Ischemic but not pharmacological preconditioning elicits a gene expression profile similar to unprotected myocardium

Rafaela da Silva1,2,*, Eliana Lucchinetti1,2,*, Thomas Pasch2, Marcus C. Schaub1 and Michael Zaugg1,2

1 Institute of Pharmacology and Toxicology, University of Zurich
2 Institute of Anesthesiology, University Hospital Zurich, Zurich, Switzerland


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
Pharmacological (PPC) and ischemic preconditioning (IschPC) provide comparable protection against ischemia in the heart. However, the genomic phenotype may depend on the type of preconditioning. Isolated perfused rat hearts were used to evaluate transcriptional responses to PPC and IschPC in the presence (mediator/effector response) or absence (trigger response) of 40 min of test ischemia using oligonucleotide microarrays. IschPC was induced by 3 cycles of 5 min of ischemia, and PPC by 15 min of 2.1 vol% isoflurane. Unsupervised analysis methods were used to identify gene expression patterns. PPC and IschPC were accompanied by marked alterations in gene expression. PPC and IschPC shared only ~25% of significantly up- and downregulated genes after triggering. The two types of preconditioning induced a more uniform genomic response after ischemia/reperfusion. Numerous genes separated preconditioned from unprotected ischemic hearts. Three stable gene clusters were identified in the trigger response to preconditioning, while eight stable clusters related to cytoprotection, inflammation, remodeling, and long interspersed nucleotide elements (LINEs) were delineated after prolonged ischemia. A single stable sample cluster emerged from cluster analysis for both IschPC and unprotected myocardium, indicating a close molecular relationship between these two treatments. Principal component analysis revealed differences between PPC vs. IschPC, and trigger vs. mediator/effector responses in transcripts predominantly related to biosynthesis and apoptosis. IschPC and PPC similarly but distinctly reprogram the genetic response to ischemic injury. IschPC elicits a postischemic gene expression profile closer to unprotected myocardium than PPC, which may be therefore more advantageous as therapeutic strategy in cardioprotection.

preconditioning; gene array analysis; ischemia


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
PRECONDITIONING is a most powerful means of attaining myocardial protection against prolonged ischemia (27). It can be elicited by single or multiple brief episodes of sublethal ischemia (IschPC) or by pharmacological agents (PPC). This phenomenon called classic or early preconditioning establishes a transient protective cellular state in the myocardium by complex multiple fast-acting signaling steps lasting 2–3 h. Preconditioning further triggers a delayed cardiac protection, also called late preconditioning or "second window of protection," which occurs 12–24 h after the initial preconditioning stimulus and is effective for 3–4 days (25). This delayed protection relies on altered gene activity.

Volatile anesthetics emerged as a model class of agents eliciting PPC with low toxicity and high clinical applicability (51, 53). Even small doses of volatile anesthetics are capable of producing profound cardioprotection. PPC by volatile anesthetics and IschPC similarly augment postischemic functional recovery, decrease infarct size, elicit a "second window of protection" (41, 45), and, most importantly, were shown to occur in humans with coronary artery disease (20). The signaling cascades of both types of preconditioning involve several G-protein-coupled receptors and alterations in nitric oxide and free oxygen radical formation and point to the key role of protein kinase C (47) as signal amplifier and to KATP channels as the main end effectors in preconditioning (52). Conversely, despite the same degree of structural and functional protection, differences with respect to key signaling steps were also reported (5, 47). These include differential activation and translocation of protein kinase C isoforms to subcellular targets (47) as well as the role of other intracellular kinases (5) in triggering and mediating preconditioning-induced cardioprotection.

Functional genomics aims at analyzing the regulation of genes in response to changes in physiological parameters. Microarray technology revolutionized the analysis of gene expression in biological processes by enabling to assess gene activity on a genome-wide scale in a single experiment. Both early and late preconditioning affect gene expression in the heart. While protection by late preconditioning directly depends on altered gene expression, early preconditioning modulates the adverse consequences of prolonged ischemia at the gene expression level. Given the complex interactions in cardioprotection by preconditioning, preconditioning might be better characterized by the expression patterns of protective and antiprotective genes. Despite distinct signaling pathways between different types of preconditioning, there may exist overlapping genetic modifiers. Thus the transcriptional comparison might unravel novel protective genes expressed in a coordinated manner and shared across different types of preconditioning. In the quest for novel mechanisms underlying preconditioning, IschPC and PPC elicited by the volatile anesthetic isoflurane were compared with respect to their pre- and postischemic genomic responses.

The data presented herein provide additional new insights into the molecular similarities of the transcriptional responses between different types of preconditioning in the myocardium and ultimately aid to conceptualize the molecular events surrounding the remarkable protection achieved by preconditioning.


    MATERIALS AND METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
All experimental protocols used in this investigation were approved by the Animal Care and Use Committee of the University of Zurich. All experimental procedures conformed to the Guiding Principles in the Care and Use of Animals of the American Physiological Society and were in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH publication 85-23, revised 1996).

Langendorff Isolated Heart Preparation
Male Wistar rats (250 g) were heparinized (500 U ip) and 20 min later were decapitated without prior anesthesia. The hearts were rapidly removed and perfused in a noncirculating Langendorff apparatus with Krebs-Henseleit buffer (in mmol/l: 155 Na+, 5.6 K+, 138 Cl, 2.1 Ca2+, 1.2 PO43–, 25 HCO3, 0.56 Mg2+, and 11 glucose) gassed with 95% O2-5% CO2 and maintained at a pH of 7.4 and a temperature of 37°C. Perfusion pressure was set to 80 mmHg. A water-filled balloon was inserted into the left ventricle and inflated to set an end-diastolic pressure of 0–5 mmHg during the initial equilibration. Data were recorded as previously described in detail (5, 47).

Perfusion Protocols and Hemodynamics
Hearts were allowed to equilibrate for 10 min and to beat spontaneously in all experiments (Fig. 1). PPC was induced by the volatile anesthetic isoflurane (APC-TRI, APC) administered for 15 min at 1.5 MAC (minimum alveolar concentration, 2.1 vol%). The buffer solution was equilibrated with isoflurane using an Isotec 3 vaporizer (Datex-Ohmeda, Tewksbury, MA) with an air bubbler. Applied concentration of isoflurane was measured in the buffer solution using a gas chromatograph (PerkinElmer, Norwalk, CT): isoflurane 2.1% (vol/vol) (1.5 MAC in rats at 37°C), 0.52 mM (SD 0.04). IschPC (IPC-TRI, IPC) was induced by 3 cycles of 5 min ischemia interspersed by 5 min of reperfusion. Preconditioned hearts were subjected to 40 min of ischemia followed by 180 min of reperfusion (APC, IPC: mediator/effector responses) or followed by 220 min of perfusion only (APC-TRI, IPC-TRI: trigger responses). Nonpreconditioned hearts subjected to ischemia and reperfusion served as ischemic control (ISCH). Control group (CTL) consisted of time-matched perfused hearts (a total of 270 min of perfusion). For each experimental group, five hearts were prepared and functional parameters were recorded (Fig. 1). Repeated-measures analysis of variance was used to evaluate differences over time between groups. Paired t-tests were used to compare within groups over time, and unpaired t-tests were used to compare groups at identical time points (SigmaStat v. 2.0; SPSS Science, Chicago, IL). Post-hoc Bonferroni test for multiple comparisons was used. Corrected P < 0.05 was considered to be statistically significant. Data are presented as means with SD in parentheses.



