Down-regulation of miR-1/miR-133 Contributes to Re-expression of Pacemaker Channel Genes HCN2 and HCN4 in Hypertrophic Heart*♦

Cardiac hypertrophy is characterized by electrical remolding with increased risk of arrhythmogenesis. Enhanced abnormal automaticity of ventricular cells contributes critically to hypertrophic arrhythmias. The pacemaker current If, carried by the hyperpolarization-activated channels encoded mainly by the HCN2 and HCN4 genes in the heart, plays an important role in determining cardiac automaticity. Their expressions reportedly increase in hypertrophic and failing hearts, contributing to arrhythmogenesis under these conditions. We performed a study on post-transcriptional regulation of expression of HCN2 and HCN4 genes by microRNAs. We experimentally established HCN2 as a target for repression by the muscle-specific microRNAs miR-1 and miR-133 and established HCN4 as a target for miR-1 only. We unraveled robust increases in HCN2 and HCN4 protein levels in a rat model of left ventricular hypertrophy and in angiotensin II-induced neonatal ventricular hypertrophy. The up-regulation of HCN2/HCN4 was accompanied by pronounced reduction of miR-1/miR-133 levels. Forced expression of miR-1/miR-133 by transfection prevented overexpression of HCN2/HCN4 in hypertrophic cardiomyocytes. The serum-responsive factor protein level was found significantly decreased in hypertrophic hearts, and silencing of this protein by RNA interference resulted in increased levels of miR-1/miR-133 and concomitant increases in HCN2 and HCN4 protein levels. We conclude that down-regulation of miR-1 and miR-133 expression contributes to re-expression of HCN2/HCN4 and thereby the electrical remodeling process in hypertrophic hearts. Our study also sheds new light on the cellular function and pathological role of miR-1/miR-133 in the heart.

The adult heart is susceptible to stress (such as hemodynamic alterations associated with myocardial infarction, hypertension, etc.) by undergoing remodeling process and hypertrophic growth to adapt to altered workloads and impaired cardiac function. The remolding process in hypertrophic hearts includes electrical remodeling that is characterized by a reprogramming of cardiac gene expression and the reactivation of "fetal" cardiac genes, with increased risk of arrhythmogenesis partially due to enhanced automaticity.
In the mammalian cardiac sinus node, the pacemaker current, termed hyperpolarization-activated current (I f ), plays a crucial role in setting the heart rate and sensing its autonomic control. In addition to the automatic cells, I f also exists in nonautomatic (non-pacemaker) regions of the heart, such as atria and ventricles, where it contributes to abnormal automatic activities (1)(2)(3). In ventricular myocytes, I f is abundantly expressed during fetal and neonatal life but progressively decreases toward adulthood in terms of the number of cells expressing I f and of the density of expressed I f as well (4 -8). This results in a loss of the capacity of adult ventricular cells to generate spontaneous activity. Strikingly, substantial up-regulation of I f expression has been observed in a variety of animal models of cardiac hypertrophy and heart failure and in human failing hearts as well (1-3, 9 -16). In such circumstances, I f is at least doubled in left ventricular (LV) 3 cells, depending upon severity of conditions (3,11,14), and reaches values comparable with those observed in the neonatal stage. Consistently, prominent re-expression of the I f -encoding genes, belonging to the hyperpolarization-activated cyclic nucleotide-gated channel (HCN) gene family in ventricular myocytes of hypertrophic heart has also been documented (17)(18)(19). The HCN family is composed of at least four members (HCN1-HCN4), of which HCN4 is believed to be the major molecular component for sinus I f and HCN2 is believed to be the dominant isoform for ventricular I f . However, mechanisms underlying the pathological re-expression of I f and HCN2/HCN4 remained yet to be elucidated.
Discovery of microRNAs (miRNAs) has revolutionized our understanding of the mechanisms that regulate gene expression (20 -22). miRNAs are endogenous ϳ22-nucleotide noncoding RNAs that anneal to inexact complementary sequences in the 3Ј-UTRs of target mRNAs of protein-coding genes to specify translational repression or/and mRNA cleavage. Among Ͼ500 mammalian miRNAs identified thus far, miR-1 and miR-133 are believed to specifically express in adult cardiac and skeletal muscle tissues where miR-1 and miR-133 importantly regulate myogenesis (21,23,24). Recent studies demonstrated that both miR-1 and miR-133 are significantly downregulated in hypertrophic and failing hearts (25)(26)(27), and down-regulation of either miR-1 or miR-133 can determine the pathogenesis of cardiac hypertrophy. More strikingly, correction of the down-regulation by forced expression of these miRNAs reversed the pathological process. This study was designed to shed light on the underlying mechanisms for re-expression of HCN2 and HCN4 in hypertrophic heart, focusing on the potential role of miR-1 and miR-133 in the post-transcriptional repression of HCN2 and HCN4 genes.
