Reprogramming of the Human Atrial Transcriptome in Permanent Atrial Fibrillation
http://www.100md.com
循环研究杂志 2005年第5期
The Departments of Medicine I (A.S.B., E.A., L.Z., P.K., S.K., M.H., H.K., G.S., M.N.) and Cardiac Surgery (E.K.), University Hospital Grosshadern
the Department of Medical Informatics, Biometrics
Epidemiology (S.M., M.D.), Ludwig-Maximilians-University, Munich, Germany
Aventis Pharma Deutschland GmbH (M.G., K.S., M.B.), Frankfurt am Main, Germany
the Institute of Physiology (M.B.), Christian- Albrechts-University, Kiel, Germany.
Abstract
Atrial fibrillation is associated with increased expression of ventricular myosin isoforms in atrial myocardium, regarded as part of a dedifferentiation process. Whether reexpression of ventricular isoforms in atrial fibrillation is restricted to transcripts encoding for contractile proteins is unknown. Therefore, this study compares atrial mRNA expression in patients with permanent atrial fibrillation to atrial mRNA expression in patients with sinus rhythm and to ventricular gene expression using Affymetrix U133 arrays. In atrial myocardium, we identified 1434 genes deregulated in atrial fibrillation, the majority of which, including key elements of calcium-dependent signaling pathways, displayed downregulation. Functional classification based on Gene Ontology provided the specific gene sets of the interdependent processes of structural, contractile, and electrophysiological remodeling. In addition, we demonstrate for the first time a prominent upregulation of transcripts involved in metabolic activities, suggesting an adaptive response to increased metabolic demand in fibrillating atrial myocardium. Ventricular-predominant genes were 5 times more likely to be upregulated in atrial fibrillation (174 genes upregulated, 35 genes downregulated), whereas atrial-specific transcripts were predominantly downregulated (56 genes upregulated, 564 genes downregulated). Overall, in fibrillating atrial myocardium, functional classes of genes characteristic of ventricular myocardium were found to be upregulated (eg, metabolic processes), whereas functional classes predominantly expressed in atrial myocardium were downregulated (eg, signal transduction and cell communication). Therefore, dedifferentiation with adoption of a ventricular-like signature is a general feature of the fibrillating atrium.
Key Words: atrial fibrillation gene expression atrial myocardium functional genomics
Introduction
Atrial fibrillation (AF), the most common sustained arrhythmia, is associated with extensive structural, contractile, and electrophysiological remodeling,1 stabilizing AF in the long run. Because current pharmacological treatment of AF is limited by ventricular proarrhythmia and inefficacy to prevent recurrences of AF, understanding the molecular events of these remodeling processes is essential for the development of new targeted therapeutic interventions. Even though knowledge of genes involved in remodeling processes exists on a gene-by-gene basis including transcripts of the extracellular matrix compartment,2 ion channels,1,3 and signal transduction molecules,4 microarray analysis offers a broader and unbiased approach to identify pathophysiologically relevant pathways. Therefore, we performed a systematic functional analysis of gene expression in permanent atrial fibrillation (pmAF) based on information from the Gene Ontology (GO) database to relate expression changes of single transcripts to known remodeling processes.
Moreover, atrial fibrillation has been shown to be associated with an atrial-to-ventricular switch by expressing ventricular myosin isoforms, which is regarded as part of a dedifferentiation process of fibrillating atrial myocardium.5 To explore the hypothesis that pmAF leads to upregulation of ventricular transcripts other than those encoding for contractile proteins, we used a genome-wide approach to compare atrial mRNA expression of patients with pmAF to atrial mRNA expression of patients with sinus rhythm (SR) as well as to ventricular (LV) gene expression. Our findings indicate that pmAF leads to repression of atrial transcriptional activity and induces a ventricular-like genomic signature in atrial myocardium.
Materials and Methods
Patients and Tissue Selection
Right atrial appendages were obtained from 30 patients undergoing open heart surgery for valve repair or coronary artery bypass grafting. Of these, 10 patients had pmAF defined as duration of AF longer than 3 months as documented by ECG, whereas 20 patients had no history of AF and were in SR when open heart surgery was performed. All patients gave written informed consent. Transmural LV tissue sections spanning the entire myocardial wall from subepicardium to subendocardium were obtained from 5 nonfailing donor hearts that were not transplanted because of palpable coronary calcifications. After excision, all tissue specimens were immediately frozen in liquid nitrogen and stored at eC80°C. The investigation was approved by the Institutional Review Board. Detailed patient characteristics are listed in online Table I (available at http://circres.ahajournals.org).
mRNA Preparation and Hybridization to Affymetrix U133A+B Oligonucleotide Microarrays
Total RNA was purified from homogenized deep-frozen tissue samples following the TRIZOL standard protocol as described by the manufacturer (GibcoBRL) and quantified by photometry. Then, 150 ng total RNA was used for quality control using the RNA 6000 Nano LabChip kit and Agilent Bioanalyzer 2100 (Agilent Technologies).
Synthesis of double-stranded cDNA and of biotinylated cRNA probes as well as subsequent steps of hybridization, washing, and staining followed instructions provided by Affymetrix.
Data Extraction and Statistical Analysis
One set of U133A+B chips (together representing 44 928 probe sets) was used for each human heart sample. The original data files have been deposited in Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo) and are accessible through GEO Series accession number GSE2240. All U133 GeneChips passed quality control to eliminate scans with abnormal characteristics, ie, abnormal low or high dynamic range, high perfect match saturation, high pixel noise, grid misalignment problems, and low mean signal-to-noise ratio. Low level and statistical analysis was done using R 1.8.1. Microarray data of all atrial and ventricular samples was normalized simultaneously using robust multiarray average (RMA)6 implemented in Bioconductor’s "affy"-package. Probe sets with "absent" calls in more than 50% of tissue samples in either group (SR, pmAF, or LV) were filtered out and omitted from further analysis. To determine differentially expressed genes between the groups, Welch t test was applied. Considering multiple hypothesis testing, differences in expression were regarded as statistically significant if a false discovery rate (FDR) of q <0.03 was achieved.7 Duplicate probe sets within the gene lists were eliminated by means of the "MatchMiner" software tool.8 Hierarchical clustering was performed by using an average linkage algorithm with Euclidean distance measure comprised in the Genesis software.9 Functional classification was based on the "cellular component," "biological process," and "molecular function" of GO using "FatiGO."10 After accounting for uncharacterized transcripts, up to 70% of all known genes were annotated across all GO levels. As many genes are not annotated at a particular level of ontology, three adjacent ontology levels are presented to increase the amount of information.
Validation by Semiquantitative Real-Time Reverse Transcription PCR
Expression patterns of 13 genes, selected with respect to the central role in the proposed working model of AF (Figure 3) were validated by real-time SYBR-Green PCR. A detailed list of the genes examined by RT-PCR including the design of the primers and a description of the protocols used for RT-PCR are given in the expanded Materials and Methods section in the online data supplement.
Results
Permanent Atrial Fibrillation Is Characterized by General Transcriptional Downregulation
We identified 1434 genes to be differentially expressed between pmAF and SR samples. Of these, 452 and 982 genes were up- and down-regulated in pmAF, respectively. A complete list of all transcripts deregulated in pmAF is provided in online Table 3. Downregulation of transcriptional processes was substantiated by GO analysis as functional classes of "transcription," "transcription factor activity," and "transcription cofactor activity" were downregulated (online Table 4, "molecular function" and "biological process," level 5). Downregulation was particularly noted for key elements of calcium-dependent signaling pathways, exemplified by the calcineurin-NFATc signaling pathway with downregulation of central components like calcineurin A and NFATc3, and concomitant upregulation of NFATc3-inhibitory transcripts, GSK3B and DSCR1. Moreover, deregulation of other calcium-dependent pathways, including CaMK- and MAPK-pathways, was observed, as 7 of 9 transcripts encoding for genes of MAPK pathways with altered expression in pmAF were downregulated (online Table 3). In addition, downregulation of several additional transcription factors like NF-B, CREB1, STAT3, and STAT5 emphasized the general repression of transcriptional processes in pmAF.