View larger version (21K):
[in this window]
[in a new window]
 
Fig. 1. Scheme of treatment protocols. APC-TRI, administration of the preconditioning stimulus isoflurane; IPC-TRI, administration of the preconditioning stimulus consisting of 3 cycles of ischemia interspersed by 5 min of reperfusion; APC, preconditioning with isoflurane followed by 40 min of test ischemia; IPC, preconditioning with ischemia followed by 40 min of ischemia; ISCH, nonpreconditioned hearts exposed to 40 min of ischemia; CTL, time-matched perfusion. Isoflurane was administered at 2.1 vol% for 15 min followed by 10 min of washout before 40 min of test ischemia. In each group, 5 hearts were used.

 
RNA Isolation and cDNA Synthesis
Left ventricular tissue was rapidly frozen in liquid nitrogen and stored at –80°C. Hearts were powdered in liquid nitrogen and homogenized in TRIzol LS reagent (Invitrogen, Basel, Switzerland) and chloroform/isoamyl alcohol (Fluka, Buchs, Switzerland). The aqueous phase was mixed with isopropanol and precipitated over night at –20°C. The pellet was washed with isopropanol, dried at 37°C, and eluted in diethyl pyrocarbonate-treated H2O. In the presence of the T7-(T)24 RNA polymerase promoter primer (Microsynth, Balgach, Switzerland), single-stranded cDNA synthesis from total RNA was performed using SuperScript II (Invitrogen). Double-stranded cDNA was synthesized with SuperScript kit (Invitrogen). Biotin-labeled antisense cRNA was synthesized in vitro using a high-yield RNA transcript labeling kit (BioArray; Enzo, Farmingdale, NY).

Microarray Hybridization and Scanning
Affymetrix Rat Genome U34A array (Affymetrix, Santa Clara, CA) was used for gene expression profiling. The U34A GeneChip contains a total of 8,799 probe sets representing ~7,000 known rat genes and 1,000 expressed sequence tags (ESTs). Five independent GeneChips for each group were used, resulting in a total of 30 GeneChips analyzed. The biotin-labeled cRNA was fragmented in fragmentation buffer (200 mM Tris-acetate, 50 mM KOAc, 150 mM MgOAc, at pH 8.1) and hybridized to the oligonucleotides in hybridization solution containing 15 µg fragmented cRNA in MES buffer (0.1 M MES, 1.0 M NaCl, 0.01% Triton X-100, at pH 6.7) and herring sperm DNA. GeneChips were placed in a hybridization oven at 60 rpm and 45°C for 16 h. Afterward, arrays were first washed at 22°C with SSPE-T (0.9 M NaCl, 60 mM NaH2PO4, 6 mM EDTA, 0.005% Triton X-100, at pH 7.6) and subsequently with 0.1 MES at 45°C for 30 min. The GeneChips were then stained with a streptavidin-phycoerythrin conjugate (Molecular Probes, Leiden, The Netherlands) and washed. Additional staining with anti-streptavidin antibody and streptavidin-phycoerythrin conjugate was used to enhance the signals. GeneChips were scanned at a resolution of 3 µm using a confocal scanner (model 900154; Affymetrix). From the 30 U34A GeneChips analyzed, one GeneChip of the IPC-TRI group did not satisfy the stringent quality criteria and was therefore excluded from further analysis. For all other experimental groups, five GeneChips, each resulting from an individual experiment, were of high quality and entered the subsequent bioinformatics analysis. The data are available at the Gene Expression Omnibus (GEO) web site under the series number GSE1616 (http://www.ncbi.nlm.nih.gov/geo/).

Analysis of Gene Expression Data
A flowchart illustrating the individual steps of data analysis can be viewed in Supplemental Fig. S1 (the Supplemental Material for this article is available online at the Physiological Genomics web site).1

Step 1: Normalization and computation of expression values.
Normalization and computation of expression values were performed using the robust multichip average (RMA) method (19) implemented in the module affy (10) of the BioConductor open-source bioinformatics software (http://www.bioconductor.org) in the R programming environment. R (http://www.r-project.org; Ref. 18) is a widely used open-source language for statistical computing and graphics. RMA performs the following operations: 1) probe-specific background correction to compensate for nonspecific binding using perfect-match (PM) distribution rather than PM-mismatch (MM) values, 2) probe level multichip quantile normalization to unify PM distributions across all GeneChips, 3) and robust probe set summary of the log-normalized probe-level data by median polishing.

Step 2: Statistical filtering and unsupervised clustering methods.
To select probe sets with a statistical significant alteration in signal intensity, the gene expression matrix was filtered using analysis of variance (ANOVA with P value = 0.01). To investigate similarities of the expression pattern across treatments, unsupervised clustering methods (principal component analysis, hierarchical clustering) were applied to the filtered data. Principal component analysis was performed using both the entire filtered data matrix and the gene lists according to functional classifications in Gene Ontology (GO, http://www.geneontology.org; Ref. 1). Hierarchical clustering was performed using the coupled two-way clustering (CTWC) algorithm (2, 11, 12). The concept and underlying philosophy of this algorithm is based on an analogy to the physics of inhomogeneous ferromagnets and has been previously described in detail (2). The method identifies submatrices of the total expression matrix, whose clustering analysis reveals partitions of genes and samples into stable classes. The transcripts analyzed were rearranged as ordered by the clustering algorithm, so that transcripts with the most similar expression patterns, as measured by the Euclidean distance, were placed adjacent to each other. Gene and sample clusters were regarded as stable according to specified size and stability index. The following parameters were used to optimize the resolution of the clustering process: gene cluster size ≥15, sample cluster size ≥4, stability threshold of gene and sample clusters {Delta}T ≥6K with one dropout for samples and 3 dropouts for genes at a single increment in T. Expression data were preprocessed using an iterative scaling and merging algorithm described in detail elsewhere (14).

Step 3: Determination of differentially expressed genes and Venn diagrams.
Both Significance Analysis of Microarrays algorithm (SAM, Ref. 46) and the LIMMA ("linear models for microarray data"; Ref. 38) analysis package are software packages for the statistical analysis of gene expression microarray data particularly designed for the assessment of differential gene expression. SAM and LIMMA both provide ranking of genes. A false discovery rate <1% was used in SAM analyses, and P = 0.01 was used in LIMMA analyses to obtain the ranked lists of differentially expressed genes. The gene lists obtained by SAM were used to generate Venn diagrams. GenMAPP, a new tool for viewing and analyzing microarray data on biological pathways was additionally used (http://www.GenMAPP.org; Refs. 6 and 8).

Validation of Selected Gene Expression Levels By Quantitative Real-Time RT-PCR
As an independent method of measuring levels of gene expression, RT-PCR was performed for 13 selected genes to confirm microarray data. The primers are listed in Table 1. For each amplification, 20 µl of cDNA were diluted in water (1:10) before using as template for the QuantiTect SYBR Green RT-PCR kit (Qiagen, Hilden, Germany). RT-PCR quantification and determination of expression levels were performed on ABI Prism 7700 sequence detector real-time PCR machine (PerkinElmer, Foster City, CA). Amplification reactions were conducted with an initial step at 90°C for 3 min followed by 20–35 cycles. All PCR reactions were performed in triplicates, and {alpha}-tubulin and aminopeptidase were used as reference controls. Predicted size of PCR products was confirmed by agarose gel electrophoresis. For all controlled genes, the direction (up- and downregulation) as well as the strength of regulation agreed with RT-PCR results.