Construction of Luciferase Reporter Vector Carrying miRNA Target Site-To construct reporter vectors bearing miRNA target sites, we synthesized (by Invitrogen) fragments containing the exact target sites for miR-1 and miR-133 or the mutated target sites, the 3Ј-UTRs of HCN2 and HCN4 without or with mutations by PCR amplification (see Fig. 1, B and C). The oligonucleotides were inserted into the cloning sites downstream of the luciferase gene in the pMIR-REPORT TM luciferase miRNA expression reporter vector (Ambion, Inc.) (28 -31).
Mutagenesis-Deletion mutations and base-substitution mutations were created by direct oligomer synthesis or by PCRbased methods. Mutations were made to the 3Ј-UTRs of HCN2 and HCN4 genes (see Fig. 1D) (28 -31).
Cell Culture-HEK293 (human embryonic kidney cell line) used in this study was purchased from American Type Culture Collection (ATCC, Manassas, VA) and cultured in Dulbecco's modified Eagle's medium.
Hypertrophic Model-LV hypertrophy was induced by an aortic stenosis technique, similar to that described by Fernández-Velasco et al. (3,12). Briefly, adult male Sprague-Dawley rats weighing 180 -200 g were used. The animals were anesthetized via an intraperitoneal injection of sodium pentobarbital (50 mg/kg). A silver clip was placed around (0.3-mm aperture) the abdominal aorta above the left renal artery. The LV pressure was progressively developed as the animals grew, and LV hypertrophy was established at 8 weeks after surgery. Sham-operated and age-matched rats were used as control. Hypertrophied hearts were defined as the ratio of heart weight (wet) to body weight Ͼ25% of control hearts. The use of animals was in accordance with the guidelines of the Institutional Animal Care and Use Committee of the Harbin Medical University.
Myocyte Isolation and Primary Cell Culture-Neonatal rat ventricular cardiomyocytes were isolated and cultured as described previously (28).
Angiotensin II-induced Hypertrophy and Cell Surface Area Analysis of Cardiomyocytes-Isolated neonatal rat ventricular cells were cultured for 24 h in serum-containing medium after which they were serum-starved for 24 h before treatment. To induce hypertrophy, cells were stimulated by angiotensin II (AngII, 1 M) for 36 h in serum-free culture medium (32,33). Cell surface area was analyzed using a Leica inverted microscope equipped with a Polaroid digital camera at ϫ200 magnification.
Small Interference RNAs (siRNAs) Synthesis and Treatment-An siRNA directed against serum-response factor (SRF) was designed to target the coding region of the SRF mRNA (position 1864, TGGGACAGTGCAGATCCCTGT-TTCA): 5Ј-UGAAACAGGGAUCUGCACUGUCCCA-3Ј (SRF-siRNA). A negative control siRNA (NC siRNA) was also designed by replacing eight nucleotides as indicated below by the bold and italic letters: 5Ј-UTCCGTAGGGAU-CUTAACUGUCCCC-3Ј. These siRNAs were synthesized by Invitrogen. The siRNAs were delivered into cultured neonatal rat ventricular cells by Lipofectamine 2000 (Invitrogen). Cell lysates were prepared 24 h after transfection for Western blot and real-time RT-PCR analyses.
Real-time RT-PCR-The mirVana TM qRT-PCR miRNA detection kit (Ambion) was used in conjunction with real-time PCR with SYBR Green I for quantification of miR-1 and miR-133 transcripts in our study, as previously described in detail (28 -31).
