1 Department of Pathology, Anatomy and Cell Biology, Thomas Jefferson University, Philadelphia, Pennsylvania
2 Department of Electrical Engineering, University of Delaware, Newark, Delaware
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
microarray study of gene expression; functional annotation analysis; quantitative RT-PCR validation; disturbance of homeostasis
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Two salient features of the alcohol-adapted state phenotype are 1) an acquired tolerance to the effects of ethanol on behavioral parameters and 2) a physical dependence on the presence of ethanol for normal function. The latter aspect of the addicted phenotype is responsible for the illness observed when alcohol is suddenly removed from the diet, called the alcohol withdrawal syndrome (AWS). Littleton (62), in his review, notes that C. K. Himmelsbach suggested that homeostatic mechanisms are responsible for drug tolerance and withdrawal symptoms, and that this theory provides the framework for a majority of neurochemical investigations of alcohol adaptation and withdrawal. The vulnerability of homeostatic mechanisms occurs at the molecular and cellular level (53) as well as the behavioral and physiological level. For example, alcohol withdrawal can lead to seizures and disturbances of cardiorespiratory function that are potentially fatal (69). The effects of withdrawal may contribute to the self-perpetuating aspects of alcohol abuse, as individuals may self medicate the adverse physiological consequences with more alcohol and thereby seal the addiction cycle. The goal in the present paper was to evaluate regulation in the expression of genes relevant to neuronal function that could help us characterize the chronic alcohol-adapted state and point toward mechanisms that could explain the vulnerability to dysfunction caused by alcohol withdrawal.
Chronic ethanol consumption in human alcoholics and rodent models changes the activities of enzymes involved in widespread cellular functions, such as ethanol metabolism [notably, cytochrome P-450, family 2, subfamily E, polypeptide 1 (CYP2E1), and alcohol dehydrogenase (ADH)], mitochondrial energy conservation, protein synthesis, and lipid metabolism (60). Importantly, chronic alcohol treatment also results in significant changes in intracellular signaling pathways (1618, 27, 33, 44, 48, 49, 59, 68, 71, 76, 77, 109). To gain insight into the modulation of neuronal processes, expression profiling with DNA microarrays has been used to identify alcohol-responsive genes in the brain (59, 68). The results show that, as mentioned above, functionally diverse sets of genes are altered. Notably, however, only modest changes in expression level, almost entirely less than twofold, were observed, and relatively modest numbers of genes were affected, 2% of the total number of genes studied (32). Patterns of differential gene expression in response to chronic alcohol exposure vary radically across brain regions (33, 55) and rodent strain (24, 55).
The application of microarray technologies to physiological brain function presents special challenges. For example, the heterogeneity of neuronal cell types means that it is crucial to sample from functionally specific nuclei or neuronal subpopulations. In the present case, we have met this requirement by sampling the NTS, a small anatomical region of specific function and interest. Furthermore, in the brain, we do not expect dramatic changes in gene expression to be observed in the regulation of neural function, but rather previous studies have shown that more subtle changes are expected in genes that are of primary interest (112, 116). Specifically, we anticipate that the transcriptional response to chronic alcohol in the NTS involves genes that 1) may be expressed at low levels and 2) have moderate, physiologically relevant changes in expression that are typically less than twofold. For these reasons, we placed great emphasis on the development of experimental and analytical microarray techniques that support acquisition of high-quality data at this level of quantitative precision (discussed in EXPERIMENTAL PROCEDURES). These issues have motivated our use of methods that allow for a common reference design with reliable normalization (described below) and experimental design to support sensitive statistically driven methods for data analysis.
The NTS is the focus of the present study because of its role as an integrative center that regulates and coordinates homeostatic function. It is the visceral afferent nucleus for organs that process and are affected by alcohol, e.g., the stomach, liver, heart, lungs (94). Within the brain it participates broadly in the regulation of autonomic and endocrine homeostatic coordination of processes disturbed by alcohol withdrawal, such as cardiorespiratory reflex regulation (82, 8688, 91, 92), the renin-angiotensin-aldosterone system, the vasopressin system, the cortisol system, and sodium sensitivity (54). It is richly interconnected with the autonomic forebrain and is associated with the interaction of anxiety and stress with visceral states. In addition, the NTS has been associated with substance withdrawal (97, 106). Acute administration of ethanol (intraperitoneal injections of 2 g/kg) causes the induction of the transcription factor c-Fos in the NTS (104). Alcohol consumption attenuates baroreflex function through effects on the NTS (111). Alcohol withdrawal elicits cardiovascular symptoms such as elevations of blood pressure and heart rate (54), arrhythmias (including sudden cardiac death), and disruptions of respiratory pattern that frequently can be fatal (31). The NTS may contribute to these dysfunctions, given its prominent involvement in cardiorespiratory regulation.
