Contribution of HCN1 variant to sinus bradycardia: A case report

2.1 Human studies

The study protocol and blood drawing procedures were reviewed and approved by West Virginia University Institutional Review Board (IRB1508782471). Written consent forms were sufficiently explained in detail to, agreed, and signed by study participants.

2.2 Electrocardiography of subjects

Heart rate was recorded by a standard 12-lead ECG from the patient (P), the patient's son (PS), and a family unrelated healthy volunteer (V).

2.3 Genomic DNA isolation, HCN4 amplification, and direct sequencing

Genomic DNA was isolated from P, PS, and V using Qiagen DNA isolation kit for blood sample (QIAamp DNA Blood Mini Kit). HCN4 exons were amplified using the primers specified in Table S1 and purified by gel electrophoresis. The WVU Genomic Core performed direct sequencing of HCN4 PCR products using sequencing primers described in Table S2.

2.4 Whole-exome sequencing and variation analysis

Whole-exome sequencing was performed on genomic DNA samples from P, PS, and V by Applied Biological Materials Inc (ABM). Variation-call-format (VCF) files were generated by ABM, then visualized and analyzed using Integrative Genomic Viewer²⁸ and gnomAD²⁹ for missense variations.

2.5 GFP-tagged HCN1 mutant constructs

Wild-type human HCN1 (WT or wtHCN1) in pcDNA3 was used for the construction of variants for ⁷²GGG⁷⁴ and P851A. wtHCN1 and mutants were then cloned into the N-terminus of pEGFP-C1 mammalian expression vector.

2.6 Cell culture and transfection

Human embryonic kidney (HEK293) cells were used for the expression of wild-type and mutant HCN1 channels. Transient plasmid transfection was performed using Lipofectamine_3000 (Invitrogen) following the vendor's instructions.

2.7 Western blotting

HEK293 cells transfected with wtHCN1 and mutant channel plasmids were harvested in RIPA buffer with fresh protease and phosphatase inhibitor cocktail (Sigma) on ice using a cell scraper. Detailed procedures are similar to our previous report⁹ and are provided in Data Supplements.

2.8 Flow cytometry

HEK293 cells expressing wtHCN1 and mutants were harvested 2 days after transfection using 0.25% trypsin with EDTA. Samples were analyzed in the WVU Flow Cytometry & Single Cell Core Facility on a four laser BD LSRFortessa. Details are provided in Data Supplements.

2.9 Confocal imaging of HEK293 cells expressing HCN1 mutant channels

HEK293 cells transfected with wild-type and mutant HCN1 channels were incubated on coverslips and fixed in 4% paraformaldehyde/PBS. The cells were imaged by the Nikon Sweptfield Confocal microscopy.

2.10 Whole-cell patch-clamp electrophysiology studies

Whole cell patch-clamp studies were carried out in HEK293 cells at 35-37℃. Details of patch-clamp studies are similar to our previous reports^{9, 30} and are provided in Data Supplements.

Threshold voltage (V_th) for HCN channel activation was defined as the least negative voltage at which HCN channels begin to open, generating a time-dependent inward current of at least 10pA was observed detected (a value that allowed clear discrimination from background noise, as seen in our previous publication³¹).

2.11 Sinoatrial pacemaker activity cell model

OpenCOR 0.1 software³² was used to simulate the effect of HCN1 variant changes in sinoatrial pacemaker activity using an updated sinoatrial pacemaker activity cell model.³³

2.12 Statistical analysis

Statistical analyses were performed by one-way ANOVA followed by post hoc test Bonferroni (all means compared to wtHCN1) or Tukey (multiple comparisons with each other). The averaged data were presented as means ± SD. A P-value of less than.05 was considered as statistically significant.

3.1 Patient studies

The patient, a 42-year-old African American, was not on any medication nor had electrolyte abnormalities. Despite no evidence of structural heart disease by echocardiography, a 12-lead EKG showed profound symptomatic sinus bradycardia in his 30s. His mother had been previously diagnosed with sick sinus syndrome and underwent pacemaker implantation. The patient received a dual-chamber pacemaker and experienced complete relief of symptoms.

The patient's son was a healthy teenage boy (age 15 years), and a family unrelated healthy volunteer (male, 26 years) who is an amateur long-distance runner, served as a volunteer control. Resting heart rate was obtained from a 12-lead electrocardiograph (ECG) for P (Figure 1A, HR = 31 bpm, with complete heart block), PS (Figure 1B, HR = 67 bpm), and V (Figure 1C, HR = 50 bpm).

3.2 Direct sequencing of HCN4 gene

Previous studies have established HCN4 mutations as one of the genetic causes in patients with sinus bradycardia. HCN4 gene contains 8 exons.^{10, 34} We used amplification primers (Table S1) to obtain each exon from genomic PCR and subsequent gel electrophoresis, and sequencing primers (Table S2) to sequence the HCN4 gene. We found no missense mutations in HCN4 gene from the patient, the son, and the volunteer.

