig. 4), five suggestive QTLs for TP, a suggestive QTL for LF (Table 3), and one significant and three suggestive QTLs for HF (Fig. 5 and Table 3). The significant HR QTL [designated HR 1 (Hr1)] was found on chromosome 6 between 52 and 56 megabases (Mb), with the peak linkage at marker rs6263715 (53.564327 Mb), and the LOD score of 3.763 exceeded the significance threshold as determined empirically by the permutation test (Fig. 4). The positive additive effect at this and neighboring markers indicates that A/J alleles accounted for the difference in heart rate. A significant HF QTL [designated HR variability-HF 1 (Hrvhf1)] was identified on chromosome 5 between 46 and 56 Mb, with the peak linkage at marker rs6263715 (53.564327 Mb) and a LOD score of 3.107 (Fig. 5). The negative additive effect for this QTL indicates that B6 alleles are responsible for the difference in HF (Table 3). Composite interval mapping was done to estimate the influence of the HF suggestive QTLs on Hrvhf1, but controlling for each did not affect Hrvhf1.
We studied 30 inbred strains and 29 RI strains of mice to determine the genetic contribution to the variation in HR and HRV. Significant between-strain differences were found in HR, HRV, and pulmonary function between the two strain sets. Furthermore, the comparison of HR and HRV SDPs indicates that the traits do not cosegregate, suggesting that different sets of genes may determine the two phenotypes (Fig. 3). To dissect specific chromosomal regions responsible for the variation in these phenotypes, a genomewide QTL analysis was performed for the HR and HRV phenotypes. Significant and suggestive QTLs were found for HR, LF, HF, and TP (Figs. 4 and 5; and Table 3).
Pulmonary Function: nV̇e, Vt, and f
Significantly greater interstrain differences in nV̇e, Vt, and f were found compared with within-strain variance, suggesting a genetic contribution to pulmonary function (Table 1). This observation was not in agreement with previous reports, which found no significant differences in V̇e among five of the inbred strains used in this study (37, 38). Furthermore, after the values are normalized for body mass, the nV̇e values in this study were notably higher than those reported by Tankersley et al. (37). The principal cause of the differences in nV̇e between these studies was breathing frequency since Vt values were similar. A number of factors may account for the differences in nV̇e between studies, including seasonal variation, differential behavioral characteristics between strains, and equipment sensitivity. Nonetheless, the new data suggest that genetic background accounts for a significant portion of the interstrain variation in nV̇e (Table 1).
HR and HRV
The strain most commonly used to investigate murine HR and HRV has been C57BL/6J, and reported HRs have not been consistent. Values for mean (±SE) HR recorded in this strain include 662 ± 12 (21) and 564 ± 24 beats/min (7). This lack of consistency in the published data makes estimates of normal murine HR values difficult and may be in part due to differences in animal arousal levels. Mean resting HR values for C57BL/6J mice (584 ± 12 beats/min) in the present study were similar to those reported by Campen et al. (7), who provided an excellent standard because their HR values were calculated over a 24-h period. These HR values are ∼10 times higher than those in the normal human HR range.
The multiple inbred strain SDP for HR revealed at least two distinct phenotypes (Fig. 1A). C57L and CAST HRs were significantly higher when compared with the remaining strains despite all ECG records being made during periods of rest. The mechanisms related to this difference are currently unclear but may become evident when candidate genes are identified and investigated. However, continuous distribution of HR values was found for the majority of the strains studied, which suggests that HR regulation is a complex trait and influenced by multiple genes. A continuous distribution for HR in the RI strain SDP supports this notion. Interestingly, we did not find a significant difference in HR between the AXB and BXA parental strains A/J and B6, where only a 10 beats/min difference was observed (Fig. 1A); this differs from previous reports of much larger differences between A/J and B6 mice (19). Hoit et al. (19) reported higher HR values for both strains during tail-cuff measurements compared with echocardiography. These conflicting values may be attributable to the differing conditions under which the measures were taken. It is assumed, although not reported, that a form of anesthesia was used in their study during echocardiography which could have depressed HR. Conversely, the tail-cuff procedure presents the potential for increased stress to the mouse, thus stimulating HR.
Differences in HR between BALB/cJ and CBA/CaJ mice have been associated with QTLs on chromosomes 2 (Hrq1) and 15 (Hrq2) (35). Sugiyama et al. (35) also demonstrated a significant gene-gene interaction for HR between Hrq1 and D1Mit10, referred to as Hrq3. However, no HR differences were found between BALB/cByJ and CBA/J in the present study. Again, these contrasting results could be due to differences in the methods used for data collection since Sugiyama et al. (35) used the tail-cuff method.