We discovered widespread gene expression alterations in the NTS, showing that the alcohol-adapted state is significantly at variance with the control state. The adapted state is characterized not by large changes in a few genes but rather by patterns of gene regulation in a range that would be expected to reflect physiological alterations, not pathological distortions. By testing for associations between our list of differentially expressed genes and Gene Ontology (GO) functional class assignments and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, we identified several gene expression patterns that provide hypotheses about the functional systems, networks, or mechanisms disturbed in the NTS by adaptation to chronic alcohol consumption.
![]() |
EXPERIMENTAL PROCEDURES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
NTS Identification, Microdissection, RNA Extraction
Animals were killed by decapitation according to a protocol approved by the Thomas Jefferson University Institutional Animal Care and Use Committees, and tissue was immediately collected from two littermate pairs at a time on three separate days: each pair was an alcohol-adapted animal and a diet-matched non-alcohol-treated control. Brainstems were excised, and blocks containing the NTS were placed in ice-cold artificial cerebral spinal fluid (ACSF; 10 mM HEPES, pH 7.4, 140 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1 mM CaCl2, 24 mM D-glucose). The blocks were cut into 250-µm coronal sections using a McIlwain Tissue Chopper (Campden Instruments, Lafayette, IN). Sections were floated in ice-cold ACSF, and those containing the NTS were selected. The NTS was identified based on anatomical landmarks and was punched out using a 750-µm diameter micropunch (Stoelting, Chicago, IL). To preserve quality of RNA, the NTS was obtained within 10 min postmortem. Total RNA was extracted using Qiagens RNeasy Mini Kit (Qiagen, Valencia, CA), yielding 200900 ng of total RNA. RNA quality was assessed using an RT-PCR protocol for high- and low-copy-number genes (ß-actin and tyrosine hydroxylase, respectively). Tyrosine hydroxylase was selected because it is specific to the NTS at the slice level, confirming that the punches contained NTS.
Microarray Manufacture, RNA Amplification, Labeling, Hybridization
Microarrays were fabricated using a rat clone set (GF200; ResGen, Huntsville, AL) for cDNA microarrays consisting of 17,000 sequence-verified nonredundant clones (as of UniGene build 78). The cDNA clones from all rat cDNA targets were prepared from freshly grown overnight bacterial cultures by PCR amplification using GF200 primers (Invitrogen, Carlsbad, CA). PCR products were purified and verified by agarose gel electrophoresis, and the yield was determined spectrophotometrically (NanoDrop, Wilmington, DE). cDNAs were mixed with an equal volume of DMSO (1070 ng/µl) and printed onto FMB cDNA slides (Full Moon Biosystems, Sunnyvale, CA) using a MicroGrid II arrayer (Genomic Solutions, Ann Arbor, MI). Microarrays were air-dried for 30 min and cross-linked by UV irradiation. We printed arrays of 18,240 spots representing
9,000 clones in adjacent duplicate spots. The arrays used in the present study had
5,300 annotated (LocusLink ID) genes.
Total RNA (70400 ng) was amplified using two rounds of the antisense RNA (aRNA) technique (107), yielding on average 180 µg of aRNA (MessageAmp; Ambion, Austin, TX). aRNA (1.125 µg) was reverse transcribed (Superscript II, Invitrogen) using random primers to generate single-stranded cyanine-5 (Cy5)-amino-allyl-labeled cDNA from the experimental samples.