Transcriptional Basis of Distinct Remodeling Pathways in Fibrillating Atrium
Based on GO, distinct functional classes could be identified for differentially expressed genes (online Tables 4 and 5), supporting the notion of several distinct areas of remodeling: consistent with prominent fibrosis and structural remodeling in pmAF, we noted changes in the expression of genes related to extracellular matrix composition and turnover with upregulation of collagen transcripts and extracellular matrix-cleaving proteases, like MMP9 and ADAM19. Downregulation of the corresponding tissue inhibitors of matrix metalloproteinases (eg, TIMP3, showing the most significant change of all genes deregulated in pmAF) also adds to adverse matrix remodeling. In addition, transforming growth factor (TGF)-1, a particularly important profibrotic cytokine, was upregulated. Downregulation of the TGF-1eCinhibitory transcript decorin (DCN) and upregulation of the insulin-like growth factor 2 receptor (IGF2R), which controls the extracellular bioavailability of TGF- is expected to augment TGF-1 levels. Deregulation of TGF- signaling is also exemplified by downregulation of signaling molecules downstream of TGF-1, including TGF- receptor 2, fibrillin 1, and SMAD4. Besides TGF-, we found evidence for transcriptional activation of additional profibrotic growth factors (upregulation of platelet-derived growth factor, PDGF, and concomitant downregulation of the PDGF receptor; downregulation of the fibroblast growth factor receptors 1 and 2, FGFR1+2) and neurohumoral signaling pathways (downregulation of endothelin receptors A and B, EDNRA and EDNRB; online Tables 3 and 5, "cellular component" and "molecular function," level 5).
There is also evidence for contractile remodeling in pmAF, as troponin I1 (TNNI1) was downregulated. Moreover, deregulation of several genes linked to a cardiomyopathic phenotype was observed (downregulation of ErbB2, ErbB3, and ErbB4; downregulation of lamin A/C (LMNA) and several genes involved in the formation of the dystrophin glycoprotein complex, including dystrophin, dystroglycan 1, -sarcoglycan, and 2-syntrophin; online Table 3).
Regarding electrophysiological remodeling, 5 transcripts encoding for subunits of L-type calcium channels were found to exhibit downregulation (online Table 3). The repolarization process was also disturbed, reflected in deregulation of several potassium channel transcripts (KCNIP2, KCNJ3, KCNJ5, KCNJ2, and KCNK3). In addition to changes of sarcolemmal ion channels, sarcoplasmic Ca2+ handling is expected to be altered by upregulation of RYR3 (a dominant-negative splice variant of the ryanodine receptor) and downregulation of ATP2A2 (sarcoplasmic Ca2+-ATPase). Changes in expression levels of KCNK3 and 12 additional transcripts were validated using real-time RT-PCR assays, confirming expression changes for 10 of 13 selected transcripts (Table 1).
Energy processes emerged as further area of pronounced changes, as 5 of 12 GO classes for biological processes found to be deregulated in pmAF serve metabolic functions (online Table 4, "biological process," level 5). Except for fatty acid and glycoprotein metabolism, there was a trend toward upregulation for all metabolic classes, as identified by GO (Figure 1 and online Table 4, "biological process," level 5). Upregulation of genes involved in the formation of the mitochondrial inner membrane equally demonstrates adaptation to an increased metabolic demand in pmAF (online Table 4, "cellular component," level 4).
Permanent AF Leads to Expression of a Ventricular-Like Signature in Atrium
Unsupervised clustering of genes deregulated in pmAF for all 30 atrial (pmAF+SR) and 5 LV samples revealed that atrial pmAF samples clustered with ventricular samples, rather than atrial SR samples (Figure 1 in the online data supplement). This is unexpected, as we identified 4 times more transcripts to be differentially expressed between atrium and ventricle than between pmAF and SR using identical technology and statistical methods: 3300 and 2974 transcripts showed higher expression values in atria and ventricles, respectively.11 Comparison of atrial- and ventricular-predominant transcripts with the genes deregulated in pmAF revealed that the ventricular-type transcription pattern of fibrillating atrial myocardium is mainly the result of downregulation of atrial-specific transcripts and only to a lesser degree of upregulation of ventricular-specific transcripts (Table 2).
Expression of a ventricular signature is also reflected in GO classes, as functional classes previously shown to be characteristic of atrial myocardium, such as those involved in signal transduction and cell communication,11 were downregulated in pmAF, whereas functional classes overrepresented in ventricular myocardium (eg, metabolic processes11) were found to be upregulated in pmAF (Figure 1 and online Table 4, "biological process").
Discussion
The present study is the first analysis of gene expression changes on a whole-genome level in human atrial fibrillation, providing specific gene sets underlying distinct remodeling processes and uncovering, for the first time, a prominent adaptation to long-term metabolic stress. Upregulation of ventricular-predominant metabolic genes as well as repression of atrial-specific transcriptional activity both contributed to the ventricular-like genomic signature in fibrillating atrial myocardium.
Transcriptional Basis of Distinct but Interconnected Remodeling Pathways in Fibrillating Atrium
Remodeling in AF has been recognized to consist of 3 major processes, affecting structural, contractile, and electrophysiological properties.1 Analysis based on GO identified major changes in several genes connecting these seemingly distinct areas of remodeling. For instance, downregulation of TIMP3 represents a possible link between extracellular matrix remodeling and depressed contractile function, as TIMP3-deficient mice have recently been shown to develop a cardiomyopathic phenotype with chamber dilatation and contractile dysfunction.12 Likewise, conditional mutations of the ErbB2 receptor in myocardium lead to dilated cardiomyopathy, and prolonged exposure of cultured human and rat cardiomyocytes with anti-ErbB2 antibody causes loss of contractile function.13 Thus, downregulation of the ErbB2 receptor might not only provide an explanation for extracellular matrix remodeling with dilatation in pmAF, but may also be linked to the contractile dysfunction that persists for some time after cardioversion of AF. Similarly, downregulation of several genes involved in the formation of the dystrophin glycoprotein complex, which is essential in providing the mechanical link from the intracellular cytoskeleton to the extracellular matrix, would also promote a cardiomyopathic phenotype.14
Other mechanisms, like endothelin activation, initiated to compensate for the contractile insufficiency in pmAF, may also be maladaptive in the long run by promoting cardiac fibrosis through activation of TGF-. Alterations of TGF- signaling have been connected to a number of cardiovascular and fibrotic diseases.15 Loss-of-function mutations in TGF- receptor 2 and fibrillin 1 have recently been linked to Marfan syndrome, an extracellular matrix disorder associated with atrial dilatation and atrial fibrillation.16 This suggests that the observed downregulation of both TGF- receptor 2 (TGFBR2) and fibrillin 1 (FBN1) in pmAF might provide the link to extracellular matrix remodeling. Interestingly, one of our AF patients had been diagnosed with Marfan syndrome based on the clinical phenotype. Figure 2 demonstrates individual measurements for FBN1 and TGFBR2 in all tissue specimens, illustrating that downregulation of these genes was not restricted to the patient with Marfan syndrome but common to all AF patients. Considering further genes for which mutations are associated with hereditary atrial fibrillation, it is important to note that Seebillon et al17 identified a lamin A/C missense mutation with early-onset AF as presenting symptom, preceding the development of dilated cardiomyopathy. Based on the observation that the same genes that have been shown to lead to hereditary forms of dilated atrial chambers and contractile dysfunction are also downregulated in our patients with acquired, valvular AF (eg, LMNA, TGFBR2, FBN1, TIMP3, and members of the dystrophin complex), it is tempting to speculate that these genes also contribute significantly to atrial dilatation and contractile dysfunction in acquired AF. Although transcriptional changes observed in the present study are the consequence rather than the cause of pmAF, it appears that some of the changes that occur initially to provide a substrate for pmAF are similar to alterations in the course of AF supporting the concept that AF promotes its own substrate ("AF begets AF").18
In addition to the well-established remodeling processes affecting contractile, electrophysiological, and structural properties, we identified prominent deregulation of genes belonging to metabolic classes, suggesting changes in atrial bioenergetics with an increased energy demand in pmAF. In line with this finding, cellular hypertrophy is among the most consistent histological findings in fibrillating atrial tissue, suggested to reflect an adaptation to altered environmental conditions.19 Induction of AF has been shown to increase atrial oxygen consumption and coronary flow almost 3-fold,20 possibly the result of high wall stress and the dramatic increase in atrial depolarization frequency with enhanced energy demand for ion transport processes including calcium-handling to maintain physiological cytosolic ion levels. In a goat model of AF, phosphocreatine decreased by 60% in atrial myocytes after 1 week of AF, indicating enhanced demand for high-energy phosphates.21 As phosphocreatine levels increased thereafter and returned to baseline levels after 16 weeks of AF,21 compensatory mechanisms are assumed to be activated to restore the metabolic balance between energy demand and supply in pmAF. Our finding of prominent upregulation of metabolic processes as identified by GO strongly supports this concept (Figure 1). The well-established atrial-to-ventricular switch of myosin isoforms might also be related to the increase in metabolic demand in pmAF: contraction has been shown to be 5-fold more economical in ventricular than atrial myocardium, mainly because of the greater relative contribution of the slow ventricular -myosin heavy chain (MHC) isoform compared with the fast atrial -MHC transcript in ventricular muscle.22 Thus, increased expression of the ventricular-predominant slow -MHC isoform in fibrillating atrial myocardium may represent an attempt to improve economy of contraction.
