Genes controlling multiple functional pathways are transcriptionally regulated in connexin43 null mouse heart

Dumitru A. Iacobas1, Sanda Iacobas1, W. E. I. Li1, Georg Zoidl3, Rolf Dermietzel3 and David C. Spray1,2

1 Department of Neuroscience Albert Einstein College of Medicine, Bronx, New York
2 Department of Cardiology, Albert Einstein College of Medicine, Bronx, New York
3 Department of Neuroanatomy and Molecular Brain Research, Ruhr-University, Bochum, Germany


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
We have used mouse 27k cDNA arrays to compare gene expression patterns in four sets of three hearts each of neonatal wild types and four sets of three hearts each of littermates lacking the major cardiac gap junction protein, connexin43 (Cx43). Each individual set of hearts was hybridized against aliquots of an RNA standard prepared from selected mouse tissues, allowing calculation of variability and coordination of gene expression among the samples from both genotypes. Overall variance of gene expression was found to be markedly higher in wild-type hearts than in those from Cx43 null littermates. Expression levels of 586 of 5,613 adequately quantifiable distinct genes with known protein products were statistically altered in the Cx43 null hearts, 38 upregulated and 548 downregulated compared with wild types. Downregulation was confirmed for seven tested genes by quantitative RT-PCR. Functions of proteins encoded by the altered genes encompassed all functional categories, with largest percent changes in genes involved in intracellular transport and transcription factors. Among the downregulated genes in the Cx43 null hearts were those related to neuronal and glial function, suggesting that cardiac innervation might be compromised as a consequence of Cx43 deletion. This was supported by immunodetection of sympathetic innervation, using antibodies to the synaptic vesicle protein synaptophysin and to the adrenergic nerve terminal marker tyrosine hydroxylase. These findings reinforce the proposal that the cardiac abnormality in Cx43 null animals may be contributed by altered innervation and indicate that Cx43 deletion has consequences in addition to reduced intercellular communication.

cDNA array; expression coordination; cardiac innervation; transcription control


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
CONNEXIN43 (Cx43; also termed gja1 or gap junction {alpha}1-protein) was the second gap junction protein to be discovered, by a strategy whereby cardiac cDNA libraries were screened at low stringency with cDNA sequence corresponding to the liver gap junction protein connexin32 (4). Cx43 is now known to be the most widely expressed gap junction protein in mammals, where it occurs in almost every tissue (4, 8). Cx43 is a primary component of intercellular gap junction channels in cardiac tissue, where it is found between ventricular and atrial myocytes and in both smooth muscle and endothelial cells of the vessel wall (see Ref. 50). Reduced gap junctional coupling and Cx43 expression or phosphorylation have been associated with both acute ischemia and chronic cardiac disease. For example, changes in electrical coupling induced by acute myocardial ischemia have been correlated with a marked dephosphorylation and internalization of Cx43, although the total Cx43 protein level was unchanged (31). Patients with healed myocardial infarcts and animal models of postinfarct myocardial injury display EKG abnormalities that are attributable to reduced junctional conductance, due to both rearrangement of junctional contacts (termed gap junction remodeling) and globally reduced Cx43 expression (2, 16, 31, 41, 44, 55). Reduced Cx43 expression has also been detected in patients with ischemic cardiomyopathy and other chronic myocardial disease states such as end-stage aortic stenosis (32, 45), congestive heart failure (14), and diabetic cardiomyopathy (42) as well as in a mouse dilated cardiomyopathy model (19). Junctional localization of Cx43 (but not Cx43 expression level) and contraction synchrony are also altered in cardiac myocytes infected with Trypanosoma cruzi, the protozoan parasite responsible for Chagasic cardiomyopathy (10, 11).

Coding region mutations in Cx43 have been reported in patients with visceroatrial heterotaxia (VAHT) (5), rare cases of nonsyndromic deafness (37), hypoplastic left heart syndrome (9), and occulodentodigital dysplasia (43). The first two of these reports are controversial, as large scale studies have not replicated the VAHT results (see Ref. 54), and the mutations associated with nonsyndromic deafness appear to be in the Cx43 pseudogene (cited in Ref. 43). Mice in which Cx43 is totally deleted (Cx43 null mice) die at birth due to a developmental cardiac abnormality, where ventricular outflow obstruction blocks blood flow to the lungs (48); additional developmental abnormalities include coronary artery patterning defects (36). Mice with cardiac-specific Cx43 deletion exhibit ventricular conduction slowing and sudden cardiac death (18).

Gene expression profiling has indicated that levels of Cx43 mRNA are altered in both acute and chronic stages of multiple sclerosis (39), in Alzheimer disease (40), in Huntington disease (56), and in other neural disorders (17). Although analogous studies on cardiac disease are only just beginning, gene arrays have revealed that Cx43 is downregulated in failing human hearts after implantation of ventricular assist devices (7), and it is noteworthy that expression of the transcription factor Nkx2.5, which may be involved in hypoplastic left heart syndrome, downregulates Cx43 (33).

Because of the critical role that Cx43 plays in normal cardiac function and the reports that its expression may be altered in pathological cardiac conditions, we have undertaken a gene profiling study in which gene expression was compared in hearts from Cx43 null mice and in those from wild-type littermates. Using the 27k cDNA array produced by the Albert Einstein College of Medicine (AECOM) Microarray Core Facility, we have found that expression of ~10% of the quantifiable known genes was significantly altered (primarily downregulated) in the Cx43 null heart. These genes encode proteins with a wide range of biological functions, indicating that Cx43 expression in the heart affects numerous processes, not all of which are readily explainable by the deficit in intercellular communication.


    MATERIALS AND METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Mice.
Adult C57BL/6j wild-type (used for reference RNA isolation, see below) and heterozygous Cx43(+/–) mice were obtained from Jackson Laboratory (Bar Harbor, ME) and bred and maintained in our Association for Assessment and Accreditation of Laboratory Animal Care-accredited animal facility. These transgenic mice were generated by homologous recombination as described (48). Because Cx43 null mice do not survive long after birth, heterozygotes were mated and offspring screened by PCR of tail DNA (12). Hearts of twelve wild-type (W) and twelve Cx43 null (K) neonatal mice were used in this study, where four sets of three hearts of each genotype were compared to obtain enough total RNA to avoid RNA amplification and/or use of indirect labeling that might introduce bias toward a subpopulation of mRNA. The mice were decapitated under aseptic conditions according to protocols approved by the AECOM Animal Use Committee, and hearts were removed and immediately processed. A mouse reference RNA mixture (hereafter denoted by S) was prepared at one time in sufficient quantity for the entire projected experiment from selected amounts of total RNA extracted from several adult male and female mouse tissues (aorta, brain, heart, kidney, liver, lung, ovary/testicles, spleen, and stomach) in a combination that provided a high diversity of genes expressed in the midrange of the detection system for the AECOM mouse cDNA microarrays (23).

RNA sources and experimental design.
Sixty micrograms of total RNA, extracted according to the AECOM protocol (http:// microarray1k.aecom.yu.edu/) from S and from each of the four sets of three hearts isolated from W and K mice, were reverse transcribed into cDNA incorporating fluorescent dUTPs [Cy3-dUTP (green) and Cy5-dUTP (red)]. The labeled cDNAs were hybridized overnight at 50°C on 10 aminosilane-coated Corning glass slides, spotted with 27k selected mouse DNA sequences produced by the Microarray Core Facility of the AECOM (http://microarray1k.aecom.yu.edu), in the combinations indicated in Fig. 1A. Eight arrays were used for hybridization of sets of W and K hearts against S, and two additional slides were used for the so-called "yellow test" by hybridization of differently labeled S extracts against each other to validate the protocol and determine the printing quality (Fig. 1B). These additional slides were scanned several times at different photomultiplier tube (PMT) voltages to optimize scanning parameters so that signals recorded by the two channels of the scanner were globally balanced and the numbers of spots eliminated because of low signal and saturated pixels was minimized.



