Exercise training reduces resting heart rate via downregulation of the funny channel HCN4

Abstract

Endurance athletes exhibit sinus bradycardia, that is a slow resting heart rate, associated with a higher incidence of sinus node (pacemaker) disease and electronic pacemaker implantation. Here we show that training-induced bradycardia is not a consequence of changes in the activity of the autonomic nervous system but is caused by intrinsic electrophysiological changes in the sinus node. We demonstrate that training-induced bradycardia persists after blockade of the autonomous nervous system in vivo in mice and in vitro in the denervated sinus node. We also show that a widespread remodelling of pacemaker ion channels, notably a downregulation of HCN4 and the corresponding ionic current, If. Block of If abolishes the difference in heart rate between trained and sedentary animals in vivo and in vitro. We further observe training-induced downregulation of Tbx3 and upregulation of NRSF and miR-1 (transcriptional regulators) that explains the downregulation of HCN4. Our findings provide a molecular explanation for the potentially pathological heart rate adaptation to exercise training.

Figure 1. Exercise training induces sinus bradycardia and remodelling of the sinus node.

(a) Increase in in the rat following the 12-week training period. Normalized of sedentary and trained rats before and after the 12-week training period shown (n=11 per group). In all bar graphs: black bars represent data from sedentary animals and hatched bars represent data from trained animals. (b) Significant correlation between heart rate in vitro and normalized in sedentary and trained rats (measured after 12-week training period). Each point corresponds to a different animal. Data fit with a straight line by linear regression (n=5/9; R² and P values shown). (c) Representative ECG traces obtained from sedentary and trained (unrestrained and conscious) rats and mice demonstrating longer RR intervals in trained animals. (d,e) Mean (+s.e.m.) RR intervals (d) and corresponding heart rates (e) measured in vivo (rats, n=5/9; mice, n=6/8) and in vitro (from isolated sinus node preparations; rats, n=5/9; mice, n=6/7) in sedentary and trained animals. (f) Mean (+s.e.m.) heart rate measured in vivo in conscious mice at baseline (n=9/11) and after complete autonomic block with propranolol and atropine (n=7/14). (g) Mean (+s.e.m.) sympathetic tone and vagal tone in conscious sedentary and trained mice (n=6/6). (h) Expression of transcripts in the sinus node (grey bars) and atrial muscle (open bars) of trained rats as a percentage of that of sedentary rats. The vertical line corresponds to 100%, that is, no change. Values <100% correspond to a decrease on training and >100% an increase. Student’s t-test used to test differences between data from trained and sedentary animals. Normal distribution of data was tested using the Shapiro–Wilk W-test and equal variance was tested using the F-test. When the null hypothesis of normality and/or equal variance was rejected, the non-parametric Mann–Whitney U-test was used. *P<0.05.

Figure 2. HCN channels in sinus node are downregulated by training.

(a) Downregulation of mRNA for HCN channels in trained rats. Mean (+s.e.m.) expression of mRNA for HCN1, HCN2 and HCN4 (normalized to expression of 18S) in the sinus node of sedentary and trained rats is shown (n=5/8; FDR-corrected Limma test). (b) Significant correlations between HCN4 mRNA and normalized (left) and heart rate in vitro and HCN4 mRNA (right) in sedentary and trained rats. Each point corresponds to a different animal. In all scatter plots: filled circles represent data from sedentary animals, and open circles represent data from trained animals. Data fit with straight lines by linear regression (n=4/8; R² and P values shown). (c) Downregulation of HCN4 protein in trained rats. Representative images of HCN4 immunolabelling (red signal) in the sinus node of sedentary and trained rats are shown in the left and middle panels, respectively (scale bar, 100 μm). The right panel shows the mean (+s.e.m.) HCN4 protein expression quantified using ImageJ from the NIH (n=4). (d) Downregulation of mRNA for HCN4 in trained mice. The left panel shows the mean (+s.e.m.) expression of HCN4 mRNA (normalized to expression of 18S) in the sinus node of sedentary and trained mice (n=6/7) and the right panel shows a significant correlation between heart rate in vitro and HCN4 mRNA in sedentary and trained mice. Each point corresponds to a different animal. Data fit with a straight line by linear regression (n=5/7; R² and P values shown). (e) Downregulation of HCN4 protein in trained mice. Representative images of HCN4 immunolabelling (green signal) in the sinus node of sedentary and trained mice are shown in the left and middle panels, respectively (scale bar, 10 μm). The right panel shows mean (+s.e.m.) HCN4 protein expression (n=4). Student’s t-test used to test differences between data from sedentary and trained animals. Normal distribution of data was tested using the Shapiro–Wilk W-test and equal variance was tested using the F-test. When the null hypothesis of normality and/or equal variance was rejected, the non-parametric Mann–Whitney U-test was used. *P<0.05.

Figure 3. Downregulation of I_f in the sinus node can explain the training-induced resting bradycardia.

