Physiol. Genomics Add DOIs to your references at manuscript stage!
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Physiol. Genomics 19: 93-105, 2004. First published August 3, 2004; doi:10.1152/physiolgenomics.00040.2004
1094-8341/04 $5.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
19/1/93    most recent
00040.2004v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (13)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Zhao, M.
Right arrow Articles by Bernstein, D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Zhao, M.
Right arrow Articles by Bernstein, D.
Received 18 February 2004; accepted in final form 6 July 2004.
Physiological Genomics 19:93-105 (2004)
1094-8341/04 $5.00 © 2004 American Physiological Society

Microarray analysis of gene expression after transverse aortic constriction in mice

Mingming Zhao, Amy Chow, Jennifer Powers, Giovanni Fajardo and Daniel Bernstein

Department of Pediatrics, Stanford University, Stanford, California 94304


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
Cardiac hypertrophy is a compensatory response initially beneficial to heart function but can ultimately lead to cardiac decompensation. It is an integrated process involving multiple cellular signaling pathways and their cross talk. Microarray GeneChip technology is a powerful new tool to identify gene expression profiles of cardiac hypertrophy. To identify well-characterized as well as novel adaptive mechanisms, we utilized a murine model of compensated pressure overload hypertrophy (transverse aortic constriction, TAC). At 48 h, 10 days, and 3 wk, hearts were harvested and total RNA hybridized to Affymetrix U74Av2 GeneChips, which contain a 12,488-gene/EST probe set. Verification of gene expression was performed by SYBR quantitative real-time RT-PCR (QRT-PCR) for selected genes. A rigorous evaluation of the adequacy of the control condition was also performed. For statistical analysis we generated a four-step filtering criteria. Our results show an upregulation of 38 genes (48 h), 269 genes (10 days), and 203 genes (3 wk) and downregulation of 15 genes (48 h), 160 genes (10 days), and 124 genes (3 wk). Transcripts differentially expressed after TAC were categorized into 12 functional groups and revealed the presence of several intriguing transcripts, e.g., cell proliferation-related Ki-67 and several apoptosis-related genes. Overall changes in QRT-PCR were in accordance with GeneChip data, with the highest correlation for genes with the largest up- or downregulation with TAC. Thus TAC results in altered expression of genes in several pathways regulating both cardiac structure and function. However, for in vivo gene microarray experiments, it is critical to define adequate controls, perform rigorous statistical analysis, and provide validation by alternative methods.

cardiac hypertrophy; signal transduction; apoptosis; proliferation


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
CARDIAC HYPERTROPHY IS A COMPENSATORY response to a variety of internal or external stimuli that may initially be beneficial to heart function but which eventually can lead to heart failure. Cardiac myocytes respond to afterload stress by initiating sets of altered gene expression and activating a complex cross talk between signaling pathways, which leads to a state of "compensated" cardiac hypertrophy. The mechanisms responsible for cardiac hypertrophy have been extensively investigated in the hope of finding new therapeutic treatments to prevent progression to the "decompensated" state (4, 16, 22).

Many techniques have been utilized over the past two decades for exploring differentially expressed genes in cardiac hypertrophy, e.g., Northern blot analysis or RT-PCR (1, 5, 20). While informative, these techniques have been limited by the ability to examine only known pathways (targeted approach) and limited numbers of transcripts selected beforehand. Discovery of genes not already suspected of exhibiting differential expression is impossible using these methods. Recently, the application of gene microarray technology has provided us with the ability to rapidly screen and monitor differentially expressed genes on a broader, genomic scale and, more importantly, to identify novel genes that were not suspected to be altered in the process of cardiac hypertrophy (nontargeted approach).

In providing an all-encompassing picture of gene expression changes, microarray technology has great potential. Nevertheless, determining the validity of a list of putative differentially expressed genes has proved to be a challenge. Problems arise in the separation of real changes from experimental variations due to inherent variability between samples, variability between subjects (age, strain, sex, natural variation in gene expression), and in the response to experimental manipulations. These problems are multiplied when studies are performed in vivo. Finally, there is still considerable debate regarding the appropriate methods for statistical analysis of such large data sets, where the potential for making a type I error is high.

To examine the pattern of altered gene expression during the development of cardiac hypertrophy, we utilized a well-established murine model of compensated pressure overload hypertrophy, transverse aortic constriction (TAC). We performed microarray GeneChip analysis on TAC and sham-operated littermates, after periods of 48 h, 10 days, and 3 wk. To address issues of sample variability, we utilized a four-step statistical protocol to rigorously identify differentially expressed transcripts. We validated our results for several previous described "hypertrophy genes" as well as several novel genes discovered in our study using SYBR quantitative real-time RT-PCR (QRT-PCR). Finally, we evaluated several methods for providing for adequate control conditions, including the effects of anesthesia, surgery, and secondary experimental manipulations. We demonstrate that controlling for these sources of potential data error is critically important for in vivo gene microarray studies.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
Animal preparation.
Three-month-old male FVB mice (body wt 28–33 g) were used in all studies to reduce variability in gene expression due to gender or strain. Mice were housed in similar conditions in the research animal facility at Stanford University. All protocols were approved by the Stanford Administrative Panel for Laboratory Animal Care and were consistent with the Guide for the Care and Use of Laboratory animals published by the National Institutes of Health.

Anesthesia was induced with 3% isoflurane and maintained with 1.5% isoflurane. TAC was performed (n = 5 in each group) via a left medial thoracotomy incision, avoiding the pleural space and hence the need for artificial ventilation, as described by Rockman et al. (38). A 7- silk suture was placed around the transverse aorta between the left common carotid artery and the brachiocephalic trunk and tied tight around both the aorta and a 27-gauge needle, which was then removed, yielding a reproducible degree of constriction. Sham-operated controls consisted of age-matched littermates which underwent an identical surgical procedure including isolation of the aorta, only without placement of the suture.

At 48 h, 10 days, or 3 wk after surgery, mice were killed, and hearts quickly were removed, weighed, and placed in RNALater solution (Qiagen, Valencia, CA) for 1 h to prevent RNA degradation. Heart weights and body weights were recorded.

In a first control study to determine the acute and chronic effects of anesthesia and surgery, we examined gene expression in a separate group of sham mice 2 h, 48 h, 10 days, and 3 wk after operation and compared gene expression profiles with mice that had not undergone any surgery (n = 4–6 in each group). In a second control study, to determine whether a commonly used method to check the adequacy of the TAC procedure affected gene expression, we allowed TAC and sham mice (n = 4 in each group) to recover for 1 wk, then measured the blood pressure gradient across the band by direct cannulation of both carotid arteries. Mice were killed within 40 min after this procedure.

GeneChip preparation.
At no time were samples pooled between mice in any experimental group. Total RNA was isolated from 45–60 mg heart tissue using the RNeasy Mini Kit from Qiagen, and 5–10 µg RNA was then reverse transcribed to double-stranded cDNA. cDNAs were purified via Phase Lock Gel-phenol/chloroform extraction followed by ethanol precipitation. Labeled cRNA was synthesized by incubation of 1 µg cDNA with biotin-labeled ribonucleotides and RNA polymerase for 5 h at 37°C using the BioArray High Yield RNA transcript labeling kit from Enzo Diagnostics (Farmingdale, NY). At each step, concentration and purity of RNA samples were checked by measuring absorbency in a spectrophotometer at 260 nm and the 260 nm/280 nm ratio, respectively. Integrity of RNA was determined using formaldehyde agarose gel electrophoresis. cRNA transcripts were purified from the in vitro transcription (IVT) reaction using the RNeasy Mini Kit. Biotin-labeled cRNAs were then fragmented by heating. Hybridization of these fragments to the mouse genome array U74Av2 (Affymetrix, Santa Clara, CA) and scanning for signal intensity were carried out by the Protein and Nucleic Acid Biotechnology Facility at Stanford.

