Anesthesia & Analgesia:
Anesthetic Pharmacology: Research Report
A Microarray Analysis of Potential Genes Underlying the Neurosensitivity of Mice to Propofol
Lowes, Damon A. PhD*; Galley, Helen F. PhD, FIMLS*; Lowe, Peter R. PhD*; Rikke, Brad A. PhD†; Johnson, Thomas E. PhD†; Webster, Nigel R. PhD FRCP, FRCA*
*Academic Unit of Anaesthesia and Intensive Care, University of Aberdeen, Scotland, United Kingdom; and †Institute for Behavioral Genetics, University of Colorado at Boulder, Boulder, Colorado
This work was supported by the Association of Anaesthetists of Great Britain and Ireland.
Accepted for publication February 7, 2005.
Address correspondence and reprint requests to Dr. Damon Lowes, Academic Unit of Anesthesia and Intensive Care, Institute of Medical Sciences, Foresterhill, Aberdeen, AB25 2ZD, Scotland, UK. Address e-mail to email@example.com.
Establishing the mechanism of action of general anesthetics at the molecular level is difficult because of the multiple targets with which these drugs are associated. Inbred short sleep (ISS) and long sleep (ILS) mice are differentially sensitive in response to ethanol and other sedative hypnotics and contain a single quantitative trait locus (Lorp1) that accounts for the genetic variance of loss-of-righting reflex in response to propofol (LORP). In this study, we used high-density oligonucleotide microarrays to identify global gene expression and candidate genes differentially expressed within the Lorp1 region that may give insight into the molecular mechanism underlying LORP. Microarray analysis was performed using Affymetrix MG-U74Av2 Genechips® and a selection of differentially expressed genes was confirmed by semiquantitative reverse transcription-polymerase chain reaction. Global expression in the brains of ILS and ISS mice revealed 3423 genes that were significantly expressed, of which 139 (4%) were differentially expressed. Analysis of genes located within the Lorp1 region showed that 26 genes were significantly expressed and that just 2 genes (7%) were differentially expressed. These genes encoded for the proteins AWP1 (associated with protein kinase 1) and “BTB (POZ) domain containing 1,” whose functions are largely uncharacterized. Genes differentially expressed outside Lorp1 included seven genes with previously characterized neuronal functions and thus stand out as additional candidate genes that may be involved in mediating the neurosensitivity differences between ISS and ILS.
General anesthetics affect a variety of molecular targets within the central nervous system, but precisely how and where these drugs act is still uncertain (1–3). The inbred long sleep (ILS) and short sleep (ISS) mouse strains were selectively bred for sensitivity to large doses of ethanol (4.2 g/kg) and are hugely different (2 h versus 20 min) in their duration of loss-of-righting reflex (LORR). These mice are also differentially sensitive to a variety of hypnotic sedatives, including propofol (4) with LORR of 9.3 and 4.8 min, respectively (5). Although most studies investigating the mechanisms of action of anesthetics are target-oriented at putative sites of action, especially the γ-aminobutyric acid (GABA)A receptor (6,7), when ILS and ISS mice are administered propofol, there is no difference in GABAA-activated chloride channel activation (8). Therefore, an unbiased approach using quantitative trait locus (QTL) analysis of the ISS and ILS mice revealed that the duration of LORR in response to ethanol is influenced by seven major QTLs (regions of a chromosome that contain one or more genes contributing to a phenotypic difference) (7–9). Further studies of recombinant inbred LSXSS mice identified an ethanol-sensitivity QTL, Lore7 (LORR caused by ethanol on chromosome 7) that also colocalized with a QTL that explained >80% of the genetic variation for LORR in response to propofol. Mapping of this QTL, Lorp1 (LORR caused by propofol), localized the candidate gene region to an estimated 99% confidence interval of just 2.5 centiMorgans (cM), between 71.4–89.7 Mb on chromosome 7 (5). In addition, an etomidate-sensitivity QTL has also been identified in this chromosome region (10,11). Although Lorp1 is very tightly linked to the Tyr albino mutation (c), neither propofol sensitivity nor sensitivity to other sedative hypnotics was attributed to albinism (12,13).
In this study, we sought to identify gene(s) within the Lorp1 region that are expressed at markedly different levels in the brains of ISS and ILS mice, because these genes are likely to be polymorphic in their regulatory regions. Microarray analysis in particular allowed us to simultaneously assess gene expression differences across the entire Lorp1 region and additionally to identify any interesting differences between ISS and ILS outside the Lorp1 region.