A DNA fragment (220 bases) from the vector pT3T7PAC was amplified by PCR with GF200 primers and converted to RNA by in vitro transcription (MEGAscript, Ambion). We refer to this RNA as vRNA. To generate Cy3-amino-allyl-labeled cDNA, 2.5 µg of vRNA were reverse transcribed as above. Because the PCR product from each clone contains the small portion of vector sequence, this Cy3-vDNA hybridizes to every clone PCR product on the microarrays. A dilution series was performed, and a nonsaturating amount of labeled vcDNA was used as a normalizing reference for the microarrays in which the experimental samples are always labeled with Cy5. We refer to this procedure as the vRNA reference approach. It is useful for spot finding and quality control for absent or weak spots, but in particular provides a reliable reference for greatly improved quantitative precision in cross-slide measurements used in the present reference design. We have extensively validated this method (data not shown).
Microarrays were prehybridized in 1% bovine serum albumin, 5x SSC, and 0.1% SDS for 45 min at 42°C; washed in H2O; and dried by centrifugation. Cy3-vDNA and Cy5-cDNA samples were mixed with 50 µl of DIG Easy Hybridization Buffer (Roche, Indianapolis, IN) containing 25 µg each of yeast tRNA and calf thymus DNA and applied to the microarrays for hybridization at 37°C for 16 h in a hybridization chamber (Corning, Corning, NY) in the dark with gentle agitation. Slides were washed for 10 min at 50°C in 1x SSC and 0.1% SDS in shaking incubator, followed by a 1-min wash in 1 x SSC, three 1-min washes in 0.1x SSC, and one rinse in H2O, at room temperature. Slides were dried by centrifugation and scanned with the ScanArray 5000XL (PerkinElmer, Wellesley, MA). Image analysis was performed using ScanArray Express v2.2 software.
Experimental Design, Replicates, and Microarray Data Analysis
We are acutely aware of the noise issues affecting microarray data (66) and thus employed the vector reference approach we have developed (see above), which supports acquisition of quantitative and reliable data. To ensure high-quality arrays, we developed a quality control (QC) protocol involving 1) PCR QC of a representative sample of cDNA by gel electrophoresis and digital image analysis and 2) spot QC by uniformity/morphology analysis using the vector reference. Hybridized microarrays were evaluated for QC, and poor arrays were discarded; arrays were processed in several steps: 1) quantitation: ScanArray Express v2.2 using the Adaptive Threshold segmentation; 2) filtering: spots tainted by artifacts (dust, high background, poor hybridization) were manually flagged and ignored; 3) normalization: MA plots to check for systematic bias, data were normalized by median log ratio (scale normalization) and median log ratio print tip normalization. Students t-test P values were calculated for the mean log ratios (alcohol adapted vs. paired control) of each gene. Five hundred seventy-five genes were classified as differentially expressed [P < 0.009, false discovery rate (FDR) < 0.12; Ref. 10]. All microarray data are deposited in a minimum information about a microarray experiment (MIAME)-compliant format to the Gene Expression Omnibus (GEO) with accession number GSE2718.
We used six paired RNA samples to increase the power of the t-test by increasing the signal-to-noise ratio of the sample mean. Supplemental Fig. S1 shows an MA plot of a representative pair. The levels of technical variability and sample variability (including punch variability) in our system require a data set of at least six replicate pairs of experimental animals to detect subtle modulatory changes in gene expression with statistical confidence. For example, given a standard deviation of 0.2 log2 fold change (average estimate for our system; R. L. Khan, Supplemental Fig. S2) and a significance level of 0.05, the use of three paired samples only allows detection of differential expression at 0.5 log fold change (df = 2, critical t value = 4.3) or 41%, whereas use of six paired samples allows detection of differential expression at 0.21 log fold change (df = 5, critical t value = 2.57) or 16% fold change. While a 17% change is the average detectable level in the present study, specific gene expression changes of greater or less magnitude were reliably detected depending on each individual gene sample variability.
Functional Enrichment Analysis
All genes on the microarray were annotated with membership to each GO term or KEGG pathway (we are aware of the limitations arising from the extent of present annotation and organization of these systems). Clones were annotated by relating clone identifiers to LocusLink identifiers using UniGene and relating LocusLink identifiers to GO terms and KEGG pathways using Entrez Gene. The parents of the GO terms were found using the GO database (http://www.godatabase.org) and annotated as containing the superset of all genes contained in their children. Up- and downregulated gene groups were tested for enrichment of GO terms and KEGG pathways by means of Fisher exact test, using all of the genes on the microarray as the reference population. This method is similar to that used in expression analysis systematic explorer (EASE; Ref. 51), except we allow each spot to be annotated with multiple LocusLink identifiers, thus enabling annotations from human, mouse, and rat through HomoloGene. Mouse and human genes have more annotations for GO terms and KEGG pathways than their rat counterparts. HomoloGene allows us to add much richer annotation to our clones than available from the rat databases alone. Those GO terms or KEGG pathways that are associated with the clone and its orthologs in multiple species are only counted once to ensure correct counts in the enrichment analysis.