Data analysis.
To determine differentially expressed genes in the mouse heart after TAC, raw data was first analyzed using Affymetrix Microarray Suite 5.0 software. Briefly, the U74Av2 oligonucleotide GeneChip contains 12,488 known genes and expressed sequence tags (ESTs). Each gene/EST probe set is represented on the GeneChip by 20 pairs of perfectly matched (PM) and mismatched (MM) oligos. The number of instances where the PM signal was greater than the MM signal was determined, and the average of the logarithm of the PM:MM ratio was used in a matrix-based algorithm that determined whether a particular cRNA was actually detected (present), not detected (absent), or marginally detected (marginal). The results were next published via the Affymetrix Micro DB software, and the Data Mining Tools 3.0 analysis package was used to display the query results, evaluate and compare replicate data, and calculate fold changes.

An individual gene had to be called "present" in at least two of the total (8–9) samples for inclusion in our study. Comparisons were then performed between each experimental and matched sham group (48 h TAC vs. 48 h sham; 10 day TAC vs. 10 day sham; 3 wk TAC vs. 3 wk sham) by a two-tailed, unpaired Student’s t-test to identify those differentially expressed genes. Analyses were then performed to determine the effects of choosing more or less conservative P values (<0.001, <0.01 and <0.05). Based on these comparisons, a P value <0.05 was adopted for final studies. Additionally, a minimum value of "fold change" of 1.5 was used as another filter factor (see RESULTS).

The filtered data sets were then uploaded into the NetAffx Gene Ontology (GO) Analysis Mining Tool (Affymetrix) to review a graph of GO terms associated with those data sets. GO Mining Tool software provided the readily available GO terms for annotated genes and graphical, interactive views of the biological process and molecular function. Transcripts in our data sets that were identified as significantly changed (P < 0.05), and with a fold change >1.5 in the TAC group compared with the sham, were categorized into 12 functional groups.

Confirmation of the GeneChip data.
Verification of altered gene expression was performed by SYBR QRT-PCR using the same total RNA used for the microarray analyses. Seven genes previously well described as up- or downregulated by the hypertrophic process, and three novel genes found in our study, were selected for QRT-PCR: atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), sarcoplasmic reticulum calcium ATPase (SERCA), {alpha}-skeletal muscle actin, early growth response 1 (EGR-1), Jun oncogene (c-Jun), calcium/calmodulin-dependent serine protein kinase (CaM kinase), pigment epithelium-derived factor (PEDF), proliferation-related Ki-67 antigen (Ki-67), and secreted frizzled-related protein 3 (SFRP3). The SYBR QRT-PCR reaction was performed in 96-well plates in reaction buffer containing 5 mM MgSO4 (QuantiTect SYBR Green RT-PCR Kit, Qiagen), 0.5 µM gene-specific primers (Table 1), and 50–100 ng/well of total RNA. Assays were performed in an ABI Prism 5500 sequence detection system (Applied Biosystems, Foster City, CA). Samples from each mouse were run in triplicate and averaged for final RNA quantitation. The results were compared with GeneChip data by Student’s t-test using Statview software (SAS, Cary, NC).


View this table:
[in this window]
[in a new window]
  		Table 1. List of the primer pairs used in QRT-PCR

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 GRANTS
 REFERENCES
 
Acute and chronic effects of anesthesia and surgery.
There were marked changes in gene expression due to the stress of anesthesia and surgery, beginning as early as 2 h postoperatively, with many persisting much longer than anticipated. When comparing sham-operated mice with nonstressed controls (without any surgery and anesthesia) at 48 h postoperatively (using a filtering algorithm, described below, including transcripts with at least two "present" or "marginal" calls out of all samples and a value of P < 0.05), there were a total of 805 genes upregulated and 1,376 genes downregulated, indicating a very strong influence on cardiac gene expression induced by surgical stress in the mouse (Fig. 1). Unexpectedly, some of these changes persisted for weeks. Using the same filtering algorithm, there were 702 genes upregulated and 1,088 genes downregulated in the sham-operated animals at 3 wk.



View larger version (16K):
[in this window]
[in a new window]
  		Fig. 1. Effects of anesthesia, sham surgery, and double carotid cannulation on cardiac gene expression. The open bars show upregulated genes when comparing sham-operated mice with nonstressed controls (without any experimental manipulation) at 2 h, 48 h, 10 days, and 3 wk. The solid bars show downregulated genes. Sham-operated mice showed a large number of changes in gene expression as early as 2 h which persisted for several weeks. Double carotid cannulation also trigged a dramatic number of transcript alterations when compared with control mice.

 
Acute effects of double-cannulation procedure.
Double carotid cannulation has been routinely utilized to evaluate the degree of TAC by measuring the blood pressure gradient between the right and left carotid arteries. This procedure is performed under anesthesia, but is considered less invasive than the TAC procedure, as it does not require entry into the chest cavity and can be accomplished fairly rapidly by a trained technician. Analyzing the results of 12 mice undergoing double cannulation, we found an excellent correlation between peak systolic pressure gradient and heart weight-to-body weight ratio (data not shown). However, the stress of the double-cannulation procedure triggered the alteration of a large set of genes (Fig. 1) when compared with non-operated controls, even though heart samples were obtained within 30 min of the initiation of anesthesia.

Thus, based on the results of these two preliminary studies, to accurately control for gene expression changes associated solely with the hypertrophic process, we used as controls mice that had undergone sham surgeries at the same time point as the TAC mice and avoided the use of double cannulation prior to gene analysis. Because of similar concerns about the effects of stress on gene expression, we also avoided the use of echocardiography for assessment of the banding procedure. Instead we relied on visual examination of the band at the time of necropsy, the presence of a size discrepancy between the right and left carotid arteries, and most importantly an increase in heart weight/body weight ratio as an indication of the success of the TAC procedure. TAC induced a reproducible degree of cardiac hypertrophy as quantified by heart weight-to-body weight ratios, which increased by 30% at 10 days and 64% at 3 wk (Table 2).


View this table:
[in this window]
[in a new window]
  		Table 2. Heart weights and body weights after various durations of TAC

 
Quality control and data analysis.
To determine whether a specific gene is actually present in the sample, Affymetrix Microarray Suite software categorizes genes into present (P), marginal (M), or absent (A) based on the relative signal intensity of the transcripts on the GeneChip. Out of 12,488 probe sets on the Affymetrix GeneChip, an average of 5,225 (SD 327) genes and ESTs were identified as "present" in our samples, for an average of 42% (SD 2.6) (this serves as a quality control for the sample, with the expected percentage being 35–50%). Genes with low signal intensity, generally signifying a low expression level, are identified as "absent." Analysis of fold changes in a specific gene in which an absent call was made across all of the samples would appear to be meaningless, although there is disagreement on whether this labeling algorithm should even be used. Thus, to further evaluate the effect of this algorithm, we examined the total number of genes identified as up- or downregulated under increasingly stricter criteria: first all genes, then genes with a minimum of one, two, three, four, or five "P" or "M" calls in both TAC and sham groups (Fig. 2). Based on this analysis, it was evident that most of the filtering occurred when a requirement for a minimum of 1 "P" call was included. Thus we chose to filter our expression data so that only transcripts with at least two P or M calls out of all samples were included. Using these initial filtering criteria, 5,982 transcripts were selected at 48 h, 5,797 transcripts were selected at 10 days, and 5,593 transcripts were selected at 3 wk for further analysis.