ILS and ISS mice were produced at the Specific Pathogen Free facility at the Institute for Behavioral Genetics, University of Colorado at Boulder. ILS and ISS were derived from LS and SS mice by 20 rounds of brother-sister mating in the absence of any behavioral selection (14). Permission was obtained from, and animals were treated in accordance with, the institute’s animal care and use committee. The mice were killed by cervical dislocation and the brains were immediately removed and frozen on dry ice. The ISS and ILS mice were all age-matched females (65–72 days of age).
Total RNA was extracted from a homogenate of total brain from each mouse using TRIzol reagent (Invitrogen Ltd., Paisley, UK) according to the manufacturer’s instructions. The integrity of total RNA from each brain was established using the 2100 Bioanalyser and RNA 6000 Nano Labchips® (Agilent Technologies UK, Cheshire, UK).
Complementary RNA (cRNA) targets for hybridization were prepared and hybridized separately for each brain preparation (3 for ISS and 3 for ILS) according to the Affymetrix protocol, as described by Lockhart et al. (15). Briefly, a primer encoding the T7 RNA polymerase promoter linked to oligo dT24 was used to prime double-stranded cDNA synthesis from each total RNA sample using Superscript II reverse transcriptase (Invitrogen). Each double-stranded cDNA sample was purified by phenol-chloroform extraction and ethanol precipitated. cRNA was in vitro transcribed using the Bio Array High Yield RNA transcript labeling kit (Enzo, Farmingdale, NY), incorporating biotin-UTP and biotin-CTP into the resulting cRNA. cRNA transcripts were precipitated and fragmented at 95°C for 35 min in 30 mM magnesium acetate, 100 mM potassium acetate, and 40 mM Tris-acetate to a mean size of approximately 50–200 nucleotides, added to hybridization buffer, and hybridized to the MG-U74Av2 Genechip® for 16 h at 45°C. GeneChips® were washed, stained with streptavidin-R-phycoerythrin, and scanned at 3-μm resolution using an Agilent gene array scanner. Non-eukaryotic (bioB, C, and D from Escherichia coli and cre from bacteriophage P1) biotinylated and fragmented target cRNA was added to the hybridization buffer before hybridization. The MG-U74Av2 Genechip® contains probe sets for these transcripts which serve as controls for hybridization, washing, and staining procedures.
Scanned image files (.cel) were visually inspected for artifacts and analyzed after normalization using Affymetrix Microarray Suite 5.0 and dChip™ 1.3 (http://www.dchip.org) (16). Model-based expression values and standard errors were calculated using dChip™ after normalization via the program defaults. Genes called “outliers” by the model-based expression were treated as blank values. Differential gene expression was determined by measuring the difference in fluorescence of combined probe sets in a sample (the ISS mouse strain) relative to a baseline (the ILS mouse strain) and only included combined probe sets that had a mean intensity value of 100 fluorescent units and a mean fold-change of ±2 (P < 0.1, t-test). A second analysis was performed on all genes using the software, Significance Analysis of Microarrays (SAM)1 version 1.21 (http://www-stat.stanford.edu/∼tibs/SAM/) (17). The output criteria selected for SAM was a significance threshold expected to produce a false discovery rate (FDR)2 of <15 differentially expressed genes.
The differential expression of selected genes determined to be differentially expressed by Genechip® microarray analysis was confirmed by two-step semiquantitative RT-PCR. Single-strand cDNA was synthesized using 5 μg of total RNA,100 pmol oligo dT24 primer, and Superscript II reverse transcriptase. In the second step of PCR, gene-specific primers (0.3 μM) were added to the PCR reaction to amplify the cDNA of interest. Hypoxanthine phosphoribosyl transferase (HPRT) was used as an endogenous control. The level of gene-specific product was normalized against the amplified HPRT. Specifically, 1.0 μL from the RT reaction was added to a total volume of 20 μL of DyNAmo SYBR® Green qPCR mix (Finzymes, Espoo, Finland). The PCR profile consisted of an initial denaturation of 10 min at 95°C, then 39 cycles of amplification under the following conditions: denaturation at 94°C for 30 s, annealing at 56°C for 30 s, and extension at 72°C for 30 s. Fluorescence was measured at 72° and 75°C after each round of amplification using the DNA Engine Opticon 2 continuous fluorescence system (MJ Bioworks).