For GO terms, P values from Fisher exact test were adjusted using the FDR method (10), using an FDR cutoff of 0.2 for significant enrichment. For KEGG pathways gene annotation was sparser, and therefore the P values were higher. We have chosen to further analyze the pathways with P value < 0.05.
Quantitative Real-Time RT-PCR Validation
To verify some of the results from the microarray data, we used quantitative real-time RT-PCR (qRT-PCR), using aRNA from four of the six matched pairs of animals. We used four replicate pairs to optimize the 96-well-plate format of the ABI Prism 7000 (Applied Biosystems, Foster City, CA). We used SYBR Green I as a fluorescent reporter (SybrGreen I Mastermix, Applied Biosystems), taking triplicate measures to determine the threshold cycle (CT) and employing the CT method to detect differential gene expression (63). For pairwise comparisons, each aRNA sample was converted to cDNA using RT (SuperScript II, Invitrogen). Aliquots (6.25 ng) of each cDNA population were used as template for replicate PCRs (n = 3) for primers corresponding to both the gene of interest and the housekeeping gene NADH-ubiquinone oxidoreductase 13 kDa-B subunit (NDUFA5). This gene was chosen because it was highly expressed but not differentially expressed in the microarray results, in contrast, for example, to other genes involved in oxidative phosphorylation (including one encoding another subunit of this complex) discussed below. CT for each reaction was determined using the ABI Prism 7000 SDS software. For each sample (control or ethanol), CT is normalized to the Ndufa5 CT (CTc CTNdufa5 = CTc1, or CTe CTNdufa5 =
CTe1). Next, these normalized CTs were subtracted to quantify the differential expression (
CTc1
CTe1 =
CT1) and significance (one-sample Students t-test, P value < 0.05). Finally, fold change was estimated from the
CT (assuming efficiency of two, 2
CT = fold change).
![]() |
RESULTS AND DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
Table 1 shows the top 30 genes in order of decreasing P value that either increase or decrease expression in response to chronic alcohol consumption, and there are clearly genes on this list that are related to neuronal functions and alcohol metabolism. However, the grouping of genes by functional classes and pathways can, if extended to the full list of all significant genes in the NTS data set, be productive and provide functional insight beyond that possible looking at individual genes. To this end, we used a bioinformatics/computational analysis approach as described above in EXPERIMENTAL PROCEDURES: we performed Fisher exact tests for enrichment of GO terms and KEGG pathways within groups of regulated genes. By using these tools and approaches, we can more efficiently manage the size of the significant gene list and more reliably and objectively derive hypotheses as to NTS physiological processes affected by chronic alcohol consumption.
|
|
|
Several of the other genes in these GO clusters have not been reported previously as responding to alcohol exposure but may nevertheless significantly affect NTS neuronal function and, in particular, changes in plasticity: neuropilin 1 (NRP1), apolipoprotein E (APOE), and CXC chemokine ligand 12 (CXCL12). Neuropilin/semaphorin regulation is prominent in status epilepticus (reinforcing the seizure-prone NCDN-related function above), axon sprouting and synaptic reorganization (7, 50), and in seizures and in the formation and function of neural circuits (89). Apolipoprotein E is a glycoprotein commonly associated with Alzheimers but related to neuronal plasticity due to its regulation of axon extension, synaptic plasticity, synaptic transmission, and cholinergic function and may have a role in the rewarding properties of alcohol in mice (9). CXCL12 is a G protein-coupled agonist highly expressed in brain (41) and has been described as having highly discrete neuroanatomical localization and to be colocalized in dopaminergic and cholinergic neurons (6). Because both of these neuron types are present in NTS, the present results suggest that the effects of chronic alcohol consumption on CXCL12 may also cause dopamine and acetylcholine neuronal function to be altered in this nucleus. Disturbances of cholinergic/dopaminergic function would contribute greatly to the vulnerability of NTS functions in alcohol withdrawal.