View this table:
[in this window]
[in a new window]
 
Table 1. Primers used for quantitative real-time RT-PCR

 

    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
Physiological Changes in Hearts Subjected to PPC and IschPC
PPC with isoflurane at 1.5 MAC over 15 min or IschPC with three cycles of 5 min of ischemia before prolonged test ischemia of 40 min equally improved postischemic functional recovery compared with nonpreconditioned hearts (Table 2). In the triggering protocols, PPC (APC-TRI) transiently increased coronary flow, but decreased developed pressure, heart rate, and inotropy, whereas IschPC (IPC-TRI) resulted in transient complete functional loss followed by rapid recovery of contractility to baseline values, which was accompanied by a compensatory and transient increase in coronary flow.


View this table:
[in this window]
[in a new window]
 
Table 2. Hemodynamics

 
Pre- and Postischemic Cardiac Gene Expression is Markedly and Distinctly Modulated by PPC and IschPC
To characterize genomic response to PPC (APC-TRI, APC) and IschPC (IPC-TRI, IPC), gene expression profiles were determined in hearts subjected to the preconditioning stimulus only (genomic trigger response: APC-TRI, IPC-TRI), and in preconditioned hearts subjected to ischemia/reperfusion (genomic mediator/effector response: APC, IPC). PPC and IschPC profoundly affected gene regulation. APC-TRI regulated a higher number of genes than IPC-TRI (APC-TRI 1,363, IPC-TRI 545). Twenty-five percent of significantly up- and downregulated genes were jointly regulated in APC-TRI and IPC-TRI compared with time-matched perfusion (CTL) (see Venn diagrams in Fig. 2, A and B). Only 15 genes were exclusively upregulated and only 12 genes exclusively downregulated in IPC-TRI compared with APC-TRI.



View larger version (29K):
[in this window]
[in a new window]
 
Fig. 2. The Venn diagrams show the numbers of differentially up- and downregulated transcripts, as obtained by Significance Analysis of Microarrays (SAM), and the number of overlapping transcripts between groups separated according to the trigger (APC-TRI, IPC-TRI) (A and B) and mediator/effector (APC, IPC) responses (CF).

 
Fewer genes were jointly regulated in APC and IPC compared with the trigger responses (APC 409, IPC 318). APC and IPC jointly regulated 37% of upregulated and 45% of downregulated genes compared with CTL. When IPC was compared with APC, only one transcript was upregulated and 41 genes were exclusively downregulated. Numerous genes separated APC (200) and IPC (296) from nonpreconditioned hearts (ISCH) (Fig. 2, CF). The individual lists of ranked genes obtained by SAM can be viewed in detail in the Supplemental Tables S4–S6. LIMMA produced similar results (data not shown). Among the most prominent ranked genes were the following. For IPC-TRI vs. APC-TRI: upregulated = heat shock protein (Hsp) 40 and Hsp70; downregulated = Hsp90 and cytochrome P-450. For IPC vs. ISCH: upregulated = hexokinase, zf36, rhoB, and Hsp10. For APC vs. ISCH: upregulated = zf36, rhoB, aldoketoreductase, uncoupling protein 2, and Hsp20; downregulated = long interspersed nucleotide elements (LINEs). For IPC vs. APC: upregulated = nuclear targeting protein phosphatase 1; downregulated = endothelin receptor type B and caspase 3. Collectively, PPC and IschPC are characterized by similar but distinct pre- and postischemic gene expression profiles in the myocardium.

Gene Ontology-Based Comparisons Between PPC and IschPC Identifies Distinct Genomic Responses
To determine differences between genomic trigger and mediator/effector responses and between the two types of preconditioning, principal component analysis was applied using previously established GO terms (Supplemental Figs. S2–S4). The widest separation was attained by trigger vs. mediator/effector responses for genes involved in apoptosis, growth factor activity, response to external stimuli, inflammatory response, electron transport, oxidoreductase activity, biosynthesis, and protein transport (Supplemental Fig. S2). Also, although less pronounced, differences within the trigger and mediator/effector responses were observed between PPC and IschPC (Supplemental Figs. S3 and S4). Depending on the GO category, CTL was more similar to protocols with prolonged ischemia (APC, IPC, ISCH) or more similar to trigger responses (APC-TRI, IPC-TRI).

IschPC but not PPC Elicits a Postischemic Gene Expression Profile Similar to Unprotected Ischemic Myocardium
Clustering analysis including all treatment groups.
CTWC was applied to the ANOVA-filtered RMA data for all six treatment groups [APC-TRI (n = 5), IPC-TRI (n = 4), APC (n = 5), IPC (n = 5), ISCH (n = 5), CTL (n = 5)]. The main cluster G1 (2,212 genes) broke into eight stable clusters as follows (Fig. 3 and 4): cluster G2 (69 genes, predominantly upregulated in APC-TRI and IPC-TRI including cell surface receptors, ion channels, Gadd45{alpha}, and many ESTs), cluster G3 (129 genes, markedly upregulated in APC-TRI and downregulated in IPC and ISCH including mitochondria-related genes such as uncoupling protein 2, carnitine palmitoyl transferase 1b and 2, and many cell defense-related genes such as Hsp8, Hsp27, crystallin {alpha}B), cluster G8 and associated subcluster G4 (103 and 83 genes, respectively, exclusively downregulated in APC, IPC, and ISCH including genes related to inflammation such as interleukin 15, tumor necrosis factor {alpha}, nuclear factor {kappa}B, vascular cell adhesion molecule, selectin), cluster G5 (58 genes, exclusively upregulated in APC, IPC, and ISCH, including many chaperones such as Hsp10, Hsp40, Hsp70, and Hsp86), cluster G6 [171 genes, upregulated in ISCH and IPC but downregulated in APC-TRI and IPC-TRI including many transcription factors such as cAMP responsive element regulator (CREB), E2F transcription factor 5, DEAF-1 related transcriptional regulator (NUDR)], cluster G7 (17 genes, exclusively upregulated in APC including many ribosomal proteins S5, S7, S8, S15a, L9, L32, L36, L37), and cluster G9 (17 genes, exclusively upregulated in ISCH including LINE, SPARC-like 1, and many genes associated with cardiac remodeling including various types of collagens, vimentin, and matrix metalloproteinase 2). The main sample cluster S1 broke into six stable clusters as follows (see Fig. 6A): cluster S2 (APC, n = 5), cluster S3 (IPC+ISCH, n = 10), cluster S4 (CTL, n = 5), cluster S5 (CTL+APC, n = 10), cluster S6 (APC-TRI, n = 5), and cluster S7 (IPC-TRI, n = 4). Principal component analysis of sample clusters established a close genomic relationship between IPC and ISCH, while APC was close to CTL (nonischemic healthy myocardium) (Fig. 3B).



View larger version (39K):
[in this window]
[in a new window]
 
Fig. 3. Coupled two-way clustering (CTWC) for all treatment groups. A: global gene expression matrix (heat map) of the 2,212 ANOVA-filtered genes. Rows correspond to genes, and the columns correspond to samples. The color scale denotes the quantification for gene expression based on normalized and centered robust multichip average (RMA) values. Blue indicates least and red the greatest degree of expression. CTWC provides dendrograms of gene and sample clusters. Each node of the dendrogram represents a cluster. The gene dendrogram with the eight stable gene clusters G2–G9 is shown on the left of the matrix. Note that cluster G4 is a subcluster of cluster G8, showing a common parent in the tree. The sample dendrogram with the six stable sample clusters is shown on the bottom of the matrix. Major splits indicating stable clusters of the dendrograms are identified by circles. Red numbers indicate stability of gene and sample clusters. B: principal component analysis of the various treatment groups. Note that IPC is close to nonpreconditioned unprotected hearts (ISCH), whereas APC clusters with virgin healthy myocardium (CTL). Corresponding reordered distance matrix can be viewed in Supplemental Fig. S5A, available at the Physiological Genomics web site.