To date, approximately 30,000 genes are estimated to exist in the murine genome, and the U74Av2 GeneChip® contains a total of 12,500 genes/expressed sequence tag probe sets taken from the mouse UniGene database (Build 74) and offer reliability, consistency (18,19), and substantial genome-wide coverage. We found that 3423 (28%) genes were being significantly expressed in whole brain according to predetermined criteria (see Methods). Using regression analysis, we found that overall the expression levels of these genes in ISS and ILS mouse brains were highly similar as expected (r2 = 0.93, P < 0.001, Fig. 1).
To identify the differentially expressed genes from among the 3423 genes being expressed, we used a dual strategy. First, the dChip™-normalized hybridization data from ISS and ILS were compared using the dChip™ software, which yielded 141 genes that were differentially expressed by more than twofold. Second, the normalized hybridization data were exported into SAM, which yielded 370 differentially expressed genes. There were 139 genes (4%) identified by both dChip™ and SAM software analysis as being differentially expressed. Of these 139 genes, 99 corresponded to full-length known gene sequences and 40 associated with expressed sequence tag clusters with no known function or given gene name, as verified by our own BLAST searches (Table 1). Curiously, the microarray data indicated that all but 2 of these 139 genes were being expressed at lower levels in ISS than ILS (even though comparison of the mean expression level of all 3423 expressed genes between ISS and ILS indicated no systematic difference, P = 0.94). To test this, we conducted semiquantitative RT-PCR on seven of these genes and found that all seven were indeed expressed at lower levels in ISS (Table 1). We also conducted semiquantitative PCR on one of the genes expressed at higher levels in ISS according to the microarray data, and this higher level of expression was also confirmed.
Of the 152 genes that lie within the 99% confidence interval for Lorp1, 109 genes (72%) were represented on the MG-U74Av2 Genechip®. The majority of genes missing on the array for this region represented genes encoding for olfactory receptors. Of these 109 genes, we found 26 that were significantly expressed (Table 2), with 2 (7%) exhibiting significant differential expression. The “associated with protein kinase 1”Awp1 gene, located at 77 Mb, was expressed 2.5 times lower in ISS, and encodes for the protein AWP1. AWP1 has two zinc-finger motifs associated with protein binding and associates with protein kinase 1 (PRK1) (20). PRK1 has been shown to control several cellular processes including the endocytic machinery and to phosphorylate a variety of cytoskeletal structural proteins such as vimentin, glial fibrillary acidic protein (21), α-actinin (22), and neurofilament protein (23,24). The other gene, Btbd1, located at 74 Mb, was expressed 2.0 times lower in ISS brains and encodes for a “BTB (POZ) domain containing 1” protein. This protein contains a zinc-finger domain and belongs to a family of proteins that can interact with topoisomerase I (25) and may be involved in myogenic differentiation (26).
Genes outside the Lorp1 region could also be important for specifying some of the other neurosensitivity differences between ILS and ISS and could be involved in mediating some of the downstream effects of Lorp1 as well. To explore what kinds of genes were exhibiting marked differences in expression outside the QTL region, we identified 137 genes that showed a consistent pattern of differential expression of at least twofold. These sequences were further analyzed using the Affymetrix (http://www.affymetrix.com) and NCBI database resources (http://www.ncbi.nlm.nih.gov/) to obtain information on their potential function. These genes are in four general classifications: protein metabolism, DNA metabolism, cell signaling, and RNA transcription. Twenty-four of these genes have previously been experimentally characterized as being expressed within the brain, with the majority coding for proteins that are involved in cell signaling. Seven of these genes have been assigned a neuronal function, and thus warrant special consideration as potential modulators of neurosensitivity. These genes encoded: 1) adenosine triphosphatase, Ca2+ transporting, plasma membrane protein; 2) 14-3-3 θ polypeptide; 3) peptidylglycine α amidating monooxygenase; 4) tachykinin 1; 5) synaptotagamin 4; 6) K+ voltage gated channel β subunit; and 7) glutamine synthetase.
