| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Fraunhofer Institute of Toxicology and Experimental Medicine, Center for Drug Research and Medical Biotechnology, 30625 Hannover, Germany
1 Correspondence: Fraunhofer Institute of Toxicology and Experimental Medicine (ITEM), Center for Drug Research and Medical Biotechnology, Nicolai-Fuchs-Str. 1, D-30659 Hannover, Germany. E-mail: Borlak{at}item.fraunhofer.de
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: ion channels • gene expression • repressors • heart failure • assist device
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2)
. Altered expression of ion channels and calcium regulatory proteins is likely to play a major role in the onset and/or progression of arrhythmic events; frequently patients suffer from ventricular extrasystoles, tachyarrythmias, atrioventricular blockings, atrial fibrillation, and prolongation of action potential.
Several lines of evidence are suggestive for single ion channels to be down-regulated in heart failure (3
4
5)
, but a comprehensive study of the transcriptional regulation of major inward and outward ion channels in various regions (atria, ventricles) of the human heart, as well as genes coding for calcium binding proteins, ion receptors, ion exchangers, calcium ATPases, and calcium/calmodulin-dependent protein kinases, has not been reported. The selection of genes we investigated was based on their established role in ion homeostasis and the control of heart muscle cell contraction; we studied expression of ANP and BNP for its diagnostic value in cardiac disease and heart failure.
This study aimed for a better understanding of the gene expression of ion channels and other regulatory proteins pertinent to electrical conduction and normal functioning of cardiomyocytes. We link pharmacotherapy and ECG measurements to specific gene expression profiles and propose a correlation that might be causally related. Based on hierarchical gene cluster analysis, we observe expression patterns that can be linked to the various anatomical regions of the heart; the applied algorithm provides valuable information on the regulated networks of genes.
We also studied the expression of transcriptional and translational repressors in end-stage heart failure and performed in silico promotor analysis of deregulated genes to obtain further insight into a potential mechanism of disease.
Finally, we investigated the effects of left ventricular assist devices in patients with ventricular dysfunction and demonstrate novel therapeutic benefit at the transcript level of cardiomyocyte-specific genes.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Immediately after explantation and during heart transplantation, biopsy material of 
5 g was removed from the left and right ventricle and the left and right atria. Excised tissue was immediately shock-frozen in liquid nitrogen and stored at -80°C until analyzed. Patients that suffered from ischemic or dilatative cardiomyopathy (ICM, DCM; New York Heart Association Class III-IV; n=13), of which n = 5 received a left ventricular assist device for >6 months, were included. All patients had abnormal electrocardiograms (ECG) including salves of ventricular extrasystoles, ventricular tachyarrhythmias, AV, and ventricular blocks, and prolongation of the QRS interval (see Table 1
).Wall thickness of hypertrophied cardiac tissue (n=3) was significantly higher than in assist device-supported hearts (1.65 cm±0.25 cm vs. 1.1 cm±0.1 cm). Pharmacotherapy data are reported in Table 1
. Biopsies from LV and RV of normal human hearts (n=3, male, ischemic time <2 h, no macroscopic signs of disease), originally assigned for transplantation, were used as controls (Clontech, Palo Alto, CA, USA). RNA quality and purity of all samples were inspected by capillary electrophoresis using the Bioanalyzer 2100 (Agilent, Palo Alto, CA, USA). Bands of the 18 s and 28 s rRNA were clearly detectable and there was no sign of RNA fragmentation. Thus, RNA was found to be of high quality.
|
RNA and cDNA
RNA was isolated from heart tissue using a total RNA isolation kit (Macherey-Nagel, Germany) according to the manufacturer’s recommendation. Quality of isolated RNA was checked using a 1.0% agarose gel; 2 µg total RNA from each sample was used for reverse transcription. RNA and random primer (Roche, Mannheim, Germany) were preheated for 10 min at 70°C. 5x RT-AMV-buffer, dNTPs (1 mM, Roche, Germany), 40 U RNase inhibitor (Stratagene, Amsterdam, Netherlands), and 20 U AMV-RT (Promega, Mannheim, Germany) were added, then diethyl pyrocarbonate (Sigma, Deisenhofen, Germany) -treated water was added to a final volume of 20 µL. Reverse transcription was carried out for 60 min at 42°C and was stopped by heating to 95°C for 5 min. The resulting cDNA was frozen at -20°C until further experimentation.