 


View larger version (94K):
[in this window]
[in a new window]
 
Fig. 4. Expression patterns of the eight stable gene clusters resulting from CTWC for all treatment groups. Groups are indicated on the x-axis. Number of genes of the individual clusters and corresponding stability index {Delta}T, over which the clusters remain stable, are given on the y-axis. Cluster G4 is a subcluster of cluster G8, showing a common parent in the tree (for more details see online Supplemental Table S1). The corresponding color scales denote the quantification of gene expression based on normalized and centered RMA values.

 


View larger version (34K):
[in this window]
[in a new window]
 
Fig. 6. CTWC for mediator/effector responses only (APC, IPC, ISCH, CTL). A: gene expression matrix of the 688 ANOVA-filtered genes. Rows correspond to genes, and the columns correspond to samples. Again, blue indicates least and red the greatest degree of expression. The gene dendrogram with the five stable gene clusters G2–G6 is shown on the left of the matrix. The sample dendrogram with the three stable sample clusters is shown on the bottom of the matrix. Note that IPC and ISCH (unprotected myocardium) form one single cluster and that one IPC sample clusters within ISCH, indicating close relationship to unprotected myocardium. Emerging clusters are marked with rings. Red numbers indicates stability of gene and sample clusters. B: principal component analysis separates the four treatments. Note that IPC is closer to unprotected myocardium (ISCH) than APC. Distance matrix can be viewed in Supplemental Fig. S5C.

 
Clustering analysis including only trigger responses to preconditioning.
To eliminate interference from protocols with test ischemia (APC, IPC, ISCH), additional subgroup analysis including the trigger protocols only was performed [APC-TRI (n = 5), IPC-TRI (n = 4), CTL (n = 5), ntot = 14] (Fig. 5 and Supplemental Table S2). From the main gene cluster G1 (1,011 genes) three stable gene clusters emerged: cluster G2 [59 genes, upregulated in APC-TRI and IPC-TRI including nitric oxide synthase 2 (NOS2), mitogen-activated kinase kinase 2 (MAPKK2), and many receptors and ESTs], cluster G3 (24 genes, upregulated in APC-TRI and IPC-TRI including B-cell translocation gene 1, chemokine ligand 10, best5 protein), cluster G4 (404 genes, downregulated in APC-TRI and IPC-TRI including early growth response 1, peroxiredoxin 6, cytochrome c, cytochrome P-450, oligoadenylate synthetase, LINE, enzymes involved in glucose and fatty acid metabolism, cell adhesion molecules, and many Hsps and ribosomal proteins). The main sample cluster S1 broke into two stable clusters as follows (Fig. 5A): cluster S2 (CTL, n = 5) and cluster S3 (APC-TRI + IPC-TRI, n = 10). Although principal component analysis for the samples displayed differences between APC-TRI and IPC-TRI, only one stable cluster (cluster S3) was identified for both (Figs. 5, A and B).



View larger version (29K):
[in this window]
[in a new window]
 
Fig. 5. CTWC for trigger responses only (CTL, APC-TRI, IPC-TRI). A: gene expression matrix of the 1,011 ANOVA-filtered genes. Rows correspond to genes, and the columns correspond to samples. The color scale denotes a quantification of gene expression based on normalized and centered RMA values (blue indicates least and red greatest degree of expression). The gene dendrogram with the three stable gene clusters G2–G4 is shown on the left of the matrix. The sample dendrogram with the two stable sample clusters is shown on the bottom of the matrix. Emerging clusters are marked with rings. Red numbers indicate stability of gene and sample clusters. B: principal component analysis shows separation between APC-TRI, IPC-TRI, and CTL. Corresponding reordered distance matrix can be viewed in Supplemental Fig. S5B.

 
Clustering analysis including mediator/effector responses to preconditioning.
A similar subgroup analysis was performed using mediator/effector protocols only (APC, IPC, ISCH, CTL, ntot = 20). From cluster G1 (688 genes) five stable gene clusters were obtained (Fig. 6 and Supplemental Table S3): cluster G2 (32 genes, downregulated in ISCH, IPC, APC including genes related to inflammation such as interleukin 1{alpha}, interleukin 1ß, best5 protein, 25 oligoadenylate synthetase), cluster G3 (39 genes, upregulated in APC and downregulated in IPC and ISCH including peroxiredoxin 3 and 6, aldehyde dehydrogenase, carnitine palmitoyl transferase 1b and 2, GATA-binding protein 6, vascular endothelial growth factor B), cluster G4 (55 genes, mainly upregulated in APC, IPC, and ISCH including Hsp40, Hsp70, and Hsp86, early growth response 1, B-cell translocation gene 2, rhoB gene), cluster G5 (16 genes, predominantly upregulated in ISCH including LINEs), and cluster G6 (19 genes, predominantly upregulated in ISCH including many genes associated with myocardial remodeling). The main sample cluster S1 broke into three stable clusters as follows (Fig. 6A): cluster S1 (CTL, n = 5), cluster S2 (APC, n = 5), and cluster S3 (IPC+ISCH, n = 10). Principal component analysis for the samples unraveled a close genomic relationship between IPC and ISCH with one IPC sample localizing close to the group of ISCH samples (Fig. 6B).

Using cluster analysis, we were able to demonstrate that similar but distinct pre- and postischemic gene expression patterns characterize PPC and IschPC in the heart. Importantly, IschPC but not PPC elicits a postischemic gene expression profile similar to unprotected ischemic myocardium.


    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
We have used high-density oligonucleotide microarrays to assess global changes in gene expression that result from preconditioning in the myocardium to gain insight into the underlying complex molecular mechanisms of this highly cardioprotective treatment strategy. The goal of this study was also to identify common mediators and patterns of protection between different types of preconditioning that can be subjected to further investigation for their therapeutic potential. To this end, isolated perfused rat hearts were exposed to brief ischemic cycles or to the pharmacological agent isoflurane in the presence or absence of test ischemia. Our survey revealed important molecular footprints that paradigmatically highlight the biological processes underlying cardiac protection. First, IschPC and PPC markedly affected pre- and postischemic genomic responses in the myocardium. Whereas trigger responses, preceding and ushering delayed protection, appear to be more type specific, postischemic genomic responses are more uniform. Not surprisingly, numerous regulated genes separate preconditioned (IPC, APC) from unprotected ischemic myocardium (ISCH). These observations support the concept that different types of preconditioning elicit their protection by similar molecular mechanisms and raise the possibility that postischemic genetic reprogramming by preconditioning may play a role in cardioprotection against ischemia (40). Second, clustering analysis revealed a number of unique gene patterns characteristic for triggering and mediating the preconditioning process. We here report for the first time that both PPC and IschPC similarly and effectively prevent postischemic activation of a gene cluster rich in transcripts related to cardiac hypertrophy and remodeling. Another interesting finding is the notion that LINEs, recently identified as "genomic rheostats" (4), were exclusively overexpressed in unprotected ischemic myocardium, whereas preconditioning effectively prevented upregulation. Lastly and most importantly, we were able to demonstrate that IschPC but not PPC elicits a postischemic gene expression profile more similar to unprotected myocardium. This is supported by the following observations. IschPC but not PPC shared most activity patterns with unprotected myocardium in many of the emerging clusters. In addition, one single sample cluster for both IschPC and unprotected myocardium emerged from the cluster analysis indicating a close relationship between the two groups. This genomic similarity is also reflected by the proximity between the individual hearts of these groups as demonstrated in the principal component analysis. In contrast, PPC revealed an expression pattern closer to virgin myocardium, implying that intermittent episodes of ischemia used for IschPC might actually promote a certain degree of injury.