Western Blot-The procedures for protein extracted from neonatal rat ventricular myocytes were the same as described in detail elsewhere (31). Membrane protein samples were extracted. Affinity-purified rabbit polyclonal anti-HCN2 and anti-HCN4 (Alomone Labs, Jerusalem, Israel) antibodies were used as the primary antibodies. The anti-ANF (atrial natriuretic factor), anti-␤-MHC (␤-myosin heavy chain), anti-MLC-2 (myosin light chain-2), and anti-SRF (serum-responsive factor) rabbit polyclonal antibodies (Santa Cruz Biotechnology Inc., Santa Cruz, CA) were used as the primary antibodies.
Data Analysis-Group data are expressed as mean Ϯ S.E. Statistical comparisons (performed using analysis of variance followed by Dunnett's method) were carried out using Microsoft Excel. A two-tailed p Ͻ 0.05 was taken to indicate a statistically significant difference.

RESULTS
Post-transcriptional Repression of HCN2 and HCN4 Expression by miR-1 and miR-133-Neither HCN2 nor HCN4 is listed as candidate targets for miR-1 and miR-133, according to the prediction by TargetScan hosted by Wellcome Trust Sanger Institute (34). However, by detailed analysis of the 3Ј-UTR of HCN2, we identified one putative target site for miR-1 and one for miR-133, based on complementarity to the 5Ј-ends of the miRNAs (Fig. 1B). The 3Ј-UTR of HCN4 does not have any sites with more than four complementary nucleotides to the 5Ј-ends of miR-1 and miR-133, neither does the 5Ј-UTR nor the coding region of HCN4; however, it does contain a stretch of sequence bearing extensive complementarity to the middle portion of miR-1 (Fig. 1C).
Perturbation of miRNA expression, including overexpression and silencing, is a powerful approach to study miRNA function (35). Transient overexpression of miRNAs in cellbased assays can be achieved by transfection of doublestranded RNA molecules that mimic the Dicer cleavage product. Moreover, the anti-miRNA AMOs specifically and stoichiometrically bind, and efficiently and irreversibly silence, their target miRNAs by competitive binding to miRNAs and by degrading them as well with unknown mechanisms. To verify that HCN2 is indeed the cognate target of both miR-1 and miR-133 for post-transcriptional repression, we took the following approaches.
We first inserted the 3Ј-UTR of HCN2 into the 3Ј-UTR of a luciferase reporter plasmid containing a constitutively active promoter to determine the effects of the miRNAs on reporter expression. Co-transfection of the chimeric luciferase-HCN2 3Ј-UTR vector with either miR-1 or miR-133 ( Fig. 2A) into HEK293 cells consistently demonstrated smaller luciferase activities relative to transfection of the chimeric plasmid alone or co-transfection with the mutant miR-1 or mutant miR-133 failed to produce any effects. Co-application of miR-1 with AMO-1 (anti-miR-1 antisense) or miR-133 with AMO-133 (anti-miR-133 antisense) eliminated the silencing effects on luciferase reporter activities. miR-1 elicited significant repression on luciferase activity with the 3Ј-UTR of HCN4 (Fig. 2B) despite the fact that the 3Ј-UTR of HCN4 matches the center portion but not the 5Ј end of miR-1. As expected, miR-133 did not produce any effects on luciferase expression.
We then observed that the mutated target sequences of HCN2 fused to the 3Ј-UTR of luciferase were not responsive to miR-1 and miR-133; however, they responded efficiently to the mutant miR-1 or the mutant miR-133 for the mutations generated exact complementarity between the mutant miRNAs and the mutant 3Ј-UTR of HCN2. Similarly, miR-1 failed to affect the luciferase activity when the putative site for miR-1 in the 3Ј-UTR of HCN4 was mutated, but the mutant miR-1 that matches the mutant HCN4 (Fig. 1D) elicited remarkable reduction of the luciferase activity (Fig. 2C).
To see whether HCN2/HCN4 repression by the miRNAs reported by luciferase assays has any functional implications, . B and C, complementarity between miR-1/miR-133 and their target sites in the 3Ј-UTRs of HCN2/HCN4. Both human and rat sequences are shown because our experiments were conducted with rat cells, and the GenBank TM accession numbers are included in the brackets. D, mutation of human miR-1 and miR-133 and the complementarities with the mutated target sites in the 3Ј-UTRs of HCN2/HCN4. Boldface letters indicate that the seed site is critical for miRNA-mRNA binding and interaction and the miRNA::mRNA base pairings; lowercase letters indicate that the base substitution mutations made to the sequences; italics in the AMOs indicate the locked nucleotides (the ribose ring is constrained by a methylene bridge between the 2Ј-O and the 4Ј-C atoms to confer a higher thermal stability, discriminative power, and a longer half-life). AMO-1 and AMO-133 target miR-1 and miR-133, respectively. Note that the mutation in the target sites of HCN2/HCN4 3Ј-UTRs was made so that it loses the complementarity to the wild-type miRNAs but meanwhile creates complementarity to the mutant miRNAs. Two miR-1 mutants were made, one for HCN2 (mutant miR-1a) and the other for HCN4 (mutant miR-1b).