Generalized Transcriptional Downregulation in Permanent AF, Likely Mediated by Altered Calcium Homeostasis
In pmAF, we observed prominent downregulation of transcriptional processes with more than twice as many genes being downregulated than upregulated. The predominant pattern of transcriptional repression is reflected in downregulation of functional classes linked to transcriptional processes as identified by GO. In addition, our data are in agreement with the first microarray study in an animal model of AF5 and a recent study in human myocardium,23 also suggesting predominant downregulation of transcription in AF.
Even if transcript levels of several myocyte-specific markers were unchanged between SR and pmAF samples (cardiac myosin heavy polypeptide 7, cardiac troponin T2, and calsequestrin 2; data not shown), suggesting that the cardiomyocyte compartment has not been diminished in pmAF, a reduction in cardiomyocyte number might still have occurred, as myolysis and cellular hypertrophy are well-established features of pmAF.24 Therefore, because of simultaneous reduction of myocyte number and increase in the expression of cardiac-specific genes per single cell no change in overall cardiac-specific gene expression might be apparent in pmAF samples. As laser microdissection has not been performed, we cannot determine transcription on a myocyte level. However, the degree of AF-induced myolysis is expected to be modest24 and cannot fully account for the pronounced reduction of transcriptional processes.
As to the mechanisms regulating gene transcription in myocardial cells, calcium homeostasis, contractile function, and stretch have been shown to play key roles.25,26 Because all these factors are altered after prolonged AF,27eC29 they must be considered potential initiators of altered gene transcription. Specifically, calcium is a potent activator of cardiac gene transcription. Although Ca2+ overload has clearly been established as an early event after initiation of AF,1 there is no direct experimental evidence to support the notion of a chronic calcium overload state in persistent and pmAF. In fact, after an initial increase, steady-state calcium levels begin to decrease as early as 30 minutes after initiation of AF and normalize within the first 48 hours.27 Moreover, Ausma and coworkers28 demonstrated decreased steady-state calcium levels after 16 weeks of AF in goats. Likewise, isolated canine atrial myocytes displayed decreased resting intracellular calcium levels and blunted calcium transients after a 4-week period of rapid atrial pacing.29 The idea that high intracellular calcium levels may no longer be present after prolonged AF is strongly supported by the experimental and clinical observation that calcium-channel blockers like verapamil are useful in preventing calcium-overload and electrical remodeling only when given very early or during paroxysms of AF but fail to have beneficial effects in persistent and permanent AF.30 Given our observation of transcriptional repression of several calcium-dependent pathways including calcineurin-NFATc3, MAPK, and CaMK in pmAF and the central, well-established role of calcium as a regulator of gene expression, we propose that the decrease in transcription during prolonged episodes of AF results from alterations in calcium homeostasis with alterations in the spatiotemporal properties of the calcium transients (Figure 3).
Permanent AF Leads to a Ventricular- and Fetal-Like Genomic Signature in Atrial Myocardium
Atrial and ventricular myocardium exhibit fundamental differences on a morphological, ultrastructural, and functional basis, because ventricular myocardium, above all, functions as contractile pump, thereby contrasting to the prominent neuroendocrine and regulatory functions of atrial myocardium.11,31
In AF, an atrial-to-ventricular switch has been shown for some individual genes in atrial samples of hypertrophied32 and cardiomyopathic hearts,33 eg, contractile protein isoforms like myosin heavy chains. It has been hypothesized that the atrial-to-ventricular switch of MHCs observed in these pathophysiological conditions is the result of the rise in intra-atrial pressure, similar to hemodynamic changes observed during postnatal development.32 Our approach of microarray analysis with simultaneous examination of thousands of genes extends these observations in single genes to a general pattern of regulation toward a ventricular gene program in AF. Central in the changes are genes related to energy metabolism, also reflected in ultrastructural changes occurring during long-term AF that resemble those seen in fetal cardiomyocytes and in hibernating myocardium. This led Ausma et al19 to put forward the concept of "cellular adaptation through dedifferentiation" in AF, where a fetal phenotype is thought to facilitate cardiomyocyte survival during stress conditions including stretch and ischemia.34 The coordinated transcriptional downregulation of enzymes controlling fatty acid oxidation and concomitant upregulation of enzymes involved in glucose utilization reported in our study is consistent with reinduction of a fetal energy metabolic program35 (Figure 1 and online Table 4). Equally, ultrastructural characteristics in AF like the perinuclear accumulation of glycogen are seen in fetal cardiomyocytes and are suggestive of a metabolic shift,36 as the glycogen storage is thought to act as supply for high-energyeCdemanding processes and to enhance mechanical cell stability.37 Thus, the observed expression of a ventricular and fetal gene program with upregulation of metabolic genes and downregulation of atrial-specific genes appears to be an integral part of this adaptive response to pmAF. Given the high metabolic demand of fibrillating atrial myocardium, one might also speculate that the extensive suppression of transcriptional processes and hence protein synthesis is an attempt to save energy by restricting synthesis to a smaller number of proteins, possibly similar to chronic hibernation.38
Limitation of the Present Study
Analysis of human myocardial tissue is complicated by biological variability, which is expected to be greater than in well-controlled animal models of AF. Different etiologies and duration of AF, differences in age, gender, and medications, as well as individual course of the disease of patients from which tissue specimens were obtained contribute to variability in gene expression when examining human tissue samples. In the present study, however, we were able to detect robust functional patterns extending beyond this biological variability. The validity of our data-set is corroborated by expression patterns of a wide variety of genes known to be regulated in pmAF (including NPPB,39 several transcripts of L-type calcium currents,3 KCNJ2,3 KCNJ3,3 KCNJ5,3 KCNIP2,40 ATP2A2,3 MMP9,2 vWF,41 EDNRA,42 EDNRB,42 and mortalin43). Still, many transcripts reported to be regulated in pmAF were not found to be differentially expressed between SR and pmAF in the present study (eg, different myosin isoforms5). In addition to the biological heterogeneity inherent to any study of human tissue specimens as outlined here, this is partly because of the incomplete annotational status of the human genome at the time the Affymetrix U133A+B arrays were designed. Even if we seek to present a comprehensive portrait of the transcriptional changes in pmAF, this picture can by no means be complete. Moreover, it has to be kept in mind that mRNA levels do not always directly translate into changes in protein or functional levels, as additional regulatory mechanisms on a posttranslational level exist.
AF is usually initiated and maintained in the left atrium, because the cycle length is shorter in left than in right atrium during AF. Accordingly, a recent microarray study of a pig model of AF suggested that the genomic changes are more pronounced in left than right atrium with considerable overlap in the genomic response.5 As only right atrial tissue was available for the present study, future studies will be needed to account for regional heterogeneity in AF at the atrial level.
Acknowledgments
This study was supported partly by a grant of the German Federal Ministry for Education and Research (BMBF-grant 01GS0109; S.K., G.S., M.N.) supporting the German National Genome Research Network (NGFN) and a grant from the scientific editorial board of the Mechner Medizinische Wochenschrift (MMW; A.S.B.). Data were generated and analyzed in the framework of a research collaboration with Aventis Pharma Deutschland GmbH.
Mathias Gebauer, Klaus Steinmeyer and Markus Bleich are full-time employees of Aventis Pharma Deutschland, GmbH.
This manuscript contains part of Elisabeth Arnoldi’s doctoral thesis, in preparation at the Medical Faculty of the Ludwig-Maximilians-University, Munich, Germany.
References
Allessie M, Ausma J, Schotten U. Electrical, contractile and structural remodeling during atrial fibrillation. Cardiovasc Res. 2002; 54: 230eC246.
Nakano Y, Niida S, Dote K, Takenaka S, Hirao H, Miura F, Ishida M, Shingu T, Sueda T, Yoshizumi M, Chayama K. Matrix metalloproteinase-9 contributes to human atrial remodeling during atrial fibrillation. J Am Coll Cardiol. 2004; 43: 818eC825.
Dobrev D, Ravens U. Remodeling of cardiomyocyte ion channels in human atrial fibrillation. Basic Res Cardiol. 2003; 98: 137eC148.
Goette A, Lendeckel U, Klein HU. Signal transduction systems and atrial fibrillation. Cardiovasc Res. 2002; 54: 247eC258.
Lai LP, Lin JL, Lin CS, Yeh HM, Tsay YG, Lee CF, Lee HH, Chang ZF, Hwang JJ, Su MJ, Tseng YZ, Huang SKS. Functional genomic study on atrial fibrillation using cDNA microarray and two-dimensional protein electrophoresis techniques and identification of the myosin regulatory light chain isoform reprogramming in atrial fibrillation. J Cardiovasc Electrophys. 2004; 15: 214eC223.
Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, Speed TP. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003; 4: 249eC264.
Storey JD, Tibshirani R. Statistical significance for genomewide studies. Proc Natl Acad Sci U S A. 2003; 100: 9440eC9445.
Bussey KJ, Kane D, Sunshine M, Narasimhan S, Nishizuka S, Reinhold WC, Zeeberg B, Ajay, Weinstein JN. MatchMiner: A tool for batch navigation among gene and gene product identifiers. Genome Biol. 2003; 4: R27.