Thermocycler RT-PCR
PCR reactions were carried out in a thermal cycler (T3, Biometra, Göttingen, Germany) using the melting, annealing and extension cycling conditions, as shown in Table 2
. DNA contamination was checked for by direct amplification of RNA extracts before conversion to cDNA and any contamination of RNA extracts with genomic DNA could be excluded. PCR reactions were done within the linear range of amplification and amplification products were separated using a 1.5% agarose gel and stained with ethidium bromide. Gels were photographed on a transilluminator (Kodak image station 440; see Fig. 1
).
|
|
Real-time semiquantitative PCR
cDNA was diluted 1:10 with nuclease-free water; 1 µL of the diluted cDNA was added to the Lightcycler-Mastermix (0.5 µM of specific primer, 4mM MgCl2 and 2 µL Master-SYBR-Green, Roche, Mannheim, Germany) and adjusted to a final volume of 20 µL. PCR reactions were carried out with the Lightcycler (Roche, Mannheim, Germany). After an initial denaturation step at 95°C for 30 s, the PCR reaction was initiated with an annealing temperature of 55°C for 8 s, followed by an extension phase for 14 s at 72°C and a denaturation cycle at 95°C for 1 s (for further details and oligonucleotide sequences, see Table 2
). The PCR reaction was stopped after a total of 30 to 40 cycles; at the end of each extension phase, fluorescence was observed and used for quantitative measurements within the linear range of amplification. To exclude unspecific product formation, fluorescence was measured above the primer dimer melting temperature. A melting point analysis was carried out by heating the DNA synthesis product from 65°C to 95°C and a characteristic melting point curve was obtained for each product (data not shown). Exact quantification was achieved by a serial dilution with cDNA produced from heart total RNA extracts using 1:10 dilution steps. Control samples contained water instead of cDNA; occasionally dimer production was seen, but could easily be distinguished by melting point analysis. Gene expression levels were then given as the ratio of the gene of interest (nominator) vs. an established housekeeping gene (cyclophilin A, denominator).
Hierarchical gene cluster analysis
Hierarchical gene cluster analysis was done according to Eisen et al. (ref 6
; http://rana.lbl.gov/). Gene expressions are given as ratios, where the area count of a gene of interest is divided by the area count of the housekeeping gene cyclophilin A. The ratios were log transformed and a hierarchical clustering algorithm produced a table of results wherein 1850 real-time PCR experiments are grouped together based on similarities in their expression patterns and are presented graphically as colored images. The genes are arranged as ordered by the clustering algorithm. The color image is proportional to the ratio. Black squares represent a ratio of 1; gray squares indicate expression levels below the limit of detection.
In silico promoter analysis
We used the human genome browser of UCSC Genome Bioinformatics (Santa Cruz, CA, USA) (http://genome.ucsc.edu/) to identify promoter regions (up to -1500 bases from the start of transcription). This approach was insufficient to annotate the promoter region of SCN5A. We therefore compared the sequence of SCN5A (7)
with all available human genomic sequences using the program BLAST® (Basic Local Alignment Search Tool) from the National Center for Biotechnology Information (NCBI, USA). We identified two mRNA sequences of SCN5A (accession numbers NM_000335 and M77235) and a BAC clone (accession number AC137587.2) with >99% homology. We annotated exon 1 of SCN5A and interrogated the region -1500 bases upstream up to start of exon 1 as part of our in silico promoter analysis. After curation of the sequence (base 114504 to base 116003 of the BAC clone AC137587), we used the program PromotorInspector (Genomatix Software GmbH, Germany), which specifically predicts polymerase II promoter regions within an unknown genomic DNA sequence (8)
. This approach provided evidence for high-quality promotor prediction (base 114504 to base 114826). We searched for promotor-specific transcription factor binding sites (CCAAT box, SRF, SP1, TATA box, USF, USF2) by applying the program Matrix Search for Transcription Factor Binding Sites (MATCHTM1.7; Biobase, Germany). Our previously constructed matrix (5'-NGN(A/T)CN(G/T)NNAGTTCT-3'; 9) was used to search for androgen-responsive elements in promoters of deregulated genes using the MATCHTM 1.7 program of the TRANSFAC® Professional 6.4 database (http://www.biobase.de/cgi bin/biobase/transfac/).