View larger version (40K):
[in this window]
[in a new window]
  		Fig. 2. Comparison of different filtering algorithms showing the number of transcripts at 48 h, 10 days, and 3 wk using the Affymetrix system for identifying genes as either "present" or "marginal" based on the number of matched or mismatched oligonucleotide pairs. The leftmost bars show the total number of genes on the Affymetrix U74Av2 GeneChip compared with genes identified as showing a minimum of one, two, three, four, or five present or marginal (grouped together here as "P") calls.

 
We next determined the influence of increasing levels of statistical stringency to determine an appropriate cutoff P value that would generate a manageable number of genes without an excessive risk of eliminating important positive findings (type I error). Analyses were performed gradually less stringent using P values of 0.001, 0.01, and 0.05 after 48 h, 10 days, and 3 wk of TAC on data previously filtered as noted above. The numbers of differentially expressed genes, comparing TAC with sham-operated controls at each P value, are listed in Table 3.


View this table:
[in this window]
[in a new window]
  		Table 3. Number of differentially expressed transcripts after 48 h, 10 days, or 3 wk of TAC at various P values

 
As expected, increasing the stringency by decreasing the P value from 0.05 to 0.001 had a major impact on the number of transcripts identified as significantly altered. Within the transcripts that passed the P < 0.05 cutoff criterion, only 3–15% also passed the P < 0.001 cutoff criterion (Table 3). With the most stringent P value of <0.001, we would expect any genes identified to have the highest probability of being truly differentially regulated (lowest type II error). Figure 3A illustrates the effect of different P values on the 10-day samples, plotting a total of 5,797 transcripts. Points above the top horizontal line represent the 149 transcripts identified as upregulated with the highest level of confidence, those that passed the most stringent threshold (P < 0.001).



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 3. A: visual representation of different filtering algorithms (using different P values) on the number of positively identified genes. Each point represents one of 5,797 genes with at least two "P" calls, plotted after 10 days of transverse aortic constriction (TAC). The x-axis shows the log2 fold change in gene expression. The y-axis shows the –log ratios of the various P values. The top horizontal line in the graph represents a P value of 0.001. The bottom horizontal line represents a P value of 0.05. It is evident that identification of differentially expressed genes is greatly influenced by the stringency of the cutoff P value. B: visual representation of the filtering algorithm using as a cutoff a fold change in gene expression of >1.5, represented as those points lying outside the two vertical lines. The 274 points in the top right sector are upregulated genes that passed the thresholds of a P value <0.05 and a fold change >1.5 fold. The 164 points in the top left sector are downregulated genes passing the same filtering criteria.

 
However, choosing too stringent a P value would significantly increase the potential for false negatives (type I error) and could result in data showing involvement of a novel pathway being lost. Thus we elected to use a less stringent P value of <0.05 to generate a list of eligible genes, with the knowledge that further confirmation, e.g., by RT-PCR would be required. We then applied one final filter to the data set, an analysis of fold change, representing the percent increase (or decrease) in signal between experimental and control samples. When this filter was set at 1.5-fold, this approach resulted in the selection of one-third of all previously positively identified transcripts (Fig. 3B).

There remain inherent problems with these filtering criteria. Selection of differentially expressed genes based on fold change favors genes with baseline low signal intensity. To correct for this, we utilized a minimum signal intensity value: for transcripts that were upregulated, the value of signal intensity in the TAC group was set at >500, and for transcripts that were downregulated, the value of signal intensity in the sham group was set at >500. Only a very small number of transcripts were filtered by these last steps. A summary of our approach to GeneChip data analysis is outlined in Fig. 4.



View larger version (14K):
[in this window]
[in a new window]
  		Fig. 4. Flow chart demonstrating the effects of each data mining filtering step on the number of differentially identified transcripts between TAC and sham-operated mice. All transcripts were subjected to three filtering steps: transcripts were first selected with at least two "present" (or "marginal") calls; then data sets were filtered by Student’s t-test with a P value <0.05; next, only transcripts that showed a fold change >1.5 were filtered. Application of these criteria identified 438 differentially expressed transcripts at 10 days and 333 transcripts at 3 wk, whereas only 57 genes at 48 h passed these filtering steps. Finally, signal intensity values were used to identify genes that were upregulated (signal intensity in the TAC group >500) and genes that were downregulated (signal intensity in the sham group >500).

 
Detection of genes differentially expressed after 48 h, 10 days, and 3 wk of TAC.
After 10 days of TAC, there were 269 transcripts identified as upregulated by our filtering algorithm. Of these, 175 transcripts were upregulated only at 10 days and had returned to baseline by 3 wk. After 3 wk of TAC, 203 transcripts were identified as upregulated. Of these, 109 were upregulated only at 3 wk. There were only 38 transcripts uniquely upregulated after 48 h of TAC. The expression of 83 transcripts was elevated at both 10-day and 3-wk time points, and 11 transcripts showed increased levels at all three time points (Fig. 5). Half of the differentially regulated transcripts represented known genes, and the remainder were unannotated ESTs.



View larger version (11K):
[in this window]
[in a new window]
  		Fig. 5. Differential expression of genes in TAC vs. sham-operated mice, comparing 48 h, 10 days, and 3 wk. We found 24 genes were upregulated after TAC only at 48 h, 172 upregulated only at 10 days, and 109 upregulated only at 3 wk. We found 83 genes upregulated at both 10 days and 3 wk, 3 genes upregulated at both 48 h and 10 days, and 11 upregulated at all three time points.

 
Table 4 lists a subset of upregulated genes with at least a twofold change (TAC vs. sham). List A shows those transcripts concordant for upregulation at both 10 days and 3 wk. These transcripts included those previously associated with cardiac hypertrophy, e.g., {alpha}-skeletal muscle actin, slow myosin heavy chain, osteoblast-specific factor 2 (29), and thrombospondin-1, as well as some novel genes, e.g., PEDF. List B shows those transcripts upregulated at 48 h only, including natriuretic peptide precursor type B (BNP). List C shows those transcripts upregulated at 10 days only. Among those genes were early growth response genes 1 and 2, FBJ osteosarcoma oncogene, Ki-67, and skeletal muscle ß-tropomyosin. List D shows those transcripts upregulated at 3 wk only.


View this table:
[in this window]
[in a new window]
  		Table 4. List of upregulated genes at different time points, ranked by fold change (TAC/sham >2-fold except for 48 h where TAC/sham ≥ 1.5 fold)

 
Categorization of differentially expressed genes.
We next categorized known transcripts into 12 functional groups using the GO Mining Tool, based on their biological function (Table 5). The first six categories divide transcripts based primarily on biological process (e.g., apoptosis); the next six are based primarily on molecular function (e.g., signal transduction). For example, of the 12,488 transcripts on the Affymetrix U74 mouse GeneChip, 192 were classified in the apoptosis pathway. Of these, three transcripts were upregulated after 10 days and four transcripts were upregulated at 3 wk. The absolute largest number of upregulated transcripts were in the biological function groups related to metabolism, cell growth and cell communication, and in the molecular function groups related to binding activity and enzyme activity.