3.3 Exome-sequencing discovery of HCN1 variants in the patient

To explore a comprehensive genetic cause of the patient's bradycardia, we performed next-generation exome sequencing on genomic DNA isolated from P, PS, and V. In HCN4 gene, we found no missense variant, but did find one known heterozygous synonymous variant (rs529004) in the C-terminal region, shared by PS and V (Figure 2A). In HCN2 gene (another HCN isoform expressed at a very low level compared to HCN1 and HCN4 in the sinus node^{21, 24}), we found no missense, but five known heterozygous synonymous variations (rs55659726, rs56170955, rs55780677, rs12981860, rs1054786).

In HCN1, two variants were found in the patient. One is located in the N-terminus in HCN1 exon_1, rs55852304 (now merged into rs560604803), located in Human hg19: Chr5: 45695986-45695995. It is an in-frame deletion (indel), GCCGCCGCC, which encodes ⁷²GGG⁷⁴ in the protein sequence (HGVS nomenclature: NM_021072.3:c. 187_195delGGCGGTGGC). This ⁷²GGG⁷⁴ deletion is found in three gene variant databases mentioned above and is shared by the patient's son (Figure 2B). This allele frequency in global population according to 1000Genomes is 2.2% (https://www.ncbi.nlm.nih.gov/snp/rs56064803#frequency_tab).

A missense c.2551C>G transversion variant was found in Human hg19: Chr5: 45262145, in exon8 of HCN1, causing a p. Pro851Ala [CCC (Pro) → GCC(Ala)] missense mutation (Figure 2C). The protein and nucleotide sequences alignments are shown in Figure S1 and S2). This variant is not found in (1) National Heart, Lung, Blood Institute (NHLBI) GO Exome Sequencing Project (ESP) [Exome Variant Server, NHLBI GO Exome Sequencing Project (ESP), Seattle, WA (http://evs.gs.washington.edu/EVS/) [accessed: Jan 11, 2019], containing 6403 samples; (2) Exome Aggregation Consortium (ExAC) (which has about 60 000 individuals) (http://exac.broadinstitute.org/) [accessed: Jan 11, 2019]; and (3) the genome Aggregation Database (gnomAD, an upgraded version of ExAC, covering two datasets: exome sequence data from 123 136 individuals and whole genome sequencing from 15 496 individuals, https://gnomad.broadinstitute.org/gene/ENSG00000164588) [accessed: Jan 12, 2019] (Figure S3). To our best knowledge, P851A is a novel variant that has not been reported before in bradycardia patient.

Sequencing alignment across species on MARRVEL³⁵ indicated that P851 is conserved in the polypeptide sequence of HCN1 of human (hs, Homo sapiens), mouse (mm1, Mus musculus), rat (rn1, Rattus norvegicus), Zebrafish (dr1, Danio rerio), and Fly (dm1, Drosophila melanogaster) (arrows in Figure S4). PolyPhen-2³⁶ predicted that P851A variant is “probably damaging” with a score of 0.999. The Sorting Intolerant From Tolerant (SIFT) algorithm predicts the effects of a missense variant on protein function.^{37, 38} SIFT predicted that P851A is not tolerated (Supplemental excel file HCN1 SIFT, row 62, highlighted).

Missense mutations in the voltage-dependent calcium (Cav1.2, Cav1.3, Cav3.1) and sodium (Nav1.5) channels have been reported to be linked to bradycardia.^39-42 We found no missense mutations in these channel genes SCN5A (Nav1.5), CACNA1C (Cav1.2), CACNA1D (Cav1.3), CACNA1G (Cav3.1, T-type) from the patient. We have also studied sequences of other genes associated with bradycardia (KCNK17, RYR2, CASQ2, AMK2, MYH6),⁴³ and found no missense mutations.

From VCF (variant call format) files, the patient was heterozygous for N-del and P851A. The patient's son has the N-del but not the P851A. Although additional blood samples from the patient's parents and siblings were unavailable, it is almost certain that the two mutations are on different alleles, otherwise, the son would inherit both or neither. Thus, we studied co-expression of N-del + P851A (mimic the patient's variant) and N-del + wtHCN1 (mimic the son's variant).

3.4 Protein expression of HCN1 mutants

We first examined the surface expression of HCN1 mutants transfected into HEK293 cells. In membrane preparations, a human HCN1 antibody detected protein expression of WT, N-del, and P851A HCN1 channels at the expected molecular weight (~120 kDa) (Figure 3A). Co-expression of wide-type HCN1 with the mutant channels (N-del + wtHCN1 for PS, N-del + P851A for P) are shown in Figure 3B.

