Microarray gene expression profiles in dilated and hypertrophic cardiomyopathic end-stage heart failure
Juey-Jen Hwang1,3,
Paul D. Allen2,
George C. Tseng4,
Ching-Wan Lam1,
Lameh Fananapazir5,
Victor J. Dzau1 and
Choong-Chin Liew1,6
1 Cardiovascular Genome Unit, Department of Medicine
2 Department of Anesthesiology, Brigham and Womens Hospital, Harvard Medical School, Boston 02115
3 Division of Cardiovascular Medicine, Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan
4 Department of Biostatistics, Harvard School of Public Health, Boston, Massachusetts
5 Cardiology Branch, National Heart, Lung, and Blood Institute, Bethesda, Maryland
6 Department of Laboratory Medicine and Pathobiology, University of Toronto, Ontario, Canada
 |
ABSTRACT
|
---|
Despite similar clinical endpoints, heart failure resulting from dilated cardiomyopathy (DCM) or hypertrophic cardiomyopathy (HCM) appears to develop through different remodeling and molecular pathways. Current understanding of heart failure has been facilitated by microarray technology. We constructed an in-house spotted cDNA microarray using 10,272 unique clones from various cardiovascular cDNA libraries sequenced and annotated in our laboratory. RNA samples were obtained from left ventricular tissues of precardiac transplantation DCM and HCM patients and were hybridized against normal adult heart reference RNA. After filtering, differentially expressed genes were determined using novel analyzing software. We demonstrated that normalization for cDNA microarray data is slide-dependent and nonlinear. The feasibility of this model was validated by quantitative real-time reverse transcription-PCR, and the accuracy rate depended on the fold change and statistical significance level. Our results showed that 192 genes were highly expressed in both DCM and HCM (e.g., atrial natriuretic peptide, CD59, decorin, elongation factor 2, and heat shock protein 90), and 51 genes were downregulated in both conditions (e.g., elastin, sarcoplasmic/endoplasmic reticulum Ca2+-ATPase). We also identified several genes differentially expressed between DCM and HCM (e.g.,
B-crystallin, antagonizer of myc transcriptional activity, ß-dystrobrevin, calsequestrin, lipocortin, and lumican). Microarray technology provides us with a genomic approach to explore the genetic markers and molecular mechanisms leading to heart failure.
cDNA microarray; normalization; real-time reverse transcription-polymerase chain reaction
 |
INTRODUCTION
|
---|
HEART FAILURE is a common reason for morbidity and mortality in developed countries today. The cardiovascular diseases that lead to heart failure are complex, and the transition from compensated cardiac hypertrophy to debilitating heart failure may involve different pathways (21). Cardiomyopathy often results in heart failure and is a major indication for cardiac transplantation in the United States. Dilated (DCM) and hypertrophic (HCM) cardiomyopathy are two common forms of cardiomyopathy; nevertheless, they result in end-stage heart failure through different remodeling and molecular pathways (21, 38, 39). The left ventricle in DCM is dilated and hypocontractile, and patients often present with easy fatigability, exercise intolerance, dyspnea, and right and left heart failure (38). HCM either occurs sporadically or is inherited as an autosomal dominant pattern with variable penetrance and has a prevalence of 2 in 1,000 in the general population (13, 28). In contrast to DCM, the left ventricle in HCM is characteristically hypertrophied and hypercontractile. The diastolic function is often abnormal despite high left ventricular ejection fraction. Patients with HCM are at risk of sudden death even in the absence of cardiac decompensation. Overt heart failure is uncommon until the patient develops atrial fibrillation or myocardial systolic dysfunction (39, 48).
Although the genetic defect is unknown in most cases of familial DCM and determined in only half of cases of familial HCM, it is clear that both diseases exhibit nonallelic and allelic genetic heterogeneity (6, 14, 16, 34). The HCM phenotype is caused in most patients by distinct mutations in one of several sarcomeric genes, whereas cytoskeletal mutations are more closely related to DCM (6, 21). However, recent evidence provides clues that the same cytoskeletal or sarcomeric gene mutation may be associated with either DCM or HCM in the same family, possibly operating through two different series of events that remodel the heart (38, 44). Furthermore, the link between genetic mutation and contractile dysfunction remains an enigma. Genomic technology enables us to look at the differential expression of tens of thousands of genes simultaneously, and to compare patterns of gene expression during disease development and progression (10). However, approaches to the analysis of the large database generated from microarray platforms are still evolving. In this study, we constructed a spotted cDNA microarray using clones from various cardiovascular cDNA libraries sequenced and annotated in our laboratory to test the hypothesis that DCM and HCM end-stage heart failure developed via different molecular pathways and that, therefore, the two diseases presented different gene expression profiles. We also proposed methods for slide normalization, which was slide-dependent and nonlinear, and for gene-specific expression levels assessment using a hierarchical model. The feasibility of our approach was validated by real-time reverse transcription-polymerase chain reaction (RT-PCR). Using cDNA microarray, we obtained preliminary molecular portraits of DCM- and HCM-related end-stage heart failure.
 |
MATERIALS AND METHODS
|
---|
RNA isolation.
Total RNA was isolated from human DCM, HCM, and normal adult heart samples using TRIzol reagent (GIBCO-BRL; Life Technologies, Grand Island, NY) and cleanup with Qiagen RNeasy mini kit (Qiagen, Valencia, CA) according to the manufacturers protocol. RNA quantity was determined by A260 measurement, and RNA integrity was checked by formaldehyde gel electrophoresis. The protocol was approved by the Institutional Review Board of Brigham and Womens Hospital, Boston, MA.
Building a cDNA library and construction of spotted cDNA microarray.