View this table:
[in this window]
[in a new window]
  		Table 5. Characterization of altered transcripts by functional category

 
Confirmation with SYBR QRT-PCR.
Seven previously well-characterized cardiac hypertrophy-associated genes were selected for validation by QRT-PCR, and the results were compared with GeneChip data at two time points: ANP, BNP, {alpha}-skeletal muscle actin (the predominant form in fetal hearts), CaM kinase, EGR-1, and c-Jun (immediate-early genes). At 10 days, six of the seven genes showed good correlation between QRT-PCR and GeneChip in terms of both the direction and magnitude of change (Fig. 6A). For four of these genes, there was concordance in statistical significance, whereas for one (ANP), the increase in expression was noted to be significant by PCR but not by GeneChip, although the direction of change and magnitude were similar. A larger sample size would probably have produced concordance in this case as well. The exception was for CaM kinase, which was found to be upregulated by PCR but not on the GeneChip. At 3 wk, some of the earlier differentially expressed genes (e.g., ANP, EGR-1 and c-Jun) had already returned to baseline (Fig. 6B). PCR and GeneChip again showed good concordance for six of these seven genes, with the exception being Serca 2, which was identified as downregulated on the GeneChip but not by QRT-PCR. The highest level of correlation between QRT-PCR and GeneChip was for {alpha}-actin, EGR-1, ANP, and BNP genes (Fig. 7), which all had high signal intensity on the GeneChip (high levels of expression). In contrast, there was a poor level of correlation for c-Jun and CaM kinase, which both had low signal intensity on the GeneChip (low levels of expression).



View larger version (28K):
[in this window]
[in a new window]
  		Fig. 6. Comparison of relative expression level for 7 selected genes after 10 days (A) and 3 wk (B) of TAC showing results obtained with GeneChip vs. SYBR quantitative real-time RT-PCR (QRT-PCR). Expression level for GeneChip is based on signal intensity and for QRT-PCR on calculated expression x100 (quantitative comparison of actual expression level between GeneChip and QRT-PCR is not intended). *Statistically significant at P < 0.05. Bars are means, with lines as SD.

 


View larger version (23K):
[in this window]
[in a new window]
  		Fig. 7. Comparison of gene expression from paired samples using SYBR QRT-PCR and GeneChip after 3 wk of TAC. QRT-PCR showed a high level of correlation with GeneChip results for {alpha}-actin, EGR-1, ANP, BNP, PEDF, and SFRP3 genes, representing predominantly those genes with high levels of expression on the GeneChip (with the exception of SFRP3). In contrast, there was poor correlation for c-Jun, CaM kinase, and Ki-67 genes, representing those genes with low levels of expression on the GeneChip. The x-axis represents the expression level of SYBR QRT-PCR, and the y-axis represents the signal intensity on the GeneChip. Open triangles represent sham samples, and solid circles represent TAC samples.

 
We also performed QRT-PCR for three novel genes that have not been previously associated with cardiac hypertrophy in the literature: PEDF, Ki-67, and SFRP3. Two of these showed a high degree of correlation with GeneChip results (Fig. 7). However, there was a lesser correlation with the GeneChip profile for Ki-67, which had a low level of signal intensity. Overall, these QRT-PCR results represent a 70% concordance with GeneChip data, with the highest level of concordance for those genes with high signal intensity on the GeneChip.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
GRANTS
 REFERENCES
 
In the present study, we used microarray GeneChip technology to identify a set of genes with altered cardiac expression during the development of physiological hypertrophy. Pressure overload was induced by TAC, a model that closely resembles the clinical scenario observed in patients with coarctation of the aorta and is somewhat similar to that in patients with aortic stenosis or systemic hypertension. We chose a degree and duration of TAC that does not induce cardiac failure or decompensated hypertrophy.

Delineation of an adequate control group has long been a standard of physiological research, especially for in vivo animal models of cardiovascular disease, where perturbations induced by surgical manipulations may be greater than those induced by the experimental condition. Unfortunately, it not uncommon for this to be forgotten when new technologies are first employed, and this has been the case for many early studies utilizing gene microarrays. The need for careful attention to experimental variables has led to the establishment of a set of guidelines for publication of studies using microarray technologies (2, 7, 17, 36). However, the degree to which gene array data is susceptible to alteration by common experimental variables such as anesthesia and surgery, and the duration of these alterations, has not been well characterized for cardiovascular disease models. The present study shows how critical it is to define exacting controls when evaluating gene expression. An acute stress, such as induced by anesthesia, a major surgical procedure, or even a "minor" procedure, such as vessel cannulation, can have significant effects on cardiac gene expression, beginning within the first few minutes or hours. The shortest interval between the initiation of the surgical procedure and gene expression analysis that we studied was 45–60 min, although it is likely that even shorter durations of stress could alter expression of some early response genes (35). The breadth of gene expression changes related to stress was dramatic, with a large number of genes altered after the double-cannulation procedure. Furthermore, some of these changes can persist for as long as a week after surgery, so that sham operations where the procedure is as nearly identical to the experimental condition are the most robust controls. To minimize the likelihood of interference from recent procedural stress, we used heart weight-to-body weight ratio to confirm the results of banding (31, 39). Whether newer techniques of nonanesthetized echocardiography in mice result in less (or more) stress and the induction of less (or more) gene changes is not known.

The utility of microarray technology in biomedical research is highly dependent on the development of bioinformatic and statistical methods used for analysis of such large data sets. Previous reports have utilized varied statistical models based on differing assumptions (25, 34, 44); however, there is as yet no unified approach to analysis of microarray data. In the present study, we examined the effect of choosing increasingly more stringent criteria on the size of the resulting data sets. Based on our preliminary studies, we developed a four-step statistical approach to identify differentially expressed genes, illustrated in Fig. 4. An initial filtering step was based on the Affymetrix algorithm for calling a particular transcript "present" or "absent" based on a comparison of signals from perfectly matched vs. mismatched oligonucleotide pairs. As expected, the application of this filter reduced the number of genes in the data sets dramatically; however, this effect was maximized by requiring that at least two samples show a "present" or "marginal" call. This first filtering reduced the data set by 50%, from ~12,000 transcripts to ~6,000. A second filtering step was then applied using Student’s t-test for statistical comparison, with increasingly stringent P values (Fig. 3A). We chose a P value of <0.05 to reduce the possibility of filtering out relevant gene changes, although with a larger sample size, the use of a lower P value could be justified. Selection of an overly stringent P value (P < 0.001) markedly truncated the data set and introduced the risk of a type I error. The last filter applied was related to fold change. The ideal algorithm for gene microarray analysis would ensure that we neither miss biologically important genes with a small fold change, nor misidentify genes with large fold change but no statistical significance due to high inherent variability. By applying the cutoff of >1.5 fold change to our data set, we identified 53 potential genes (38 upregulated and 15 downregulated) after 48 h, 429 genes (269 upregulated and 160 downregulated) after 10 days, and 327 genes (203 upregulated and 124 downregulated) after 3 wk of TAC. Adding an additional filter related to signal intensity eliminated less than 10 genes from each of these three data sets, implying that the transcripts that "survived" our first three filtering criteria were those that already had relatively high levels of expression as illustrated by higher signal intensities.

Correlation of GeneChip results with QRT-PCR or Northern blot is an important part of confirming whether a specific transcript is up- or downregulated. Our QRT-PCR results demonstrate a 70% concordance with GeneChip data, which is consistent with the findings of Rajeevan et al. (37). Their results are also in agreement with our finding that the degree of correlation between GeneChip and QRT-PCR is related to hybridization intensity, with the highest level of concordance for those genes with the highest signal intensity on the GeneChip.