we determined the effects of the miRNAs on endogenous expression of HCN2/HCN4 at the protein level in neonatal rat ventricular myocytes by Western blot analysis. Transfection of either AMO-1 or AMO-133 at 100 nM did not cause significant changes of HCN2 protein level (Fig. 2D). Co-transfection of AMO-1 and AMO-133, however, was able to increase the level by ϳ87 Ϯ 6.5% (p Ͻ 0.05). By comparison, application of AMO-1 alone increased HCN4 protein level by 46 Ϯ 3.7%, and AMO-133 did not produce any effects (Fig. 2E). Co-transfection of AMO-1 and AMO-133 produced a similar effect as AMO-1 alone. Transfection of miR-1 or miR-133 alone pro-duced remarkable depression of HCN2 (Fig. 2D), but only miR-1 and not miR-133 decreased HCN4 level (Fig. 2E). On the other hand, AMO-1 and AMO-133, transfected alone or together, failed to produce any significant effects on HCN2 and HCN4 mRNA levels (Fig. 2, F and G), indicating that the miRNAs do not affect HCN2/HCN4 mRNA stability.
Finally, to confirm that the observed effects of AMO-1 and AMO-133 were ascribed to their anti-miR-1 or anti-miR-133 actions, we analyzed the changes of endogenous miR-1 and miR-133 upon transfection of the AMOs since AMOs have been shown to degrade their target miRNAs (28 -31, 35). Our data indeed support this notion; as shown in Fig. 2H, AMO-1 and AMO-133 significantly knocked down the levels of miR-1 and miR-133, respectively. For cross comparisons, AMO-1 did not affect miR-133 level, and neither did AMO-1 affect miR-1 level. Moreover, to ensure the specificity of the probes (primers) for quantifying miR-1 and miR-133 levels, we transfected the mutant miR-1 and miR-133 constructs (Fig. 1A) into neonatal cells. Our data showed that the mutants did not cause any significant alterations of miR-1 and miR-133 levels relative to non-transfected cells.

Role of miR-1 and miR-133 in Overexpression of f-Channel Genes in Hypertrophic Rat Left Ventricle (HLV)-HCN2
expression at the protein level was elevated by 3-fold and HCN4 protein was found 1.7-fold increased in HLV (Fig. 3A). To explore the potential roles of miR-1 and miR-133 in hypertrophic overexpression of HCN2/HCN4, we first quantified the levels of miR-1 and miR-133. The miR-133 level in HLV was found decreased by ϳ65% relative to the value from normal hearts (Fig. 3B). miR-1 level was also decreased but to a lesser extent (ϳ35%).
We went on to evaluate the possibility of reversing the hypertrophic increases of HCN2/HCN4 expression by correcting miR-1 and miR-133 levels in AngII-induced hypertrophic neonatal rat ventricular cardiomyocytes. AngII markedly increased HCN2 and HCN4 protein levels (Fig. 4A). These increases were abolished with co-transfection of cells with miR-1 and miR-133 together. In addition, the effects of the miRNAs were prevented by co-application of AMO-1 and AMO-133 to degrade miR-1/ miR-133. Consistent with the up-regulation of HCN2/HCN4 proteins, AngII also decreased miR-1 and miR-133 levels the neonatal cardiomyocytes (Fig. 4B). To confirm the transfection efficiency of miR-1 and miR-133, we demonstrated that exogenously applied miR-1/miR-133 indeed tremendously increased the cellular levels of miR-1/miR-133 (Fig. 4B).