The cDNA library was constructed using various cardiovascular tissues, including HCM in end-stage heart failure, fetal hearts (812 wk), normal adult hearts and aorta, with the lambda ZAP Express vector system (Stratagene, La Jolla, CA) as previously described (19, 20). Expressed sequence tags (ESTs) were generated using a well-established PCR and cycle-sequencing-based approach. All ESTs were searched against the nonredundant GenBank/EMBL/DDBJ and dbEST databases using the BLAST algorithm on a Unix platform (Sun Microsystems). Assignment of putative identities for ESTs required an expected value of 10-10 or less and a minimum of 95% nucleotide identity (9). ESTs matching to known genes were classified into seven different functional groups described previously (20). Individual nonredundant cDNA clones were isolated into unique pools and amplified by PCR in 96-well microplates. PCR products with a final volume of 50 µl were precipitated with 5 µl of 3 M ammonium acetate and 125 µl of 95% ethanol at -20°C overnight. The plates were centrifuged at 4,000 rpm for 30 min at 4°C, and the supernatant was decanted. Following two washes with 50 µl of 70% ethanol, the resulting DNA pellets were air-dried and resuspended in 20 µl of 3x SSC (sodium chloride/ sodium citrate buffer). A total of 10,368 nonredundant PCR products (including 96 bacteria clones as negative controls) were spotted onto Corning CMT-GAPS amino-Silane-coated glass microarray slides (Corning, Corning, NY) using the model GMS 417 arrayer (Affymetrix, Santa Clara, CA) and postprocessed using succinic anhydride blocking according to the manufacturers manual (4). The unique clones include 2,496 (24.3%) known genes, 3,296 (32.1%) matched ESTs, and 4,480 (43.6%) novel genes. The possibility of contaminated cDNA clones was reduced by prescreening with agarose gel electrophoresis.
Making cDNA probes and hybridization.
Human heart failure samples were obtained from the left ventricular free wall of explanted hearts from three idiopathic DCM patients without identifiable etiologies or antecedent myocarditic episode and from two patients with HCM (disease caused by Arg719Gln ß-myosin heavy chain mutation in both) during cardiac transplantation. Normal adult heart tissues of three donors were obtained from the left ventricular free wall of hearts not used for cardiac transplantation and pooled as reference samples. Thirty-five micrograms of total RNA from pooled normal adult heart, DCM, or HCM samples was oligo-dT primed, and probe synthesis was performed in the presence of either Cy3-dUTP (pooled normal) or Cy5-dUTP (pooled DCM or HCM) (Amersham Biosciences, Piscataway, NJ). Briefly, RNA was prepared in 8 µl of DEPC water, to which 1 µl of oligo-dT primer (0.5 µg/µl, GIBCO-BRL) was added. The mixture was incubated at 70°C for 5 min, then ice-chilled immediately. Four microliters of 5x first-strand buffer, 2 µl of 10x low-T dNTP (5 mM dATP, dCTP, and dGTP, and 2 mM dTTP), 2 µl of Cy3- or Cy5-dUTP, 2 µl of 0.1 M DTT, and 1 µl of RNaseOUT RNase inhibitor (40 U/µl, Invitrogen) were added, mixed well, and heated to 65°C for 5 min. One microliter of Superscript II (200 U/µl, GIBCO-BRL) was then added and incubated for 30 min at 42°C, followed by the addition of another 1 µl of Superscript II for 40 min at 42°C. The reverse transcription reaction was terminated with 2.5 µl of 500 mM EDTA, and the mixture was heated to 65°C for 1 min. Then, 5 µl of 1 M NaOH was added for 10 min at 65°C to hydrolyze RNA, and the pH was neutralized with 12.5 µl of 1 M Tris buffer (pH 7.5). Following purification of the labeled probe by gel exclusion chromatography (ProbeQuant G-50, Amersham), two cDNA probes of interest were mixed, reduced to a volume of about 5 µl, and combined with 30 µl of hybridization solution [stock solution containing 100 µl of DIG EasyHyb hybridization solution (Roche), 5 µl of yeast tRNA (10 mg/ml), and 5 µl of salmon sperm DNA (10 mg/ml) as blocking agents]. The probe solution was then hybridized to the arrayed slide at 37°C overnight. The next day, slides were washed first with 1x SSC to remove the coverslip; next, the slides were given three successive washes of 0.1% SDS and 1x SSC at 50°C for 15 min each, followed by a rinse with 1x SSC at room temperature. The slides were dried by a 5-min spin in a conical tube at 5001,000 rpm to remove excess fluid. Scanning of the slide was performed using the model GMS 418 scanner (Affymetrix) at 532 nm (Cy3) and 635 nm (Cy5). The DCM experiments were performed on four slides in replicate experiments performed separately (slide number N = 2 for each experiment); HCM experiments were done on three slides with one replicate experiment (N = 2) and another one separate single slide experiment. All diseased heart samples were labeled with Cy5, and normal heart samples were labeled with Cy3. Pooled normal adult heart samples were labeled with Cy3 or Cy5, and then hybridized on four slides [2 replicate experiments (N = 2 for each experiment)] that were used as calibration experiments. Replication of hybridization experiments will greatly reduce the misclassification rate (25, 27); thus we performed replicate experiments in our study to exclude the nonconsistent data.
Image acquisition and data processing.
Raw scanned images were processed using ScanAlyze 2.44 microarray image analysis software (Michael Eisen, Stanford University, CA, http://rana.lbl.gov/EisenSoftware.htm). Cy3 and Cy5 scans for each slide were superimposed onto each other, and values corresponding to the fluorescence intensity for each spot were obtained and exported to an Excel spreadsheet. Local background was subtracted from the fluorescence value of each spot to obtain a "net" value. To avoid false-positive results generated from poor quality spots with weaker fluorescence signals, we filtered the weak hybridization spots first to account for incomplete hybridization. The CH1GTB2 and CH2GTB2 background criteria, which mean the fraction of pixels with greater than 1.5 times the background intensity, were used with a cutoff values of 0.5 in both channels (Michael Eisen, http://rana.lbl.gov/EisenSoftware.htm). We also included one tray of bacterial clones as negative controls. Spots with net signal intensities on the Cy3 or Cy5 channels less than the median intensities of bacterial clones were also filtered out to exclude signals due to nonspecific binding.
Bioinformatics.