During the development of compensatory hypertrophy, the greatest number of upregulated transcripts was in the biological functional groups related to metabolism, cell growth, and cell communication, and in the molecular function groups related to binding activity and enzyme activity. As expected, TAC mice showed enhanced expression of several immediate early genes and embryonic marker genes classically associated with the development of cardiac hypertrophy, i.e., EGR-1, c-Jun, BNP, and {alpha}-skeletal muscle actin. Our study also identified several hypertrophy-related genes only more recently reported by others: osteoblast-specific factor 2 (24, 29), angiotensin converting enzyme (27), four and half LIM domains (50), Bnip3L/Nix (51), thrombospondin, fibrillin, biglycan, S100 calcium-binding protein isoform, epithelial membrane protein 1, and fibulin 2, as well as several extracellular matrix genes encoding procollagen isoforms (29, 47).

There were many differential changes in gene expression between the 48-h, 10-day, and 3-wk time points. As seen in Fig. 5, only 11 transcripts were increased at all three time points; however, there was temporal variance in their level of expression. For example, BNP expression was highest at 48 h (3.5-fold change) and decreased thereafter (1.7-fold at 10 days and 1.4-fold at 3 wk). A similar pattern was seen for interferon-ß (3.9-, 2.5-, and 2.0-fold changes at 48 h, 10 days, and 3 wk, respectively). Among the 269 transcripts upregulated at 10 days, 172 were uniquely upregulated at this time point, including early growth response gene, c-Jun, procollagen, and elastin. All had returned to baseline levels of expression by 3 wk. Our study demonstrated only a small number of genes (38) that were upregulated at 48 h compared with the sham controls, largely due to the elimination of many additional genes that were also upregulated due to surgical stress. Of these 38, 24 were limited to the 48-h time point, 3 remained upregulated at 10 days, and 11 showed persistently increased expression at both 10 days and 3 wk (Table 4).

Our results have both similarities and differences to the TAC data of the CardioGenomics group, documented on their web site (10). For example, this group also reported elevated expression of several genes at 48 h: {alpha}-skeletal muscle actin (1.7-fold increase in our data set vs. 1.9-fold in theirs), BNP (3.5-fold vs. 15.7-fold), spi2 proteinase inhibitor (2.0-fold vs. 2.5-fold), and interferon-ß (3.9-fold vs. 14.8-fold). They did not find any change for three of the novel genes reported in our study: PEDF, Ki-67, and SFRP3, even though we confirmed upregulation for two of these three by QRT-PCR. In contrast to the present study, the CardioGenomics study utilized a smaller sample size (n = 3) and a more heterogenous level of TAC (echo Doppler gradients in their study ranged from as low as 16 mmHg to as high as 82 mmHg), as well as a different algorithm for statistical analysis (cluster analysis). Notably, we filtered low-intensity signals (<500) to eliminate background noise, whereas in the CardioGenomics study, 1/3 of their 48-h samples had signal intensities <500. By applying more stringent criteria, fewer transcripts "survived" our filtering steps. Finally, we avoided stressing the animals after the initial surgery, whereas in the CardioGenomics study, TAC gradient was measured with echo Doppler. This additional stress may explain some of the differences between the two data sets, especially in the expression of early response genes.

Our analysis also identified several novel hypertrophy-associated genes, e.g., Ki-67, which showed a dramatic sixfold increase at 10 days. Although the correlation between GeneChip and QRT-PCR was low for this transcript, even by PCR four of the five TAC samples were increased compared with the sham samples. Ki-67 is a nuclear antigen expressed in proliferating but not in quiescent cells. Expression of Ki-67 occurs preferentially during late G1, S, G2, and M phases of the cell cycle, whereas in cells in G0 phase Ki-67 antigen cannot be detected (41). Since the expression of Ki-67 is strictly associated with cell proliferation, it is widely used as a "proliferation marker" for cycling cells in tumor diagnosis (40). The increased expression of this transcript would be unexpected based on the classic dogma that cardiac myocytes are terminally differentiated cells and are unable to reenter the cell cycle and divide. This contention has been challenged more recently by several studies which demonstrate that the number of ventricular myocytes does increase in the decompensated human heart (32) and document myocytes in the mitosis stage (21). Beltrami et al. (3) showed Ki-67 antigen present in the nuclei of myocytes from both control and postinfarction hearts by confocal microscopy. This result and others (26) demonstrate the potential for myocyte proliferation in the stressed heart, although this is still a matter of controversy. However, to date, there has been no single example of a Ki-67-positive cell that cannot divide (28, 41). Because of the discrepancy between GeneChip and QRT-PCR data, further studies will be required, with larger numbers of subjects and perhaps additional time points, to resolve the issue. However, if confirmed, our demonstration of marked upregulation of Ki-67 antigen suggests that myocyte proliferation may contribute to the increase in myocardial mass after TAC. Alternatively, it is possible that the increased level of Ki-67 is due to the proliferation of other cardiac cell populations, such as fibroblasts, endothelial cells, or vascular smooth muscle cells.

At 3 wk, once the state of stable hypertrophy had been reached, one of the transcripts most significantly (6-fold) upregulated was SFRP3. SFRP3 is a seven-transmembrane receptor with a large extracellular cysteine-rich domain which functions to antagonize Wnt activity by sequestering Wnt and preventing its binding to the frizzled receptor (18). Previous studies have demonstrate that Wnt/frizzled is involved in the regulation of cell proliferation and differentiation (15). However, little is known about the role of the Wnt/frizzled pathway in cardiac overload hypertrophy and apoptosis. Recently, Schumann et al. (42) detected expression of SFRP3 in cardiomyocytes by in situ RT-PCR. Comparing samples from failing and nonfailing hearts, they found differential expression of secreted frizzled-related proteins (SFRP3 and 4) in relation to apoptosis-related gene expression. Our microarray data confirms this previous report of cardiac expression of SFRP3 and its potential role in regulating hypertrophic growth. We also detected enhanced expression of cyclin D1 (1.6-fold) and c-Jun (1.8-fold) during TAC. These two components of the WNT/frizzled signaling pathway regulate cell cycle progression and have been shown to be involved in cardiac hypertrophic growth. c-Jun is also a key component of the SAPK/JNK signaling pathway (8, 9). By demonstrating increased expression of SFRP3, c-Jun, and cyclin D1, our results suggest that chronically overloaded cardiomyocytes might become more susceptible to apoptosis through the activation of Wnt and other signaling pathways.

Finally, we also identified several novel gene families not previously known to increase in expression during the development of stable cardiac hypertrophy. One example is PEDF, a member of the serine protease inhibitor (serpin) family (43), initially discovered because of its neurotrophic activity. PEDF is produced by retinal cells to maintain angiostasis in the normal vitreous and cornea. A recent study showed that PEDF and thrombospondin-1 (increased 2.9-fold with TAC) are potent natural inhibitors of angiogenesis (46). The anti-angiogenic activity of PEDF and thrombospondin-1 is dependent on the induction of Fas/FasL and resulting apoptosis, although we did not find upregulation of Fas/FasL during TAC.

Among the several mechanisms invoked in the transition from stable to decompensated cardiac hypertrophy is the loss of cardiomyocytes via the induction of apoptotic pathways (30, 48). Increased mechanical load has been proposed as a stimulant of cardiomyocyte apoptosis (19). Condorelli and colleagues (13) showed a dramatic upregulation of the pro-apoptotic gene Bax and reduced Bcl-2/Bax ratio in a rat model of TAC, predisposing cardiomyocytes to apoptosis (13). Using the GeneChip technique, we detected several differentially expressed genes involved in both pro- and anti-apoptotic pathways even during this compensated phase (Table 6).