Statistics
Statistical analysis was done using the 2-tailed t test and differences were considered to be significant at P <0.05 or P <0.01. We performed linear regression analysis and determined correlation coefficients using the program Excel 2000 (Microsoft Seattle, WA, USA).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 11)
. We show highly significant induction of ANP and BNP gene expression in end-stage heart failure. ANP gene expression was less in patients with LVAD implants (see Fig. 1
and Fig. 2
).
|
Ion channels
In the following, we report the expression of genes coding for ion channels and other regulatory proteins in end-stage heart failure according to their importance in cardiac conductivity and ion homeostasis (see Fig. 3
). We use the nomenclature of The IUPHAR Compendium of Voltage-gated Ion Channels 2002 (12)
and discuss our findings based on the role of ion channels in action potential (see Fig. 3
).
|
Phase 1 of the action potential (inward and transient outward currents; see Fig. 3
)
In tissue extracts of diseased LVs gene expression of Nav1.5 (voltage-gated sodium channel) and of Cav1.2 (L-type Ca2+ channel) was reduced to 30–50% (P<0.01) of controls, whereas no significant change was obvious with LVADs. No transcript of Cav3.2 (T-type Ca2+ channel) could be detected in healthy or diseased ventricular tissue, but its expression was evident in some atria and assist device-supported LV tissues. Gene expression of Kv4.3 (transient outward channel Ito) was not detectable in diseased LV tissue and decreased to 15% of controls in diseased RV tissue (P<0.01). With LVADs, expression of Kv4.3 was unchanged compared with controls. Expression of the N-type calcium channel was below the limit of detection.
Phase 2 of the action potential (inward and delayed rectifier currents and background currents)
With diseased ventricular tissue and when compared with controls, expression of Kv1.5 (ultrarapid delayed rectifier channel IKur) and K2p1.1 (twin pore K+ channel, background channel) was reduced to 30% (Kv1.5; P<0.05) or below the limit of detection (K2p1.1; P<0.01). With LVADs, expression was unchanged (K2p1.1) or even slightly increased (2-fold) (Kv1.5). We found expression of Kv7.1 (slow delayed rectifier channel IKs) to be 1.6-fold and 4-fold increased in diseased LV and RV tissue (P<0.05), but basically unchanged in LVAD-supported hearts. The expression of Kv11.1 (rapid component of delayed rectifier current IKr) was unchanged in diseased LV or LVAD but twofold up-regulated in diseased RV vs. healthy controls (see Fig. 2
).
Phase 3 and 4 of the action potential (delayed rectifier and inward rectifier currents)
With diseased LVs and RVs, expression of Kir2.1 (inward rectifier current IK1) and Kir3.4 (acetylcholin-gated K+ channel IKAch) was reduced to 20% of controls (P<0.05), whereas expression of Kir2.1 was basically unchanged and Kir3.4 was threefold increased in LVAD-supported hearts (P<0.01). In contrast, Kir3.1 (acetylcholin-gated K+ channel IKAch) was unchanged in diseased LVs, but repressed to 70% in RVs and threefold increased in LVADs. In diseased ventricles, Kir6.2 (ATP sensitive K+ channel) was reduced to 20% of controls, but 2- to 5-fold increased in atrial tissue (P<0.01) and LVADs (P<0.01). Kir6.1 was unchanged in diseased LVs and 3.5-fold increased in diseased RVs, but decreased to 40% in LVADs. Expression of Kir2.2 was below the limit of detection.
Pacemaker currents
HCN4 gene expression was threefold and twofold increased in diseased and LVAD-supported hearts (P<0.05). Transcripts of HCN2 were below the limit of detection.
Calcium binding proteins
With diseased hearts, expression of troponin T, calreticulin, calsequestrin, and calmodulin type I was decreased to 10–70% of controls and with LVADs, expression of calsequestrin and calreticulin was decreased to 65% and 85% of controls, whereas troponin T was threefold increased (P<0.05). With diseased hearts mRNA expression of phospholamban, troponin I, and troponin C was increased by two- to threefold, but with LVAD expression was decreased to 50% of controls. Gene expression of calmodulin type II varied considerably (range from not expressed to 15-fold increased) in RNA extracts of diseased LV, but with RV expression was decreased to 50% and unchanged in LVAD-supported heart tissue.
Calcium-dependent receptors
Gene expression of ryanodine receptor type II was reduced to 60–70% of controls in diseased ventricles and LVADs. Transcript copies of the ryanodine receptor type I and III were below the limit of detection.