On average, there was over five-fold reduction in surface expression of N-del, compared to the wild-type HCN1 channel protein, after being normalized to the loading control (Figure 3C) (WT: 2.01 ± 0.830, n = 7; N-del: 0.36 ± 0.17; n = 6, P < .001). There was a three-fold reduction of surface expression in N-del + P851A compared to the wildtype channel (WT: 2.01 ± 0.830, n = 7; N-del + P851A: 0.67 ± 0.28; n = 4, P < .01). There was no statistical difference in surface expression levels of P851A comparing to the wildtype channels (WT: 2.01 ± 0.830, n = 7; P851A: 1.11 ± 0.55; n = 4, P > .05) and N-del + wtHCN1 (WT: 2.01 ± 0.830, n = 7; N-del + wtHCN1: 1.04 ± 0.44; n = 4, P > .05) channel proteins (Figure 3C).

Total channel protein levels were examined in the whole lysate preparations. N-del (Figure 3D) and N-del + P851A (Figure 3E) showed a reduction in channel protein levels compared to wildtype HCN1. The channel protein levels are decreased by ~12-fold for N-del in comparison to the WT, while changes in P851A, N-del + P851A, and N-del + wtHCN1 were statistically not significant (Figure 3F) (WT: 1.86 ± 0.56, n = 6; N-del: 0.15 ± 0.04, n = 6; P851A: 1.80 ± 0.81, n = 5; N-del + P851A: 1.21 ± 0.44, n = 6; N-del + HCN1wt: 1.12 ± 0.40, n = 5).

Additionally, we used flow cytometry to detect the surface expression of channel proteins at the single cell level. A hemagglutinin (HA) tag in the extracellular loop between S3 and S4 transmembrane domains was introduced into the HCN1 wild-type and mutant constructs (WT, P851A, N-del). We used an HA-antibody conjugated with Alexa647 to stain surface expression of HCN1 mutants in cells without permeabilization. For intracellular staining, we used an HCN1-antibody conjugated with phycoerythrin (PE) fluorophore after cells were permeabilized. The HCN1 antibody targets an intracellular epitope (amino acid residues 790-883) in the C terminus of the protein (Thermo Scientific, PA553950). GFP-positive cells that bind to HA-APC antibody prior to permeabilization were counted (Figure 4A). The fluorescent signals are contributed by HA-tagged HCN1 channels expressing at the cell surface. The total number of channels at cell surface is the product of the count and the median or geometric mean of HA-APC fluorescent signals (Figure 4B). Surface expression of N-del + wtHCN1 channels are decreased by 69% when normalized to WT (WT: 1, N-del + wtHCN1: 0.31 ± 0.09; n = 5, P < 001). Expression levels of N-del + P851A channels at cell surface are also decreased, but to lesser extent, 24% (WT: 1; N-del + P851A: 0.76 ± 0.11; P < .01, n = 5). There is no statistical difference between the surface expression levels of P851A and WT HCN1 channels (WT: 1; P851A: 0.92 ± 0.17; P > .05, n = 5). Figure S5 shows that surface expression of N-del channels is decreased by 71% (WT: 1; N-del: 0.29 ± 0.11; P < .001, n = 5), but the surface expression of P851 channels is not statistically different from wild-type channel (WT: 1; P851A: 0.87 ± 0.31; P > .05, n = 5).

3.5 Confocal imaging studies of HCN1 mutants

We also performed confocal imaging studies of HCN1 mutant channels co-expressed with the wild-type HCN1 in HEK293 cells. Representative channel distributions (green) are shown for the volunteer (wtHCN1) (Figure 4C), the patient (N-del + P851A) (Figure 4D), and the patient's son (wtHCN1 + N-del) (Figure 4E). The relative channel distribution was estimated by calculating the total number of the expressed channels divided by the total cell area from the brightfield image, the averaged results are shown in Figure 4F. Compared to the wild-type HCN1, the total number of channels was >fivefold decreased for wtHCN1 + P851A and >3-fold decreased for wtHCN1 + N-del (wtHCN1: 0.54 ± 0.20, N-del + wtHCN1: 0.16 ± 0.02, N-del + P851A: 0.095 ± 0.079, n = 6, P < .05). There was no significant difference in the average ratio of GFP area to total cell area between N-del + P851A and N-del + wtHCN1 using one-way ANOVA followed by post hoc test Tukey (N-del + wtHCN1: 0.155 ± 0.0593, N-del + P851A: 0.0958 ± 0.0884, P > .05) (Figure 4F).

3.6 Biophysical properties of HCN1 mutants

Studies of HCN4 mutations have revealed two mechanisms by which missense mutations cause reduced channel activity: reduced channel expression and/or a negative shift of voltage-dependent activation.¹⁸ Therefore, we investigated the threshold activation (the least negative potential at which the HCN channel inward current of 10pA can be detected³¹) and the current density (the whole-cell current divided by cell capacitance), which reflects the functional channels of HCN1 mutants available at the cell surface.