View this table:
[in this window]
[in a new window]
  		Table 6. Changes in genes related to pro- and anti-apoptotic pathways at different time points after TAC

 
Although only a few apoptosis-related genes passed our selection criteria, they are all key components of well-described programmed cell death pathways, including those mediated by caspases, Fas, TNF, ATM/p53, integrins, MAPK, chemokines, and Wnt. For example, Bourdon et al. (6) showed that apoptosis induced by the p53/Scotin pathway is caspase dependent. Chen et al. (11) demonstrated that Bnip3L is upregulated in myocardial hypertrophy and that the expression of Bnip3L can overcome the effect of the anti-apoptotic BCL2 family. In contrast, the anti-apoptotic protein PEA15 is a death effector domain-containing protein. Kitsberg et al. (23) have shown that PEA15 protects astrocytes from TNF-induced apoptosis. Finally, De Smaele et al. (14) showed that upregulation of Gadd45ß blunted the activation of JNKs and suppressed apoptosis (14). Thus, in the development of cardiac hypertrophy, we found elements of both pro-apoptotic and pro-survival signaling. The relative abundance of these different proteins may determine the balance between survival and death for cells during the transition from stable to decompensated hypertrophy.

Our study has several limitations. First, we only studied three major time points in the development of cardiac hypertrophy. At none of these time points was there development of cardiac decompensation, so that the transcription changes associated with this change were not studied. However, the finding of alterations in transcripts associated with apoptotic pathways in our study of "compensated" hypertrophy suggests that these could be targets for future study using more severe degrees of TAC which lead to heart failure. Furthermore, we studied whole hearts rather than isolated left ventricles. We chose this approach to avoid pooling tissues between animals, which we felt was potentially detrimental to the statistical analysis we used. Although the degree of left ventricular hypertrophy with TAC resulted in the very vast majority of the cardiac mass being that of the left ventricle, contamination with right ventricular and atrial tissue could have diluted results for some genes, the expression of which was increased only in the stressed left ventricle. Further studies will be important to elucidate the chamber-specific gene expression changes associated with TAC in the mouse. Finally, our study examines gene changes in the myocardium as a whole; thus the contribution of any individual cell type would obviously require further study.

In conclusion, our results demonstrate that the development and maintenance of myocardial hypertrophy involves a coordinate response between several major gene programs. We also demonstrate the importance of obtaining adequate controls, especially when using in vivo models, given the marked (both rapid and persistent) gene changes associated with experimental procedures such as surgery, anesthesia, and vessel cannulation. GeneChip data have a high level of correlation with data from other methods such as QRT-PCR when the transcript in question is expressed at a relatively high level. Still, correlation of GeneChip results with QRT-PCR is warranted before any definitive conclusions can be drawn. With these caveats in mind, GeneChip analysis of cardiac hypertrophy in mice allows the analysis of genes using a nontargeted approach, will lead to the elucidation of new control mechanisms, and may potentially provide new candidates for pharmacotherapy to prevent the deleterious effects of long-standing cardiac hypertrophy.


   GRANTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
REFERENCES
 
This work was supported by National Heart, Lung, and Blood Institute Grant HL-61535 (to D. Bernstein). G. Fajardo was supported by a Fellowship from the American Heart Association, Western Affiliates.


   ACKNOWLEDGMENTS
 
All microarray data have been submitted to the NCBI Gene Expression Omnibus (GEO) database, to comply with the MIAME standards for microarray data.


   FOOTNOTES
 
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).

Address for reprint requests and other correspondence: D. Bernstein, 750 Welch Road Suite 305, Palo Alto, CA 94304 (E-mail: danb{at}stanford.edu).