Sturn A, Quackenbush J, Trajanoski Z. Genesis: Cluster analysis of microarray data. Bioinformatics. 2002; 18: 207eC208.
Al Shahrour F, Diaz-Uriarte R, Dopazo J. FatiGO: a web tool for finding significant associations of Gene Ontology terms with groups of genes. Bioinformatics. 2004; 20: 578eC580.
Barth AS, Merk S, Arnoldi E, Zwermann L, Kloos P, Gebauer M, Steinmeyer K, Bleich M, Kb S, Pfeufer A, eerfuhr P, Dugas M, Steinbeck G, Nabauer M. Functional profiling of human atrial and ventricular gene expression. Eur J Physiol. In Press.
Fedak PW, Smookler DS, Kassiri Z, Ohno N, Leco KJ, Verma S, Mickle DAG, Watson KL, Hojilla CV, Cruz W, Weisel RD, Li RK, Khokha R. TIMP-3 Deficiency Leads to Dilated Cardiomyopathy. Circulation. 2004; 110: 2401eC2409.
Garratt AN, Ozcelik C, Birchmeier C. ErbB2 pathways in heart and neural diseases. Trends Cardiovasc Med. 2003; 13: 80eC86.
Lapidos KA, Kakkar R, McNally EM. The dystrophin glycoprotein complex - Signaling strength and integrity for the sarcolemma. Circ Res. 2004; 94: 1023eC1031.
Akhurst RJ. TGFbeta signaling in health and disease. Nat Genet. 2004; 36: 790eC792.
Mizuguchi T, Collod-Beroud G, Akiyama T, Abifadel M, Harada N, Morisaki T, Allard D, Varret M, Clausters M, Morisaki H, Ihara M, Kinoshita A, Yoshiura K, Junien C, Kajii T, Jondeau G, Ohta T, Kishino T, Furukawa Y, Nakamura Y, Niikawa N, Boileau C, Matsumoto N. Heterozygous TGFBR2 mutations in Marfan syndrome. Nat Genet. 2004; 36: 855eC860.
Seebillon P, Bouchier C, Bidot LD, Bonne G, Ahamed K, Charron P, Drouin-Garraud V, Millaire A, Desrumeaux G, Benaiche A, Charniot JC, Schwartz K, Villard E, Komajda M. Expanding the phenotype of LMNA mutations in dilated cardiomyopathy and functional consequences of these mutations. J Med Gen. 2003; 40: 560eC567.
Wijffels MCEF, Kirchhof CJHJ, Dorland R, Allessie MA. Atrial fibrillation begets atrial fibrillation - a study in awake chronically instrumented goats. Circulation. 1995; 92: 1954eC1968.
Thijssen VLJL, Ausma J, Borgers M. Structural remodelling during chronic atrial fibrillation: act of programmed cell survival. Cardiovasc Res. 2001; 52: 14eC24.
White CW, Kerber RE, Weiss HR, Marcus ML. The effects of atrial fibrillation on atrial pressure-volume and flow relationships. Circ Res. 1982; 51: 205eC215.
Ausma J, Coumans WA, Duimel H, Van der Vusse GJ, Allessie MA, Borgers M. Atrial high energy phosphate content and mitochondrial enzyme activity during chronic atrial fibrillation. Cardiovasc Res. 2000; 47: 788eC796.
Narolska NA, van Loon RB, Boontje NM, Zaremba R, Eiras Penas S, Russell J, Spiegelenberg SR, Huybregts MAJM, Visser FC, de Jong JW, van der Velden J, Stienen GJM. Myocardial contraction is 5-fold more economical in ventricular than in atrial human tissue. Cardiovasc Res. 2005; 65: 221eC229.
Ohki-Kaneda R, Ohashi J, Yamamoto K, Ueno S, Ota J, Choi YL, Koinuma K, Yamashita Y, Misawa Y, Fuse K, Ikeda U, Shimada K, Mano H. Cardiac function-related gene expression profiles in human atrial myocytes. Biochem Biophys Res Comm. 2004; 320: 1328eC1336.
Schotten U, Ausma J, Stellbrink C, Sabatschus I; Vogel M; Frechen D; Schoendube F, Hanrath P, Allessie MA. Cellular Mechanisms of Depressed Atrial Contractility in Patients With Chronic Atrial Fibrillation. Circulation. 2001; 103: 691eC698.
Kubisch C, Wollnik B, Maass A, Meyer R, Vetter H, Neyses L. Immediate-early gene induction by repetitive mechanical but not electrical activity in adult rat cardiomyocytes. FEBS Lett. 1993; 335: 37eC40.
McDonough PM, Glembotski CC. Induction of atrial natriuretic factor and myosin light chain-2 gene expression in cultured ventricular myocytes by electrical stimulation of contraction. J Biol Chem. 1992; 267: 11665eC11668.
Akar JG, Everett TH, Ho RY, Craft J, Haines DE, Somlyo AP, Somlyo AV. Intracellular chloride accumulation and subcellular elemental distribution during atrial fibrillation. Circulation. 2003; 107: 1810eC1815.
Ausma J, Dispersyn GD, Duimel H, Thone F, Donck LV, Allessie MA, Borgers M. Changes in ultrastructural calcium distribution in goat atria during atrial fibrillation. J Mol Cell Cardiol. 2000; 32: 355eC364.
Kneller J, Sun H, Leblanc N, Nattel S. Remodeling of Ca2+-handling by atrial tachycardia: evidence for a role in loss of rate-adaptation. Cardiovasc Res. 2002; 54: 416eC426.
Lee SH, Yu WC, Cheng JJ, Hung CR, Ding YA, Chang MS, Chen SA. Effect of verapamil on long-term tachycardia-induced atrial electrical remodeling. Circulation. 2000; 101: 200eC206.
Piper HM, Isenberg G, eds. Isolated Cardiac Myocytes. Boca Raton, Fla: CRC Press, Inc; 1989.
Kumar C, Saidapet C, Delaney P, Mendola C, Siddiqui MAQ. Expression of ventricular-type myosin light chain messenger RNA in spontaneously hypertensive rat atria. Circ Res. 1988; 62: 1093eC1097.
Wankerl M, Bohm M, Morano I, Ruegg JC, Eichhorn M, Erdmann E. Calcium sensitivity and myosin light chain pattern of atrial and ventricular skinned cardiac fibers from patients with various kinds of cardiac disease. J Mol Cell Cardiol. 1990; 22: 1425eC1438.
Ausma J, Thone F, Dispersyn GD, Flameng W, Vanoverschelde JL, Raemaekers FCS, Borgers M. Dedifferentiated cardiomyocytes from chronic hibernating myocardium are ischemia-tolerant. Mol Cell Biochem. 1998; 186: 159eC168.
Ingwall JS, Weiss RG. Is the failing heart energy starved On using chemical energy to support cardiac function. Circ Res. 2004; 95: 135eC145.
Ausma J, Litjens N, Lenders MH, Duimel H, Mast F, Wouters L, Ramaekers F, Allessie M, Borgers M. Time course of atrial fibrillation-induced cellular structural remodeling in atria of the goat. J Mol Cell Cardiol. 2001; 33: 2083eC2094.
Knaapen MWM, Vrolijk BCM, Wenink ACG. Ultrastructural changes of the myocardium in the embryonic rat heart. Anatom Rec. 1997; 248: 233eC241.
Storey KB. Mammalian hibernation: transcriptional and translational controls. Adv Exp Med Biol. 2003; 543: 21eC38.
Tuinenburg AE, Brundel BJJM, Van Gelder IC, Henning RH, Van den Berg MP, Driessen C, Grandjean JG, van Gilst WH, Crijns HJGM. Gene expression of the natriuretic peptide system in atrial tissue of patients with paroxysmal and persistent atrial fibrillation. J Cardiovasc Electrophys. 1999; 10: 827eC835.
Barth A, Zwermann L, Kb S, Nbauer M. KCHiP2 mRNA downregulation in human atrial and ventricular myocardium in heart failure and atrial fibrillation may limit functional Ito expression. Circulation. 2002; 106: 90.
Kumagai K, Fukuchi M, Ohta J, Baba S, Oda K, Akimoto H, Kagaya Y, Watanabe J, Tabayashi K, Shirato K. Expression of the von Willebrand factor in atrial endocardium is increased in atrial fibrillation depending on the extent of structural remodeling. Circulation J. 2004; 68: 321eC327.
Brundel BJJM, Van Gelder IC, Tuinenburg AE, Wietses M, Van Veldhuisen DJ, van Gilst WH, Crijns HJGM, Henning RH. Endothelin system in human persistent and paroxysmal atrial fibrillation. J Cardiovasc Electrophys. 2001; 12: 737eC742.