Intracellular signaling.
Two other genes in this cluster may influence synaptic plasticity via effects on intracellular signaling. YWHAG is new member of the 14-3-3 family of proteins, which regulate signal transduction pathways by binding to phosphoserine-containing proteins. Among the reactions affected by this member of the 14-3-3 protein family are v-raf-1 murine leukemia viral oncogene homolog 1 (RAF1) and protein kinase C zeta (PRKCZ), which would potentially affect the regulation of a wide variety of signal transduction pathways (5). Protein tyrosine kinase 2 (PTK2) encodes a cytoplasmic protein tyrosine kinase member of the FAK subfamily with function in intracellular signal transduction pathways triggered in response to certain neural peptides or to cell interactions with the extracellular matrix. Although downregulated here, in a previous report (35), it was upregulated in frontal cortex and hippocampus after chronic cocaine exposure. PTK2 may also interact with mitogen-activated protein kinase (MAPK) signaling pathways discussed below and illustrated in Fig. 3.
|
|
Cation transport, endocytosis/exocytosis.
Cation transport, in particular potassium and sodium/potassium ion transport (ATP1A2, ATP1B1, ATP1B2, ATP1B3, ATP2B1) (see Fig. 2 GO term "metal ion transport"), is highly enriched in the upregulated genes. Previous studies on brain (4, 19, 84, 99) reported that the mRNA level and the activity of the Na+-K+-ATPase was enhanced by chronic ethanol consumption, presumably reflecting a response to a sustained acute suppression of the activity of this protein by ethanol (98, 100). In addition, genes involved in calcium transport are affected (e.g., ATP2B1), which may be relevant for calcium homeostasis and calcium signaling, as discussed in the KEGG pathway analysis below. Several genes important in calcium-dependent regulation of endocytosis and exocytosis appear to be upregulated. ANXA6, a calcium binding annexin VI has been implicated in mediating endosome aggregation and vesicle fusion in secreting epithelia during exocytosis (79). It is colocalized with Rab protein 5 (Rab5A) (79), which interacts with Rab GTPase binding effector protein 2 (RABEP2) (37) to facilitate endocytic membrane fusion. Calcium binding protein p22 (CHP) is necessary for exocytosis in a calcium-dependent manner (8) and plays a role in microtubule and endoplasmic reticulum organization (2). The protein encoded by this gene is a phosphoprotein that binds to the sodium-proton exchange proteins (NHEs). CHP serves as an essential cofactor for NHE family members where it is important for intracellular pH (pHi)-dependent regulation of Na+/H+-exchanger 1 (NHE1) via tightly bound Ca2+ ions that serve as a crucial structural elements required for this role. The protein has sequence similarity to calcineurin B, and it is also known to be an endogenous inhibitor of calcineurin activity. Another interesting observation is the very large upregulation of the Na+-K+-Cl cotransporter gene NKCC1 (SLC12A2). This activity regulates chloride permeability and thereby affects GABAergic function and neuronal excitability and may also contribute to volume control (15). To summarize, the upregulated genes seem to converge on several interrelated processes: neuronal excitability, intracellular pH and volume control, calcium signaling and calcium homeostasis, vesicle-mediated transport, and cytoskeleton organization.
Cell cycle.
Because the brain is considered to be an end-differentiated organ consisting of postmitotic fixed neurons, we were surprised to find the mitotic cell cycle GO terms and gene clusters (see Fig. 2 GO terms). However, there is literature showing that cat and rat neurons are capable of reentering the cell cycle (1, 39). We note this with interest, as ethanol affects cell cycle regulation in other organs: acutely, it can inhibit DNA synthesis (20, 29, 110) and prolong the G1 phase of the cell cycle, including basic leucine zipper and W2 domains 1 protein (BZW1), branched chain aminotransferase 1 cytosolic protein (BCAT1), integrin beta 1 (ITGB1), phosphatase 3 catalytic subunit alpha (PPP3CA), and calcium/calmodulin-dependent protein kinase II gamma (CAMK2G), resulting in a reduction in cell proliferation and regeneration (45, 46, 110). The upregulated genes that cluster in cell cycle GO terms include mostly those involved in G1-phase progression without evidence of S-phase and G2/M-phase markers being enriched. Interphase was also significantly enriched for upregulated genes (in particular, D123 protein). GO terms relevant to M phase specifically were not enriched, indicating that cell division is not occurring and suggesting the biological functional categories in this case are misleading, i.e., that the regulation of these genes is not relevant to cell-cycle, but that these genes may have other roles in the NTS/brain.