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 

  1. Akiyama Y, Ashizawa N, Seto S, Ohtsuru A, Kuroda H, Ito M, Yamashita S, and Yano K. Involvement of receptor-type tyrosine kinase gene families in cardiac hypertrophy. J Hypertens 17: 1329–1337, 1999.[CrossRef][Web of Science][Medline]
  2. Ball CA, Sherlock G, Parkinson H, Rocca-Sera P, Brooksbank C, Causton HC, Cavalieri D, Gaasterland T, Hingamp P, Holstege F, Ringwald M, Spellman P, Stoeckert CJ Jr, Stewart JE, Taylor R, Brazma A, and Quackenbush J. The underlying principles of scientific publication. Bioinformatics 18: 1409, 2002.[Free Full Text]
  3. Beltrami AP, Urbanek K, Kajstura J, Yan SM, Finato N, Bussani R, Nadal-Ginard B, Silvestri F, Leri A, Beltrami CA, and Anversa P. Evidence that human cardiac myocytes divide after myocardial infarction. N Engl J Med 344: 1750–1757, 2001.[Abstract/Free Full Text]
  4. Bocker W, Hupf H, Grimm D, Kurzidim K, and Schunkert H. Effects of indapamide in rats with pressure overload left ventricular hypertrophy. J Cardiovasc Pharmacol 36: 481–486, 2000.[CrossRef][Web of Science][Medline]
  5. Boluyt MO, O’Neill L, Meredith AL, Bing OH, Brooks WW, Conrad CH, Crow MT, and Lakatta EG. Alterations in cardiac gene expression during the transition from stable hypertrophy to heart failure. Marked upregulation of genes encoding extracellular matrix components. Circ Res 75: 23–32, 1994.[Abstract/Free Full Text]
  6. Bourdon JC, Renzing J, Robertson PL, Fernandes KN, and Lane DP. Scotin, a novel p53-inducible proapoptotic protein located in the ER and the nuclear membrane. J Cell Biol 158: 235–246, 2002.[Abstract/Free Full Text]
  7. Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P, Stoeckert C, Aach J, Ansorge W, Ball CA, Causton HC, Gaasterland T, Glenisson P, Holstege FC, Kim IF, Markowitz V, Matese JC, Parkinson H, Robinson A, Sarkans U, Schulze-Kremer S, Stewart J, Taylor R, Vilo J, and Vingron M. Minimum information about a microarray experiment (MIAME): toward standards for microarray data. Nat Genet 29: 365–371, 2001.[CrossRef][Web of Science][Medline]
  8. Bueno OF, De Windt LJ, Tymitz KM, Witt SA, Kimball TR, Klevitsky R, Hewett TE, Jones SP, Lefer DJ, Peng CF, Kitsis RN, and Molkentin JD. The MEK1-ERK1/2 signaling pathway promotes compensated cardiac hypertrophy in transgenic mice. EMBO J 19: 6341–6350, 2000.[CrossRef][Web of Science][Medline]
  9. Busk PK, Bartkova J, Strom CC, Wulf-Andersen L, Hinrichsen R, Christoffersen TE, Latella L, Bartek J, Haunso S, and Sheikh SP. Involvement of cyclin D activity in left ventricle hypertrophy in vivo and in vitro. Cardiovasc Res 56: 64–75, 2002.[Abstract/Free Full Text]
  10. CardioGenomics PGA. Pressure-Overload Induced Cardiac Hypertrophy. http://cardiogenomics.med.harvard.edu/groups/proj1/pages/band_phys.html (11 March 2004).
  11. Chen G, Cizeau J, Vande Velde C, Park JH, Bozek G, Bolton J, Shi L, Dubik D, and Greenberg A. Nix and Nip3 form a subfamily of pro-apoptotic mitochondrial proteins. J Biol Chem 274: 7–10, 1999.[Abstract/Free Full Text]
  12. Colotta F, Polentarutti N, Sironi M, and Mantovani A. Expression and involvement of c-fos and c-jun protooncogenes in programmed cell death induced by growth factor deprivation in lymphoid cell lines. J Biol Chem 267: 18278–18283, 1992.[Abstract/Free Full Text]
  13. Condorelli G, Morisco C, Stassi G, Notte A, Farina F, Sgaramella G, de Rienzo A, Roncarati R, Trimarco B, and Lembo G. Increased cardiomyocyte apoptosis and changes in proapoptotic and antiapoptotic genes bax and bcl-2 during left ventricular adaptations to chronic pressure overload in the rat. Circulation 99: 3071–3078, 1999.[Abstract/Free Full Text]
  14. De Smaele E, Zazzeroni F, Papa S, Nguyen DU, Jin R, Jones J, Cong R, and Franzoso G. Induction of gadd45ß by NF{kappa}B downregulates pro-apoptotic JNK signalling. Nature 414: 308–313, 2001.[CrossRef][Medline]
  15. Dierick H and Bejsovec A. Cellular mechanisms of wingless/Wnt signal transduction. Curr Top Dev Biol 43: 153–190, 1999.[Web of Science][Medline]
  16. Donohue TJ, Dworkin LD, Ma J, Lango MN, and Catanese VM. Antihypertensive agents that limit ventricular hypertrophy inhibit cardiac expression of insulin-like growth factor-I. J Investig Med 45: 584–591, 1997.[Medline]
  17. Firestein GS and Pisetsky DS. DNA microarrays: boundless technology or bound by technology? Guidelines for studies using microarray technology. Arthritis Rheum 46: 859–861, 2002.[CrossRef][Web of Science][Medline]
  18. Hu E, Zhu Y, Fredrickson T, Barnes M, Kelsell D, Beeley L, and Brooks D. Tissue restricted expression of two human Frzbs in preadipocytes and pancreas. Biochem Biophys Res Commun 247: 287–293, 1998.[CrossRef][Web of Science][Medline]
  19. Janssen PM, Hasenfuss G, Zeitz O, Lehnart SE, Prestle J, Darmer D, Holtz J, and Schumann H. Load-dependent induction of apoptosis in multicellular myocardial preparations. Am J Physiol Heart Circ Physiol 282: H349–H356, 2002; doi:10.1152/ajpheart.00024.2001.[Abstract/Free Full Text]
  20. Kaiser M, Kahr O, Shimada Y, Smith P, Kelly M, Mahadeva H, Adams M, Lodwick D, Aalkjaer C, Avkiran M, and Samani NJ. Differential regulation of ventricular adrenomedullin and atrial natriuretic peptide gene expression in pressure and volume overload in the rat. Clin Sci (Lond) 94: 359–365, 1998.[Medline]
  21. Kajstura J, Leri A, Finato N, Di Loreto C, Beltrami CA, and Anversa P. Myocyte proliferation in end-stage cardiac failure in humans. Proc Natl Acad Sci USA 95: 8801–8805, 1998.[Abstract/Free Full Text]
  22. Kessler-Icekson G, Barhum Y, Schlesinger H, Shohat J, Sharma H, and Schaper W. Molecular manifestations of cardiac hypertrophy in the spontaneously hypertensive rat effects of antihypertensive treatments. Adv Exp Med Biol 382: 195–203, 1995.[Medline]
  23. Kitsberg D, Formstecher E, Fauquet M, Kubes M, Cordier J, Canton B, Pan G, Rolli M, Glowinski J, and Chneiweiss H. Knock-out of the neural death effector domain protein PEA-15 demonstrates that its expression protects astrocytes from TNF{alpha}-induced apoptosis. J Neurosci 19: 8244–8251, 1999.[Abstract/Free Full Text]
  24. Kruzynska-Frejtag A, Machnicki M, Rogers R, Markwald RR, and Conway SJ. Periostin (an osteoblast-specific factor) is expressed within the embryonic mouse heart during valve formation. Mech Dev 103: 183–188, 2001.[CrossRef][Web of Science][Medline]
  25. 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 USA 97: 9834–9839, 2000.[Abstract/Free Full Text]
  26. Leri A, Barlucchi L, Limana F, Deptala A, Darzynkiewicz Z, Hintze TH, Kajstura J, Nadal-Ginard B, and Anversa P. Telomerase expression and activity are coupled with myocyte proliferation and preservation of telomeric length in the failing heart. Proc Natl Acad Sci USA 98: 8626–8631, 2001.[Abstract/Free Full Text]
  27. Li P, Chen PM, Wang SW, and Chen LY. Time-dependent expression of chymase and angiotensin converting enzyme in the hamster heart under pressure overload. Hypertens Res 25: 757–762, 2002.[CrossRef][Web of Science][Medline]
  28. MacCallum DE and Hall PA. The biochemical characterization of the DNA binding activity of pKi67. J Pathol 191: 286–298, 2000.[CrossRef][Web of Science][Medline]
  29. Mirotsou M, Watanabe CM, Schultz PG, Pratt RE, and Dzau VJ. Elucidating the molecular mechanism of cardiac remodeling using a comparative genomic approach. Physiol Genomics 15: 115–126, 2003. First published August 5, 2003; doi:10.1152/physiolgenomics.00071.2003.[Abstract/Free Full Text]
  30. Nediani C, Celli A, Formigli L, Perna AM, Fiorillo C, Ponziani V, Ibba-Manneschi L, Zecchi-Orlandini S, Nosi D, Liguri G, Modesti PA, and Nassi P. Possible role of acylphosphatase, Bcl-2 and Fas/Fas-L system in the early changes of cardiac remodeling induced by volume overload. Biochim Biophys Acta 1638: 217–226, 2003.[Medline]
  31. Ogawa T, Linz W, Stevenson M, Bruneau BG, Kuroski de Bold ML, Chen JH, Eid H, Scholkens BA, and de Bold AJ. Evidence for load-dependent and load-independent determinants of cardiac natriuretic peptide production. Circulation 93: 2059–2067, 1996.[Abstract/Free Full Text]
  32. Olivetti G, Melissari M, Balbi T, Quaini F, Sonnenblick EH, and Anversa P. Myocyte nuclear and possible cellular hyperplasia contribute to ventricular remodeling in the hypertrophic senescent heart in humans. J Am Coll Cardiol 24: 140–149, 1994.[Abstract]
  33. Oltvai ZN, Milliman CL, and Korsmeyer SJ. Bcl-2 heterodimerizes in vivo with a conserved homolog, Bax, that accelerates programmed cell death. Cell 74: 609–619, 1993.[CrossRef][Web of Science][Medline]
  34. Pan W, Lin J, and Le CT. A mixture model approach to detecting differentially expressed genes with microarray data. Funct Integr Genomics 3: 117–124, 2003.[Medline]
  35. Peng M, Huang L, Xie Z, Huang WH, and Askari A. Partial inhibition of Na+/K+-ATPase by ouabain induces the Ca2+-dependent expressions of early-response genes in cardiac myocytes. J Biol Chem 271: 10372–10378, 1996.[Abstract/Free Full Text]
  36. Pollock JD. Gene expression profiling: methodological challenges, results, and prospects for addiction research. Chem Phys Lipids 121: 241–256, 2002.[CrossRef][Web of Science][Medline]
  37. Rajeevan MS, Vernon SD, Taysavang N, and Unger ER. Validation of array-based gene expression profiles by real-time (kinetic) RT-PCR. J Mol Diagn 3: 26–31, 2001.[Abstract/Free Full Text]
  38. Rockman HA, Ross RS, Harris AN, Knowlton KU, Steinhelper ME, Field LJ, Ross J Jr, and Chien KR. Segregation of atrial-specific and inducible expression of an atrial natriuretic factor transgene in an in vivo murine model of cardiac hypertrophy. Proc Natl Acad Sci USA 88: 8277–8281, 1991.[Abstract/Free Full Text]
  39. Rockman HA, Wachhorst SP, Mao L, and Ross J Jr. ANG II receptor blockade prevents ventricular hypertrophy and ANF gene expression with pressure overload in mice. Am J Physiol Heart Circ Physiol 266: H2468–H2475, 1994.[Abstract/Free Full Text]
  40. Schluter C, Duchrow M, Wohlenberg C, Becker MH, Key G, Flad HD, and Gerdes J. The cell proliferation-associated antigen of antibody Ki-67: a very large, ubiquitous nuclear protein with numerous repeated elements, representing a new kind of cell cycle-maintaining proteins. J Cell Biol 123: 513–522, 1993.[Abstract/Free Full Text]
  41. Scholzen T and Gerdes J. The Ki-67 protein: from the known and the unknown. J Cell Physiol 182: 311–322, 2000.[CrossRef][Web of Science][Medline]
  42. Schumann H, Holtz J, Zerkowski HR, and Hatzfeld M. Expression of secreted frizzled related proteins 3 and 4 in human ventricular myocardium correlates with apoptosis related gene expression. Cardiovasc Res 45: 720–728, 2000.[Abstract/Free Full Text]
  43. Steele FR, Chader GJ, Johnson LV, and Tombran-Tink J. Pigment epithelium-derived factor: neurotrophic activity and identification as a member of the serine protease inhibitor gene family. Proc Natl Acad Sci USA 90: 1526–1530, 1993.[Abstract/Free Full Text]
  44. Troyanskaya OG, Garber ME, Brown PO, Botstein D, and Altman RB. Nonparametric methods for identifying differentially expressed genes in microarray data. Bioinformatics 18: 1454–1461, 2002.[Abstract/Free Full Text]
  45. Verhagen AM, Silke J, Ekert PG, Pakusch M, Kaufmann H, Connolly LM, Day CL, Tikoo A, Burke R, Wrobel C, Moritz RL, Simpson RJ, and Vaux DL. HtrA2 promotes cell death through its serine protease activity and its ability to antagonize inhibitor of apoptosis proteins. J Biol Chem 277: 445–454, 2002.[Abstract/Free Full Text]
  46. Volpert OV, Zaichuk T, Zhou W, Reiher F, Ferguson TA, Stuart PM, Amin M, and Bouck NP. Inducer-stimulated Fas targets activated endothelium for destruction by anti-angiogenic thrombospondin-1 and pigment epithelium-derived factor. Nat Med 8: 349–357, 2002.[CrossRef][Web of Science][Medline]
  47. Weinberg EO, Mirotsou M, Gannon J, Dzau VJ, Lee RT, and Pratt RE. Sex dependence and temporal dependence of the left ventricular genomic response to pressure overload. Physiol Genomics 12: 113–127, 2003. First published November 26, 2002; doi:10.1152/physiolgenomics.00046.2002.[Abstract/Free Full Text]
  48. Wollert KC and Drexler H. Regulation of cardiac remodeling by nitric oxide: focus on cardiac myocyte hypertrophy and apoptosis. Heart Fail Rev 7: 317–325, 2002.[CrossRef][Medline]
  49. Woo M, Hakem R, Soengas MS, Duncan GS, Shahinian A, Kagi D, Hakem A, McCurrach M, Khoo W, Kaufman SA, Senaldi G, Howard T, Lowe SW, and Mak TW. Essential contribution of caspase 3/CPP32 to apoptosis and its associated nuclear changes. Genes Dev 12: 806–819, 1998.[Abstract/Free Full Text]
  50. 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. Circulation 102: 3046–3052, 2000.[Abstract/Free Full Text]
  51. Yussman MG, Toyokawa T, Odley A, Lynch RA, Wu G, Colbert MC, Aronow BJ, Lorenz JN, and Dorn GW II. Mitochondrial death protein Nix is induced in cardiac hypertrophy and triggers apoptotic cardiomyopathy. Nat Med 8: 725–730, 2002.[Web of Science][Medline]