We used a rank-invariant method to identify nondifferentially expressed genes on each slide across various signal intensities (46). After selecting nondifferentially expressed genes and fitting them into a normalization curve (using the Lowess smoothing procedure in S-plus), we extrapolated the normalization curve to normalize genes with extremely high or low intensities. After obtaining the normalized data for each filtered spot, we use a hierarchical linear model to assess gene expression levels and incorporated the calibration experiments as prior knowledge of variance components in the analysis of comparative experiments. The 95% posterior interval for each gene was displayed as upper and lower quartiles. Genes with a score lower than 0.025 (which indicates 2.5% probability of logarithmic Cy5/Cy3 expression level below 0) were selected as upregulated (e.g., increased gene expression), whereas genes with score greater than 0.975 (which indicates 97.5% probability of logarithmic Cy5/Cy3 expression level below 0) were selected as downregulated (e.g., decreased gene expression). This model allows us to select differentially expressed genes from replicate experiments, eliminating those spots with high variation in their expression ratios across slides.
Real-time RT-PCR.
To confirm the expression patterns of upregulated or downregulated genes, we chose several genes for further analysis using quantitative real-time RT-PCR in a 96-well format. For each gene of interest, real-time RT-PCR was performed for pooled DCM (n = 3) or HCM (n = 2) heart RNA samples which were used in microarray experiments, and pooled normal adult heart RNA samples were used as reference. Triplicate aliquots of each pooled RNA sample were used in the same reactions. As an internal control, primers for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were designed and amplified in parallel with the genes of interest. One-step real-time RT-PCR was performed using 50 ng of total RNA per reaction. Primers were designed using the Primer3 program (http://www.genome.wi.mit.edu), verified for complementarity (http://www.basic.nwu.edu/biotools/oligocalc.html), and searched against the public database to confirm unique amplification products (http://www.ncbi.nlm.nih.gov). Primers were generally 20 base pairs long and were chosen to generate PCR products of 100 to 150 base pairs. The melting temperature range was between 59 and 61°C. The PCR products were checked by 2% agarose gel electrophoresis for each reaction. Water control and no reverse transcriptase control were also performed on all DCM, HCM, and normal heart samples to eliminate the possibility of significant genomic DNA contamination. The primer sequences are listed in Table 1. All reactions were carried in 50-µl volumes containing 25 µl of SYBR Green PCR Master Mix (Perkin-Elmer Applied Biosystems, Foster City, CA), 0.25 µl of Multiscribe reverse transcriptase (50 U/µl, Perkin-Elmer), 0.5 µl of RNaseOUT RNase inhibitor, 50 ng of sample RNA, and 10 pmol of each forward and reverse primer. Reactions in 96-well format were performed in the Perkin-Elmer ABI Prism 7700 sequence detection system. The cycling parameters were 30 min at 48°C (reverse transcription), heated to 95°C for 10 min, and followed by 40 cycles of PCR (15 s at 94°C and 1 min at 60°C). The threshold cycle value (CT) represents the cycle at which a statistically significant increase in the normalized reporter signal (Rn) above a chosen threshold can first be detected, according to the manufacturers manual. Threshold is defined as the average standard deviation of Rn for the early cycles, multiplied by an adjustable factor. To determine relative expression levels in each RNA population, a standard curve was plotted on the basis of expressions of GAPDH in serial dilutions of pooled normal adult heart RNA (200 ng, 100 ng, 50 ng, 25 ng, 12.5 ng, 6.25 ng, and 3.12 ng) and a no-template control. For all experimental samples, the relative RNA quantity of each sample was determined from the standard curve, divided by the corresponding amount of GAPDH control to achieve a normalized value. Fold differences were calculated by dividing the mean of DCM or HCM samples by the averaged amount of normalized mRNA generated in the normal adult samples, and standard error (SE) was calculated for each category. All reaction results for the same samples with a coefficient of variation greater than 10% were retested. Statistical significance was defined by P < 0.05 using Students t-test.
 |
RESULTS
|
---|
Normalization for cDNA microarray is slide-dependent and nonlinear.
The purpose of normalization is to minimize the systematic variations in the measured gene expression levels and to allow the comparisons across slides. Figure 1 shows the M-A plot for a representative slide of the calibration experiments and for slides of the DCM or HCM comparative experiments, where M = log(Cy5/Cy3) represents the common log ratio of the fluorescence intensities of the Cy5 and Cy3 channels, and A = [log(Cy5) + log(Cy3)]/2 is the average logarithmic fluorescence intensities of both channels (46, 51). The nondifferentially expressed genes were selected by the rank-invariant method and used for normalization. The normalization curve for each slide was nonlinear and differed among slides. The spread of gene expression patterns was more widely distributed among DCM and HCM slides than that of calibration experiments, indicating expression profiles differed among the comparative experiments.

View larger version (20K):
[in this window]
[in a new window]
|
Fig. 1. M-A plot for a representative slide of calibration experiments (pooled normal vs. normal mRNA samples) (A) and for slides of dilated cardiomyopathy (DCM, B) or hypertrophic cardiomyopathy (HCM, C) comparative experiments (diseased heart mRNA samples labeled with Cy5 and normal heart mRNA labeled with Cy3) where M = log(Cy5/Cy3) represents the common log ratio of the two dyes and A = [log(Cy5) + log(Cy3)]/2 is the average logarithmic fluorescence intensities of both channels. Note that the normalization curve for each slide was nonlinear and differed among the slides. Gene expression patterns were more widely distributed on the DCM and HCM slides than on the calibration experiment slide.
|
|
Selecting differentially expressed genes among DCM and HCM.