Clustering Analysis
We used the CTWC method (2, 12) to identify patterns of genes within the large database. CTWC is characterized by a high robustness against noise and a natural ability to identify stable clusters, providing insight that would have been impossible by simply looking at particular gene lists. Unique expression patterns emerged within the transcriptional responses and placed the expression of many transcripts into a more holistic context. The clusters included families of transcripts known to have similar function, suggesting that this method closely followed biological likeness. In this study, we were able to track a genomic similarity between unprotected myocardium and IschPC. Although test ischemia itself activated protective genes, this may merely reflect the transcriptional response of the highly stressed but yet surviving myocardial tissue. Irrespectively, the gene expression profile of unprotected myocardium was clearly coupled to poor functional recovery and cell death and therefore represents the characteristic postischemic profile of the unprotected state. Likewise, the gene expression pattern of untreated virgin myocardium should be regarded as archetypal for nonischemic healthy myocardium. Collectively, using a global gene discovery approach (CTWC clustering), our data support the concept that PPC may be less harmful to the myocardium and thus superior to IschPC as therapeutic strategy in cardiac protection. However, since many genes separated protected from unprotected myocardium, the molecular similarity between IschPC and unprotected myocardium may be regarded as relative rather than absolute, and its significance remains to be determined.

Comparison Between IschPC and PPC
In the present study, the regulation of many transcripts previously reported to be involved in preconditioning was confirmed, but some transcripts showed opposite regulation. However, preconditioning is a highly dynamic complex network of intricate mechanisms undergoing multiphase regulation. Accordingly, different mechanisms have been reported to be responsible for the protection at different time points after preconditioning (49). Also, the genomic trigger responses as measured after 3 h in this study do not necessarily reflect the transcriptional changes involved in delayed protection. To date, few studies used microarrays to uncover the complex molecular mechanisms underlying preconditioning (30, 3437, 54). Onody et al. (30) observed upregulation of chaperonin {epsilon} (TCP-1{epsilon}) and ribosomal proteins in preconditioned rat hearts after test ischemia and reperfusion. Simkhovich et al. (37) reported the activation of a protective genetic program predominantly including various heat shock proteins and transcription factors in rat hearts after brief ischemic episodes. Albeit not directly comparable, the results of the present study in principle confirm and extend these findings as well as the results of our previous microarray study, where we investigated the trigger responses of brief episodes of ischemia compared with a prolonged isoflurane exposure (110 min) but did not investigate protocols with test ischemia and reperfusion (35). In the latter study, a differential regulation of Hsp27, Hsp70, and programmed cell death 8 was observed in response to brief ischemia compared with isoflurane exposure. In another study, Rokosh et al. (34) compared trigger responses in mouse hearts exposed to brief ischemia and nitric oxide but did not compare the postischemic genomic reprogramming of the two types of preconditioning. Hence, the current study is the first comparing comprehensively pre- and postischemic genomic responses of IschPC vs. PPC.

Chaperones
Hearts exposed to global prolonged ischemia overexpressed many heat shock proteins independent of whether preconditioning was applied or not. Hsp27 scavenges cytochrome c (3), inhibits activation of caspase 3 (26), and blocks Fas-related apoptotic pathways (15). Hsp70 together with Hsp40 prevents mitochondrial release of cytochrome c and inhibits caspase 9 activation via Apaf-1 (26). Interestingly, Hsp10 was exclusively upregulated in IPC, while Hsp20 was exclusively upregulated in APC. Hsp10 acts in collaboration with Hsp60 opposing the proapoptotic Bax (15), and Hsp20 was recently found to inhibit ß-agonist-induced cardiac apoptosis (9). In the trigger responses (APC-TRI, IPC-TRI), several chaperones including Hsp8, Hsp20, and crystallin {alpha}B were jointly downregulated. Collectively, these observations provide evidence for a highly dynamic and distinct regulation of chaperones in both IschPC and PPC.

Inflammation
Surprisingly and in contrast to previous work (28), the inflammatory response was profoundly and consistently downregulated in all protocols receiving prolonged ischemia and 3 h of reperfusion independent of whether preconditioning was applied or not. Mediators of inflammation are known to be important in ischemia/reperfusion-induced myocardial damage, and their inhibition was previously implicated in the protection underlying preconditioning. In contrast, upregulation of cytokines at the late phase of IschPC may represent a cytoprotective mechanism (54). It is possible that in the Langendorff model, which is virtually devoid of blood components, the inflammatory response may be limited to a short and blunted burst of inflammatory mediators at the early reperfusion (17). Alternatively, it could be speculated that the observed delayed postischemic anti-inflammatory status reflects a counterregulatory response and simply represents the intrinsic protective response of the viable myocardium unmasked in the absence of leukocytes, macrophages, and other extrinsic proinflammatory components.

Transcription Factors
Early growth response-1 (Egr-1), an immediate-early gene zinc finger transcription factor in the vasculature (29), which triggers increased expression of transcripts encoding intercellular adhesion molecule-1, vascular cell adhesion molecule-1, and platelet-derived growth factor, was upregulated after APC but downregulated after both types of triggering. Zf36, another member of the zinc finger transcription factors, which is widely distributed in tissues, was exclusively increased in the protected myocardium. This may result in anti-inflammatory protective effects, as the zf36 knockout mouse model displays deleterious tumor necrosis factor-{alpha} overexpression (42). Likewise, activating transcription factor 3 (ATF3) was exclusively upregulated in protected myocardium. ATF3 is a member of the cAMP-responsive element binding protein family, which is known to downregulate the transcription of p53 gene, thus promoting cell survival (16). Growth arrest and DNA-damage-inducible 45{alpha} (Gadd45{alpha}) controlling DNA stability and repair clustered in both trigger responses. Interestingly, enhanced E2F activity, previously linked to apoptosis (23), was observed in ISCH and IPC, but was downregulated in the trigger responses. A similar expression pattern throughout the treatment groups was observed for deformed epidermal autoregulatory factor-1 (DEAF-1).

Metabolic Plasticity (Supplemental Fig. S6)
Enzymes involved in long-chain fatty acid ß-oxidation were increased in APC and IPC. Also, pyruvate dehydrogenase, which determines the fate of the glycolytic product pyruvate, i.e., mitochondrial oxidation or anaerobic conversion into lactate, was upregulated exclusively in protected myocardium. Interestingly, APC but not IPC upregulated carnitine palmitoyltransferase, the rate-limiting enzyme in fatty acid ß-oxidation, possibly preventing palmitate-induced myocyte apoptosis (32). An intriguing new finding was the upregulation of many phosphoprotein phosphatases, which may be due to the regulation of multiple metabolic pathways. Collectively, enhanced substrate oxidation reflects the more robust preservation of energy production in preconditioned hearts allowing better functional recovery. In line with this view is the notion that members of the mitochondrial respiratory chain, i.e., uncoupling protein 2 in APC and uncoupling factor 6 in IPC, respectively, were upregulated. Overexpression of uncoupling protein 2 was recently shown to inhibit mitochondrial death signaling in superoxide-stressed cardiomyocytes (43). Interestingly, in both types of preconditioning, although less pronounced in APC, postischemic myocardium expressed increased levels of hexokinase. Recent research has shown that specific isoforms of hexokinase bind to mitochondrial voltage-dependent anion channel, thus suppressing the release of cytochrome c and inhibiting apoptosis (24).