Figure 5 shows the representative current activation of wild-type human version HCN1 (wt-hHCN1) and the mutant channels expressed in HEK293 cells. Channel-expressing cells were voltage-clamped at −20 mV and hyperpolarized to a range of potentials marked in the figure. Compared to wild-type HCN1 (Figure 5A), N-del (Figure 5B), P851A (Figure 5C), N-del + P851A (Figure 5D), and N-del + wtHCN1 (Figure 5F) mutant channels were activated at more negative potentials. The average threshold activation is −52.0 ± 3.3 mV (n = 7) for WT, −65.0 ± 7.1 mV (n = 10) for P851A, −67.5 ± 4.5 mV (n = 10) for N-del, −68.8 ± 8.3 mV (n = 7) for N-del + wtHCN1, and −107.5 ± 12.2 mV (n = 12) for N-del + P851A, respectively (Figure 6A). There exists a statistically significant negative shift in the threshold activation for P851A (−13 mV), for N-del (−15 mV), for N-del + wtHCN1 (−16 mV), and for N-del + P851A (−55 mV), respectively, when compared to the wild-type HCN1 (Table 1). There was a significance difference of HCN1 mutant channel threshold activation between N-del + P851A and N-del + wtHCN1 using one-way ANOVA followed by post hoc test Tukey (N-del + wtHCN1: −68.75 ± 8.35 mV, N-del + P851A: −107.5 ± 12.15 mV, P < .05) (Figure 6A).

The half-voltage activation of mutant channels is also significantly shifted to hyperpolarizing potentials compared to wild-type (Table 1): N-del by −19 mV, P851A by −23 mV, N-del + P851A by −49 mV, and N-del + HCN1wt by −12 mV, respectively. Wild-type HCN1 channel has fast activation kinetics (100-300 milliseconds near threshold voltage of activation).^{8, 44} All mutant channels have slower activation kinetics near their respective threshold voltage of activation (τ_Vth) (Table 1), notably, N-del + P851A has the slowest activation kinetics near the voltage threshold of activation.

We next measured the current density (current amplitude normalized to the cell capacitance, pA/pF) of HCN1 mutant channels at the voltage where the channels are fully activated to calculate the relative functional channels available at the cell surface. Compared to WT, there is a 47% reduction of current density for N-del (WT (−110 mV): 27.3 ± 2.0 pA/pF, n = 5; N-del (−130 mV): 14.4 ± 2.7 pA/pF, n = 7), a 51% reduction of current density for N-del + P851A (WT (−110 mV): 27.3 ± 2.0 pA/pF, n = 5; N-del + P851A (−140 mV): 13.5 ± 3.4 pA/pF, n = 7), and a 44% decrease in current density for N-del + HCN1wt (WT (−110 mV): 27.3 ± 2.0 pA/pF, n = 5; N-del + HCN1wt (−110 mV): 15.3 ± 3.3 pA/pF, n = 5). On the other hand, P851A showed no statistical decrease in current density compared to WT HCN1 channels (WT (−110 mV): 27.3 ± 2.0 pA/pF, n = 7; P851A (−130 mV): 21.6 ± 1.9 pA/pF, P > .05, n = 8) (6B, Table 1). There was no significance difference of HCN1 mutant channel current density between N-del + P851A and N-del + wtHCN1 using one-way ANOVA followed by post hoc test Tukey (N-del + wtHCN1: 15.33 ± 3.33 pA/pF, N-del + P851A: 13.47 ± 3.39 pA/pF, P > .05) (Figure 6B).

One of the hallmarks of HCN channels is the functional channel blockade by ZD7288, a selective blocker for HCN pacemaker channels.⁴⁵ Figure 7 shows the ionic current inhibition of HCN1 wild-type (A), mutant channels (B, N-del; C, P851A), co-expression of N-del and P851A (D), and co-expression of N-del with wild-type channel (E) by 10 μM ZD7288 at the respective membrane potential.

3.7 Effect of HCN1 variant in the sinoatrial pacemaker activity

To evaluate the effect of HCN1 variant in the sinoatrial pacemaker activity, we employed an updated sinoatrial pacemaker activity cell model developed by DiFrancesco's group.³³ HCN1 protein levels are >15-fold higher than HCN4 and HCN2 in the sinus node.²⁷ We assumed that removing the main contributor HCN1 result in one-third decrease in I(f) channel conductance; the simulation result showed that the rate of spontaneous action potential is decreased by about 50% because of reduction in the initial phase 4 depolarization speed (control: ~110 mV/s, HCN1 variant: ~60 mV/s) (Figure 8, A-control action potential (upper) and I(f) (lower), B-reduced rate of action potential (upper) and I(f) (lower)).