Genes considered to be upregulated or downregulated were selected by the hierarchical model described previously (46). We found several spots with scores >0.975 or <0.025 which had normalized ratios only in one or two replicate slides for all DCM or HCM comparative experiments after the filtering process, reflecting the fact that the expression data for a few target genes were excluded by the filtering criteria in certain experiments. We included those spots with normalized ratios in all replicates to obtain more consistent data and minimize the possibility of false-positive spots in this study. Our result showed that 621 and 399 genes were highly expressed in DCM and HCM, respectively, including 192 (192/399, 48%) that were highly expressed in both DCM and HCM (e.g., genes for atrial natriuretic factor, CD 59 antigen, calcium activated neutral protease, decorin, elongation factor 2, and heat shock protein 90) (Tables 24). Among the downregulated genes, there were 263 in DCM, 236 in HCM, and 51 (51/236, 22%) in both [e.g., genes for sarcoplasmic/endoplasmic reticulum Ca2+-ATPase (SERCA), elastin, and phosphofructokinase]. If we used twofold change as a cutoff criterion, then the number of upregulated genes in DCM and HCM was 121 and 25, respectively, and the number of downregulated genes in DCM and HCM was 19 and 62, respectively (Table 2).
View this table:
[in this window]
[in a new window]
|
Table 2. Number of differentially expressed genes in DCM- and HCM-related end-stage heart failure samples as assessed by cDNA microarrays
|
|
View this table:
[in this window]
[in a new window]
|
Table 3. Genes with altered expression as defined by score <0.025 or >0.975 and >1.5 fold change (50th percentile) in dilated cardiomyopathy
|
|
View this table:
[in this window]
[in a new window]
|
Table 4. Genes with altered expression as defined by score <0.025 or >0.975 and >1.5 fold change (50th percentile) in hypertrophic cardiomyopathy
|
|
Differential gene expression profiles in DCM and HCM.
Among the known genes (upregulated, 103 in DCM, 66 in HCM, and 32 in both; downregulated, 101 in DCM, 97 in HCM, and 26 in both), which were grouped into subcategories such as cell division, cell signaling/communication, cell structure/motility, cell/organism defense, protein/gene expression, metabolism, and unclassified functions (20), more numbers of the transcripts of the cell/ organism defense subcategory were associated with DCM than with HCM, whereas more numbers of ribosomal proteins were found in HCM samples [HCM, 10/ 66 (15%) vs. DCM, 4/103 (4%)] in the category of upregulated genes (Fig. 2A). Furthermore, several genes in the cell/organism defense subcategory in DCM were related to the immune response (immunoglobulin heavy chain, T-cell receptor ß-chain, etc.) and genes in the same subcategory in HCM samples were stress proteins (heat shock protein 40 homolog, 70, and 90) (Tables 3 and 4). (Please note that a full list of these genes can be found in APPENDIX Tables A and B, available as Supplementary Material,1
published online at the Physiological Genomics web site.) Of the downregulated genes, more numbers of transcripts were in the DCM samples than in HCM samples in the metabolism subcategory, whereas transcripts in the cell signaling/communication and cell structure/motility subcategories were more frequently found in HCM samples (Fig. 2B).

View larger version (28K):
[in this window]
[in a new window]
|
Fig. 2. The percent of known genes in each functional category that are upregulated (A) and downregulated (B) in DCM and HCM.
|
|
Validation of differentially expressed genes by real-time RT-PCR.
Real-time RT-PCR was employed to confirm the relative expression patterns of randomly chosen genes in upregulated or downregulated categories. We arbitrarily defined two groups with different score values (group A, score <0.01 or >0.99; group B, 0.01 < score < 0.025 or 0.975 < score < 0.99). The results shown in Table 5 demonstrated the accuracy of current approach to microarray data mining. Group A included 18 DCM or HCM candidate genes, of which 16 (89%) were confirmed by real-time RT-PCR. Group B included 14 DCM or HCM candidate genes, of which 10 (71%) were confirmed by real-time RT-PCR. If we used the 50th percentile fold change of the gene-specific posterior expression as the cutoff criterion, then the confirmation rate by RT-PCR in the groups with greater than twofold upregulated or downregulated and less than twofold change of altered gene expression was 90% (9/10) and 77% (17/22), respectively (Table 5). Furthermore, the fold changes in RT-PCR were usually greater than those shown in microarray data. Based on real-time RT-PCR results, we confirmed several genes identified as commonly regulated (Fig. 3A) or differentially expressed (Fig. 3B) in DCM and HCM samples by microarray. Several genes (e.g.,
B-crystallin, antagonizer of myc transcriptional activity, ß-dystrobrevin, calsequestrin, lipocortin, and lumican) were found differentially expressed in either DCM or HCM (Fig. 3B). One gene (copper/zinc superoxide dismutase) did not display differential expression on HCM microarray data, while increased expression (1.6-fold increase in HCM, P = 0.047) was demonstrated by real-time RT-PCR (Table 5).
View this table:
[in this window]
[in a new window]
|
Table 5. Real-time RT-PCR results for randomly selected genes with differential expression in microarray experiments
|
|

View larger version (37K):
[in this window]
[in a new window]
|
Fig. 3. The real-time RT-PCR confirmed commonly upregulated or downregulated (A) and differentially expressed (B) genes in DCM and HCM. The fold change was displayed as relative to normalized normal adult heart samples. *P < 0.05. #P < 0.01. The atrial natriuretic peptide was more than 20-fold increased in both DCM and HCM (not shown in the bar graph). Calpain, calcium-activated neutral protease; EF2, elongation factor 2; HSP 90, heat shock protein 90; SOD, copper/zinc superoxide dismutase; SERCA, sarcoplasmic/ endoplasmic reticulum calcium-ATPase; B-cryst, B-crystallin; CASQ, calsequestrin; MALC, atrial myosin alkali light chain; BDTN, ß-dystrobrevin; Mad, antagonizer of myc transcriptional activity; and TRR, thioredoxin reductase.
|
|
 |
DISCUSSION
|
---|
Previous microarray studies in cardiovascular field.
Microarray technique has been used for large-scale genomic approach to decipher the molecular mechanisms involved in physiological and pathological processes in various cells, animal or human tissues (30, 37). Oligonucleotide or cDNA microarray has also been applied to cardiovascular disease-related gene changes. Friddle et al. (15) analyzed gene expression profiles involved in the pharmacological induction and regression of cardiac hypertrophy, and Liu et al. (27) reported the genes in rat cardiomyocytes responsive to insulin-like growth factor-1. Altered gene expression patterns assessed by microarray were also demonstrated in heart muscles from patients with HCM (26), in left ventricular free wall and interventricular septum from a rat myocardial infarction model (42), and in the failing human heart (50).