Metabolic remodeling was completely different in the trigger protocols. Transcripts of enzymes involved in glycolysis and tricarboxylic acid cycle, fatty acid ß-oxidation, and mitochondrial respiration were consistently downregulated. This "state of metabolic hibernation" was more pronounced in APC than IPC. Myocardial protection by preconditioning and slowed metabolism was previously reported to coincide (31). Reduced energy demand is a feature of preconditioned myocardium and may be due to protein kinase C-mediated phosphorylation of various regulatory proteins in several energy-consuming reactions. Alternatively, downregulation of key metabolic pathways may be a regulatory response to preconditioning-induced increased glucose uptake (44). Taken together, these results confirm previous observations and extend our knowledge on metabolic plasticity of preconditioned myocardium.

Remodeling
A wide range of transcripts encoding extracellular matrix and structural proteins were enriched in a cluster archetypal for unprotected ischemic myocardium. Both types of preconditioning prevented activation of the remodeling program, a process that might be mechanistically linked to improved postischemic cardiac function and a decreased propensity for arrhythmogenesis. Consistent with this notion, angina, the clinical correlate to IschPC, was recently shown to protect patients against ventricular remodeling (39). In the current study, APC and IPC were associated with increased brain natriuretic peptide expression. Brain natriuretic peptide is known to decrease collagen synthesis and cardiac remodeling. Upregulated matrix metalloproteinase 2, responsible for collagen degradation and remodeling of the extracellular matrix, was clustering with vimentin and insulin-like growth factor II in unprotected myocardium. Metalloproteinase 2 was previously reported to cleave troponin I at reperfusion, thereby reducing contractile function (48).

Long Interspersed Nucleotide Elements
We have also uncovered a retrotransposon transcriptional burst. LINEs are long interspersed repeated retrotransposable elements and found in almost all eukaryotes (13). They contain an internal promoter for RNA polymerase II (pol II), usually two open reading frames encoding proteins of unknown function and polypeptides with reverse transcriptase and DNA endonuclease activity. After translation, the nascent reverse transcriptase binds to the LINE mRNA forming a ribonucleoprotein complex, which enters the nucleus where it initiates a process called "target-primed reverse transcription" priming reverse transcription of the LINE mRNA into the chromosome. Short interspersed retrotransposable elements (SINEs) may use the LINE machinery for retrotransposition, a process called "retropositional parasitism." Notably, retrotransposable elements are integrated in many introns of many genes accounting for ~37% of the rat genome. We here report for the first time that two types of preconditioning consistently reversed ischemia-enhanced LINE expression. It is tempting to speculate that LINE activity and SINE activity may be involved in pro- and/or antiprotective gene regulation by posttranscriptional gene silencing, inhibition of transcriptional elongation, or inhibition of protein kinase R (4, 50). Interestingly, Alu, the most common LINE, is known to attract DNA fragmentation events within ORF2 at early stages of apoptosis (21). Collectively, this raises the possibility that LINEs and/or SINEs may play an important regulatory role in ischemia/reperfusion phenomena.

Clinical Implications
Clinical studies suggest that pre-infarction angina, a clinical correlate to IschPC, increases the chances of rapid reperfusion after thrombolytic therapy, and reduces the number of hypokinetic myocardial segments and infarct size. Most recently, decreased ventricular remodeling (39) and improved cardiovascular long-term outcome (22) was observed in patients with coronary artery disease and an effective preconditioning mechanism. Despite these beneficial effects, some publications raised concerns about the safety of IschPC as a therapeutic strategy (33). In support of these concerns are experimental results obtained in old Fisher 344 rats exposed to brief ischemic episodes (36). In this model, brief ischemia promoted expression of injury- and disease-related genes. Our study now shows for the first time the close molecular relationship between IschPC and unprotected myocardium. Thus PPC, as opposed to IschPC, may have a wider "therapeutic window." There are a large number of experimental studies, and more recently an increasing number of clinical studies, demonstrating the significant cardioprotection of volatile anesthetics in patients undergoing coronary artery bypass grafting. Laboratory investigations also stress the concept that volatile anesthetics may precondition endothelial and smooth muscle cells (7), implying that systemic administration of these agents may potentially protect a variety of other vital organs. Intriguingly, preconditioning by sevoflurane was reported to attenuate cardiopulmonary bypass-associated transient renal dysfunction in coronary artery bypass graft patients (20). Together, based on our experimental results and previous clinical observations, PPC should be preferentially applied as a therapeutic strategy for cardioprotection in the clinical setting.

Study Limitations
The following remarks should be added. RT-PCR may be more powerful to detect gene regulation. Also, changes in mRNA levels may be not always correlated with respective protein levels. Although genomics has demonstrated that there is more than 85% similarity in coding regions of the rat genome compared with the human genome, data from rodent studies must be always interpreted with caution. In addition, buffer-perfused hearts have a limited long-term biologic stability and may undergo short confounding ischemic periods during the isolation procedure. Finally, the observations as obtained by the volatile anesthetic isoflurane may not be applicable for all PPC inducing agents.

Conclusions
We have used an unbiased gene discovery approach to define the global transcriptional responses surrounding preconditioning. Novel key clusters containing LINEs and transcripts related to cardiac remodeling emerged after ischemia and were effectively modulated by preconditioning. Due to the genomic similarity between unprotected myocardium and IschPC, IschPC may be less desirable as therapeutic approach, specifically in high-risk patients in whom an ischemic type of preconditioning may further jeopardize diseased myocardium. The information obtained herein by comparing PPC and IschPC might help to develop novel rational therapeutic interventions targeted to specific cardioprotective mechanisms.


    GRANTS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 
This work was supported by the Functional Genomics Center, University of Zurich, Zurich, Switzerland; Grant 3200B0-103980/1 of the Swiss National Science Foundation, Berne, Switzerland; the Swiss University Conference, Berne, Switzerland; a grant of the Swiss Heart Foundation, Berne, Switzerland; a grant from Abbott Switzerland, Baar, Switzerland; and a grant from the Swiss Society of Anaesthesiology and Reanimation, Berne, Switzerland.


    FOOTNOTES
 
* R. da Silva and E. Lucchinetti contributed equally to this work. Back

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

Address for reprint requests and other correspondence: M. Zaugg, Institute of Anesthesiology, Univ. Hospital Zurich, Rämistrasse 100, CH-8091 Zurich, Switzerland (E-mail: michael.zaugg{at}usz.ch)

doi:10.1152/physiolgenomics.00166.2004.