ATPase-dependent transporters
With diseased LV expression of ATP1A1 (sodium-potassium ATPase; P<0.01), ATP1A2, PMCA1 (plasma membrane Ca2+ ATPase; P<0.01), PMCA4, and SERCA2 (sarcoplasmatic reticulum Ca2+ ATPase; P<0.01) were reduced to 40%, 70%, 30%, 50%, and 30% of controls, whereas expression levels were 60%, 40%, 50%, 50%, and 70% in diseased RVs. SERCA1 was only expressed in diseased ventricular tissue, but not in controls. With LVADs, expression of ATP1A1, ATP1A2, PMCA1, and PMCA4 was basically unchanged, but SERCA1 was up-regulated (P<0.05), whereas SERCA2 was down to 40% of controls.
In the case of PMCA3 and SERCA3, expression was only detectable in two of eight diseased LVs and in atria as well as in LVAD-supported hearts. Transcript copies of PMCA2 and ATP1A3 were not detected in any of the tissues examined.
Ion exchanger
In diseased LV and RV, expression of NCX1 (Na+/Ca2+ exchanger) and NHE (Na+/H+ exchanger) was 40% and 30% of controls (P<0.05), whereas expression of the skeletal isoform NCX3 was increased 1.5-fold and 3-fold in RNA extracts of LV and RV tissue. With LVADs no significant change in NCX1 and NHE gene expression was obvious, but NCX3 expression was threefold increased. Transcript copies of NCX2 were not detected in any of the tissues examined.
Calcium-dependent kinases
Gene expression of CaMKII gamma and CaMKII delta was unchanged in diseased LVs, but with RV expression of CaMKII delta was increased twofold. With LVAD expression of the latter kinase was reduced to 40% of controls, whereas CaMKII gamma was basically unchanged. Transcript copies of CaMKII 
or CaMKII ß were not detected in any of the tissues examined.
Transcriptional and translational repressors
The transcriptional repressor m-Bop was up to threefold (P<0.01) increased in diseased ventricles, but unchanged in LVADs and healthy controls. Similarly, expression of the translational repressor NAT1 was fivefold (P<0.01) and sevenfold (P<0.01) increased in left and right ventricles and unchanged in LVADs and healthy controls (see Fig. 4
). Expression of the transcriptional repressor CHF-1 was below the limit of detection.
|
Androgen receptor
The androgen receptor was up to fourfold (P<0.01) increased in diseased ventricles but unchanged in LVADs and healthy controls.
Gene expression and pharmacotherapy
All patients received complex cocktails of cardiovascular drugs; detailed information from n = 11 patients is available (see Table 1
). As shown in Table 1
, one of 11 patients was treated with a calcium antagonist, 8/11 with digitalis, 4/11 with ß-blockers, and 11/11 with ACE inhibitors or ACE receptor antagonists. We link gene expression profiles to drug treatment. Only > twofold differences compared with controls were considered to be significant. We emphasize that the proposed correlation may simply be circumstantial and may reflect individual variability in end-stage heart failure rather than effects caused by drug treatment.
Digitalis
N = 8 patients received digitalis treatment; gene expression of Nav1.5, CALM2, PMCA1, Kir2.1, Kv7.1, Kv11.1, Kir6.2, and Kir3.1 was increased (from 2- to 10-fold) but CaMKII gamma was repressed to 5% of appropriate controls (P<0.05), e.g., patients without digitalis treatment (n=3).
Beta adrenoceptor blocking agents
We noticed increased gene expression of PMCA1, CaMKII delta Kir1.5, Kir2.1, Kir3.4, Kir6.2, and Kv11.1 (2- to 5-fold, P<0.05) when patients receiving ß-blockers (n=4) were compared with appropriate controls, e.g., patients without this treatment (n=4). Expression of Cav1.2 (P<0.05), Nav1.5, PMCA4 (P<0.05), CALM1 (P<0.05), CALM2, CALR (P<0.05), ATP1A2 (P<0.05), and NCX3 was reduced to 15%–50% of controls.
Gene expression and cardiac hypertrophy
With hypertrophic left ventricles (n=3) expression of ATP1A2 (4-fold, P<0.05), Kir3.1 (2-fold), calreticulin (2-fold), calmodulin 1 (2-fold), and NCX3 (3-fold) was increased, but expression of PMCA1 (P<0.05), CaMKII gamma, CaMKII delta, and Kir6.2 was reduced.