View larger version (27K):
[in this window]
[in a new window]
 
Fig. 1. A: experimental design. RNA sources are indicated as S (RNA obtained from a mixture of mouse tissues used for the reference), K1–K4 [RNA from 4 connexin43 (Cx43) null hearts], and W1–W4 (RNA from 4 wild-type hearts). Differently labeled sources [red and green triangles, corresponding to whether cDNA was labeled with Cy3 (green) or Cy5 (red)] are cohybridized on microarrays (squares formed by the red and green triangles). Calib, calibration procedure used to verify the hybridization protocol, select the photomultiplier tube voltages, and test the normalization algorithm by studying the expression ratios obtained when differently labeled sample references were hybridized against each other ("yellow test"); K:S and W:S, expression ratios when the K and W extracts are compared with the reference sample; K:W, computed expression ratio in the Cx43 null heart with respect to the wild-type heart. B: yellow test to verify the quality of the hybridization protocol and of the normalization procedure (results from the 1st chip used in calibration). Sg(spot) and Sr(spot), background-subtracted normalized signals of the quantifiable spots; mean Sg and mean Sr, mean values of the above for the entire array. Note the high value of the linear regression and the approximate equality of the 2 sets of normalized signals (slope 0.95 in the linear fitting).

 
After hybridization, the slides were washed at room temperature with solutions containing 0.1% sodium dodecyl sulfate (SDS) and 1% SSC (3 M NaCl + 0.3 M sodium citrate) to remove the nonhybridized probes.

Scanning, data acquisition, and normalization.
The microarrays were scanned with an Axon GenePix 4000A scanner, and data were acquired through GenePix Pro 4.0 software (http://www.axon.com). The spots with substantial local imperfections (customarily flagged by the acquisition program), those for which the medians of the foreground signals were not twice as high the medians of the background signals in both channels, and those with saturated pixels were eliminated from the analysis to avoid false hits. The background-subtracted signals were normalized through an in house-developed iterative algorithm, alternating within-array normalization with interarray normalization, until the average corrected ratio differed by <5% from the previous one (22). The cDNA arrays used contained 27,571 spotted sequences: 15,693 spots probing 7,455 distinct mouse genes with known protein products in http://genome-www5.stanford.edu/cgi-bin/source//sourceBatchSearch, 11,686 spots corresponding to mouse expressed sequence tags (ESTs) the annotations of which were incomplete at the date of the study (eliminated from the expression analysis), and 192 spots with bacterial sequences for quality control of the arrays. We organized the subset of valid spots that probed genes with known protein products in "redundancy groups" (RGs), with each group composed of the spots probing the same gene. To each of the n(j) spots composing the RG of gene j were assigned a "weight" g(i;j) equal to the ratio between the number I(i;j) of identities with the probed gene and the length L(i;j) of the query, thereby maximizing the contribution of the most trustable sequences within the redundancy group (23).

The normalized expression ratio {chi}(i;j) and the P value of significant regulation P(i;j), obtained according to a previously described method (25), were determined for each spot i [=1,..., n(j)] within the RG of gene j. In rare situations when oppositely [e.g., {chi}(i1;j) >1 and {chi}(i2;j) <1] statistically significant [P(i1;j) <0.05 and P(i2;j) <0.05] regulations were obtained by spots (here i1 and i2) within the redundancy group, the entire RG was eliminated. As result of this analysis, we obtained 5,613 valid RGs that were used for subsequent comparisons of gene expression in wild-type and Cx43 null hearts.

Detection of significant gene regulation.
The expression ratio x(j) of a potentially regulated gene was computed as


{zh70030506850e01}

(1)
Definition in Eq. 1 assigns negative ratios to downregulated genes (e.g., x = –2 is 2-fold downregulation).

To reduce the number of false hits and also to limit the elimination of true hits, we have developed a regulation analysis that relies on both fold changes in expression ratio and the statistical significance of the two-tailed t-test for equality of two ratios (21) with a Bonferroni-type adjustment (13, 51) applied to the redundancy groups. The threshold fold change was established for each valid RG to exceed the effect of the 95% maximum error of estimate (21) of the normalized expression ratio, while the Bonferroni type adjustment multiplies the averaged P value by the number of spots composing the RG. The results were interpreted as previously described (23).

Evaluation of transcript abundance variability and control stringency.
We have used the 5%-significant relative estimated variability (REV) of transcript abundance in each genotype to evaluate the control stringency of the transcript abundance for each gene (with high REV values indicating low control) and the gene expression stability (GES; the percentile locating the REV value in an inversely ordered sequence) to categorize the genes in stability classes (25, 26). Genes with GES >75 were regarded as stably expressed and genes with GES <25 as unstably expressed. Although REV includes technical variability, our estimation of the technical noise (see below) indicates that it does not affect assignment of genes to stable and unstable expression categories.

Coordination of transcript abundances.
Pearson’s correlation coefficients {rho} between the binary logarithms of averaged normalized expression ratios of adequately quantified genes j in wild-type heart compared with the sample reference (determined as indicated in Fig. 1A by the module W/S) {psi}532(i;j;W)/{psi}635(i;j;S) within their redundancy groups n(j)

were used to determine the degree of expression coordination (2325) with Cx43 in wild-type hearts. Positive correlation indicates that expression of paired genes increases and decreases simultaneously from animal to animal (synergistically expressed), negative correlation indicates inverse expression tendencies (antagonistically expressed), and a value close to zero indicates that the variations of expression levels of the two genes are not coordinated (independently expressed). The statistical significance of the correlation coefficient is strongly dependent on the number of specimens. In our case of four biological replicas, the 5% cutoff values are 0.9 < {rho} ≤ 1 for synergism, –0.9 > {rho} ≥ –1 for antagonism, and |{rho}| < 0.05 for independence, and the 10% cut-offs are |{rho}| > 0.8 for coordination (synergistic or antagonistic) and |{rho}| < 0.1 for independence. Moderate correlations (0.1 ≥ |{rho}| ≥ 0.8) were eliminated from the analysis because of questionable significance.

Assignment of gene functions.
We have used GeneBank accession numbers and the website http://genome-www5.stanford.edu/cgi-bin/source//sourceBatchSearch for gene annotation of the spotted sequences. Annotated genes were categorized in functional classes of proteins encoded according to a simple classification scheme (see Ref. 26; derived from Ref. 1): junction-adhesion-extracellular matrix (JAE; antigens, globulins, integrins, claudins, cadherins, connexins, desmosomal components, laminin, proteoglycans, etc.), cytoskeletal (CY; intermediary filaments, microtubules, centrioles, actin, and their associated proteins), transport into the cell (T1; channels, transporters, and ionotropic receptors), transport within the cell (T2; proteins of vesicles, cellular motors, endosomes, lysosomes, nuclear transport, protein folding, etc.), cell signaling (CS; G protein-coupled receptors, protein kinases, SH2 and SH3 domain proteins, calcium-binding proteins, etc.), cell cycle-shape-differentiation-death (CSD; growth factors, apoptosis-related genes, cytokines, etc.), transcription (TR; DNA binding proteins, DNA repair, RNA, transcription factors, oncogenes, etc.), energy and metabolism (EnMet; oxidants, peroxisomes, respiratory chain, glycolysis and glycogenesis, enzymes, etc.), and genes encoding proteins for which function is presently unknown (UNKf). Percentages of regulated genes encoding proteins of each functional class were calculated from this classification and are presented with the above-mentioned in the form of a bar graph (see Fig. 3B).