Statistically Significantly Enriched KEGG Pathway Results
We used the KEGG pathway resource to test for enrichment of specific pathway components in the functionally regulated gene groups (see Supplemental Table S3) in addition to annotating the pathway maps as shown in Figs. 3 and 4. Nine KEGG pathways were significantly enriched in the upregulated gene list (Table 3), and two were significantly enriched in the downregulated gene list. Pathways with two or fewer regulated genes were not discussed further, leaving seven KEGG pathways significantly enriched in the upregulated gene list and none in the downregulated list. The results are illustrated in Fig. 2. All of the significant KEGG results were related to metabolism, which is consistent with the present bias in the KEGG pathway set toward metabolism and can lead to false negatives for signaling pathways. Because our focus in the present study is on neuronal function, we will only briefly discuss the most interesting of the metabolic pathways. We note that there are technical limits involving the complexity and redundancy of assignment of genes to functional elements in signaling maps that make Fisher exact test P values less useful for KEGG signaling pathways. We identified two signaling pathways that contained the largest number of regulated genes: the interrelated calcium and MAPK signaling pathways. These had relatively low-scoring P values (0.1 and 0.2, respectively), were neurobiologically interesting, and are discussed below.
|
Apart from the genes encoding alcohol-metabolizing enzymes, glycerolipid metabolism was highly enriched in the upregulated genes, including phosphatidic acid phosphatase type 2B (PPAP2B), platelet-activating factor acetylhydrolase (PAFAH1B1; also validated by qRT-PCR analysis), lipase (LIPG), ß-1,4-mannosyltransferase (ALG1), glyceronephosphate O-acyltransferase (GNPAT), and NADP-dependent steroid dehydrogenase like (NSDHL). It has long been known that alcohol affects glycerolipid metabolism in the liver (3, 70) and other tissues. PPAP2B has been implicated with fatty liver as a result of ethanol consumption in rats (95) and humans (26, 30). PAFAH1B1 was significantly increased in mice in maternal plasma as a result of alcohol consumption (90). PAFAH is typically found in blood plasma. LIPG has been shown to be expressed in the brain by in situ hybridization studies (80). The fact that we detected increased expression of these genes in our studies may suggest that lipid-related cell signaling processes or other phospholipid-dependent processes are affected in the NTS.
Electron transport and oxidative stress.
Although there is substantial literature on alterations in mitochondrial electron transport and oxidative stress associated with chronic ethanol treatment (42, 78, 85), our studies detected only a significant upregulation of two nuclear-encoded subunits of the NADH ubiquinone oxidoreductase (complex I) (NDUFS3, NDUFA10). In addition, COXIV (COX4I1), a nuclear-encoded subunit of cytochrome oxidase (complex IV) was upregulated. Other studies have demonstrated that chronic ethanol treatment causes a selective decrease in the expression of the mitochondrially encoded subunits of this enzyme complex (23). It is conceivable that a compensatory upregulation of some of the other subunits occurs as a response of the cell to an inadequate assembly of the complex. Surprisingly, despite the literature evidence that chronic ethanol consumption is associated with an enhanced oxidative stress, our studies detected significant changes in the expression of only a few of the oxidative stress response genes in the NTS: glutathione peroxidase, thioredoxin reductase 2, and glutathione S-transferase isoforms. Furthermore, the JNK and p38 MAPK genes, which are important cell signaling pathways that mediate the stress responses, were upregulated (see MAPK signaling).
Transcriptional regulation.
General transcription factor IIB (GTF2B) and three TATA binding protein-associated factors (TAF4B, TAF9L, and TAF13) were all upregulated in the alcohol-adapted state. The alcohol dehydrogenases are known to have TATA boxes and are regulated by TATA binding proteins and CCAAT/enhancer binding protein (C/EBP) (12, 96). However, further studies of transcription regulatory element enrichment in promoter regions of regulated genes will be necessary to assess transcriptional regulation (105).