This article has been cited by other articles:


Home page
 Am. J. Physiol. Heart Circ. Physiol.Home page
I. Banerjee, J. W. Fuseler, A. R. Intwala, and T. A. Baudino
IL-6 loss causes ventricular dysfunction, fibrosis, reduced capillary density, and dramatically alters the cell populations of the developing and adult heart
Am J Physiol Heart Circ Physiol, May 1, 2009; 296(5): H1694 - H1704.
[Abstract] [Full Text] [PDF]

Home page
 Circ Cardiovasc GenetHome page
S. P. Sheehy, S. Huang, and K. K. Parker
Time-Warped Comparison of Gene Expression in Adaptive and Maladaptive Cardiac Hypertrophy
Circ Cardiovasc Genet, April 1, 2009; 2(2): 116 - 124.
[Abstract] [Full Text] [PDF]

Home page
 J. Thorac. Cardiovasc. Surg.Home page
W. E. Stansfield, P. C. Charles, R.-h. Tang, M. Rojas, R. Bhati, N. C. Moss, C. Patterson, and C. H. Selzman
Regression of pressure-induced left ventricular hypertrophy is characterized by a distinct gene expression profile
J. Thorac. Cardiovasc. Surg., January 1, 2009; 137(1): 232 - 238.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Physiol. Heart Circ. Physiol.Home page
T. Urashima, M. Zhao, R. Wagner, G. Fajardo, S. Farahani, T. Quertermous, and D. Bernstein
Molecular and physiological characterization of RV remodeling in a murine model of pulmonary stenosis
Am J Physiol Heart Circ Physiol, September 1, 2008; 295(3): H1351 - H1368.
[Abstract] [Full Text] [PDF]

Home page
 Physiol. GenomicsHome page
C. Cerutti, M. Kurdi, G. Bricca, W. Hodroj, C. Paultre, J. Randon, and M.-P. Gustin
Transcriptional alterations in the left ventricle of three hypertensive rat models
Physiol Genomics, November 21, 2006; 27(3): 295 - 308.
[Abstract] [Full Text] [PDF]

Home page
 Physiol. GenomicsHome page
S. Schiekofer, I. Shiojima, K. Sato, G. Galasso, Y. Oshima, and K. Walsh
Microarray analysis of Akt1 activation in transgenic mouse hearts reveals transcript expression profiles associated with compensatory hypertrophy and failure
Physiol Genomics, October 11, 2006; 27(2): 156 - 170.
[Abstract] [Full Text] [PDF]

Home page
 FASEB J.Home page
L. S. Spruill and P. J. McDermott
Regulation of c-jun mRNA expression in adult cardiocytes by MAP kinase interacting kinase-1 (MNK1)
FASEB J, October 1, 2006; 20(12): 2133 - 2135.
[Abstract] [Full Text] [PDF]

Home page
 Physiol. GenomicsHome page
M. Liang and B. Ventura
Physiological genomics in PG and beyond: July to September 2005
Physiol Genomics, October 17, 2005; 23(2): 119 - 124.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
19/1/93    most recent
00040.2004v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (13)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Zhao, M.
Right arrow Articles by Bernstein, D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Zhao, M.
Right arrow Articles by Bernstein, D.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Visit Other APS Journals Online
Copyright © 2004 by the American Physiological Society.