Molecular portraits of DCM- and HCM-related end-stage heart failure.
In the genes with altered expression in DCM and HCM human heart failure samples, we demonstrated the commonly upregulated or downregulated and differentially expressed genes of DCM and HCM, despite similar clinical situations of end-stage heart failure, in our patients. In our study, the number of transcripts involved in cell/organism defense, especially the immune response subcategory, was more in the DCM than in the HCM samples, possibly reflecting the diverse etiology of DCM. In contrast, HCM is most commonly caused by sarcomeric gene mutations. Although myocardial remodeling characterizes both conditions to some extent, hypertrophic processes might be more prominent in HCM, as evidenced in this and previous studies by greater expression of ribosomal genes in HCM (19). DCM and HCM also presented different portraits of downregulated genes. As the percentage of known genes was derived from our cDNA microarray database, not dealing with the entire population of known genes, the difference in the functional classification subcategory between DCM and HCM should be further confirmed and interpreted with caution. Moreover, the expression levels, not only the numbers, of genes in a certain subcategory are more related to biological significance. Despite similar clinical features of end-stage heart failure, the gene defects leading to DCM or HCM and secondary consequences of those defects differ in these two diseases (31, 38, 39), and indeed this is what we found in this study. Our molecular portraits of DCM and HCM are the first example of predictive markers for these two diseases (17).
Commonly upregulated or downregulated genes in DCM and HCM.
Our results also show several genes commonly upregulated or downregulated in both DCM and HCM samples. Atrial natriuretic peptide and SERCA displayed increased and decreased expression levels consistently in all the slides, respectively, and therefore served as positive control in this study (45). Copper/zinc superoxide dismutase and heat shock protein 90 were upregulated, reflecting cardiac response to oxidative stress in end-stage heart failure (40). Increased elongation factor 2 expression levels contributed to the activation of protein synthesis in myocardial hypertrophy and decompensated heart failure (50), and dephosphorylation of elongation factor 2 resulted in its activation and subsequently increased protein synthesis (12). Elevated calcium-activated neutral protease activity was reported in an isoproterenol-induced cardiac hypertrophy rat model, but its role in cardiomyopathy-related heart failure has not yet been corroborated (1). Decreased elastin/collagen ratio was one of the causes of adverse extracellular matrix remodeling in heart failure (32). Decorin is an extracellular matrix proteoglycan and is a member of leucine-rich protein family. Decorin can neutralize the activity of transforming growth factor-ß, which increases in hypertrophic heart failure through collagen accumulation. Decreased protein levels of decorin were shown in a spontaneously hypertensive heart failure rat model (32). Our RT-PCR results demonstrated increased decorin mRNA expression in human end-stage heart failure, suggesting a translational regulation defect, species variation, or the difference between the disease models used. CD59 is a complement regulatory protein (33), and the implications of its increased expression in heart failure are still unknown. Decreased phosphofructokinase expression in both DCM and HCM reflected low activity of glycolysis in heart failure (2).
Differentially expressed genes in DCM and HCM.
Several genes were found to be differentially expressed in DCM and HCM. In DCM, expression levels of atrial myosin alkali light chain, calsequestrin, lipocortin, and lumican were increased; whereas thioredoxin reductase was decreased. In HCM, expression levels of
B-crystallin and desmin increased, whereas mRNA levels of the antagonizer of myc transcriptional activity and ß-dystrobrevin decreased. Calsequestrin, a sarcoplasmic reticulum Ca2+ storage protein, was highly expressed in our DCM samples. Calsequestrin may play a role in the Ca2+ regulatory pathway of heart failure, as was indicated in one transgenic mouse model with overexpression of calsequestrin developing cardiac hypertrophy and heart failure (35). Reprogramming of gene expression in the failing myocardium is denoted by the change from
-myosin heavy chain (MHC) to ß-MHC gene expression. Myosin light chain isoform changes were described in the mRNA and protein levels of failing myocardium (7, 36), but the implication of increased expression of atrial myosin alkali light chain in DCM is still not clear. Recently, increased lumican expression was demonstrated in ischemic and reperfused rat heart, suggesting its contribution to myocardial fibrosis and regulation of collagen fiber assembly in DCM (3). Lipocortin belongs to an annexin family of calcium-dependent phospholipid-binding proteins, and several annexins were elevated in the failing heart (5), although they differ in cytosolic phospholipase A2 activity (24). Lipocortin also has anti-inflammatory actions and has been shown to reduce myocardial ischemia-reperfusion injury through leukocyte recruitment (8). Thioredoxin reductase has antioxidant activity (18), but its role in DCM warrants further investigation.
B-crystallin and desmin were shown to be highly expressed in HCM samples in our study. Altered expression in their mRNA in HCM has been reported previously (19, 47). In a study on the genetic dissection of left ventricular noncompaction or Barth syndrome, Ichida et al. (22) reported a novel mutation of the gene for
-dystrobrevin, a cytoskeletal protein, in a family with this disease. Whether the decreased ß-dystrobrevin in our HCM sample has genetic implications needs further study. Proto-oncogene expression might mediate the hypertrophic mechanism in heart failure (23). Enforced expression of myc proto-oncogene invokes a proliferation stimulus, and its activity might be suppressed by the Mad (antagonizer of myc transcriptional activity) family of proteins (53). It is intriguing that we showed decreased expression of Mad in HCM, partly contributing to the sustained increased myc proto-oncogene expression in HCM.
Pitfalls in two-color system cDNA microarray study.