Quantitative real-time PCR.
RNA samples independent of those used for the cDNA array study were extracted from four hearts of neonatal wild-type and four hearts of neonatal Cx43 null mice; they were tested for RNA integrity by electrophoresis, and cDNA synthesis was performed as previously described (58). Primers were designed with the use of the Primer Express program (Applied Biosystems) for universal reaction conditions, defined by the manufacturer. The sequences of the eight primer pairs (Biomers.net, Ulm, Germany) are summarized in Supplemental Table S1 (available at the Physiological Genomics web site).1 All primers produced single amplification products of the calculated melting temperature as verified by melting point analysis. SYBR Green I reaction conditions were as recommended by the manufacturer (Eurogentec). The experiments were performed, using a GenAmp 5700 Sequence Detection System (Applied Biosystems), in triplicate. A total of 12 data points were generated from four hearts of each genotype.

Expression ratios were calculated on the basis of primer efficiencies and the crossing point deviation of unknown samples vs. a control (46). Primer efficiencies (E) for each primer pair were determined from standard curves by the formula E = 101/slope (47). The crossing point value (Cp) was defined as the point where the fluorescence signal reached a value of 0.1 above background during the exponential phase of the reaction. The relative expression software tool (REST; Ref. 46) was applied for group-wise comparison and statistical analysis of relative expression results in real-time PCR using default settings.

Immunocytochemistry and Western blot analysis.
Cx43 null and wild-type newborn mice were killed by fluoquene overdose immediately after birth. Hearts were promptly removed and fixed for 3 h in 4% paraformaldehyde. Hearts were then washed in phosphate-buffered saline, cryoprotected in 15% sucrose, and cut into 15-µm sections using a cryostat. These sections were subsequently stained for Cx43 using a rabbit polyclonal antibody (35), for synaptophysin using a monoclonal anti-synaptophysin antibody (Sigma), and for tyrosine hydroxylase using a monoclonal antibody (Chemicon) according to a previously described protocol (35). Sections were subsequently stained by use of fluorescent secondary antibodies conjugated with Alexa 488 for the Cx43 polyclonal antibody and Alexa 594 for the monoclonal antibodies (Molecular Probes, Eugene, OR) and examined with a Nikon TE300 epifluorescence microscope equipped with a 1.3 NA 40x oil immersion objective. Photomicrographs were obtained with a Spot digital camera (Diagnostic Instruments, Sterling Heights, MI) and composed with the use of Adobe Photoshop software.

Western blotting analysis was conducted according to a protocol previously described (35). Briefly, left ventricles from wild-type and Cx43 null pups were excised and homogenized by sonication in a lysis buffer containing 50 mM Tris buffer (pH 7.4), 1% Nonidet P-40, 1x protease inhibitors, 10 mM sodium fluoride, and 1 mM sodium orthovanadate. Protein contents were determined using a Pierce BCA kit (Rockford, IL) to ensure equal loading for Western blotting analysis. After electrophoresis and transblotting, membranes were blocked in Tris-buffered saline solution containing 0.2% Tween 20 (TBST) and 5% skim milk for an hour before the overnight incubation with the above antibodies against Cx43 or synaptophysin, and subsequent horseradish peroxidase-conjugated secondary antibodies at room temperature. Western blots were quantified by measuring the relative density of the 38-kDa bands recognized by the synaptophysin antibody using Scion Image software (Frederick, MD). The difference in synaptophysin A expression was determined with the Student’s t-test.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Quality control.
The study was performed according to the standards of the Microarray Gene Expression Data Society, and data complying with the Minimum Information About Microarray Experiments have been deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo; platform GPL1698, series GSE1961, samples GSM35025–35032). The main features and representative examples are considered below.

We have shown previously (28) that the reproducibility of the AECOM mouse microarrays is within reasonable limits for this type of platform (average interarray coefficient of technical variability of the unnormalized, background-subtracted signals is ~30% without any significant difference between the 2 channels) (27). The power of the normalization procedure was evidenced by the reduction of the coefficient of variability of the red background-subtracted signals of the sample reference across the arrays from 32 ± 15% in the unnormalized data to 15 ± 8% after normalization. The quality of the incorporation of fluorescent dyes and of the hybridization of the labeled cDNA was checked by the yellow test (Fig. 1B); for the selected PMT, we obtained the average expression ratio x = 0.98 ± 0.05 and the correlation coefficient between the sets of normalized background subtracted signals in the two channels {rho} = 0.99. We conclude from this high correlation coefficient that technical variability is comparable in both channels. After elimination of all spots for which gene annotation did not recover known proteins and the invalid RGs due to significant opposite regulation, redundancy groups probing 5,613 distinct genes were adequately quantified.

Expression stability and control stringency of transcript abundance.
We calculated the REV values for transcript abundance of each adequately quantified gene with known protein product and analyzed the distributions of the REV values in wild-type and Cx43 null hearts. Individual genes in both genotypes exhibited a wide range of REV values (Fig. 2A), indicating a great variability in the stringency of transcriptional control and posttranscriptional mRNA stabilization/destabilization mechanisms leading to the actual steady-state mRNA levels measured by our approach.



View larger version (22K):
[in this window]
[in a new window]
 
Fig. 2. Expression variability. A: histogram of relative estimated variability (REV) values in wild-type (WT) and Cx43 null (KO) hearts. Note the lower mean value in Cx43 null vs. wild-type hearts. Skew, skewness; Kurt, kurtosis. B: REV values for each spot in the 2 genotypes plotted each against other. Observe that data set is bent toward the x-axis, with a very weak correlation between the 2 sets of REV data (r = 0.23 overall, and r = 0.07 for REV values in wild types above 100%).

 
For wild-type hearts, the mean REV value was 90.5% with a large standard deviation (62.6%). The distribution (Fig. 2A) is positively skewed (skewness = 1.3) and relatively peaked (kurtosis = 1.7) compared with that expected for a normal distribution (for which skewness and kurtosis are both 0). The distribution of REV values in hearts with disrupted expression of Cx43 (Fig. 2A) exhibited a significant (P<0.001) 36% lower mean value (57.9%), reduced standard deviation (34.4%), and increased asymmetry (skewness = 2.0) and peakedness (kurtosis = 8.8) compared with the wild-type hearts. Variability of expression for each quantifiable gene was compared between genotypes by plotting the REV values for Cx43 null hearts against those for wild-type hearts. The resulting graph (Fig. 2B) indicates a strong tendency for genes with high variability in wild types (REV >100%) to have lower values in transgenic mice, whereas genes having high REV values in the transgenics exhibited a wide range of REV values in wild-type hearts. Correlation between REV values for individual genes between wild-type and Cx43 null hearts was very low ({rho} = 0.23 over the entire range, {rho} = 0.07 for REV >100), indicating the virtual absence of a consonant change of transcriptional control stringency in the case of highly variable genes.

To further compare the variability of transcript abundance in wild-type and Cx43 null hearts, individual genes were ranked according to their expression stability (GES). Table 1 presents examples of stably and unstably transcribed genes in the wild-type hearts that either preserved (stable-stable and unstable-unstable) or reversed (stable-unstable and unstable-stable) their stability in the Cx43 null hearts. Although each gene chosen for illustration showed significant regulation of expression in Cx43 null compared with wild-type hearts (see Expression regulation), similar examples were also abundant in the case of nonregulated genes. We have included in Table 1 a classification of the proteins encoded by these genes as well as their chromosomal locations (see Distributions of regulated genes within functional classes and chromosomes). When stability of all regulated genes was considered in the context of functions, a larger proportion of CSD (41 vs. 31%) and TR (30 vs. 17%) and a smaller proportion of T1 (21 vs. 33%) and JAE (26 vs. 39%) regulated genes were unstable (GES <25%) in wild-type compared with Cx43 null hearts, and more JAE (25 vs. 14%) and fewer TR genes (23 vs. 29%) were stable (GES >75%) in wild-type hearts. T2 and T1 have the highest (34 and 33%, respectively) proportion of stable genes in both wild-type and Cx43 null hearts; CSD is the least-stable functional class in wild types (41%) and JAE the least-stable class (39%) in the Cx43 null hearts.