Role of Decreased SRF in Overexpression of HCN2 and HCN4 in Hypertrophic Hearts-It has been shown that expression of miR1/miR-133 is dependent upon binding of SRF to their promoter regions (23,24). We therefore investigated the role of SRF in linking miR1/miR-133 to HCN2/HCN4 expression. SRF expression at both mRNA and protein levels was found significantly decreased in hypertrophic left ventricles relative to healthy hearts of adult rats (Fig. 5A). To verify that SRF indeed produces down-regulation of miR-1/miR-133 expression, we assessed the effects of silencing of SRF by an siRNA directed against SRF (SRF-siRNA) on miR-1/miR-133 levels in cultured neonatal rat ventricular myocytes. As shown in Fig. 5B, SRF-siRNA significantly reduced miR-1 and miR-133 levels. Con- comitantly, SRF-siRNA treatment robustly increased HCN2 and HCN4 protein levels (Fig. 5C). The efficacy of SRF-siRNA to silence SRF expression was verified at both mRNA and protein levels in rat cells (Fig. 5D). The negative control siRNA (NC siRNA) failed to alter miR-1/miR-133, HCN2/HCN4, and SRF expressions (Fig. 5, B-D). As an additional control for the SRF-siRNA actions, we performed experiments testing the effects of the SRF-siRNA on the mRNA levels of two housekeeping genes GAPDH and Sp1 (stimulating protein) (36). As shown in Fig. 5E, the SRF-siRNA failed to alter the mRNA levels of these genes. In addition, as a positive control, we also assessed the effects of the SRF-siRNA on the mRNA level of ␣-MHC since it has been shown that the promoter region of ␣-MHC gene contains an SRF binding site and that SRF can activate ␣-MHC transcription (37). The data in Fig. 5F confirm that the SRF-siRNA indeed reduced the transcript level of ␣-MHC, whereas the NC siRNA failed to do so. These results are consistent with the long recognized down-regulation of ␣-MHC in hypertrophic hearts in light of the fact that SRF is also down-regulated in hypertrophic heart as shown in this study (Fig. 5A).
Finally, to confirm that the myocytes treated with angiotensin II indeed developed hypertrophic conditions, we measured the changes of cell size and expression of several biomarkers for cardiac hypertrophy including ␤-MHC, MLC-2, and ANF (38). As depicted in Fig. 6, A and B, the cell surface area was significantly enlarged, and ␤-MHC, MLC-2, and ANF were all up-reg-  ulated in their expression. To exclude the possibility of repression of ␤-MHC, MLC-2, and ANF by miR-1/miR-133 underlying the hypertrophic changes of HCN2/HCN4 expression, we demonstrated the lack of ability of miR-1/miR-133 to alter the protein levels of ␤-MHC, MLC-2, and ANF in cultured rat myocytes (Fig. 6C). In agreement, bioinformatics analysis using miRBase and miRanda websites revealed the absence of miR-1 or miR-133 binding sites in the whole sequences of ␤-MHC, MLC-2, and ANF genes.

DISCUSSION
One of the characteristic alterations of hypertrophic hearts is the re-expression or overexpression of pacemaker channel genes, and thereby I f current, which likely contributes importantly to the enhanced abnormality and the risk of arrhythmias under such pathological conditions (1-3, 9 -16). However, the mechanisms remained poorly understood. In this study, we demonstrated significant overexpression of HCN2 and HCN4 in LV hypertrophy induced by an aortic stenosis technique and in neonatal cardiomyocyte hypertrophy induced by angiotensin II. We experimentally established HCN2 and HCN4 genes as targets for post-transcriptional repression by the muscle-specific miRNAs miR-1 and miR-133. The up-regulation of HCN2 and HCN4 was accompanied by substantial decreases in miR-1/miR-133 mRNA levels. Moreover, correction of miR-1/miR-133 down-regulation prevented overexpression of HCN2/ HCN4 in hypertrophy myocytes. We conclude that downregulation of miR-1 and miR-133 expression contributes to the re-expression of HCN2/HCN4, and thereby the electrical remodeling process in hypertrophic hearts. Our study also sheds new light on the cellular function of miR-1 and miR-133 in the heart and on the regulation of ion channel genes by these miRNAs.  miR-1 and miR-133 are expressed in a chamber-specific manner during cardiogenesis and are activated during the period of differentiation (23,24). Increasing expression of miR-1 and miR-133 has been found in neonatal hearts, and substantially higher levels are maintained in adult cardiac tissues (24). This may at least partially account for the decreased I f density in adult ventricular cells (1-3) for enhanced miR1-/ miR-133 expression limits HCN2/HCN4 expression. When this limiting factor becomes weaker as in hypertrophic myocytes where miR1-/miR-133 levels drop, HCN2/HCN4 re-express. The inhibitory effects of miR-1/miR-133 on HCN2/ HCN4 expression are primarily ascribed to posttranscriptional inhibition of translation of HCN2/HCN4 mRNAs into their respective protein products. Decreases in miR-1 and miR-133 hypertrophic hearts found in this study are consistent with the recent studies (25)(26)(27). The present study represents a novel insight into the pathophysiological role of miR-1/miR-133 by regulating HCNs in the heart.