1 The following additional data files are available with the online version of this article: Excel sheets containing complete lists of differentially regulated genes separated according to trigger (APC-TRI, IPC-TRI) and mediator/effector responses (APC, IPC) (Tables S4–S6), lists of genes of the individual clusters resulting from CTWC (Tables S1–S3), a flowchart illustrating the individual steps of data analysis (Fig. S1), additional principal component analysis results (Figs. S2–S4), reordered gene expression matrices of the various clustering analyses (Fig. S5), and representative GenMAPP pathways (Fig. S6). This is available online at http://physiolgenomics.physiology.org/cgi/content/full/00166.2004/DC1. Back


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 GRANTS
 REFERENCES
 

  1. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT, Harris MA, Hill DP, Issel-Tarver L, Kasarskis A, Lewis S, Matese JC, Richardson JE, Ringwald M, Rubin GM, and Sherlock G. Gene Ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 25: 25–29, 2000.[CrossRef][ISI][Medline]
  2. Blatt M, Wiseman S, and Domany E. Superparamagnetic clustering of data. Phys Rev Lett 76: 3251–3254, 1996.[CrossRef][ISI][Medline]
  3. Bruey JM, Ducasse C, Bonniaud P, Ravagnan L, Susin SA, Diaz-Latoud C, Gurbuxani S, Arrigo AP, Kroemer G, Solary E, and Garrido C. Hsp27 negatively regulates cell death by interacting with cytochrome c. Nat Cell Biol 2: 645–652, 2000.[CrossRef][ISI][Medline]
  4. Bushman F. Gene regulation: selfish elements make a mark. Nature 429: 253–255, 2004.
  5. da Silva R, Grampp T, Pasch T, Schaub MC, and Zaugg M. Differential activation of mitogen-activated protein kinases in ischemic and anesthetic preconditioning. Anesthesiology 100: 59–69, 2004.[CrossRef][ISI][Medline]
  6. Dahlquist KD, Salomonis N, Vranizan K, Lawlor SC, and Conklin BR. GenMAPP, a new tool for viewing and analyzing microarray data on biological pathways. Nat Genet 31: 19–20, 2002.[CrossRef][ISI][Medline]
  7. de Klaver MJ, Buckingham MG, and Rich GF. Isoflurane pretreatment has immediate and delayed protective effects against cytokine-induced injury in endothelial and vascular smooth muscle cells. Anesthesiology 99: 896–903, 2003.[CrossRef][ISI][Medline]
  8. Doniger SW, Salomonis N, Dahlquist KD, Vranizan K, Lawlor SC, and Conklin BR. MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-expression profile from microarray data. Genome Biol 4: R7, 2003.[CrossRef][Medline]
  9. Fan GC, Chu G, Mitton B, Song Q, Yuan Q, and Kranias EG. Small heat-shock protein Hsp20 phosphorylation inhibits beta-agonist-induced cardiac apoptosis. Circ Res 94: 1474–1482, 2004.[Abstract/Free Full Text]
  10. Gautier L, Cope L, Bolstad BM, and Irizarry RA. affy: Analysis of Affymetrix GeneChip data at the probe level. Bioinformatics 20: 307–315, 2004.[Abstract/Free Full Text]
  11. Getz G and Domany E. Coupled two-way clustering server. Bioinformatics 19: 1153–1154, 2003.[Abstract/Free Full Text]
  12. Getz G, Levine E, and Domany E. Coupled two-way clustering analysis of gene microarray data. Proc Natl Acad Sci USA 97: 12079–12084, 2000.[Abstract/Free Full Text]
  13. Gibbs RA et al.; Rat Genome Sequencing Project Consortium. Genome sequence of the Brown Norway rat yields insights into mammalian evolution. Nature 428: 493–521, 2004.[CrossRef][ISI]
  14. Greenbaum D, Jansen R, and Gerstein M. Analysis of mRNA expression and protein abundance data: an approach for the comparison of the enrichment of features in the cellular population of proteins and transcripts. Bioinformatics 18: 585–596, 2002.[Abstract/Free Full Text]
  15. Gupta S and Knowlton AA. Cytosolic heat shock protein 60, hypoxia, and apoptosis. Circulation 106: 2727–2733, 2002.[Abstract/Free Full Text]
  16. Hashimoto Y, Zhang C, Kawauchi J, Imoto I, Adachi MT, Inazawa J, Amagasa T, Hai T, and Kitajima S. An alternatively spliced isoform of transcriptional repressor ATF3 and its induction by stress stimuli. Nucleic Acids Res 30: 2398–2406, 2002.[Abstract/Free Full Text]
  17. Hiasa G, Hamada M, Ikeda S, and Hiwada K. Ischemic preconditioning and lipopolysaccharide attenuate nuclear factor-kappaB activation and gene expression of inflammatory cytokines in the ischemia-reperfused rat heart. Jpn Circ J 65: 984–990, 2001.[CrossRef][ISI][Medline]
  18. Ihaka R and Gentleman R. R: A language for data analysis and graphics. J Comput Graph Stat 5: 299–314, 1996.
  19. Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, and Speed TP. Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res 31: e15, 2003.[Abstract/Free Full Text]
  20. Julier K, da Silva R, Garcia C, Bestmann L, Frascarolo P, Zollinger A, Chassot PG, Schmid ER, Turina MI, von Segesser LK, Pasch T, Spahn DR, and Zaugg M. Preconditioning by sevoflurane decreases biochemical markers for myocardial and renal dysfunction in coronary artery bypass graft surgery: a double-blinded, placebo-controlled, multicenter study. Anesthesiology 98: 1315–1327, 2003.[ISI][Medline]
  21. Khodarev NN, Bennett T, Shearing N, Sokolova I, Koudelik J, Walter S, Villalobos M, and Vaughan AT. LINE L1 retrotransposable element is targeted during the initial stages of apoptotic DNA fragmentation. J Cell Biochem 79: 486–495, 2000.[CrossRef][ISI][Medline]
  22. Laskey WK and Beach D. Frequency and clinical significance of ischemic preconditioning during percutaneous coronary interventions. J Am Coll Cardiol 42: 988–1003, 2003.
  23. Liu K, Luo Y, Lin FT, and Lin WC. TopBP1 recruits Brg1/Brm to repress E2F1-induced apoptosis, a novel pRb-independent and E2F1-specific control for cell survival. Genes Dev 18: 673–686, 2004.[Abstract/Free Full Text]
  24. Majewski N, Nogueira V, Robey RB, and Hay N. Akt inhibits apoptosis downstream of BID cleavage via a glucose-dependent mechanism involving mitochondrial hexokinases. Mol Cell Biol 24: 730–740, 2004.[Abstract/Free Full Text]
  25. Marber MS, Latchman DS, Walker JM, and Yellon DM. Cardiac stress protein elevation 24 hours after brief ischemia or heat stress is associated with resistance to myocardial infarction. Circulation 88: 1264–1272, 1993.[Abstract]
  26. Michels AA, Kanon B, Bensaude O, and Kampinga HH. Heat shock protein (Hsp) 40 mutants inhibit Hsp70 in mammalian cells. J Biol Chem 274: 36757–36763, 1999.[Abstract/Free Full Text]