View this table:
[in this window]
[in a new window]
 
Table 1. Examples of stably and unstably expressed genes

 
When expression stability was compared for genes on each chromosome, chromosome 16 was found to have the highest (71.4%) and chromosome 4 the lowest (42.7% mean) stability in wild-type hearts. In the Cx43 null heart, the distribution was altered, chromosome X having the highest (64.6%) and chromosome 5 the lowest (36.5%) average stability.

Our observation that disruption of Cx43 alters the variability of transcript abundance and that alterations are not uniform with regard to either functional classes or chromosomal location suggests a complex pattern of interactions between the expression level of Cx43 and the control mechanisms of transcript abundance of other genes.

Expression regulation.
As illustrated in Fig. 3A, when ratio values of Cx43 null and wild-type hearts were compared, 1,277 spots probing well-annotated sequences exhibited significant (P < 0.05) hybridization ratio [identified using the Student’s t-test for the equality of two ratios (K/S:W/S)].



View larger version (19K):
[in this window]
[in a new window]
 
Fig. 3. Characteristics of quantified and regulated genes. A: plot of significantly (P < 0.05) regulated hybridization of well-annotated genes (log2 ratios) in set of arrays hybridized with cDNA obtained from Cx43 null heart (K) and reference sample (S) against hybridization (log2 ratios) in set of arrays hybridized with cDNA obtained from the wild-type heart (W) and reference sample. The line marks the equality of the 2 log ratios. Note the asymmetry toward downregulation. B and C: distributions of quantified and regulated genes within each of 9 functional classes (B; for description of classes, see MATERIALS AND METHODS) and chromosomes represented on the array (C). Note that the most highly regulated class of genes is that of transcription factors (TR). D and E: percentage distribution of regulated genes on each functional class (D) and chromosome (E). Note the highly nonuniform distribution of the absolute numbers of regulated genes in functional classes and the approximately uniform distribution of the percentages of regulated genes in functional classes, and that regulation of genes located on chromosome 10, which contains Cx43, is not markedly different from the percentage of regulated genes found on the other chromosomes.

 
These spots corresponded to the significant regulation of 586 (10.4%) of the total quantified distinct genes on the array, 38 (6.5%) genes being upregulated and 548 (93.5%) being downregulated. Of the significantly regulated genes, 156 (2.8%) were downregulated and 10 (0.2%) were upregulated by twofold or more. Table 2A presents examples of the most highly downregulated genes, and Table 2B presents all upregulated genes in the limit of P < 0.05. The full data set from which the examples and quantitation are derived is available as Supplemental Table S2.


View this table:
[in this window]
[in a new window]
 
Table 2. Downregulated and upregulated genes in Cx43 null mouse hearts

 
Distribution of regulated genes within functional classes and chromosomes.
Column charts illustrating the functions and chromosomal locations of all quantified and of all regulated genes in the Cx43 null hearts are shown in Fig. 3 B, and C. When affected genes were categorized according to function, the greatest numbers were transcription factors (128 TR genes, 22% of the total), followed by genes functioning in energy metabolism (107 EnMet genes, 18%). Thirty-four genes (5.8%) were classified as encoding proteins with unknown functions.

Expression of 80 CS genes was statistically different in the Cx43 null hearts (14% of the total regulated genes), including upregulation of apelin, an agonist of angiotensin receptor-like-1 and a potent inotropic agonist in myocardial cells (6), and downregulation of the ß2-adrenergic receptor and the estrogen, opioid, and transforming growth factor-ß receptors. Of the remaining 75 CS genes, a remarkably high proportion encoded protein kinases (12 genes) and phosphatases (13 genes). The regulated protein kinases include PKA regulatory subunits-ß1 and -ß2 and six that are related to mitogen-activated protein kinase pathways.

Fifty-nine CDS genes (10% of the total) were differentially expressed between the genotypes, including several growth factors: vascular endothelial growth factor (VEGF) and macrophage migration inhibitory factor were about twofold upregulated in the Cx43 null hearts, whereas insulin-like growth factor-1 (and its receptor), leukemia inhibitory factor, the glial cell neurotrophic factors-{alpha}1 and -{alpha}2, nerve growth factor receptor, and interleukin-4 receptor-{alpha} were all downregulated. Of the remaining genes, nine were cell cycle or mitosis related, and eight were related to apoptosis.

Twenty-four CY genes showed expression ratios that were different (4.1%), including three associated with myosin isoforms (myosin Va, phosphorylatable myosin light chain, and myosin-binding protein H) and five with functions related to cell motility (actin-ß, actinin-{alpha}4, destrin, band 4.1-like, proline-serine-threonine phosphatase-interacting protein-1, and drebrin-1).

Fifty-six JAE genes exhibited differential expression (9.7%), including decreases in 10 components of gap, tight, and adherens junctions and desmosomes (cingulin, cadherin-23, claudins-3 and -12, integrin-ß5, periplakin, carcinoembryonic antigen-related cell adhesion molecule, stromal interaction molecule-1, testicular cell adhesion molecule, and Cx43 itself); decreases in genes encoding three proteins related to myelin maintenance (myelin basic protein, proteolipid protein, and periaxin); and decreases in genes associated with pre- or postsynaptic specializations of nerve endings (e.g., calsyntenin) as well as genes related to neuronal migration (neuropilin, a semaphorin receptor, and trefoil). Extracellular matrix-regulated genes included chondroitin sulfate proteoglycan (>4-fold upregulated) and syndecan-1 (slightly downregulated).

Twenty-four transmembrane flux and transport (T1) genes were affected, more than one-half of which were involved in carrier-mediated transport (11 genes) or ion flux (amiloride-sensitive cation channel-3, voltage-gated potassium channel Kcns3, chloride channel-7, and the peripheral benzodiazopine receptor). Seventy-four genes related to intracellular transport were regulated (T2, 12%), including those encoding eight presynaptic proteins [bassoon, piccolo, SNAP23, synaptotagmin (Syt)-3 and -13 and Syt-like-1, syntaxin 3, and syntaxin-binding protein-1]. Caveolin, to which Cx43 has been shown to bind (49), was found to be 1.7-fold upregulated.

We further evaluated whether there was a pattern evident in the functional classes and/or chromosomal location of genes whose regulation was affected by Cx43 deletion. As shown in Fig. 3 B and C,, genes from all functional classes (from 271 CY genes to 1,266 TR genes) and, with the exception of the missing chromosomes 20–22 and only two genes from chromosome Y, genes from all other chromosomes were quantified on the arrays (from 175 genes located on chromosome 18 to 620 genes located on chromosome 11). Disruption of the Cx43 gene affected transcription of genes from all functional classes (from 24 CY genes to 128 TR genes) and located on all chromosomes (from 13 unigenes on chromosome 14 to 56 unigenes on chromosome 11). Remarkably, despite highly nonuniform distributions of absolute numbers, rather uniform percentages of quantified genes were regulated in each functional class (8.7 ± 1.1%) or chromosome (8.0 ± 1.3%) (Fig. 3 D, and E). This uniformity, consistent with that observed in Cx43 null and Cx43 heterozygous mouse brains (23), suggests that Cx43 expression may represent a central node in the regulation of gene expression patterns. Perhaps surprisingly, chromosome 10, containing the Cx43 gene, was ranked in the twelfth position with 8.1% regulated genes, whereas chromosomes 18 and 12 with 11.4% and 9.4% regulated genes, respectively, occupied the first two positions. This finding indicates that the pattern of gene regulation in the Cx43 null heart is not governed by proximity to the Cx43 gene locus and suggests a complex network of expression coordination.