Microarray experiments using two-color comparisons present potential pitfalls for data analysis. We do not measure gene expression level directly, but rather fluorescence intensity recorded by a scanner. Many factors influence the observed intensity levels including the following: differences of the amount of overall mRNA between two samples, concentration, brightness, dye labeling efficiency, exposure time, hybridization, washing stringencies, and scanning camera sensitivity (49). These factors may produce a multiplicative effect, creating a need for bias correction, or normalization, between the two color systems. We demonstrated that the normalization in two-color cDNA microarray experiments was slide-dependent and nonlinear, especially among those genes with lower fluorescence intensities. For selecting differentially expressed genes, many reports arbitrarily use 1.5-, 2.0-, or 3.0-fold changes as cutoff criteria, which may overlook differentially expressed genes with lesser, but statistically significant, fold changes (27).
Our real-time RT-PCR (41, 43, 52) results supported the hierarchical model we used to assess expression level and revealed that the accuracy rate of differentially expressed genes obtained from microarray data depended on fold change and the level of statistical significance. The chance of false-negative selection could not be assessed from this study, as we did not randomly pick up the nondifferentially expressed genes for RT-PCR confirmation. However, the expression of copper/zinc superoxide dismutase (which was increased in RT-PCR, but not on microarray analysis) in HCM samples may exemplify this caveat.
Study limitations.
In this study, we were limited by small DCM and HCM sample sizes, and the molecular mechanisms involved in cardiac hypertrophy and decompensated heart failure are complex. Furthermore, gene expression may be confounded by drug treatment in these end-stage heart failure patients. Therefore, we used pooled DCM, HCM, and adult normal heart samples for hybridization to minimize the effect of biological variation, and thus we tested the feasibility of our data mining algorithm. We also used the concept of confidence interval and statistical strength to minimize the possibility of false-positive results and to overcome the variability in expression levels across experiments (29, 49). Our study included only end-stage heart failure samples, and this was an obstacle to the observation of gene expression changes as heart failure developed.
In conclusion, an approach to select differentially expressed genes in cDNA microarray experiments was proposed and validated in this study. This offers the basis for advanced microarray data mining. Based on our results, we suggest that investigators could narrow down the gene lists for further study by choosing genes with more stringent levels of statistical significance and adding fold change criteria based on standard deviation of preselected reference genes (11). Several candidate genes with differential expression in DCM or HCM open the avenue for further diagnostic or therapeutic targets. The DNA microarray technology greatly facilitates the molecular characterization of DCM- and HCM-related heart failure on a genomic scale and provides us a preliminary molecular portrait of these two diseases in end-stage heart failure.
 |
ACKNOWLEDGMENTS
|
---|
We thank Dimitrios Stamatiou, James Ip, Youyuan Xu, Man Zhang, and Ming Zhu for assistance in the construction of the array, and Wing-Hung Wong at Harvard School of Public Health for instruction in the statistical analysis. We also acknowledge J. David Barrans, Leonard Anderson, and Richard E. Pratt for insight on analyzing microarray data, and Isolde Prince for manuscript preparation.
This work was supported in part by grants from the Heart and Stroke Foundation of Ontario, the Medical Research Council of Canada, the Canadian Genome Analysis and Technology Program (CGAT), and the National Institutes of Health (Grants 5RO1-HL-5851603, 5P5O-HL-5931603, and 5RO1-HL-6166102). J. J. Hwang was supported by the National Taiwan University Hospital, Taipei, Taiwan. C.-W. Lam was supported by the Croucher Foundation, Hong Kong and The Chinese University of Hong Kong. V. J. Dzau is the recipient of National Institutes of Health Merit Award 5R37-HL-3561016.
Part of the content was presented at the 5th Annual Scientific Meeting of the Heart Failure Society of America, Washington DC, September 912, 2001.
 |
FOOTNOTES
|
---|
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: C. C. Liew, Cardiovascular Genome Unit, Dept. of Medicine, Brigham and Womens Hospital, Harvard Medical School, 75 Francis St., Thorn 1326, Boston, MA 02115 (E-mail: cliew{at}rics.bwh.harvard.edu; URL, http://tcgu.bwh.harvard.edu).
10.1152/physiolgenomics.00122. 2001.
1 Supplementary Material to this article (APPENDIX Tables A and B) is available online at http://physiolgenomics.physiology.org/cgi/content/full/10/1/31/DC1. 
 |
REFERENCES
|
---|
- Arthur GD and Belcastro AN. A calcium stimulated cysteine protease involved in isoproterenol induced cardiac hypertrophy.
Mol Cell Biochem176
:241
248,1997
.[ISI][Medline]
- Auffermann W, Wu ST, Parmley WW, and Wikman-Coffelt J. Glycolysis in heart failure: a 31P-NMR and surface fluorometry study.
Basic Res Cardiol85
:342
357,1990
.[ISI][Medline]
- Baba H, Ishiwata T, Takashi E, Xu G, and Asano G. Expression and localization of lumican in the ischemic and reperfused rat heart.
Jpn Circ J65
:445
450,2001
.[ISI][Medline]
- Barrans JD, Stamatiou D, and Liew CC. Construction of a human cardiovascular cDNA microarray: portrait of the failing heart.
Biochem Biophys Res Commun280
:964
969,2001
.[ISI][Medline]
- Benevolensky D, Belikova Y, Mohammadzadeh R, Trouve P, Marotte F, Russo-Marie F, Samuel JL, and Charlemagne D. Expression and localization of the annexins II, V, and VI in myocardium from patients with end-stage heart failure.
Lab Invest80
:123
133,2000
.[ISI][Medline]
- Bowles NE, Bowles KR, and Towbin JA. The "final common pathway" hypothesis and inherited cardiovascular disease. The role of cytoskeletal proteins in dilated cardiomyopathy.
Herz25
:168
175,2000
.[ISI][Medline]
- Corbett JM, Why HJ, Wheeler CH, Richardson PJ, Archard LC, Yacoub MH, and Dunn MJ. Cardiac protein abnormalities in dilated cardiomyopathy detected by two-dimensional polyacrylamide gel electrophoresis.