(a) Density of I_f is reduced in trained mice. Top, representative I_f traces, normalized to cell capacitance, during steps to -55 to -135 mV (20 mV increments; holding potential −35 mV) from single sinus node cells isolated from sedentary (left) and trained (right) mice. Bottom, mean (±s.e.m.) I_f IV curves recorded at steady state from sedentary (filled circles; n=17/5 cells per mice) and trained (open circles; n=18/5) mice. Student’s t-test used to test differences between sedentary and trained mice; *P<0.05. (b) Whole-cell conductance of I_f is reduced in trained mice. Top, sample current traces recorded with the protocol used to measure the fully activated IV relationship of I_f from single sinus node cells isolated from sedentary (left) and trained (right) mice. Conditioning activating/deactivating voltages were −125 and −35 mV, and pairs of steps to the same test voltage were applied in sequence from these two levels; records shown for test voltages of −100, −40 and 0 mV. Bottom, mean (±s.e.m.) fully activated I/V relationship, normalized to cell capacitance, from sedentary (filled circles; n=9/5 cells per mice) and trained (open circles; n=11/5) mice. Linear fit of data (straight lines) yielded conductances of 653 and 356 pS/pF and reversal potentials of -26.3 and -27.3 mV for sedentary and trained mice. The two curves are statistically different (F-test; P<0.05). (c,d), Spontaneous rate of isolated rat (c; n=7) and mouse (d; n=6/7) sinus node preparations from sedentary and trained animals before and after I_f block with Cs⁺. (e,f) Heart rate of sedentary and trained mice when conscious (e; n=9/13) and when anaesthetized and after autonomic block (f) before and after I_f block with ivabradine (n=7/14). (g–j) Decrease in heart rate (beats min⁻¹) on blocking I_f in corresponding experiment above (c–f). In c-j, means+s.e.m. shown. Student’s t-test used to test differences between sedentary and trained animals. Normal distribution of data tested using Shapiro–Wilk W-test and equal variance tested using F-test. When null hypothesis of normality and/or equal variance was rejected, non-parametric Mann–Whitney U-test used. *P<0.05.

Figure 4. Three regulators of HCN4 in the sinus node are altered by training.

(a) No changes in the expression of three regulators of HCN4 in the sinus node in trained rats, but significant changes in another three. Mean (+s.e.m.) mRNA expression of Tbx18 (sinus node, n=6/6; right atrial muscle, n=7/6), Mef2 (sinus node, n=6/5; right atrial muscle, n=7/6), Sp1 (sinus node and right atrial muscle, n=6/6), Tbx3 (sinus node, n=5/8; right atrial muscle, 8/7), NRSF (sinus node and right atrial muscle, n=7/6) and miR-1 (sinus node, n=4/7; right atrial muscle, n=6/5) (normalized to expression of 18S in the case of mRNAs and RNU1A1 in the case of miR-1) in the sinus node and right atrial muscle of sedentary and trained rats shown. (b,c) Significant correlations between HCN4 mRNA (b) and heart rate in vitro (c) and Tbx3 mRNA, NRSF mRNA and miR-1 in sedentary and trained rats. Each point corresponds to a different animal. Data fit with straight lines by linear regression (n=4–8, R² and P values shown). Student’s t-test used to test differences between data from sedentary and trained rats. Normal distribution of data was tested using the Shapiro–Wilk W-test and equal variance was tested using the F-test. When the null hypothesis of normality and/or equal variance was rejected, the non-parametric Mann–Whitney U-test was used. *P<0.05.

Figure 5. Sinus bradycardia and sinus node remodelling in the mouse are reversed by detraining.

(a) Representative ECG traces obtained from sedentary, trained and detrained mice (unrestrained and conscious). (b) Mean (+s.e.m.) heart rate measured in vivo (n=7/8/5) and in vitro (from isolated sinus node preparations; n=7/8/5) in sedentary, trained and detrained mice. (c) Restoration of the contribution of I_f to pacemaking in the detrained mouse. Mean (+s.e.m.) percentage decrease in heart rate of isolated sinus node preparations from sedentary, trained and detrained mice on blocking I_f using 2 mM CsCl shown (n=6/7/4). (d–f) Restoration of normal levels of HCN4 and Tbx3 mRNA and miR-1 in the sinus node in the detrained mouse. Mean (+s.e.m.) expression of HCN4 and Tbx3 mRNA (normalized to 18S) and miR-1 (normalized to RNU1A1) in the sinus node of sedentary, trained and detrained mice shown (n=4). Student’s t-test used to test differences. Normal distribution of data was tested using the Shapiro–Wilk W-test and equal variance was tested using the F-test. When the null hypothesis of normality and/or equal variance was rejected, the non-parametric Mann–Whitney U-test was used. *P<0.05, trained versus sedentary mice; †P<0.05, detrained versus trained mice.