To validate our cDNA microarray analysis, we used quantitative (q)RT-PCR as an independent methodological approach. Genes encoding adrenergic receptor-ß2 (Adrb2), bassoon (Bsn), piccolo (Pclo), proteolipid protein (myelin; Plp), stratifin (Sfn), solute carrier family 43, member 1 (Slc43a1), and tektin-2 (Tekt2) were selected for experimental verification. Note that five of these are related to innervation (Adrb2, Bsn, Pclo, Plp, and Tekt2). As shown in Fig. 4, the significant downregulation observed in the microarray study was confirmed for all seven genes by the qRT-PCR analysis.



View larger version (14K):
[in this window]
[in a new window]
 
Fig. 4. Gene expression downregulation confirmation by quantitative (q)RT-PCR. Note that all 7 tested significant downregulations detected by microarray were confirmed by qRT-PCR.

 
Coordination of transcript abundances.
The altered gene profile in the Cx43 null heart might represent a unique pattern that only arises from the complete absence of Cx43 gene expression. However, it could also be the case that the Cx43 null phenotype reflects an exaggeration of the coordination among the genes that occurs in normal heart. To test this second hypothesis, we have exploited the approximately twofold variability found in the log2 expression ratio of Cx43 in wild-type heart vs. reference to calculate correlation coefficients between these ratios and those of each gene whose expression was found to be statistically regulated in the Cx43 null hearts. Figure 5A presents examples of two genes either synergistically or antagonistically expressed with Cx43 in hearts of wild-type mice. Note the strong degree of coordination indicated by the high {rho} values in these examples.



View larger version (16K):
[in this window]
[in a new window]
 
Fig. 5. Coordination of the expression of other genes with that of Cx43 in wild-type hearts and relation of the coordination to regulation seen in Cx43 null hearts. A: binary logarithm of the average expression ratios within the redundancy group for Cx43 (Gja1) in wild-type hearts with respect to the sample reference

was plotted against that of each other gene j

for the 4 separate arrays. Examples are shown of synergistic expressions (Olfm1, olfactomedin-1; Cbr2, carbonyl reductase-2) and of antagonistic expressions [Ndufa8, NADH dehydrogenase (ubiquinone)-1{alpha} subcomplex-8; Prickle1, prickle-like-1 (Drosophila)]. Note the high correlation coefficients ({rho} values). B: coordination analysis of other genes with Cx43 in wild-type hearts plotted as a function of the log2 ratio of gene expression in Cx43 null hearts compared with wild-type hearts. For 347 of the regulated genes, spots were quantifiable on all 4 arrays. Of these, 69 coordinations were significant synergisms ({rho} > 0.8) and 5 were significant antagonisms ({rho} < –0.8) at the 10% level (see MATERIALS AND METHODS) and are plotted on the ordinate as synergistic or antagonistic coordinations together with 31 independent coordinations (|{rho}| < 0.1). As indicated by the arrow, 64 (87.8%) of these coordinations accurately predicted the type of regulation in the Cx43 null hearts.

 
To determine whether the degree of coordination matched the gene regulation seen in the Cx43 null heart, we have plotted the significant correlation coefficients of each gene with Cx43 expression level obtained in wild-type hearts as a function of the log2 K:W ratio for each of the genes. As indicated in Fig. 5B, the analysis of transcription coordination with Cx43 in the wild-type hearts predicted the type of regulation in Cx43 null hearts with surprising accuracy. For 347 of the regulated genes, spots were quantifiable on all four arrays. Of these, 77 coordinations were significant at the 10% level (|{rho}| > 0.8; see MATERIALS AND METHODS), and 65 (87.8%) of these accurately predicated the type of regulation (up or down) in the Cx43 null hearts. This finding supports the hypothesis that the altered gene regulation in the Cx43 null heart may in large part reflect interactions between the expression of Cx43 and other genes that exist in the heart under normal and/or pathophysiological conditions.

Downregulation of the synaptic protein synaptophysin in Cx43 null heart.
It has been demonstrated previously (15, 20) that Cx43 expression is critical for the migration of cardiac neural crest cells and for the formation of peripheral ganglia, which are derivatives of these cells. However, it has not been established whether Cx43 is indeed crucial for normal cardiac innervation. Thus it was particularly interesting to note that mRNAs encoding several synaptic proteins were downregulated in Cx43 null mouse hearts (see above), suggesting anomalous synaptogenesis. To determine whether innervation of the wild-type and Cx43 null hearts differs, using an independent methodological approach, we examined the distribution of the synaptic vesicle marker synaptophysin (which was not quantified by our cDNA microarrays) in wild-type and Cx43 null ventricles. Ventricular cardiac myocytes of wild-type heart exhibited abundant Cx43 immunolabeling (Fig. 6A), which was totally absent in Cx43 null heart (Fig. 6B). As expected, synaptophysin antibody labeled numerous puncta in wild-type heart sections (Fig. 6C; representative immunostaining images obtained from 5 wild-type hearts). However, immunostaining of synaptophysin was markedly lower in Cx43 null ventricle (Fig. 6D; representative images from 5 Cx43 null hearts). We further examined sympathetic innervation in the heart by tyrosine hydroxylase immunostaining. In wild-type heart, tyrosine hydroxylase immunoreactivity had a fibrous appearance and was present throughout the ventricle sections examined (arrows in Fig. 6E). In contrast, tyrosine hydroxylase immunostaining was absent in most areas of the left ventricles of Cx43 null hearts (Fig. 6F), except in the area adjacent to the epicardium. The specificity of monoclonal antibodies, including synaptophysin and tyrosine hydroxylase, is supported by the virtual absence of immunoreactivity in sections probed only with secondary antibody (not shown). Western blot analysis comparing wild-type and Cx43 null hearts confirmed the reduced synaptophysin expression seen with immunostaining (6.21 ± 0.62 arbitrary densitometric units for wild type vs. 2.32 ± 0.99 for Cx43 null; P < 0.02, data not shown). These results support the suggestion based on the microarray data that there is a deficiency in cardiac innervation in Cx43 null hearts.



View larger version (134K):
[in this window]
[in a new window]
 
Fig. 6. Immunostaining of left ventricles from wild-type and Cx43 null mice for Cx43 (A and B), for the nerve terminal marker synaptophysin (C and D), and for the sympathetic nerve marker tyrosine hydroxylase (E and F). Note that Cx43 immunolabeling is extensive in wild-type heart (A), Cx43 immunolabeling is absent in Cx43 null heart (B), punctate synaptophysin immunostaining was extensive throughout the ventricle of wild-type heart (C), punctate synaptophysin staining was much lower in Cx43 null heart than in wild-type heart (D), extensive sympathetic nerve fibers were revealed by fibrous immunostaining of tyrosine hydroxylase in Cx43 wild-type heart (E; arrows), and tyrosine hydroxylase staining in Cx43 null heart is sparse and mostly restricted to the region near the heart surface (F). Bar = 70 µm.

 

    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
We have used cDNA arrays to evaluate the gene expression profiles of hearts from neonatal Cx43 null mice. Using an experimental design where Cy3-labeled RNA extracted from both wild-type and Cx43 null hearts was cohybridized to a constant Cy5-labeled reference RNA has allowed us to measure biological variability in expression of each gene and to use this variability to determine whether apparent differences in gene expression ratios in Cx43 null hearts were statistically significant. Moreover, this method allowed calculation of correlation between Cx43 expression level and that of all other genes in wild-type and Cx43 null hearts.