Kirmanoglou K, Hannekum A, Schafler AE. Expression of mortalin in patients with chronic atrial fibrillation. Basic Res Cardiol. 2004; 99: 404eC408.(Andreas S. Barth, Sylvia )
the Department of Medical Informatics, Biometrics
Epidemiology (S.M., M.D.), Ludwig-Maximilians-University, Munich, Germany
Aventis Pharma Deutschland GmbH (M.G., K.S., M.B.), Frankfurt am Main, Germany
the Institute of Physiology (M.B.), Christian- Albrechts-University, Kiel, Germany.
Abstract
Atrial fibrillation is associated with increased expression of ventricular myosin isoforms in atrial myocardium, regarded as part of a dedifferentiation process. Whether reexpression of ventricular isoforms in atrial fibrillation is restricted to transcripts encoding for contractile proteins is unknown. Therefore, this study compares atrial mRNA expression in patients with permanent atrial fibrillation to atrial mRNA expression in patients with sinus rhythm and to ventricular gene expression using Affymetrix U133 arrays. In atrial myocardium, we identified 1434 genes deregulated in atrial fibrillation, the majority of which, including key elements of calcium-dependent signaling pathways, displayed downregulation. Functional classification based on Gene Ontology provided the specific gene sets of the interdependent processes of structural, contractile, and electrophysiological remodeling. In addition, we demonstrate for the first time a prominent upregulation of transcripts involved in metabolic activities, suggesting an adaptive response to increased metabolic demand in fibrillating atrial myocardium. Ventricular-predominant genes were 5 times more likely to be upregulated in atrial fibrillation (174 genes upregulated, 35 genes downregulated), whereas atrial-specific transcripts were predominantly downregulated (56 genes upregulated, 564 genes downregulated). Overall, in fibrillating atrial myocardium, functional classes of genes characteristic of ventricular myocardium were found to be upregulated (eg, metabolic processes), whereas functional classes predominantly expressed in atrial myocardium were downregulated (eg, signal transduction and cell communication). Therefore, dedifferentiation with adoption of a ventricular-like signature is a general feature of the fibrillating atrium.
Key Words: atrial fibrillation gene expression atrial myocardium functional genomics
Introduction
Atrial fibrillation (AF), the most common sustained arrhythmia, is associated with extensive structural, contractile, and electrophysiological remodeling,1 stabilizing AF in the long run. Because current pharmacological treatment of AF is limited by ventricular proarrhythmia and inefficacy to prevent recurrences of AF, understanding the molecular events of these remodeling processes is essential for the development of new targeted therapeutic interventions. Even though knowledge of genes involved in remodeling processes exists on a gene-by-gene basis including transcripts of the extracellular matrix compartment,2 ion channels,1,3 and signal transduction molecules,4 microarray analysis offers a broader and unbiased approach to identify pathophysiologically relevant pathways. Therefore, we performed a systematic functional analysis of gene expression in permanent atrial fibrillation (pmAF) based on information from the Gene Ontology (GO) database to relate expression changes of single transcripts to known remodeling processes.
Moreover, atrial fibrillation has been shown to be associated with an atrial-to-ventricular switch by expressing ventricular myosin isoforms, which is regarded as part of a dedifferentiation process of fibrillating atrial myocardium.5 To explore the hypothesis that pmAF leads to upregulation of ventricular transcripts other than those encoding for contractile proteins, we used a genome-wide approach to compare atrial mRNA expression of patients with pmAF to atrial mRNA expression of patients with sinus rhythm (SR) as well as to ventricular (LV) gene expression. Our findings indicate that pmAF leads to repression of atrial transcriptional activity and induces a ventricular-like genomic signature in atrial myocardium.
Materials and Methods
Patients and Tissue Selection
Right atrial appendages were obtained from 30 patients undergoing open heart surgery for valve repair or coronary artery bypass grafting. Of these, 10 patients had pmAF defined as duration of AF longer than 3 months as documented by ECG, whereas 20 patients had no history of AF and were in SR when open heart surgery was performed. All patients gave written informed consent. Transmural LV tissue sections spanning the entire myocardial wall from subepicardium to subendocardium were obtained from 5 nonfailing donor hearts that were not transplanted because of palpable coronary calcifications. After excision, all tissue specimens were immediately frozen in liquid nitrogen and stored at eC80°C. The investigation was approved by the Institutional Review Board. Detailed patient characteristics are listed in online Table I (available at http://circres.ahajournals.org).
mRNA Preparation and Hybridization to Affymetrix U133A+B Oligonucleotide Microarrays
Total RNA was purified from homogenized deep-frozen tissue samples following the TRIZOL standard protocol as described by the manufacturer (GibcoBRL) and quantified by photometry. Then, 150 ng total RNA was used for quality control using the RNA 6000 Nano LabChip kit and Agilent Bioanalyzer 2100 (Agilent Technologies).
Synthesis of double-stranded cDNA and of biotinylated cRNA probes as well as subsequent steps of hybridization, washing, and staining followed instructions provided by Affymetrix.
Data Extraction and Statistical Analysis
One set of U133A+B chips (together representing 44 928 probe sets) was used for each human heart sample. The original data files have been deposited in Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo) and are accessible through GEO Series accession number GSE2240. All U133 GeneChips passed quality control to eliminate scans with abnormal characteristics, ie, abnormal low or high dynamic range, high perfect match saturation, high pixel noise, grid misalignment problems, and low mean signal-to-noise ratio. Low level and statistical analysis was done using R 1.8.1. Microarray data of all atrial and ventricular samples was normalized simultaneously using robust multiarray average (RMA)6 implemented in Bioconductor’s "affy"-package. Probe sets with "absent" calls in more than 50% of tissue samples in either group (SR, pmAF, or LV) were filtered out and omitted from further analysis. To determine differentially expressed genes between the groups, Welch t test was applied. Considering multiple hypothesis testing, differences in expression were regarded as statistically significant if a false discovery rate (FDR) of q <0.03 was achieved.7 Duplicate probe sets within the gene lists were eliminated by means of the "MatchMiner" software tool.8 Hierarchical clustering was performed by using an average linkage algorithm with Euclidean distance measure comprised in the Genesis software.9 Functional classification was based on the "cellular component," "biological process," and "molecular function" of GO using "FatiGO."10 After accounting for uncharacterized transcripts, up to 70% of all known genes were annotated across all GO levels. As many genes are not annotated at a particular level of ontology, three adjacent ontology levels are presented to increase the amount of information.
Validation by Semiquantitative Real-Time Reverse Transcription PCR
Expression patterns of 13 genes, selected with respect to the central role in the proposed working model of AF (Figure 3) were validated by real-time SYBR-Green PCR. A detailed list of the genes examined by RT-PCR including the design of the primers and a description of the protocols used for RT-PCR are given in the expanded Materials and Methods section in the online data supplement.
Results
Permanent Atrial Fibrillation Is Characterized by General Transcriptional Downregulation
We identified 1434 genes to be differentially expressed between pmAF and SR samples. Of these, 452 and 982 genes were up- and down-regulated in pmAF, respectively. A complete list of all transcripts deregulated in pmAF is provided in online Table 3. Downregulation of transcriptional processes was substantiated by GO analysis as functional classes of "transcription," "transcription factor activity," and "transcription cofactor activity" were downregulated (online Table 4, "molecular function" and "biological process," level 5). Downregulation was particularly noted for key elements of calcium-dependent signaling pathways, exemplified by the calcineurin-NFATc signaling pathway with downregulation of central components like calcineurin A and NFATc3, and concomitant upregulation of NFATc3-inhibitory transcripts, GSK3B and DSCR1. Moreover, deregulation of other calcium-dependent pathways, including CaMK- and MAPK-pathways, was observed, as 7 of 9 transcripts encoding for genes of MAPK pathways with altered expression in pmAF were downregulated (online Table 3). In addition, downregulation of several additional transcription factors like NF-B, CREB1, STAT3, and STAT5 emphasized the general repression of transcriptional processes in pmAF.
Transcriptional Basis of Distinct Remodeling Pathways in Fibrillating Atrium
Based on GO, distinct functional classes could be identified for differentially expressed genes (online Tables 4 and 5), supporting the notion of several distinct areas of remodeling: consistent with prominent fibrosis and structural remodeling in pmAF, we noted changes in the expression of genes related to extracellular matrix composition and turnover with upregulation of collagen transcripts and extracellular matrix-cleaving proteases, like MMP9 and ADAM19. Downregulation of the corresponding tissue inhibitors of matrix metalloproteinases (eg, TIMP3, showing the most significant change of all genes deregulated in pmAF) also adds to adverse matrix remodeling. In addition, transforming growth factor (TGF)-1, a particularly important profibrotic cytokine, was upregulated. Downregulation of the TGF-1eCinhibitory transcript decorin (DCN) and upregulation of the insulin-like growth factor 2 receptor (IGF2R), which controls the extracellular bioavailability of TGF- is expected to augment TGF-1 levels. Deregulation of TGF- signaling is also exemplified by downregulation of signaling molecules downstream of TGF-1, including TGF- receptor 2, fibrillin 1, and SMAD4. Besides TGF-, we found evidence for transcriptional activation of additional profibrotic growth factors (upregulation of platelet-derived growth factor, PDGF, and concomitant downregulation of the PDGF receptor; downregulation of the fibroblast growth factor receptors 1 and 2, FGFR1+2) and neurohumoral signaling pathways (downregulation of endothelin receptors A and B, EDNRA and EDNRB; online Tables 3 and 5, "cellular component" and "molecular function," level 5).