MAPK signaling.
Figure 3 shows genes in the MAPK KEGG pathways that were significantly regulated (light gray) and potentially regulated genes that did not reach significance (P < 0.05). Both the classic MAPK pathway (ERK pathway) and the stress-activated JNK and p38 MAPK pathways contained a number of regulated genes. Some of these may be evidence of oxidative stress-cell stress response induced by alcohol. All of these pathways have been implicated in response to alcohol-induced oxidative stress in the liver (74). ERK and p38 are activated in neurons under oxidative stress (115). Activation of these pathways may also be due to an inflammatory response (75), which may explain the regulation of interleukin-1ß (IL1B) and interleukin-1 receptor, type II (IL1R2), and various other cell surface receptors (CD24, CD9). Chronic ethanol consumption is associated with increased levels of circulating endotoxins and proinflammatory cytokines that affect liver function (47), notably tumor necrosis factor (TNF)-. Interestingly, a recent preliminary report (21) suggests that the endotoxin-induced elevation of TNF-
levels in the circulation may penetrate the blood-brain barrier and sustain elevations of cytokines in the CNS. This may also explain why MAPK pathways are elevated in chronic alcohol exposure.
Calcium signaling.
A large number of upregulated genes represent calcium signaling and calcium homeostasis pathways (Fig. 4). There is considerable evidence that ethanol inhibits NMDA (ROC in Fig. 4) receptor function (22). Furthermore, alcohol increases NMDA sensitivity (11). NMDA receptor activation enhances calcium influx and promotes excitotoxicity (36). The cells may respond by the regulation of genes involved in calcium homeostasis (Fig. 4). In addition, upregulated genes include several G protein-coupled receptor (GPCR) systems (e.g., endothelin receptor, shaded in Fig. 4) and receptor tyrosine kinases (e.g., platelet-derived growth factor receptor, PTK in Fig. 4) that activate phospholipase C and initiate inositol 1,4,5-trisphosphate-dependent Ca2+ signaling pathways which are affected by ethanol exposure (34, 64). There is evidence that ethanol inhibits calcium reuptake into the endoplasmic reticulum (ER) through the sarco(endo)plasmic reticulum Ca2+-ATPase (SERCA) pump (ATP2A1) (72, 101) and enhances ER Ca2+ release through nonspecific leak pathways. Alcohol withdrawal induces cytotoxicity in neuron cultures (73). A dysregulation of cellular calcium homeostasis may contribute to this response.
Central cardiorespiratory control.
The molecular and genetic basis of central cardiorespiratory control is largely unknown, but both the calcium and MAPK pathways affected by chronic alcohol in the present study are involved. The NTS plays a key role in baroreflex function and sensitivity, central respiratory drive, and response to acute hypoxia. The baroreflex beat-to-beat regulation of blood pressure is inhibited by c-Fos activity and, by implication, its downstream cascade of intracellular events in NTS (14). Likewise, c-Fos activity in the NTS is associated with respiratory response to hypercapnic challenge (102, 103, 113). Alcohol elicits dose-dependent attenuation of baroreflex sensitivity associated with significant increases in c-jun mRNA in the NTS of normotensive rats, showing the NTS to be a target for alcohol action on baroreflexes (111). The alcohol-induced perturbation of JNK pathway signaling observed in the present study (Fig. 3) would lead to subsequent AP-1 (c-Fos/c-Jun heterodimer or c-Jun homodimer)-dependent transcriptional regulation in the NTS that is associated with and is very likely to lead to dysregulation of ANG II-dependent processes, such as the baroreflex and blood pressure regulation, both established targets of alcohol abuse. These results suggest the hypothesis that there are changes in the expression of c-fos and c-jun as well (neither of these were included in the array used in the present study), and this should be tested in subsequent studies. ANG II acting on ANG II type 1 (AT1) receptors in the NTS depresses the baroreflex (81), and it is likely the present RGS2 regulation also affects ANG II functions (see Intracellular signaling). As a result, the calcium signaling network (Fig. 4) is also involved: CAMKII mediates critical components of the hypoxic ventilatory response within the NTS (83) and contributes to ANG II-mediated depression of baroreflex function (114). The present results begin to provide data on the effects of chronic alcohol consumption on the molecular and genetic mechanisms involved in central cardiorespiratory regulation by the NTS.