Evidence for coregulation
We observed statistically significant negative correlation between the expression of the androgen receptor and Kir6.2 (correlation coefficient r=–0.81), Nav1.5 (r=–0.86), and Cav1.2 (r=–0.79). We used in silico promotor analysis of Kir6.2, Nav1.5, and Cav1.2 and searched for binding sites of the androgen receptor. As the promoter sequence of SCN5A is unknown, we predicted its sequence based on an in silico approach (see Materials and Methods). Using a specific matrix (see Materials and Methods), we identified several androgen-responsive elements in promotor regions of the aforementioned genes, as shown in Fig. 5
.
|
A positive relationship was computed when the expression of the repressor m-Bop and the androgen receptor was compared (r=0.91; see Fig. 6
). Likewise, a significant negative relationship between the expression of m-Bop and the calcium binding proteins calreticulin (r=–0.71), calmodulin 1 (r=–0.85), and the ATPase-dependent transporter ATP1A1 (r=–0.78) was observed, whereas expressions of the androgen receptor (r=0.91) and translational repressor NAT1(r=0.85) were positively correlated (see Fig. 6
).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
We observed significant repression of genes coding for ion channels and ATP-driven ion transporters/exchangers. Many of the deregulated genes returned to normal in assist device-supported hearts, and this provides novel information for the therapeutic benefit of implant devices. We link pressure load, stretch force, and gene expression in end-stage failing and assist device-supported hearts and propose novel regulations, as detailed below. Table 3A
demonstrates good concordance between gene expression and protein function in cardiac disease, and thus our findings may well translate to activity of coded proteins.
|
Excessive expression of ANP and BNP are diagnostic markers for cardiac hypertrophy and end-stage heart failure (10
, 11)
. We observed up to 300-fold increases in ANP and BNP gene expression. Patients with LVADs had reduced expression of ANP; this demonstrates the effectiveness of LVAD implants in unloading the failing heart (see Figs. 1
and 2
).
Ion channels and regulatory proteins
A variety of genes encoding -subunits of major ion channels have been cloned, but little is known about their expression during heart disease. Opening of the cardiac Na+ channel is essential for initiation of the action potential (phase I, see Fig. 3
). Certain mutations within in the SCN5A gene, which encodes the -subunit of INa (Nav1.5), lead to arrhythmic events, as observed with the LQT3 and Brugada syndrome (13
14
15)
. We observed a 50% reduction in mRNA expression of SCN5A. It remains speculative whether reduced gene expression can be linked to fibrillation and sudden death (2)
. With LVADs, no change in SCN5A mRNA expression was obvious. This provides evidence for ion channel gene expression to be directly regulated by pressure load and stretch force.
We investigated the gene expression of ICa(L) (Cav1.2), ICa(T) (Cav3.2), and ITo (Kv4.3), other key players in the phase I action potential, and observed significant reduction of transcript levels in diseased hearts (see Figs. 1
and 2
). The cardiac Ca2+ current plays a major role in the initiation of heart muscle cell contraction. It is well established that the L-type Ca2+ channel contributes to the characteristically long action potential in the heart. Although T-type Ca2+ channels are predominant in atria, sinus, and AV node, the latter channels play a role in determining automaticity of contraction. Previous studies reported down-regulation of L-type Ca2+ channel in atrial fibrillation and heart failure animal models (16)
as a result of disturbed calcium homeostasis (17)
. We extended these studies to ventricles tissue and demonstrated a 60% decrease in L-type calcium channel expression, but no changes in LVAD assisted hearts were noted.
Overall, the preponderance of reduced expression levels of ion channels initiating phase I of the action potential (see Fig. 3
) is an important finding. Further studies are on the way to obtain an in-depth understanding of the molecular events leading to repressed Nav1.5, Cav1.2, Cav3.2, and Kv4.3 ion channels.