There is also evidence for contractile remodeling in pmAF, as troponin I1 (TNNI1) was downregulated. Moreover, deregulation of several genes linked to a cardiomyopathic phenotype was observed (downregulation of ErbB2, ErbB3, and ErbB4; downregulation of lamin A/C (LMNA) and several genes involved in the formation of the dystrophin glycoprotein complex, including dystrophin, dystroglycan 1, -sarcoglycan, and 2-syntrophin; online Table 3).
Regarding electrophysiological remodeling, 5 transcripts encoding for subunits of L-type calcium channels were found to exhibit downregulation (online Table 3). The repolarization process was also disturbed, reflected in deregulation of several potassium channel transcripts (KCNIP2, KCNJ3, KCNJ5, KCNJ2, and KCNK3). In addition to changes of sarcolemmal ion channels, sarcoplasmic Ca2+ handling is expected to be altered by upregulation of RYR3 (a dominant-negative splice variant of the ryanodine receptor) and downregulation of ATP2A2 (sarcoplasmic Ca2+-ATPase). Changes in expression levels of KCNK3 and 12 additional transcripts were validated using real-time RT-PCR assays, confirming expression changes for 10 of 13 selected transcripts (Table 1).
Energy processes emerged as further area of pronounced changes, as 5 of 12 GO classes for biological processes found to be deregulated in pmAF serve metabolic functions (online Table 4, "biological process," level 5). Except for fatty acid and glycoprotein metabolism, there was a trend toward upregulation for all metabolic classes, as identified by GO (Figure 1 and online Table 4, "biological process," level 5). Upregulation of genes involved in the formation of the mitochondrial inner membrane equally demonstrates adaptation to an increased metabolic demand in pmAF (online Table 4, "cellular component," level 4).
Permanent AF Leads to Expression of a Ventricular-Like Signature in Atrium
Unsupervised clustering of genes deregulated in pmAF for all 30 atrial (pmAF+SR) and 5 LV samples revealed that atrial pmAF samples clustered with ventricular samples, rather than atrial SR samples (Figure 1 in the online data supplement). This is unexpected, as we identified 4 times more transcripts to be differentially expressed between atrium and ventricle than between pmAF and SR using identical technology and statistical methods: 3300 and 2974 transcripts showed higher expression values in atria and ventricles, respectively.11 Comparison of atrial- and ventricular-predominant transcripts with the genes deregulated in pmAF revealed that the ventricular-type transcription pattern of fibrillating atrial myocardium is mainly the result of downregulation of atrial-specific transcripts and only to a lesser degree of upregulation of ventricular-specific transcripts (Table 2).
Expression of a ventricular signature is also reflected in GO classes, as functional classes previously shown to be characteristic of atrial myocardium, such as those involved in signal transduction and cell communication,11 were downregulated in pmAF, whereas functional classes overrepresented in ventricular myocardium (eg, metabolic processes11) were found to be upregulated in pmAF (Figure 1 and online Table 4, "biological process").
Discussion
The present study is the first analysis of gene expression changes on a whole-genome level in human atrial fibrillation, providing specific gene sets underlying distinct remodeling processes and uncovering, for the first time, a prominent adaptation to long-term metabolic stress. Upregulation of ventricular-predominant metabolic genes as well as repression of atrial-specific transcriptional activity both contributed to the ventricular-like genomic signature in fibrillating atrial myocardium.
Transcriptional Basis of Distinct but Interconnected Remodeling Pathways in Fibrillating Atrium
Remodeling in AF has been recognized to consist of 3 major processes, affecting structural, contractile, and electrophysiological properties.1 Analysis based on GO identified major changes in several genes connecting these seemingly distinct areas of remodeling. For instance, downregulation of TIMP3 represents a possible link between extracellular matrix remodeling and depressed contractile function, as TIMP3-deficient mice have recently been shown to develop a cardiomyopathic phenotype with chamber dilatation and contractile dysfunction.12 Likewise, conditional mutations of the ErbB2 receptor in myocardium lead to dilated cardiomyopathy, and prolonged exposure of cultured human and rat cardiomyocytes with anti-ErbB2 antibody causes loss of contractile function.13 Thus, downregulation of the ErbB2 receptor might not only provide an explanation for extracellular matrix remodeling with dilatation in pmAF, but may also be linked to the contractile dysfunction that persists for some time after cardioversion of AF. Similarly, downregulation of several genes involved in the formation of the dystrophin glycoprotein complex, which is essential in providing the mechanical link from the intracellular cytoskeleton to the extracellular matrix, would also promote a cardiomyopathic phenotype.14
Other mechanisms, like endothelin activation, initiated to compensate for the contractile insufficiency in pmAF, may also be maladaptive in the long run by promoting cardiac fibrosis through activation of TGF-. Alterations of TGF- signaling have been connected to a number of cardiovascular and fibrotic diseases.15 Loss-of-function mutations in TGF- receptor 2 and fibrillin 1 have recently been linked to Marfan syndrome, an extracellular matrix disorder associated with atrial dilatation and atrial fibrillation.16 This suggests that the observed downregulation of both TGF- receptor 2 (TGFBR2) and fibrillin 1 (FBN1) in pmAF might provide the link to extracellular matrix remodeling. Interestingly, one of our AF patients had been diagnosed with Marfan syndrome based on the clinical phenotype. Figure 2 demonstrates individual measurements for FBN1 and TGFBR2 in all tissue specimens, illustrating that downregulation of these genes was not restricted to the patient with Marfan syndrome but common to all AF patients. Considering further genes for which mutations are associated with hereditary atrial fibrillation, it is important to note that Seebillon et al17 identified a lamin A/C missense mutation with early-onset AF as presenting symptom, preceding the development of dilated cardiomyopathy. Based on the observation that the same genes that have been shown to lead to hereditary forms of dilated atrial chambers and contractile dysfunction are also downregulated in our patients with acquired, valvular AF (eg, LMNA, TGFBR2, FBN1, TIMP3, and members of the dystrophin complex), it is tempting to speculate that these genes also contribute significantly to atrial dilatation and contractile dysfunction in acquired AF. Although transcriptional changes observed in the present study are the consequence rather than the cause of pmAF, it appears that some of the changes that occur initially to provide a substrate for pmAF are similar to alterations in the course of AF supporting the concept that AF promotes its own substrate ("AF begets AF").18
In addition to the well-established remodeling processes affecting contractile, electrophysiological, and structural properties, we identified prominent deregulation of genes belonging to metabolic classes, suggesting changes in atrial bioenergetics with an increased energy demand in pmAF. In line with this finding, cellular hypertrophy is among the most consistent histological findings in fibrillating atrial tissue, suggested to reflect an adaptation to altered environmental conditions.19 Induction of AF has been shown to increase atrial oxygen consumption and coronary flow almost 3-fold,20 possibly the result of high wall stress and the dramatic increase in atrial depolarization frequency with enhanced energy demand for ion transport processes including calcium-handling to maintain physiological cytosolic ion levels. In a goat model of AF, phosphocreatine decreased by 60% in atrial myocytes after 1 week of AF, indicating enhanced demand for high-energy phosphates.21 As phosphocreatine levels increased thereafter and returned to baseline levels after 16 weeks of AF,21 compensatory mechanisms are assumed to be activated to restore the metabolic balance between energy demand and supply in pmAF. Our finding of prominent upregulation of metabolic processes as identified by GO strongly supports this concept (Figure 1). The well-established atrial-to-ventricular switch of myosin isoforms might also be related to the increase in metabolic demand in pmAF: contraction has been shown to be 5-fold more economical in ventricular than atrial myocardium, mainly because of the greater relative contribution of the slow ventricular -myosin heavy chain (MHC) isoform compared with the fast atrial -MHC transcript in ventricular muscle.22 Thus, increased expression of the ventricular-predominant slow -MHC isoform in fibrillating atrial myocardium may represent an attempt to improve economy of contraction.
Generalized Transcriptional Downregulation in Permanent AF, Likely Mediated by Altered Calcium Homeostasis
In pmAF, we observed prominent downregulation of transcriptional processes with more than twice as many genes being downregulated than upregulated. The predominant pattern of transcriptional repression is reflected in downregulation of functional classes linked to transcriptional processes as identified by GO. In addition, our data are in agreement with the first microarray study in an animal model of AF5 and a recent study in human myocardium,23 also suggesting predominant downregulation of transcription in AF.