The analysis of REV revealed that variability in individual transcript abundance among the four wild-type heart samples was higher than the variability in Cx43 null hearts. This difference between wild-type and Cx43 null hearts was most apparent for genes with high REV values (>100%) in wild-type heart, which showed less variable expression in hearts of Cx43 null littermates. This finding that deletion of Cx43 diminished the average expression variability among animals indicates that there is a higher degree of control of transcript abundances in Cx43-deficient compared with wild-type hearts. One consequence of this tighter control may be to compensate for the loss of Cx43 or to minimize the alteration of other genes linked to Cx43.

In an attempt to gain insight into possible mechanisms underlying this difference in REV in wild-type and Cx43 null hearts, we calculated whether genes encoding proteins within different functional classes showed more or less variability and also calculated the relative variability of genes found on each chromosome. From the standpoint of function, it may be noteworthy that JAE genes are the most unstable class in Cx43 null hearts, whereas genes involved in membrane and intracellular transport (T1 and T2) are the most stable in both wild-type and Cx43 null hearts. We have previously hypothesized that gene stability might be highest for genes determining differentiated cellular phenotype, thereby more readily allowing environmental impact on expression of genes with "housekeeping" cell functions (25, 26). The analysis performed here indicates profound alteration in the stability of JAE genes by disruption of a gap junction gene and also indicates that gene expression related to other functions (T1 and T2, transport of ions and molecules into and within cells) is only minimally destabilized. Chromosomal association with expression stability revealed a less than twofold difference between highest and lowest total REV values. However, chromosomal expression variance differed markedly in wild-type and Cx43 null hearts, further emphasizing the complexity of the control processes involved in gene regulation.

Because significant reductions of median REVs were also observed in brains of neonatal Cx43 null and Cx43 heterozygous mice (23), but not in cortical astrocytes cultured from Cx43 null mice (25), and both knockout and wild-type neonates used in the experiments reported here and in Refs. 23 and 25 were offsprings of the same heterozygous mothers, we hypothesize that the reduction in transcript abundance variability may represent a compensatory effect within the brain that is not necessarily present in a single isolated cell type. Quantification of expression ratios and their variances allowed identification of 586 distinct genes that showed significantly different expression in the Cx43 null compared with wild-type hearts. These regulated genes represented >10% of the identified, nonredundant, quantifiable spots on the arrays, indicating that disruption of Cx43 expression led to widespread changes in the expression of other genes.

Among the identified regulated genes in Cx43 null hearts are those that seem likely to result in altered cardiac function and organization. First, alterations detected in Ca2+-related genes (e.g., hippocalcin like-1, frizzled-2), apelin, and ß-adrenergic receptors suggest the possibility that inotropic responses may be altered in Cx43 null hearts. Second, a rather large number of growth factors and their receptors were found to be downregulated, with the exception of upregulation of a macrophage inhibitory factor and VEGF. Downregulation of growth factors (as well as changes in cell motility genes) could contribute to the decreased migration of cardiac neural crest cells that has been described in Cx43 mice (20, 38), whereas the upregulation of VEGF might contribute to the coronary patterning deficits in these animals (36). Third, alterations observed in neuron and glial-related genes and in a smaller number of fibroblast and smooth muscle genes may represent altered cardiac innervation and angiogenesis. Although cardiac innervation has not been investigated previously in Cx43 null mice, overexpression of Cx43 in neural crest has been shown to lead to altered sympathetic and sensory nerve development (15), and the Cx43 null heart shows increased smooth muscle and fibroblast infiltration (35). The results with synaptophysin and tyrosine hydroxylase immunostaining shown here and Western blot of synaptophysin (not illustrated) indicate that sympathetic innervation in the Cx43 null heart is substantially reduced. This is consistent with array results, confirmed by qRT-PCR, that myelination and presynaptic and cytoskeletal neuronal genes are downregulated in the Cx43 null heart.

Finally, differences were detected in the expression of several ion channel types. Previous studies on heterozygous Cx43(+/–) mice have indicated that Na+ channel activity may be increased, thereby compensating for the slowing of impulse propagation that would be expected to result from reduced Cx43 expression (29, 53). Measurements of ventricular conduction in Cx43 heterozygotes, both in situ and in culture, have generally not detected substantially slowed propagation (see Refs. 3, 53, 57), although adult myocytes with cardiac-targeted Cx43 deletion exhibit substantial conduction deficits (18). The concept of channel remodeling mentioned above (52), supported by the T1 gene expression changes detected in our array analysis, might lead to partial compensation for the loss of Cx43 that may restore conduction to normal levels in the heterozygote but be inadequate in the null heart.

Coordination analysis in wild-type hearts of expression levels of Cx43 and each of the other regulated genes indicated that a high percentage of the strong synergistic and antagonistic coordinations predicted the up- or downregulation of the genes in the Cx43 null hearts. This finding has far-ranging implications, suggesting that the changes seen after disruption of a gene may reflect the existence of interlinkage of other genes with the deleted one under normal conditions and that major changes might result from pathophysiological conditions affecting any of the linked partners.

The most obvious phenotype of the Cx43 null mouse is perinatal death due to blockage of ventricular output to the lungs (48). However, more subtle phenotypes have been detected as various physiological processes have been explored in detail, including altered growth rate and sensitivity to apoptotic stimuli (reviewed in Ref. 25), increased velocity of hippocampal spreading depression and increased locomotor behavior (52), and increased vulnerability of cardiac tissue to ischemic injury that has been observed in Cx43 null mice (30, 34). Some of these phenotypic differences between wild-type and Cx43 null mice may be completely attributable to deficiencies in intercellular communication. However, the large number of gene expression differences resulting from Cx43 deletion reported here and the lack of any obvious functional linkage between intercellular communication and some of these processes (such as migration rate or sensitivity to apoptotic stimuli) suggest that an additional component of the phenotype may result from alterations in expression of genes regulated by Cx43 expression per se in addition to the intercellular coupling that it provides.


    ACKNOWLEDGMENTS
 
We are grateful to Aldo Massimi and Dr. Kate Milova for helpful, collegial discussions regarding microarray technology and analysis.

This work was supported in part by National Institutes of Health Grants NS-42807, NS-41282, and MH-65495.


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

Address for reprint requests and other correspondence: D. A. Iacobas, 915C Kennedy Center, 1410 Pelham Parkway South, Albert Einstein College of Medicine, Bronx, NY 10461 (E-mail: diacobas{at}aecom.yu.edu).

10.1152/physiolgenomics.00229.2003.