A characteristic finding in the failing human heart is prolongation of the action potential. This has been attributed to a reduction of the transient outward current ITO (Kv4.3). Further evidence stems from the investigations by Kaab et al. (4)
and Beuckelmann et al. (18)
, who reported reduced mRNA level and channel density of ITO in ventricular tissue. The latter studies are therefore highly suggestive for a pivotal role of ITO in ion channel function and cardiac disease. The Kv4.3 gene encodes a subunit of this transient outward current, and we observed a significant repression in ventricular tissue of diseased hearts. Gene transfer of a dominant negative Kv4.3 construct (functional knockout) in guinea pig myocytes led to action potential and QT interval prolongation; similar findings were observed in human heart failure studies (19)
. Clearly, this demonstrates the importance of Kv4.3 for the transient outward current. In strong contrast, we found slight up-regulation of Kv4.3 in LVAD-supported human hearts. Several genes coding for potassium channels (Kv1.5, K2p1.1, Kir2.1, Kir3.4) were down-regulated in diseased ventricular tissue but, once again, not in LVADs. Thus, reduced pressure load and stretch force after (long-term) LVAD implantation will contribute to a reversal of the electrogenic remodeling process in failing hearts. In the study by Harding et al. (20)
, electrophysiological alterations were investigated in patients with mechanical circulatory support and advanced cardiac failure, and the authors report significant decreases in action potential duration after LVAD implantation. Again, this points to a reversal of the electrophysiological remodeling process in heart failure and our findings agree well with those of Harding et al. (20)
.
Brundel et al. (21)
reported reduced expression of Kir6.2 in atria of patients with atrial fibrillation. We extend the study of Brundel et al. to diseased human ventricular tissue and demonstrate a significant negative correlation of Kir6.2 and the androgen receptor. Expression of Kir6.1 is predominantly found in endothelial and smooth muscle cells and is potentially is involved in the regulation of vascular tonus (22
, 23)
. We also observed expression of this potassium channel in heart tissue and measured a significant (P<0.05), 3.5-fold up-regulation in diseased RVs, but reduced expression in LVAD-supported hearts (see Fig. 2
).
We observed an increase (3-fold) in HCN4 gene expression in end-stage heart failure. HCN4 encodes for the pacemaker current If, and overexpression of this protein may result in enhanced automaticity and arrhythmic events, as observed in the cohort of our patients (see Table 1
). Supporting evidence stems from Qu et al. (24)
, who demonstrated altered voltage dependency in adult rat cardiomyocytes transfected with the heterologue HCN2.
In conclusion, we report significant down-regulation of genes coding for certain potassium channels and propose a direct relationship between altered expression and abnormal repolarization.
Next to ion channels, a variety of other regulatory proteins are important in ion homeostasis. Calsequestrin and calreticulin are calcium binding chaperones of the sarcoplasmic reticulum, and overexpression of these genes leads to cardiac hypertrophy and decreased systolic function in transgenic mice (25
, 26)
. The reduced expression of genes coding for calcium binding proteins of the sarcoplasmatic reticulum might be viewed as an adaptive response due to reduced calcium availability.
Expression of slow skeletal troponin I in transgenic mice resulted in partial substitution of cardiac troponin I in heart muscle cells, which led to impaired cardiomyocyte relaxation and diastolic function (27)
. We observed a twofold up-regulation of the genes coding for the contraction inhibitory proteins troponin C and troponin I, whereas its expression was down to 50% of controls in LVAD-supported hearts. Once again, this demonstrates the reversal of the remodeling process after LVAD implantation. In contrast, troponin T mRNA expression was reduced to 10% and 50% in diseased LV and RV, but unchanged in LVAD-supported hearts. This is an important finding as certain mutations of the troponin T gene may lead to severe cardiac sarcomere defects and myocyte disarray, as recently shown by Sehnert et al. (28)
. Whether repression of troponin T mRNA translates to reduced protein levels in heart failure requires further investigation.
Calcium-dependent receptors play a major role in the regulation of ion balance and contractility. Ryanodine receptors are intracellular calcium ion release channels responsible for the release of Ca2+ from intracellular stores after transduction of many different extracellular stimuli. The ryanodine receptor type II is the major isoform in the heart and the major source of calcium required for cardiac muscle excitation-contraction coupling. We report decreased expression of RYR2 in end-stage heart failure, and this agrees well with the earlier findings of Go et al. (29)
. We investigated RYR2 mRNA expression levels in LVAD-supported hearts, but found no differences in expression between diseased and assist device-supported left ventricles. We speculate that decreased expression of RYR2 might be linked to decreased cytosolic calcium availability, but further studies are needed to investigate its effect on contractility. The phosphorylation status of RYR2 and/or its interaction with FK506 binding proteins will also regulate function of this receptor (30)
. Future studies are needed to understand the role of post-translational changes and of interactions with regulatory proteins in heart failure.