It is important to note that QTL analysis using RI strain sets does not require a significant trait difference between the parental strains but rather a broad range of trait values between the RI strains (Figs. 1 and 3). The recombination of alleles that occurs during derivation of the RI strain set enables different gene × gene interactions that may give rise to trait values that are beyond the range observed in the parental strains. This wide range of trait values can then be analyzed for QTLs that associate with the phenotype of interest. Linkage analysis with the AXB/BXA RI set identified Hr1 on chromosome 6, which contains a number of potentially important genes in association with HR regulation. These candidate genes include corticotropin-releasing factor receptor 2 (Crhr2) and neuropeptide Y (Npy), both of which have been associated with heart rate regulation (34, 40). Although speculative, it is important to note that the effects of these genes may be strain dependent and associate with the higher or lower inbred and RI strain HRs in the present study, warranting further investigation.
In this study, higher LF relative to the HF HRV in 15 inbred strains and in 21 RI strains was found. A number of studies have indicated that the ratio between LF and HF is an index of the relative sympathetic and vagus nerve tone in regulating HR (2, 28, 39a). Previous studies using C57BL/6J mice also reported a higher level of LF relative to HF power (7, 21). This relationship between LF and HF power is the reverse of that expected in resting humans (27) at HRs lower than the anticipated human intrinsic heart rate of 120 beats/min. However, the ratio between LF and HF as an index of sympathovagal balance has been criticized in recent years. The interpretation of the LF and HF components of HRV is a subject of current debate (13, 25, 26, 39a). The LF and HF values most likely indicate only the degree of variability within the HR and not nerve activity.
The continuous distribution of HRV among both strain sets (Fig. 2) suggests that HRV is a complex physiological trait. A significant QTL on chromosome 5 (Hrvhf1) and 3 suggestive QTLs were identified for HF HRV in the RI strains. Candidate genes in Hrvhf1 include Drd5, which encodes the D5 dopamine receptor, and mice deficient in Drd5 develop hypertension caused by a CNS defect increasing sympathetic outflow (20). Peroxisome proliferative-activated receptor-γ coactivator-1α (Pcg1α) has a critical role in cardiac mitochondrial function, and hearts from Pcg1α−/− mice had a reduced capacity for increasing cardiac output in response to stimulation (3), and the overexpression of PCG1α led to cardiomyopathy (30). Endothelial nitric oxide synthase (eNOS) has also been shown to be important of cardiac function and pathology (4, 8). With the consideration of the effect of these genes on cardiovascular function, it is reasonable to suggest an effect on HF. However, given the size of Hrvhf1 and other QTLs identified in this study, it is possible that other genes important in HR and HRV regulation may be identified after fine mapping of these regions.
Genetic analysis of factors regulating HF HRV did not produce evidence of genes involved in respiratory sinus arrhythmia, the principal determinant of HRV at these higher frequencies (18). However, no differences in baseline breathing frequency were observed between the RI strains (Table 2). Therefore, respiratory sinus arrhythmia did not account for the differences in HRV in these data, and other factors were perhaps more important in these resting mice. QTLs containing genes associated with respiratory sinus arrhythmia may only be found if variability in breathing frequency is used a phenotype.
Figure 2 highlights the potential differences between murine and human cardiac regulation. It is well known that in resting humans higher HF HRV is observed relative to LF HRV (39a). However, differences in the HRV characteristics were found between these strains. Fifteen of the inbred strains had higher LF compared with HF HRV than the others (Fig. 2B). Moreover, in PL and BALB mice, higher HF HRV levels were observed, but these strains also had relatively low HR and TP (Fig. 3A). Conversely, in the C57L and CAST strains, high HRs but different levels of HRV were found. The present data demonstrate the complexity of between-strain differences in the autonomic nervous system regulation of the murine heart at rest, but its relationship between murine cardiac responses to acute cardiovascular perturbations remains unclear.
This study reports a clear genetic influence on HR, HRV, and nV̇e using a broad range of inbred mouse strains. However, the influence of genotype on HR and, in particular, HRV is highly complex. This is supported by the continuous distribution of HR and HRV among the strain sets and the multiple QTLs for each phenotype. The complexity of cardiovascular regulation is well known (17) and should be kept in mind when discussing the importance of candidate genes that may play a role in cardiac regulation. The candidate genes discussed were selected based on previously reported associations with cardiovascular function. However, this list of genes was not intended to be definitive. The QTLs identified may contain many other genes that are involved in HR and HRV regulation. This study provides the basis to investigate the precise interaction between genotype and the underlying mechanisms associated with murine heart regulation.
This research was funded by the Intramural Research program of the National Institute of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services.
We thank Drs. Donald Cook and Abraham Nyska for reviewing the manuscript.
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 by the American Physiological Society