1 The supplemental Material for this article (Supplemental Tables S1 and S2) is available online at http://physiolgenomics.physiology.org/cgi/content/full/00229.2003/DC1. Back


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Alberts B, Bray D, Lewis J, Raff M, Roberts K, and Watson JD. Molecular Biology of the Cell (3rd ed.). New York: Garland, 1994.
  2. Beardslee MA, Lerner DL, Tadros PN, Laing JG, Beyer EC, Yamada KA, Kléber AG, Schuessler RB, and Saffitz JE. Dephosphorylation and intracellular redistribution of ventricular Cx43 during electrical uncoupling induced by ischemia. Circ Res 87: 656–662, 2000.[Abstract/Free Full Text]
  3. Betsuyaku T, Kovacs A, Saffitz JE, and Yamada KA. Cardiac structure and function in young and senescent mice heterozygous for a connexin43 null mutation. J Mol Cell Cardiol 34: 175–184, 2002.[CrossRef][ISI][Medline]
  4. Beyer EC, Paul DL, and Goodenough DA. Connexin43: a protein from rat heart homologous to a gap junction protein from liver. J Cell Biol 105: 2621–2629, 1987.[Abstract]
  5. Britz-Cunningham SH, Shah MM, Zuppan CW, and Fletcher WH. Mutations of the Connexin43 gap-junction gene in patients with heart malformations and defects of laterality. N Engl J Med 332: 1323–1329, 1995.[Abstract/Free Full Text]
  6. Chen MM, Ashley EA, Deng DXF, Tsalenko A, Deng A, Tabibiazar RT, Ben-Dor A, Fenster B, Yang E, King JY, Fowler M, Robbins R, Johnson FL, Bruhn L, McDonagh T, Dargie H, Yakhini Z, Tsao PS, and Quetermous T. Novel role for the potent endogenous inotrope apelin in human cardiac dysfunction. Circulation 108: 1432–1439, 2003.[Abstract/Free Full Text]
  7. Chen Y, Park S, Li Y, Missov E, Hou M, Han X, Hall JL, Miller LW, and Bache RJ. Alterations of gene expression in failing myocardium following left ventricular assist device support. Physiol Genomics 14: 251–260, 2003.[Abstract/Free Full Text]
  8. Chen ZQ, Lefebvre D, Bai XH, Reaume A, Rossant J, and Lye SJ. Identification of two regulatory elements within the promoter region of the mouse connexin 43 gene. J Biol Chem 270: 3863–3868, 1995.[Abstract/Free Full Text]
  9. Dasgupta C, Martinez AM, Zuppan CW, Shah MM, Bailey LL, and Fletcher WH. Identification of connexin43 ({alpha}1) gap junction gene mutations in patients with hypoplastic left heart syndrome by denaturing gradient gel electrophoresis (DGGE). Mutat Res 479: 173–186, 2001.[ISI][Medline]
  10. de Carvalho AC, Masuda MO, Tanowitz HB, Wittner M, Goldenberg RC, and Spray DC. Conduction defects and arrhythmias in Chagas’ disease: possible role of gap junctions and humoral mechanisms. J Cardiovasc Electrophysiol 5: 686–698, 1994.[ISI][Medline]
  11. de Carvalho AC, Tanowitz HB, Wittner M, Dermietzel R, Roy C, Hertzberg EL, and Spray DC. Gap junction distribution is altered between cardiac myocytes infected with Trypanosoma cruzi. Circ Res 70: 733–742, 1992.[Abstract]
  12. Dermietzel R, Gao Y, Scemes E, Vieira D, Urban M, Kremer M, Bennett MVL, and Spray DC. Connexin43 null mice reveal that astrocytes express multiple connexins. Brain Res Rev 32: 45–56, 2000.[CrossRef][ISI][Medline]
  13. Draghici S. Data Analysis Tools for DNA Microarrays. Boca Raton, FL: Chapman and Hall, 2003.
  14. Dupont E, Matsushita T, Kaba RA, Vozzi C, Coppen SR, Khan N, Kaprielian R, Yacoub MH, and Severs NJ. Altered connexin expression in human congestive heart failure. J Mol Cell Cardiol 33: 359–371, 2001.[CrossRef][ISI][Medline]
  15. Ewart JL, Cohen MF, Meyer RA, Huang GY, Wessels A, Gourdie RG, Chin AJ, Park SMJ, Lazatin BO, Villabon S, and Lo CW. Heart and neural tube defects in transgenic mice overexpressing the Cx43 gap junction gene. Development 124: 1281–1292, 1997.[Abstract/Free Full Text]
  16. Gardner PI, Ursell PC, Fenoglio JJ, and Wit AL. Electrophysiologic and anatomic basis for fractionated electrograms recorded from healed myocardial infarcts. Circulation 72: 596–611, 1985.[Abstract]
  17. Gutmann DH, Hedrick NM, Hagarajan J, Li R, Perry A, and Watson MA. Comparative gene expression profile analysis of neurofibromatosis 1-associated and sporadic pilocytic astrocytomes. Cancer Res 62: 2085–2091, 2002.[Abstract/Free Full Text]
  18. Gutstein DE, Morley GE, Tamaddon H, Vaidya D, Schneider MD, Chen K, Chien KR, Stuhlmann H, and Fishman GI. Conduction slowing and sudden arrhythmic death in mice with cardiac-restricted inactivation of connexin43. Circ Res 88: 333–339, 2001.[Abstract/Free Full Text]
  19. Hall DG, Morley GE, Vaidya D, Ard M, Kimball TR, Witt SA, and Colbert MC. Early onset heart failure in transgenic mice with dilated cardiomyopathy. Pediatr Res 48: 36–42, 2000.[Abstract/Free Full Text]
  20. Huang GY, Cooper ES, Waldo K, Kirby ML, Gilula NB, and Lo CW. Gap junction-mediated cell-cell communication modulates mouse neural crest migration. J Cell Biol 143: 1725–1734, 1998.[Abstract/Free Full Text]
  21. Iacobas DA. Medical Biostatistics (3rd English ed.). Bucharest, Romania: Bucura Mond, 1997.
  22. Iacobas DA, Iacobas S, and Spray DC. Use of cDNA arrays to explore gene expression in genetically manipulated mice and cell lines. In: Practical Methods in Cardiovascular Research, edited by Dhein S, Mohr FW, and Delmar M. New York: Springer-Verlag, 2004, p. 887–895.
  23. Iacobas DA, Iacobas S, Urban-Maldonado M, and Spray DC. Sensitivity of the brain transcriptome to connexin ablation. Biochimica et Biofisica Acta, December, 2004 [ePub ahead of print: doi 10.1016/j.bbamem.2004.12.002].
  24. Iacobas DA, Scemes E, and Spray DC. Gene expression alteration in connexin43 null astrocytes extended beyond the gap junction. Neurochem Int 45: 243–250, 2004.[CrossRef][ISI][Medline]
  25. Iacobas DA, Urban M, Iacobas S, Scemes E, and Spray DC. Array analysis of gene expression in connexin43 null astrocytes. Physiol Genomics 15: 177–190, 2003.[Abstract/Free Full Text]
  26. Iacobas DA, Urban M, Iacobas S, and Spray DC. Control and variability of gene expression in mouse brain and in a neuroblastoma cell line. Rom J Physiol 39/40: 53–67, 2003.
  27. Iacobas DA, Urban M, Massimi A, Iacobas S, and Spray DC. Hits and misses from gene expression ratio measurements in cDNA microarray studies. J Biomol Tech 13: 143–157, 2002.[Abstract/Free Full Text]
  28. Iacobas DA, Urban M, Massimi A, and Spray DC. Improved methods to mine data obtained from spotted cDNA arrays. J Biomol Tech 13: 5–19, 2002.[Abstract/Free Full Text]
  29. Johnson CM, Green KG, Kante EM, Bou-Abboud E, Saffitz JE, and Yamada KA. Voltage-gated Na+ channel activity and connexin expression in Cx43-deficient cardiac myocytes. J Cardiovasc Electrophysiol 10: 1390–1401, 1999.[ISI][Medline]