Even if transcript levels of several myocyte-specific markers were unchanged between SR and pmAF samples (cardiac myosin heavy polypeptide 7, cardiac troponin T2, and calsequestrin 2; data not shown), suggesting that the cardiomyocyte compartment has not been diminished in pmAF, a reduction in cardiomyocyte number might still have occurred, as myolysis and cellular hypertrophy are well-established features of pmAF.24 Therefore, because of simultaneous reduction of myocyte number and increase in the expression of cardiac-specific genes per single cell no change in overall cardiac-specific gene expression might be apparent in pmAF samples. As laser microdissection has not been performed, we cannot determine transcription on a myocyte level. However, the degree of AF-induced myolysis is expected to be modest24 and cannot fully account for the pronounced reduction of transcriptional processes.
As to the mechanisms regulating gene transcription in myocardial cells, calcium homeostasis, contractile function, and stretch have been shown to play key roles.25,26 Because all these factors are altered after prolonged AF,27eC29 they must be considered potential initiators of altered gene transcription. Specifically, calcium is a potent activator of cardiac gene transcription. Although Ca2+ overload has clearly been established as an early event after initiation of AF,1 there is no direct experimental evidence to support the notion of a chronic calcium overload state in persistent and pmAF. In fact, after an initial increase, steady-state calcium levels begin to decrease as early as 30 minutes after initiation of AF and normalize within the first 48 hours.27 Moreover, Ausma and coworkers28 demonstrated decreased steady-state calcium levels after 16 weeks of AF in goats. Likewise, isolated canine atrial myocytes displayed decreased resting intracellular calcium levels and blunted calcium transients after a 4-week period of rapid atrial pacing.29 The idea that high intracellular calcium levels may no longer be present after prolonged AF is strongly supported by the experimental and clinical observation that calcium-channel blockers like verapamil are useful in preventing calcium-overload and electrical remodeling only when given very early or during paroxysms of AF but fail to have beneficial effects in persistent and permanent AF.30 Given our observation of transcriptional repression of several calcium-dependent pathways including calcineurin-NFATc3, MAPK, and CaMK in pmAF and the central, well-established role of calcium as a regulator of gene expression, we propose that the decrease in transcription during prolonged episodes of AF results from alterations in calcium homeostasis with alterations in the spatiotemporal properties of the calcium transients (Figure 3).
Permanent AF Leads to a Ventricular- and Fetal-Like Genomic Signature in Atrial Myocardium
Atrial and ventricular myocardium exhibit fundamental differences on a morphological, ultrastructural, and functional basis, because ventricular myocardium, above all, functions as contractile pump, thereby contrasting to the prominent neuroendocrine and regulatory functions of atrial myocardium.11,31
In AF, an atrial-to-ventricular switch has been shown for some individual genes in atrial samples of hypertrophied32 and cardiomyopathic hearts,33 eg, contractile protein isoforms like myosin heavy chains. It has been hypothesized that the atrial-to-ventricular switch of MHCs observed in these pathophysiological conditions is the result of the rise in intra-atrial pressure, similar to hemodynamic changes observed during postnatal development.32 Our approach of microarray analysis with simultaneous examination of thousands of genes extends these observations in single genes to a general pattern of regulation toward a ventricular gene program in AF. Central in the changes are genes related to energy metabolism, also reflected in ultrastructural changes occurring during long-term AF that resemble those seen in fetal cardiomyocytes and in hibernating myocardium. This led Ausma et al19 to put forward the concept of "cellular adaptation through dedifferentiation" in AF, where a fetal phenotype is thought to facilitate cardiomyocyte survival during stress conditions including stretch and ischemia.34 The coordinated transcriptional downregulation of enzymes controlling fatty acid oxidation and concomitant upregulation of enzymes involved in glucose utilization reported in our study is consistent with reinduction of a fetal energy metabolic program35 (Figure 1 and online Table 4). Equally, ultrastructural characteristics in AF like the perinuclear accumulation of glycogen are seen in fetal cardiomyocytes and are suggestive of a metabolic shift,36 as the glycogen storage is thought to act as supply for high-energyeCdemanding processes and to enhance mechanical cell stability.37 Thus, the observed expression of a ventricular and fetal gene program with upregulation of metabolic genes and downregulation of atrial-specific genes appears to be an integral part of this adaptive response to pmAF. Given the high metabolic demand of fibrillating atrial myocardium, one might also speculate that the extensive suppression of transcriptional processes and hence protein synthesis is an attempt to save energy by restricting synthesis to a smaller number of proteins, possibly similar to chronic hibernation.38
Limitation of the Present Study
Analysis of human myocardial tissue is complicated by biological variability, which is expected to be greater than in well-controlled animal models of AF. Different etiologies and duration of AF, differences in age, gender, and medications, as well as individual course of the disease of patients from which tissue specimens were obtained contribute to variability in gene expression when examining human tissue samples. In the present study, however, we were able to detect robust functional patterns extending beyond this biological variability. The validity of our data-set is corroborated by expression patterns of a wide variety of genes known to be regulated in pmAF (including NPPB,39 several transcripts of L-type calcium currents,3 KCNJ2,3 KCNJ3,3 KCNJ5,3 KCNIP2,40 ATP2A2,3 MMP9,2 vWF,41 EDNRA,42 EDNRB,42 and mortalin43). Still, many transcripts reported to be regulated in pmAF were not found to be differentially expressed between SR and pmAF in the present study (eg, different myosin isoforms5). In addition to the biological heterogeneity inherent to any study of human tissue specimens as outlined here, this is partly because of the incomplete annotational status of the human genome at the time the Affymetrix U133A+B arrays were designed. Even if we seek to present a comprehensive portrait of the transcriptional changes in pmAF, this picture can by no means be complete. Moreover, it has to be kept in mind that mRNA levels do not always directly translate into changes in protein or functional levels, as additional regulatory mechanisms on a posttranslational level exist.
AF is usually initiated and maintained in the left atrium, because the cycle length is shorter in left than in right atrium during AF. Accordingly, a recent microarray study of a pig model of AF suggested that the genomic changes are more pronounced in left than right atrium with considerable overlap in the genomic response.5 As only right atrial tissue was available for the present study, future studies will be needed to account for regional heterogeneity in AF at the atrial level.
Acknowledgments
This study was supported partly by a grant of the German Federal Ministry for Education and Research (BMBF-grant 01GS0109; S.K., G.S., M.N.) supporting the German National Genome Research Network (NGFN) and a grant from the scientific editorial board of the Mechner Medizinische Wochenschrift (MMW; A.S.B.). Data were generated and analyzed in the framework of a research collaboration with Aventis Pharma Deutschland GmbH.
Mathias Gebauer, Klaus Steinmeyer and Markus Bleich are full-time employees of Aventis Pharma Deutschland, GmbH.
This manuscript contains part of Elisabeth Arnoldi’s doctoral thesis, in preparation at the Medical Faculty of the Ludwig-Maximilians-University, Munich, Germany.
References
Allessie M, Ausma J, Schotten U. Electrical, contractile and structural remodeling during atrial fibrillation. Cardiovasc Res. 2002; 54: 230eC246.
Nakano Y, Niida S, Dote K, Takenaka S, Hirao H, Miura F, Ishida M, Shingu T, Sueda T, Yoshizumi M, Chayama K. Matrix metalloproteinase-9 contributes to human atrial remodeling during atrial fibrillation. J Am Coll Cardiol. 2004; 43: 818eC825.
Dobrev D, Ravens U. Remodeling of cardiomyocyte ion channels in human atrial fibrillation. Basic Res Cardiol. 2003; 98: 137eC148.
Goette A, Lendeckel U, Klein HU. Signal transduction systems and atrial fibrillation. Cardiovasc Res. 2002; 54: 247eC258.
Lai LP, Lin JL, Lin CS, Yeh HM, Tsay YG, Lee CF, Lee HH, Chang ZF, Hwang JJ, Su MJ, Tseng YZ, Huang SKS. Functional genomic study on atrial fibrillation using cDNA microarray and two-dimensional protein electrophoresis techniques and identification of the myosin regulatory light chain isoform reprogramming in atrial fibrillation. J Cardiovasc Electrophys. 2004; 15: 214eC223.
Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, Speed TP. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003; 4: 249eC264.
Storey JD, Tibshirani R. Statistical significance for genomewide studies. Proc Natl Acad Sci U S A. 2003; 100: 9440eC9445.
Bussey KJ, Kane D, Sunshine M, Narasimhan S, Nishizuka S, Reinhold WC, Zeeberg B, Ajay, Weinstein JN. MatchMiner: A tool for batch navigation among gene and gene product identifiers. Genome Biol. 2003; 4: R27.