The capacity to restore low resting calcium levels during diastole is dependent on the activity of the sarcoplasmic reticulum calcium ATPase. SERCA2 expression is known to be decreased during heart failure both at the transcript and protein level (31)
, whereas SERCA1 was shown to be increased in the failing human heart (32)
. We observed similar changes in diseased LV and RV tissue, as well as in assist device-supported left ventricles, which suggests ventricular implants not to influence mRNA expression levels of SERCA2. Next to SERCA, the plasma membrane calcium ATPase (PMCA) is a calcium extruding enzyme controlling Ca2+ homeostasis in nonexcitable cells, but its function in heart muscle cells is far from clear because of an additional availability of the Na+/Ca2+ exchanger. The PMCA has little relevance for beat-to-beat regulation of contraction/relaxation in adult animals, but potentially plays a role in regulating myocardial growth (33)
. Four isoforms of PMCA are currently known, of which PMCA1 and PMCA4 protein was identified in heart tissue, whereas protein expression of PMCA2 and PMCA3 appears to be confined to neuronal tissue (34)
. We observed repressed gene expression levels of PMCA1 and PMCA4 in diseased ventricular tissue as well as in LVAD-supported hearts, and our study is the first report on PMCA3 expressed predominantly in atria as well as in LVAD-supported hearts.
The Na+/K+-ATPase (ATP1A) plays an important role in the maintenance of electrolyte balance in heart muscle cells. We observed reduced expression of ATP1A1 and ATP1A2 in end-stage heart failure; this may result in decreasing Ca2+ extrusion by the Na+/Ca2+ exchanger, which increases the cellular content of Ca2+. Confirmation stems from the study of Schwinger et al. (35)
, who reported reduced protein expression of ATP1A1 in left ventricles of failing heart transplants. We demonstrated reduced mRNA expression in the failing RV and show normalization of expression in patients with LVAD implants.
The Na+/Ca2+ exchanger (NCX) is thought to be the main calcium extrusion system in cardiomyocytes, and NCX knockout mice demonstrate marked cardiac hypertrophy in response to pressure overload (36)
. We observed reduced expression of NCX1; conceivably this would lead to elevated intracellular Ca2+ concentration, as it is frequently observed in cardiac hypertrophy and heart failure (37)
. We report increased NCX3 mRNA expression in the failing heart. This isoform was thought to be expressed in skeletal muscle only (38)
, and induction of the skeletal/fetal isoform of NCX may be viewed as cellular dedifferentiation that may lead to the expression of fetal genes. A similar switch from adult to fetal isoform gene expression was reported for the myosin heavy chain gene (39)
.
Our observation of reduced NHE (Na+/H+ exchanger) mRNA expression provided further evidence for the severe deregulation of ion exchangers leading to electrolytic imbalance in end-stage heart failure.
The calcium/calmodulin-dependent protein kinase class II is a multifunctional enzyme involved in the regulation of gene expression, cell cycle control, and differentiation (40)
. Whereas members of the and ß classes are mainly expressed in neuronal tissue, gamma and delta isoforms are more abundantly expressed in some tissues, including heart (41)
. We did not observe significant differences in the expression levels of CaMKII delta in ventricular tissue compared with healthy human hearts, but observed repression of CaMKII gamma to 40% of controls with LVADs. As CaMKII is implicated in the progression of cardiac hypertrophy (42)
, its down-regulation might provide a good biomarker for the reversal of cardiac hypertrophy after LVAD implantation.
Gene expression and pharmacotherapy
As discussed above, it is difficult to propose direct relationships between gene expression and pharmacotherapy in a multigenic and epigenetically driven disease. Nonetheless, we searched for correlation to obtain clues for hypothesis-driven research.
Fan et al. (43)
reported transcriptional up-regulation of the L-type calcium channel after treatment of myocytes with ß-adrenergic agonists. We observed strong repression of the L-type calcium channel in patients receiving ß-blockers, which agrees well with the aforementioned study.
Measurements of ion currents in myocytes from hypertrophic and failing hearts point to a decreased outward potassium current (3)
. We observed increased expression of several outward potassium channels upon pharmacotherapy with ß-blockers. This increase in gene expression fits well with the well-known normalization of the repolarization phase of contracting cardiomyocytes upon treatment of patients with ß-blockers. The beneficial effects of ß-blockers regarding electrogenic stability are proven (44)
.