Similar to many general anesthetics and sedative hypnotics, propofol enhances GABAergic neurotransmission and directly activates GABAA receptors, which could account for its anesthetizing action (20). The ILS and ISS mouse strains, however, do not exhibit differences in their GABAA-activated chloride channels (7). Toward identifying the basis for their differential sensitivity, we have used microarray expression differences as an unbiased method of identifying candidate genes within the Lorp1 region (5,6,14). The present study shows that there are just two genes within the Lorp1 region that exhibit marked, whole-brain lower expression differences in the ISS strain. Both encoded proteins are involved in interactions with other proteins that could have wide-ranging effects, which might explain the colocalization of several LORR QTLs to this same region (ethanol, etomidate, and propofol) (10,11). In this study, we focused on the propofol QTL region because it is much more narrowly defined than the other QTL regions (5). Unfortunately, similar to the molecular mechanism of anesthesiology and the functioning of most genes, there is very little biochemical information available to postulate how these genes might be modulating neurosensitivity. AWP1 associates, and therefore, may possibly regulate PRK1 signal transduction. This protein is a member of the protein kinase C (PKC) superfamily of serine/threonine kinases (PKN) (20), which modulates a variety of cellular processes within the central nervous system that include organization of neuronal intermediate filaments (21–24), secretory vesicle transport (27), and transcription factor activation (28). AWP1 may, therefore, affect propofol neurosensitivity by regulating cell signaling networks by modifying PRK1 substrate phosphorylation. This hypothesis is supported by studies showing that genetic reduction of protein kinase A activity in mouse brain makes these animals more sensitive to the sedative effects of ethanol (29). Likewise, Btbd1, a relatively uncharacterized protein, can interact with topoisomerase I, a ubiquitously expressed protein involved in DNA replication, transcription, and repair. However, this protein has also been shown to interact with other proteins that include transcription factors, p53, nucleolin, RING finger proteins, RNA splicing factors, and the RNA polymerase II complex (25). Therefore, Btbd1 may be involved in the regulation of topoisomerase I and gene transcription.
More importantly, the marked difference in the expression of Awp1 and Btbd1 in ISS versus ILS suggests polymorphic differences in the regulatory regions of these genes. Such differences are a prerequisite for genes to be plausible candidates for underlying a QTL. However, microarray studies do not assess the activity of a gene product, because polymorphisms within the coding region of a gene could affect protein function resulting from changes in protein translation, posttranslational modification, or protein interactions and are therefore important considerations when interpreting any microarray study. Although sequence analysis of all the genes in the Lorp1 region (if practical) would undoubtedly identify additional polymorphic genes, we would not expect many more. Previous sequence analysis comparing ISS and ILS has shown that only about 1 in 8 genes on average are polymorphic (30). Therefore, given our finding that only 26 genes in the Lorp1 region are expressed in the brain, we would only expect to find 3–4 genes of potential interest. This suggests then that there is a realistic chance—perhaps as much as 30%—that either Awp1 or Btbd1 is the gene responsible for the Lorp1 QTL. In addition, finding two differentially expressed genes within the Lorp1 region is slightly more than expected by chance alone. This is because the observed frequency of differentially expressed genes relative to the number of genes significantly expressed was 4%. Therefore, the Lorp1 region, in which 26 genes were differentially expressed, would be expected to have just 1 differentially expressed gene by chance. Similarly, if one considers the ratio of differentially expressed genes to the total number of genes on the array, the expectation for the Lorp1 region containing 109 genes would be to find only 1 differentially expressed gene. Therefore, both estimates suggest that finding two differentially expressed genes in the Lorp1 region is more than expected, which is consistent with one of these genes being the Lorp1 gene.
This study was undertaken to evaluate the QTL on chromosome 7 but not only limited to the analysis of the QTL. Our data showed that the mouse brain expresses a large number of genes that have no known function within the brain, with relatively few being differentially expressed. Most of these genes are expressed lower in ISS brain and analyses of overall gene expression showed no significant difference between ISS and ILS samples (P = 0.94). We do not know why these genes exhibited a strong bias for lower expression in ISS brain, as the animals were sex-matched, closely aged-matched, harvested, processed, and analyzed at the same time. The lower ISS expression was not an artifact of chip hybridizations, as confirmed by quantitative PCR performed on a random sample of genes. The neuronal functions of many of these genes shown from our study are not known but could be important in our model. Consequently, this study is a significant advance toward identifying which genes should be directly tested for effects on propofol sensitivity.