Electrophoresis19
:2031
2042,1998
.[ISI][Medline]
- DAmico M, Di Filippo C, La M, Solito E, McLean PG, Flower RJ, Oliani SM, and Perretti M. Lipocortin 1 reduces myocardial ischemia-reperfusion injury by affecting local leukocyte recruitment.
FASEB J14
:1867
1869,2000
.[Abstract/Free Full Text]
- Dempsey AA, Pabalan N, Tang HC, and Liew CC. Organization of human cardiovascular-expressed genes on chromosome 21 and 22.
J Mol Cell Cardiol33
:587
591,2001
.[ISI][Medline]
- Dempsey AA, Ton C, and Liew CC. A cardiovascular EST repertoire: progress and promise for understanding cardiovascular disease.
Mol Med Today6
:231
237,2000
.[ISI][Medline]
- DeRisi J, Penland L, Brown PO, Bittner ML, Meltzer PS, Ray M, Chen Y, Su YA, and Trent JM. Use of a cDNA microarray to analyse gene expression patterns in human cancer.
Nat Genet14
:457
460,1996
.[ISI][Medline]
- Everett AD, Stoops TD, Nairn AC, and Brautigan D. Angiotensin II regulates phosphorylation of translation elongation factor-2 in cardiac myocytes.
Am J Physiol Heart Circ Physiol281
:H161
H167,2001
.[Abstract/Free Full Text]
- Fananapazir L and Epstein ND. Prevalence of hypertrophic cardiomyopathy and limitations of screening methods.
Circulation92
:7000
7004,1995
.
- Fananapazir L. Advances in molecular genetics and management of hypertrophic cardiomyopathy.
JAMA281
:1746
1752,1999
.[Free Full Text]
- Friddle CJ, Koga T, Rubin EM, and Bristow J. Expression profiling reveals distinct sets of genes altered during induction and regression of cardiac hypertrophy.
Proc Natl Acad Sci USA97
:6745
6750,2000
.[Abstract/Free Full Text]
- Gollob MH, Green MS, Tang A, Ahmad F, Hassan A, Gollob T, Lozado R, Gonzales O, Tapscott T, Karibe A, Fananapazir L, Bachinski L, and Roberts R. Identification of a gene responsible for Wolff-Parkinson-White Syndrome.
N Engl J Med344
:1823
1831,2001
.[Abstract/Free Full Text]
- Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri MA, Bloomfield CD, and Lander ES. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring.
Science286
:531
537,1999
.[Abstract/Free Full Text]
- Holmgren A. Antioxidant function of thioredoxin and glutaredoxin systems.
Antioxid Redox Signal2
:811
820,2000
.[Medline]
- Hwang DM, Dempsey AA, Lee CY, and Liew CC. Identification of differentially expressed genes in cardiac hypertrophy by analysis of expressed sequence tags.
Genomics66
:1
14,2000
.[ISI][Medline]
- Hwang DM, Dempsey AA, Wang RX, Rezvani M, Barrans JD, Dai KS, Wang HY, Ma H, Cukerman E, Liu YQ, Gu JR, Zhang JH, Tsui SKW, Waye MMY, Fung KP, Lee CY, and Liew CC. A genome-based resource for molecular cardiovascular medicine: toward a compendium of cardiovascular genes.
Circulation96
:4146
4203,1997
.[Abstract/Free Full Text]
- Hwang JJ, Dzau VJ, and Liew CC. Genomics and the pathophysiology of heart failure.
Current Cardiol Rep3
:198
207,2001
.
- Ichida F, Tsubata S, Bowles KR, Haneda N, Uese K, Miyawaki T, Dreyer WJ, Messina J, Li H, Bowles NE, and Towbin JA. Novel gene mutations in patients with left ventricular noncompaction or Barth syndrome.
Circulation103
:1256
1263,2001
.[Abstract/Free Full Text]
- Kai H, Muraishi A, Sugiu Y, Nishi H, Seki Y, Kuwahara F, Kimura A, Kato H, and Imaizumi T. Expression of proto-oncogenes and gene mutation of sarcomeric proteins in patients with hypertrophic cardiomyopathy.
Circ Res83
:594
601,1998
.[Abstract/Free Full Text]
- Kim S, Ko J, Kim JH, Choi EC, and Na DS. Differential effects of annexins I, II, III, and V on cytosolic phospholipase A2 activity: specific interaction model.
FEBS Lett489
:243
248,2001
.[ISI][Medline]
- Lee ML, Kuo FC, Whitmore GA, and Sklar J. Importance of replication in microarray gene expression studies: statistical methods and evidence from repetitive cDNA hybridizations.
Proc Natl Acad Sci USA97
:9834
9839,2000
.[Abstract/Free Full Text]
- Lim DS, Roberts R, and Marian AJ. Expression profiling of cardiac genes in human hypertrophic cardiomyopathy: insight into the pathogenesis of phenotypes.
J Am Coll Cardiol38
:1175
1180,2001
.[ISI][Medline]
- Liu TJ, Lai HC, Wu W, Chinn S, and Wang PH. Developing a strategy to define the effects of insulin-like growth factor-1 on gene expression profiles in cardiomyocytes.
Circ Res88
:1231
1238,2001
.[Abstract/Free Full Text]
- Maron BJ, Gardin JM, Flack JM, Gidding SS, Kurosaki TT, and Bild DE. Prevalence of hypertrophic cardiomyopathy in a general population of young adults: echocardiographic analysis of 4111 subjects in the CARDIA study.
Circulation92
:785
789,1995
.[Abstract/Free Full Text]
- Miller RA, Galecki A, and Shmookler-Reis RJ. Interpretation, design, and analysis of gene array expression experiments.
J Gerontol Biol Sci56A
:B52
B57,2001
.[Abstract/Free Full Text]
- Mills JC, Roth KA, Cagan RL, and Gordon JI. DNA microarrays and beyond: completing the journey from tissue to cell.
Nat Cell Biol3
:E175
E178,2001
.[ISI][Medline]
- Molkentin JD and Dorn GW II. Cytoplasmic signaling pathways that regulate cardiac hypertrophy.