Global vs. knowledge-based gene expression studies
Recently, Barrans et al. (45)
reported global gene expression profiling in end-stage dilated hearts. Of the 10,848 genes investigated, only ANP, CCAAT box binding factor as well as two EST clones were found to be >fivefold up-regulated. Another 58 genes were identified to be up-regulated (1.09–3.78), and this represents 0.53% of all genes investigated. A total of 81 genes were found to be repressed (1.03–3.19), of which a large number were EST clones with uncertain annotations. This study clearly demonstrates the limitation of microarray studies with EST clones and (as yet) unknown genomic sequences. However, global gene expression studies are important tools for hypothesis generation and downstream hypothesis-driven research.
Hierarchical gene cluster analysis
We applied a hierarchical clustering method to group genes on the basis of similarity of their expression. The cluster diagram is shown in Fig. 2
and is based on a total of 1850 gene expressions. Despite its complex nature, the clustering analysis provided a remarkable order. With few exceptions, the expression segregates to various anatomical regions of the heart. The dentogram shown on the left hand of Fig. 2
indicates a complex cluster that groups distinct sets of genes, and this includes calcium binding chaperones of the sarcoplasmatic reticulum and a large group of ion channels and exchangers/transporters. A group of calcium binding proteins group together, as do two sets of potassium channels separated by troponins. Finally, the stress markers ANP and BNP clustered together, as does SERCA1 and the T-type calcium channel. This clearly demonstrates the usefulness of gene cluster analysis in identifying potential networks of regulated genes and studies are under way to determine the promoter activity of deregulated genes. Further evidence stems from the study of Tan et al. (46)
, who investigated expression of >6600 genes in failing and nonfailing human hearts using oligonucleotide microarrays. Indeed, a total of 103 genes were found to be differentially expressed. ANP, BNP and genes coding for extracellular matrix were increased whereas genes coding for metabolic processes, such as alcohol dehydrogenase or fatty acid synthase, were repressed in heart failure. By applying a hierarchical clustering method, the authors were able to distinguish between different etiologies of cardiomyopathy.
Evidence for coregulation
Recently, we reported increased androgen receptor expression in cardiac hypertrophy (9)
and demonstrated statistically significant negative correlations between androgen receptor and ion channel expression (see Fig. 6
). As K(ATP) channels can be regulated by estradiol (47)
ligands for the androgen receptor may also regulate, at least in part, ion channel expression. With an in silico approach, we identified several androgen-responsive elements in promotors of deregulated ion channels; this provides further evidence for cross-talk between the androgen receptor and the ion channels Kir6.1, Nav1.5, and Cav1.2.
m-Bop, a histone deacetylase-dependent transcriptional repressor, regulates a critical step in the development of cardiomyocytes (48)
. m-Bop recruits histone deacetylases, which deacetylate specifically lysin residues on histone tails, thus leading to chromatin condensation and silencing of gene transcription. We observed highly significant increased expression of m-Bop in cardiac disease and show statistically significant negative coregulation between m-Bop and calcium binding proteins and the ATP pump 1A1 (see Fig. 6
). Thus, up-regulation of m-Bop coincides with repression of genes in cardiac disease.
NAT1 functions as a translational repressor during early postnatal cardiac development and appears to be elevated after birth during the first 2 wk, but its expression declines thereafter (49)
. Enhanced expression in diseased hearts results in repressed ion channel gene expression, as observed in this study and by others (3
, 4
, 18)
. The highly significant up-regulation of the repressors m-BOP and NAT1 (r=0.85; see Fig. 6
) correlated well with repressed gene expression of ion channels, and thus we suggest overexpression of repressors and silencers to be important events in cardiac hypertrophy. As the DNA binding sites for the aforementioned repressors are unknown, no electromobility shift assays could be done to confirm interaction of the repressors based on consensus binding site of the respective target genes.
In conclusion, we report a comprehensive survey of gene expression profiles of ion channels and other regulatory proteins involved in conduction and cardiomyocyte biology. We demonstrate very significant alterations of genes handling ion homeostasis and cardiomyocyte contraction in end-stage heart failure. We propose repression of gene expression in heart disease to be mediated by transcriptional and translational repressors, which in turn are regulated by the androgen receptor. We link altered expression levels with channel dysfunction and disease and demonstrate reversal of altered gene expression in patients with assist device implants.
|
|
ACKNOWLEDGMENTS |
|---|
Received for publication October 7, 2002. Accepted for publication May 8, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