  30. Kanno S, Kovacs A, Yamada KA, and Saffitz JE. Connexin43 as a determinant of myocardial infarct size following coronary occlusion in mice. J Am Coll Cardiol 41: 681–686, 2003.[CrossRef][ISI][Medline]
  31. Kanno S and Saffitz JE. The role of myocardial gap junctions in electrical conduction and arrhythmogenesis. Cardiovasc Pathol 10: 169–177, 2001.[CrossRef][ISI][Medline]
  32. Kaprielian RR, Gunning M, Dupont E, Sheppard MN, Rothery SM, Underwood R, Pennell DJ, Fox K, Pepper J, Poole-Wilson PA, and Severs NJ. Downregulation of immunodetectable connexin43 and decreased gap junction size in the pathogenesis of chronic hibernation in the human left ventricle. Circulation 97: 651–660, 1998.[Abstract/Free Full Text]
  33. Kasahara H, Ueyama T, Wakimoto H, Liu MK, Maguire CT, Converso KL, Kang PM, Manning WJ, Lawitts J, Paul DL, Berul CI, and Izumo S. Nkx2.5 homeoprotein regulates expression of gap junction protein connexin 43 and sarcomere organization in postnatal cardiomyocytes. J Mol Cell Cardiol 35: 243–56, 2003.[CrossRef][ISI][Medline]
  34. Lerner DL, Yamada KA, Schuessler RB, and Saffitz JE,. Accelerated onset and increased incidence of ventricular arrhythmias induced by ischemia in Cx43-deficient mice. Circulation 101: 547–552, 2000.[Abstract/Free Full Text]
  35. Li WEI and Nagy JL. Connexin43 phsophroylation and intercellular communication in cultured astrocytes following hypoxia and protein phosphatase inhibition. Eur J Neurosci 12: 2644–2650, 2000.[CrossRef][ISI][Medline]
  36. Li WEI, Waldo K, Linask KL, Chen T, Wessels A, Parmacek MS, Kirby ML, and Lo CW. An essential role for connexin43 gap junctions in mouse coronary development. Development 129: 2031–2042, 2002.[ISI][Medline]
  37. Liu XZ, Xia XJ, Adams J, Chen ZY, Welch KO, Tekin M, Ouyang XM, Kiristiansen A, Pandya A, Balkany T, Arnos KS, and Nance WE. Mutations in GJA1 (connexin43) are associated with non-syndromic autosomal recessive deafness. Hum Mol Genet 10: 2945–2951, 2001.[Abstract/Free Full Text]
  38. Lo CW. Role of gap junctions in cardiac conduction and development. Insights from the connexin knockout mice. Circ Res 87: 346–348, 2000.[Free Full Text]
  39. Lock C, Hermans G, Pedotti R, Brendolan A, Schadt E, Garren H, Langer-Gould A, Strober S, Cannella B, Allard J, Klonowski P, Austin A, Lad N, Kaminski N, Gall SJ, Oksenberg JR, Raine CS, Heller R, and Steinman L. Gene-microarray analysis of multiple sclerosis lesions yields new targets validated in autoimmune encephalomyelitis. Nat Med 8: 500–508, 2002.[CrossRef][ISI][Medline]
  40. Loring JF, Wen X, Lee JM, Seilhamer J, and Somogyi R. A gene expression profile of Alzheimer’s disease. DNA Cell Biol 20: 683–695, 2001.[CrossRef][ISI][Medline]
  41. Luke RA and Saffitz JE. Remodeling of ventricular conduction pathways in healed canine infarct border zones. J Clin Invest 87: 1594–1602, 1991.[ISI][Medline]
  42. Okruhlicova L, Tribulova N, Misejkova M, Kucka M, Stetka R, Slezak J, and Manoach M. Gap junction remodelling is involved in the susceptibility of diabetic rats to hypokalemia-induced ventricular fibrillation. Acta Histochem 104: 387–391, 2002.[ISI][Medline]
  43. Paznekas WA, Boyadjiev SA, Shapiro RE, Daniels O, Wollnik B, Keegan CE, Innis JW, Dinulos MB, Christian C, Hannibal MC, and Jabs EW. Connexin 43 (GJA1) mutations cause the pleiotropic phenotype of oculodentodigital dysplasia. Am J Hum Genet 72: 408–418, 2003.[CrossRef][ISI][Medline]
  44. Peters NS, Coromilas J, Severs NJ, and Wit AL. Disturbed connexin43 gap junction distribution correlates with the location of reentrant circuits in the epicardial border zone of healing canine infarcts that cause ventricular tachycardia. Circulation 95: 988–996, 1997.[Abstract/Free Full Text]
  45. Peters NS, Green CR, Poole-Wilson PA, and Severs NJ. Reduced content of connexin43 gap junctions in ventricular myocardium from hypertrophied and ischemic human hearts. Circulation 88: 864–875, 1993.[Abstract]
  46. Pfaffl MW, Horgan GW, and Dempfle L. Relative expression software tool (REST) for group-wise comparison and statistical analysis of relative expression results in real-time PCR. Nucleic Acids Res 30: e36, 2002.[Abstract/Free Full Text]
  47. Rasmussen R. Quantification on the LightCycler. In: Rapid Cycle Real-Time PCR: Methods and Applications, edited by Meuer S, Wittwer C, and Nakagawara K. Heidelberg, Germany: Springer, 2001, p. 21–34.
  48. Reaume AG, de Sousa AP, Kulkarni S, Langille BL, Zhu D, Davies TC, Juneja SC, Kidder GM, and Rossant J. Cardiac malformation in neonatal mice lacking connexin43. Science 267: 1831–1834, 1995.[ISI][Medline]
  49. Schubert AL, Schubert W, Spray DC, and Lisanti MP. Connexin family members target to lipid raft domains and interact with caveolin-1. Biochemistry 41: 5754–5764, 2002.[CrossRef][ISI][Medline]
  50. Spray DC, Suadicani S, Srinivas M, Gutstein DE, and Fishman GI. Gap junctions in the cardiovascular system. In: Handbook of Physiology. The Cardiovascular System. The Heart. Bethesda, MD: Am Physiol Soc, 2001, sect. 2, vol. I, p. 169–212.
  51. Stekel D. Microarray Bioinformatics. Cambridge, MA: Cambridge Univ. Press, 2003.
  52. Theis M, Jauch R, Zhuo L, Speidel D, Wallraff A, Doring B, Frisch C, Sohl G, Teubner B, Euwens C, Huston J, Steinhauser C, Messing A, Heinemann U, and Willecke K. Accelerated hippocampal spreading depression and enhanced locomotory activity in mice with astrocyte-directed inactivation of connexin43. J Neurosci 23: 766–776, 2003.[ISI][Medline]
  53. Thomas SP, Kucera JP, Bircher-Lehmann L, Rudy Y, Saffitz JE, and Kleber AG. Impulse propagation in synthetic strands of neonatal cardiac myocytes with genetically reduced levels of connexin43. Circ Res 92: 1209–1216, 2003.[Abstract/Free Full Text]
  54. Toth T, Hajdu J, Marton T, Nagy B, and Papp Z. Connexin43 gene mutations and heterotaxy. Circulation 97: 117–118, 1998.[Free Full Text]
  55. Ursell PC, Gardner PI, Albala A, Fenoglio JJ, and Wit AL. Structural and electrophysiological changes in the epicardial border zone of canine myocardial infarcts during infarct healing. Circ Res 56: 436–451, 1985.[Abstract]
  56. Vis JC, Nicholson LF, Faull RL, Evans WH, Severs NJ, and Green CR. Connexin expression in Huntington’s diseased human brain. Cell Biol Int 22: 837–847, 1998.[CrossRef][ISI][Medline]
  57. Yao JA, Gutstein DE, Liu F, Fishman GI, and Wit AL. Cell coupling between ventricular myocyte pairs from connexin43-deficient murine hearts. Circ Res 93: 736–743, 2003.[Abstract/Free Full Text]
  58. Zoidl G, Bruzzone R, Weickert S, Kremer M, Zoidl C, Mitropoulou G, Srinivas M, Spray DC, and Dermietzel R. Molecular cloning and functional expression of zfCx52.6: a novel connexin with hemichannel-forming properties expressed in horizontal cells of the zebrafish retina. J Biol Chem 279: 2913–2921.