Annu Rev Physiol63
:391
426,2001
.[ISI][Medline]
- Mujumdar VS and Tyagi SC. Temporal regulation of extracellular matrix components in transition from compensatory hypertrophy to decompensatory heart failure.
J Hypertens17
:261
270,1999
.[ISI][Medline]
- Murray KP, Mathure S, Kaul R, Khan S, Carson LF, Twiggs LB, Martens MG, and Kaul A. Expression of complement regulatory proteins-CD 35, CD 46, CD 55, and CD 59-in benign and malignant endometrial tissue.
Gynecol Oncol76
:176
182,2000
.[ISI][Medline]
- Olson TM, Doan TP, Kishimoto NY, Whitby FG, Ackerman MJ, and Fananapazir L. Inherited and de novo mutations in the cardiac actin gene cause hypertrophic cardiomyopathy.
J Mol Cell Cardiol32
:1687
1694,2000
.[ISI][Medline]
- Sato Y, Ferguson DG, Sako H, Dorn GW, Kadambi VJ, Yatani A, Hoit BD, Walsh RA, and Kranias EG. Cardiac-specific overexpression of mouse cardiac calsequestrin is associated with depressed cardiovascular function and hypertrophy in transgenic mice.
J Biol Chem273
:28470
28477,1998
.[Abstract/Free Full Text]
- Schaub MC and Hirzel HO. Atrial and ventricular isomyosin composition in patients with different forms of cardiac hypertrophy.
Basic Res Cardiol 82, Suppl2
:357
367,1987
.
- Schena M, Shalon D, Davis RW, and Brown PO. Quantitative monitoring of gene expression patterns with a complementary DNA microarray.
Science270
:467
470,1995
.[Abstract]
- Schonberger J and Seidman CE. Many roads lead to a broken heart: the genetics of dilated cardiomyopathy.
Am J Hum Genet69
:249
260,2001
.[Medline]
- Seidman CE. Hypertrophic cardiomyopathy: from man to mouse.
J Clin Invest106
:S9
S13,2000
.[ISI]
- Siwik DA, Tzortzis JD, Pimental DR, Chang DL, Pagano PJ, Singh K, Sawyer DB, and Colucci WS. Inhibition of copper-zinc superoxide dismutase induces cell growth, hypertrophic phenotype, and apoptosis in neonatal rat cardiac myocytes in vitro.
Circ Res85
:147
153,1999
.[Abstract/Free Full Text]
- Somura F, Izawa H, Iwase M, Takeichi Y, Ishiki R, Nishizawa T, Noda A, Nagata K, Yamada Y, and Yokota M. Reduced myocardial sarcoplasmic reticulum Ca2+-ATPase mRNA expression and biphasic force-frequency relations in patients with hypertrophic cardiomyopathy.
Circulation104
:658
663,2001
.[Abstract/Free Full Text]
- Stanton LW, Garrard LJ, Damm D, Garrick BL, Lam A, Kapoun AM, Zheng Q, Protter AA, Schreiner GF, and White RT. Altered patterns of gene expression in response to myocardial infarction.
Circ Res86
:939
945,2000
.[Abstract/Free Full Text]
- Steuerwald N, Cohen J, Herrera RJ, and Brenner CA. Analysis of gene expression in single oocytes and embryos by real-time rapid cycle fluorescence monitored RT-PCR.
Mol Hum Reprod5
:1034
1039,1999
.[Abstract/Free Full Text]
- Sussman MA, Welch S, Walker A, Klevitsky R, Hewett TE, Price RL, Schaefer E, and Yager K. Altered focal adhesion regulation correlates with cardiomyopathy in mice expressing constitutively active rac1.
J Clin Invest105
:875
886,2000
.[Abstract/Free Full Text]
- Takahashi T, Allen PD, and Izumo S. Expression of A-, B-, and C-type natriuretic peptide genes in failing and developing human ventricles: correlation with expression of the Ca2+-ATPase gene.
Circ Res71
:9
17,1992
.[Abstract]
- Tseng GC, Oh MK, Rohlin L, Liao JC, and Wong WH. Issues in cDNA microarray analysis: quality filtering, channel normalization, models of variations and assessment of gene effects.
Nucleic Acids Res29
:2549
2557,2001
.[Abstract/Free Full Text]
- Wang X, Osinska H, Klevitsky R, Gerdes AM, Nieman M, Lorenz J, Hewett T, and Robbins J. Expression of R120G-
B-crystallin causes aberrant desmin and
B-crystallin aggregation and cardiomyopathy in mice.
Circ Res89
:3
5,2001
.[Free Full Text]
- Wigle ED, Rakowski H, Kimball BP, and Williams WG. Hypertrophic cardiomyopathy: clinical spectrum and treatment.
Circulation92
:1680
1692,1995
.[Free Full Text]
- Wu TD. Analysing gene expression data from DNA microarrays to identify candidate genes.
J Pathol95
:53
65,2001
.
- Yang J, Moravec CS, Sussman MA, DiPaola NR, Fu D, Hawthorn L, Mitchell CA, Young JB, Francis GS, McCarthy PM, and Bond M. Decreased SLIM1 expression and increased gelsolin expression in failing human hearts measured by high-density oligonucleotide arrays.
Circulation102
:3046
3052,2000
.[Abstract/Free Full Text]
- Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J, and Speed TP. Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation.
Nucleic Acids Res30
: e15 (p.1
10),2002
.[Abstract/Free Full Text]
- Yuan A, Yu CJ, Luh KT, Chen WJ, Lin FY, Kuo SH, and Yang PC. Quantification of VEGF mRNA expression in non-small cell lung cancer using a real-time quantitative reverse transcription-PCR assay and a comparison with quantitative competitive reverse transcription-PCR.
Lab Invest80
:1671
1680,2000
.[ISI][Medline]
- Zhou ZQ and Hurlin PJ. The interplay between Mad and Myc in proliferation and differentiation.
Trends Cell Biol11
:S10
S14,2001
.[ISI][Medline]