Sturn A, Quackenbush J, Trajanoski Z. Genesis: Cluster analysis of microarray data. Bioinformatics. 2002; 18: 207eC208.
Al Shahrour F, Diaz-Uriarte R, Dopazo J. FatiGO: a web tool for finding significant associations of Gene Ontology terms with groups of genes. Bioinformatics. 2004; 20: 578eC580.
Barth AS, Merk S, Arnoldi E, Zwermann L, Kloos P, Gebauer M, Steinmeyer K, Bleich M, Kb S, Pfeufer A, eerfuhr P, Dugas M, Steinbeck G, Nabauer M. Functional profiling of human atrial and ventricular gene expression. Eur J Physiol. In Press.
Fedak PW, Smookler DS, Kassiri Z, Ohno N, Leco KJ, Verma S, Mickle DAG, Watson KL, Hojilla CV, Cruz W, Weisel RD, Li RK, Khokha R. TIMP-3 Deficiency Leads to Dilated Cardiomyopathy. Circulation. 2004; 110: 2401eC2409.
Garratt AN, Ozcelik C, Birchmeier C. ErbB2 pathways in heart and neural diseases. Trends Cardiovasc Med. 2003; 13: 80eC86.
Lapidos KA, Kakkar R, McNally EM. The dystrophin glycoprotein complex - Signaling strength and integrity for the sarcolemma. Circ Res. 2004; 94: 1023eC1031.
Akhurst RJ. TGFbeta signaling in health and disease. Nat Genet. 2004; 36: 790eC792.
Mizuguchi T, Collod-Beroud G, Akiyama T, Abifadel M, Harada N, Morisaki T, Allard D, Varret M, Clausters M, Morisaki H, Ihara M, Kinoshita A, Yoshiura K, Junien C, Kajii T, Jondeau G, Ohta T, Kishino T, Furukawa Y, Nakamura Y, Niikawa N, Boileau C, Matsumoto N. Heterozygous TGFBR2 mutations in Marfan syndrome. Nat Genet. 2004; 36: 855eC860.
Seebillon P, Bouchier C, Bidot LD, Bonne G, Ahamed K, Charron P, Drouin-Garraud V, Millaire A, Desrumeaux G, Benaiche A, Charniot JC, Schwartz K, Villard E, Komajda M. Expanding the phenotype of LMNA mutations in dilated cardiomyopathy and functional consequences of these mutations. J Med Gen. 2003; 40: 560eC567.
Wijffels MCEF, Kirchhof CJHJ, Dorland R, Allessie MA. Atrial fibrillation begets atrial fibrillation - a study in awake chronically instrumented goats. Circulation. 1995; 92: 1954eC1968.
Thijssen VLJL, Ausma J, Borgers M. Structural remodelling during chronic atrial fibrillation: act of programmed cell survival. Cardiovasc Res. 2001; 52: 14eC24.
White CW, Kerber RE, Weiss HR, Marcus ML. The effects of atrial fibrillation on atrial pressure-volume and flow relationships. Circ Res. 1982; 51: 205eC215.
Ausma J, Coumans WA, Duimel H, Van der Vusse GJ, Allessie MA, Borgers M. Atrial high energy phosphate content and mitochondrial enzyme activity during chronic atrial fibrillation. Cardiovasc Res. 2000; 47: 788eC796.
Narolska NA, van Loon RB, Boontje NM, Zaremba R, Eiras Penas S, Russell J, Spiegelenberg SR, Huybregts MAJM, Visser FC, de Jong JW, van der Velden J, Stienen GJM. Myocardial contraction is 5-fold more economical in ventricular than in atrial human tissue. Cardiovasc Res. 2005; 65: 221eC229.
Ohki-Kaneda R, Ohashi J, Yamamoto K, Ueno S, Ota J, Choi YL, Koinuma K, Yamashita Y, Misawa Y, Fuse K, Ikeda U, Shimada K, Mano H. Cardiac function-related gene expression profiles in human atrial myocytes. Biochem Biophys Res Comm. 2004; 320: 1328eC1336.
Schotten U, Ausma J, Stellbrink C, Sabatschus I; Vogel M; Frechen D; Schoendube F, Hanrath P, Allessie MA. Cellular Mechanisms of Depressed Atrial Contractility in Patients With Chronic Atrial Fibrillation. Circulation. 2001; 103: 691eC698.
Kubisch C, Wollnik B, Maass A, Meyer R, Vetter H, Neyses L. Immediate-early gene induction by repetitive mechanical but not electrical activity in adult rat cardiomyocytes. FEBS Lett. 1993; 335: 37eC40.
McDonough PM, Glembotski CC. Induction of atrial natriuretic factor and myosin light chain-2 gene expression in cultured ventricular myocytes by electrical stimulation of contraction. J Biol Chem. 1992; 267: 11665eC11668.
Akar JG, Everett TH, Ho RY, Craft J, Haines DE, Somlyo AP, Somlyo AV. Intracellular chloride accumulation and subcellular elemental distribution during atrial fibrillation. Circulation. 2003; 107: 1810eC1815.
Ausma J, Dispersyn GD, Duimel H, Thone F, Donck LV, Allessie MA, Borgers M. Changes in ultrastructural calcium distribution in goat atria during atrial fibrillation. J Mol Cell Cardiol. 2000; 32: 355eC364.
Kneller J, Sun H, Leblanc N, Nattel S. Remodeling of Ca2+-handling by atrial tachycardia: evidence for a role in loss of rate-adaptation. Cardiovasc Res. 2002; 54: 416eC426.
Lee SH, Yu WC, Cheng JJ, Hung CR, Ding YA, Chang MS, Chen SA. Effect of verapamil on long-term tachycardia-induced atrial electrical remodeling. Circulation. 2000; 101: 200eC206.
Piper HM, Isenberg G, eds. Isolated Cardiac Myocytes. Boca Raton, Fla: CRC Press, Inc; 1989.
Kumar C, Saidapet C, Delaney P, Mendola C, Siddiqui MAQ. Expression of ventricular-type myosin light chain messenger RNA in spontaneously hypertensive rat atria. Circ Res. 1988; 62: 1093eC1097.
Wankerl M, Bohm M, Morano I, Ruegg JC, Eichhorn M, Erdmann E. Calcium sensitivity and myosin light chain pattern of atrial and ventricular skinned cardiac fibers from patients with various kinds of cardiac disease. J Mol Cell Cardiol. 1990; 22: 1425eC1438.
Ausma J, Thone F, Dispersyn GD, Flameng W, Vanoverschelde JL, Raemaekers FCS, Borgers M. Dedifferentiated cardiomyocytes from chronic hibernating myocardium are ischemia-tolerant. Mol Cell Biochem. 1998; 186: 159eC168.
Ingwall JS, Weiss RG. Is the failing heart energy starved On using chemical energy to support cardiac function. Circ Res. 2004; 95: 135eC145.
Ausma J, Litjens N, Lenders MH, Duimel H, Mast F, Wouters L, Ramaekers F, Allessie M, Borgers M. Time course of atrial fibrillation-induced cellular structural remodeling in atria of the goat. J Mol Cell Cardiol. 2001; 33: 2083eC2094.
Knaapen MWM, Vrolijk BCM, Wenink ACG. Ultrastructural changes of the myocardium in the embryonic rat heart. Anatom Rec. 1997; 248: 233eC241.
Storey KB. Mammalian hibernation: transcriptional and translational controls. Adv Exp Med Biol. 2003; 543: 21eC38.
Tuinenburg AE, Brundel BJJM, Van Gelder IC, Henning RH, Van den Berg MP, Driessen C, Grandjean JG, van Gilst WH, Crijns HJGM. Gene expression of the natriuretic peptide system in atrial tissue of patients with paroxysmal and persistent atrial fibrillation. J Cardiovasc Electrophys. 1999; 10: 827eC835.
Barth A, Zwermann L, Kb S, Nbauer M. KCHiP2 mRNA downregulation in human atrial and ventricular myocardium in heart failure and atrial fibrillation may limit functional Ito expression. Circulation. 2002; 106: 90.
Kumagai K, Fukuchi M, Ohta J, Baba S, Oda K, Akimoto H, Kagaya Y, Watanabe J, Tabayashi K, Shirato K. Expression of the von Willebrand factor in atrial endocardium is increased in atrial fibrillation depending on the extent of structural remodeling. Circulation J. 2004; 68: 321eC327.
Brundel BJJM, Van Gelder IC, Tuinenburg AE, Wietses M, Van Veldhuisen DJ, van Gilst WH, Crijns HJGM, Henning RH. Endothelin system in human persistent and paroxysmal atrial fibrillation. J Cardiovasc Electrophys. 2001; 12: 737eC742.
Kirmanoglou K, Hannekum A, Schafler AE. Expression of mortalin in patients with chronic atrial fibrillation. Basic Res Cardiol. 2004; 99: 404eC408.(Andreas S. Barth, Sylvia )