  27. Murry CE, Jennings RB, and Reimer KA. Preconditioning with ischemia: a delay of lethal cell injury in ischemic myocardium. Circulation 74: 1124–1136, 1986.[Abstract]
  28. Newby LK. Markers of cardiac ischemia, injury, and inflammation. Prog Cardiovasc Dis 46: 404–416, 2004.[CrossRef][ISI][Medline]
  29. Okada M, Wang CY, Hwang DW, Sakaguchi T, Olson KE, Yoshikawa Y, Minamoto K, Mazer SP, Yan SF, and Pinsky DJ. Transcriptional control of cardiac allograft vasculopathy by early growth response gene-1 (Egr-1). Circ Res 91: 135–142, 2002.[Abstract/Free Full Text]
  30. Onody A, Zvara A, Hackler L Jr, Vigh L, Ferdinandy P, and Puskas LG. Effect of classic preconditioning on the gene expression pattern of rat hearts: a DNA microarray study. FEBS Lett 536: 35–40, 2003.[CrossRef][ISI][Medline]
  31. Opie LH and Sack MN. Metabolic plasticity and the promotion of cardiac protection in ischemia and ischemic preconditioning. J Mol Cell Cardiol 34: 1077–1089, 2002.[ISI][Medline]
  32. Paumen MB, Ishida Y, Muramatsu M, Yamamoto M, and Honjo T. Inhibition of carnitine palmitoyltransferase I augments sphingolipid synthesis and palmitate-induced apoptosis. J Biol Chem 272: 3324–3329, 1997.[Abstract/Free Full Text]
  33. Perrault LP, Menasche P, Bel A, de Chaumaray T, Peynet J, Mondry A, Olivero P, Emanoil-Ravier R, and Moalic JM. Ischemic preconditioning in cardiac surgery: a word of caution. J Thorac Cardiovasc Surg 112: 1378–1386, 1996.[Abstract/Free Full Text]
  34. Rokosh G, Guo Y, Williams J, Wu WJ, Zhu XP, Tan W, and Bolli R. Gene expression after ischemic and pharmacologic preconditioning: analysis by DNA microarrays. Circulation 106: 284, 2002.[Free Full Text]
  35. Sergeev P, da Silva R, Lucchinetti E, Zaugg K, Pasch T, Schaub MC, and Zaugg M. Trigger-dependent gene expression profiles in cardiac preconditioning: evidence for distinct genetic programs in ischemic and anesthetic preconditioning. Anesthesiology 100: 474–488, 2004.[CrossRef][ISI][Medline]
  36. Simkhovich BZ, Marjoram P, Poizat C, Kedes L, and Kloner RA. Age-related changes of cardiac gene expression following myocardial ischemia/reperfusion. Arch Biochem Biophys 420: 268–278, 2003.[CrossRef][ISI][Medline]
  37. Simkhovich BZ, Marjoram P, Poizat C, Kedes L, and Kloner RA. Brief episode of ischemia activates protective genetic program in rat heart: a gene chip study. Cardiovasc Res 59: 450–459, 2003.[CrossRef][ISI][Medline]
  38. Smyth GK. Linear models and empirical Bayes methods for assessing differential expression in microarray experiments. In: Statistical Applications in Genetics and Molecular Biology. Berkeley, CA: Berkeley Electronic Press, vol. 3, issue 1, article 3, 2004, p. 1–25 (http://www.bepress.com/sagmb/vol3/iss1/art3/).
  39. Solomon SD, Anavekar NS, Greaves S, Rouleau JL, Hennekens C, and Pfeffer MA. Angina pectoris prior to myocardial infarction protects against subsequent left ventricular remodeling. J Am Coll Cardiol 43: 1511–1514, 2004.[CrossRef][ISI][Medline]
  40. Strohm C, Barancik M, von Bruehl ML, Strniskova M, Ullmann C, Zimmermann R, and Schaper W. Transcription inhibitor actinomycin-D abolishes the cardioprotective effect of ischemic preconditioning. Cardiovasc Res 55: 602–618, 2002.[CrossRef][ISI][Medline]
  41. Tanaka K, Ludwig LM, Kersten JR, Pagel PS, and Warltier DC. Mechanisms of cardioprotection by volatile anesthetics. Anesthesiology 100: 707–721, 2004.[CrossRef][ISI][Medline]
  42. Taylor GA, Carballo E, Lee DM, Lai WS, Thompson MJ, Patel DD, Schenkman DI, Gilkeson GS, Broxmeyer HE, Haynes BF, and Blackshear PJ. A pathogenetic role for TNF alpha in the syndrome of cachexia, arthritis, and autoimmunity resulting from tristetraprolin (TTP) deficiency. Immunity 4: 445–454, 1996.[ISI][Medline]
  43. Teshima Y, Akao M, Jones SP, and Marban E. Uncoupling protein-2 overexpression inhibits mitochondrial death pathway in cardiomyocytes. Circ Res 93: 192–200, 2003.[Abstract/Free Full Text]
  44. Tong H, Chen W, London RE, Murphy E, and Steenbergen C. Preconditioning enhanced glucose uptake is mediated by p38 MAP kinase not by phosphatidylinositol 3-kinase. J Biol Chem 275: 11981–11986, 2000.[Abstract/Free Full Text]
  45. Tonkovic-Capin M, Gross GJ, Bosnjak ZJ, Tweddell JS, Fitzpatrick CM, and Baker JE. Delayed cardioprotection by isoflurane: role of KATP channels. Am J Physiol Heart Circ Physiol 283: H61–H68, 2002. First published March 7, 2002; doi:10.1152/ajpheart.01040.2001.[Abstract/Free Full Text]
  46. Tusher VG, Tibshirani R, and Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 98: 5116–5121, 2001.[Abstract/Free Full Text]
  47. Uecker M, da Silva R, Grampp T, Pasch T, Schaub MC, and Zaugg M. Translocation of protein kinase C isoforms to subcellular targets in ischemic and anesthetic preconditioning. Anesthesiology 99: 138–147, 2003.[CrossRef][ISI][Medline]
  48. Wang W, Schulze CJ, Suarez-Pinzon WL, Dyck JRB, Sawicki G, and Schulz R. Intracellular action of matrix metalloproteinase-2 accounts for acute myocardial ischemia and reperfusion injury. Circulation 106: 1543–1549, 2002.[Abstract/Free Full Text]
  49. Wang Y, Kodani E, Wang J, Zhang SX, Takano H, Tang XL, and Bolli R. Cardioprotection during the final stage of the late phase of ischemic preconditioning is mediated by neuronal NO synthase in concert with cyclooxygenase-2. Circ Res 95: 84–91, 2004.[Abstract/Free Full Text]
  50. Weiner AM. SINEs and LINEs: the art of biting the hand that feeds you. Curr Opin Cell Biol 14: 343–350, 2002.[CrossRef][ISI][Medline]
  51. Zaugg M, Lucchinetti E, Garcia C, Pasch T, Spahn DR, and Schaub MC. Anaesthetics and cardiac preconditioning. II. Clinical implications. Br J Anaesth 91: 566–576, 2003.[Abstract/Free Full Text]
  52. Zaugg M, Lucchinetti E, Spahn DR, Pasch T, and Schaub MC. Volatile anesthetics mimic cardiac preconditioning by priming the activation of mitochondrial K(ATP) channels via multiple signaling pathways. Anesthesiology 97: 4–14, 2002.[ISI][Medline]
  53. Zaugg M, Lucchinetti E, Uecker M, Pasch T, and Schaub MC. Anaesthetics and cardiac preconditioning. I. Signalling and cytoprotective mechanisms. Br J Anaesth 91: 551–565, 2003.[Abstract/Free Full Text]
  54. Zubakov D, Hoheisel JD, Kluxen FW, Brändle M, Ehring T, Hentsch B, and Frohme M. Late ischemic preconditioning of the myocardium alters the expression of genes involved in inflammatory response. FEBS Lett 547: 51–55, 2003.[CrossRef][ISI][Medline]