Cumulatively, the above demonstrates that the NTS responds to chronic alcohol consumption with widespread alteration of gene expression levels for genes that participate in functional pathways centrally involved in neuronal behavior and response to inputs. These functional relationships suggest the hypothesis that these alterations in expression may be associated with the homeostatic dysregulation of AWS. The localization of these processes to the NTS further suggests that these adaptations by NTS neurons could make them vulnerable to the highly undesirable effects of alcohol withdrawal, including disturbances in cardiorespiratory reflex regulation. These altered gene functions could prove to be fruitful targets for prevention of these symptoms.
qRT-PCR QC Expression Validation
We analyzed 16 positive GO function-related genes and 8 other genes by qRT-PCR, and, of these, 20 were confirmed as differentially expressed (Table 4; gene primers in Supplemental Table S4). These genes were chosen to represent a wide distribution of P values for significance and of relative signal magnitudes (amount of mRNA in the sample). These results indicate that our microarray results are robust. Regression analysis of the relationship between the microarray and qRT-PCR results yields a slope of 1.35 (P < 0.03; plot not shown). There is a slight compression of the fold change (35%) by the microarray data compared with the qPCR data. The bias (y-intercept) was not significantly different from zero (P > 0.3). For most of these genes, the direction (up- or downregulation) and strength of the microarray results agree with the qRT-PCR results (Table 4).
|
A key feature of the study is the use of methods and analyses that support a high level of quantitative precision. This has made it possible to study low-abundance genes and genes that exhibit differential gene expression changes of less than twofold, making the analysis of the alcohol-accommodated state of the NTS possible. The consistency and low noise obtainable with the vRNA approach together with the use of six replicate pairs supported statistically significant results to, on average, the 20% differential expression change level, arguably as low as is likely to be physiologically meaningful.
Our findings are specific to the NTS and as such are interesting in that they are quite distinct to those previously reported in structures of the forebrain and cerebellum. The genes affected in the NTS may contribute to altered functions relevant to the role of the NTS in maintenance of homeostatic functions and in particular to cardiorespiratory regulation. The hypothesized involvement of the MAPK and calcium signaling pathways is particularly meaningful, since they are strongly associated with the NTS function in the baroreceptor reflexes and respiratory regulation. These NTS-specific changes in signaling, neurogenesis, and neuronal plasticity may be critical to the maintenance of these NTS functions in the face of chronic alcohol consumption. The altered state of these groups of genes may be the alcohol-adapted phenotypic state for NTS-associated functions.
At the same time, the altered neuronal functions of NTS neurons in the alcohol-adapted state are highly likely to make the NTS vulnerable to the noxious and life-threatening effects of alcohol withdrawal. Effects on cellular and organismic homeostasis are likely to be experienced as highly unpleasant. Because the NTS is richly connected with the limbic forebrain, in particular the central nucleus of the amygdala (52, 93), the altered state of NTS will certainly influence emotional state. The negative emotional experience of alcohol withdrawal is a major deterrent to quitting drinking for alcoholics. In addition, disturbances of cardiorespiratory reflexes very plausibly may contribute to the arrhythmias and disturbances of respiration associated with acute withdrawal (31). These hypotheses of a causal relationship between the alcohol-adapted state of NTS and the symptoms of withdrawal may provide explanations for the symptoms to be tested by direct study of the role of the systems and pathways reported in the present study in withdrawal.
![]() |
GRANTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
ACKNOWLEDGMENTS |
---|
![]() |
FOOTNOTES |
---|
Address for reprint requests and other correspondence: J. S. Schwaber, Daniel Baugh Institute for Functional Genomics/Computational Biology, Dept. of Pathology, Anatomy and Cell Biology, Thomas Jefferson Univ., 1020 Locust St., Rm. 381, Philadelphia, PA 19107 (e-mail: james.schwaber{at}mail.dbi.tju.edu).
10.1152/physiolgenomics.00184.2005.
* M. Y. Covarrubias and R. L. Khan contributed equally to this work and are listed in alphabetical order.
1 The Supplemental Material for this article (Supplemental Tables S1S4 and Supplemental Figs. S1 and S2) is available online at http://physiolgenomics.physiology.org/cgi/content/full/00184.2005/DC1.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Visit Other APS Journals Online |