Our results indicate that as suggested previously, complementarity with six consecutive nucleotides matching the 5Ј-end of a miRNA is adequate for a target mRNA to interact with the miRNA leading to translational repression (22). This form of complementarity has been described as "seed site." Notably, the 3Ј-UTR of HCN4 does not contain such complementary sequence; instead, it has a stretch of nucleotides well complementary to the middle portion of miR-1, and this form of base pairing obviously is sufficient for miR-1 to bind to HCN4 and to elicit the consequence. The fact that such a complementarity confers similar degrees of repression as seen with the seed site strongly suggests that besides the parameters of base pairing and free energy, there are some other yet-to-bedefined factors contributing to miRNA-mRNA interactions. Noticeably, to obtain significant effects on the endogenous HCN2 protein level, co-transfection of AMO-1 and AMO-133 is required, and either of the two AMOs alone failed to affect HCN2 level. This is likely ascribed to the fact that HCN2 is simultaneously repressed by miR-1 and miR-133, and removal of either of the miRNAs is not sufficient to reverse the repressive effect produced by the remaining miRNA; only when both miRNAs are concomitantly silenced can the repression be relieved.
The present study aimed at understanding the underlying mechanisms for the re-expression of HCN2 and HCN4 in hypertrophic conditions, and our findings indicate the altered expression of miR-1/miR-133 as one of the important factors. The lack of effects of miR-1 and miR-133 on expression of ANF, ␤-MHC, and MLC-2 indicates that the link between downregulation of miR-1/miR-133 and up-regulation of HCN2/ HCN4 are likely direct and that other hypertrophic factors may not be directly involved in the relationships. It should be emphasized, however, that regulation of HCN2/HCN4 expression must be a complex process, and our study merely suggests the potential contributions of miR-1/miR-133 to this process but does not at all exclude other regulatory factors.
We have also quantified the changes of HCN2 and HCN4 expression at the mRNA level in hypertrophic hearts. Our data showed that the mRNA levels of HCN2 and HCN4 were also elevated but to much lesser extents relative to the increases at the protein level, with 172 Ϯ 21 and 50 Ϯ 13% increases in HCN2 and HCN4 mRNA levels, respectively, when compared with 300 Ϯ 26 and 173 Ϯ 12% increases in their respective protein levels. This discrepancy can be easily explained by a relief of the post-transcriptional repression of HCN2 and HCN4 expression due to down-regulation of miR-1 and miR-133. However, our study does not provide the mechanisms for the up-regulation of HCN2 and HCN4 at the transcriptional level.
Our study also revealed down-regulation of SRF expression as a potential mechanism for the up-regulation of HCN2 and HCN4 expression via down-regulation of miR-1/miR-133 in hypertrophic myocytes. Expression of miR-1/miR-133 has previously been shown to critically depend upon SRF binding to their promoter regions (23,24). In agreement with the changes of SRF expression in hypertrophic heart, in failing hearts of both humans and rabbits, a substantial decrease in expression of full-length SRF accompanied by robust expression of a low molecular-mass dominant negative SRF isoform has been documented (39,40).
Our findings indicate that forced expression of miR-1 and miR-133 might be a novel approach for suppressing overexpression of f-channels and the associated arrhythmias in hypertrophic hearts. An alternative strategy is the so-called miRNA Mimic techniques (miR-Mimic) that we have developed recently (31). This approach employs artificially designed nonnatural miRNAs complementary specifically to the target genes such as HCN2 and HCN4. We have demonstrated the efficacy of the miR-Mimics for HCN2 and HCN4 in repressing f-channel expression and function in a gene-specific manner.