1. Franks NP, Lieb WR. Molecular and cellular mechanisms of general anaesthesia. Nature 1994;364:607–14.
2. Franks NP, Leib WR. Which molecular targets are most relevant to general anaesthesia? Toxicol Lett 1998;100–101:1–8.
3. Little HL. How has molecular pharmacology contributed to our understanding of the mechanism(s) of general anaesthesia? Pharmacol Ther 1996;69:37–58.
4. McClearn GE, Kakihana R. Selective breeding for ethanol sensitivity: short-sleep and long-sleep mice. In: McClearn GE, Deitrich RA, Erwin VG, eds. Development of animal models as pharmacokinetics tools. PHHS Publ. No. ADM 81-113. Washington, DC: US Government Printing Office, 147–159.
5. Simpson VJ, Rikke BA, Costello JM, et al. Identification of a genetic region in mice that specifies sensitivity to propofol. Anesthesiology 1998;88:379–89.
6. Belelli D, Pistis M, Peters JA, et al. The interaction of general anaesthetics and neurosteroids with GABAA
and glycine receptors. Neurochem Int 1999;34:447–52.
7. Krasowski MD, Harrison NL. General anaesthetic actions on ligand-gated ion channels. Cell Mol Life Sci 1999;55:1278–303.
8. Simpson VJ, Blendov Y. Propofol produces differences in behavior but not chloride channel function between selected lines of mice. Anesth Analg 1996;82:327–31.
9. De Fries JC, Wilson JR, Erwin VG, Peterson DR. LS x SS recombinant inbred strains of mice: initial characterization. Alcohol Clin Exp Res 1989;13:196–200.
10. Christensen SC, Johnson TE, Markel PD, et al. Quantitative trait locus analyses of sleep-times induced by sedative-hypnotics in LSXSS recombinant inbred strains of mice. Alcohol Clin Exp Res 1996;20:543–50.
11. Downing C, Shen EH, Simpson VJ, Johnson TE. Mapping quantitative trait loci mediating sensitivity to etomidate. Mamm Genome 2003;14:367–75.
12. Johnson TE, De Fries JC, Markel PD. Mapping quantitative trait loci for behavioral traits in the mouse. Behav Genet 1992;22:635–53.
13. Rikke BA, Simpson VJ, Montoliu L, Johnson TE. No effect of albinism on sedative-hypnotic sensitivity to ethanol and anaesthetics. Alcohol Clin Exp Res 2001;25:171–6.
14. Wehner JM, Pounder JI, Parham C, Collins AC. A recombinant inbred strain analysis of sleep-time responses to several sedative-hypnotics. Alcohol Clin Exp Res 1992;16:522–8.
15. Lockhart DJ, Dong H, Byrne MC, et al. Expression monitoring by hybridisation to high-density oligonucleotide arrays. Nat Biotechnol 1996;14:1675–80.
16. Li C, Wong WH. Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection. Proc Natl Acad Sci USA 2001;98:31–6.
17. Tusher VG, Tibshirani R, Chu G. Significant analysis of microarrays applied to the ionising radiation response. Proc Natl Acad Sci USA 2001;98:5116–21.
18. Bakay M, Chen YW, Borup R, et al. Sources of variability and effect of experimental approach on expression profiling data interpretation. BMC Bioinformatics 2002;3:4.
19. Yuen T, Wurmbach E, Pfeffer RL, et al. Accuracy and calibration of commercial oligonucleotide and custom cDNA microarrays. Nucleic Acids Res 2002;30:e48.
20. Duan W, Sun B, Li TW, et al. Cloning and characterization of AWP1, a novel protein that associates with serine/threonine kinase PRK1 in vitro. Gene 2000;256:113–21.
21. Matsuzawa K, Kosako H, Inagaki N, et al. Domain-specific phosphorylation of vimentin and glial fibrillary acidic protein by PKN. Biochem Biophys Res Commun 1997;234:621–5.
22. Mukai H, Toshimori M, Shibata H, et al. Interaction of PKN with alpha-actinin. J Biol Chem 1997;272:4740–6.
23. Mukai H, Toshimori M, Shibata H, et al. PKN associates and phosphorylates the head-rod domain of neurofilament protein. J Biol Chem 1996;271:9816–22.
24. Smythe E, Ayscough KR. The Ark1/Prk1 family of protein kinases: regulators of endocytosis and the actin cytoskeleton. EMBO Rep 2003;4:246–51.
25. Xu L, Yang L, Hashimoto K, et al. Characterization of BTBD1 and BTBD2, two similar BTB-domain-containing Kelch-like proteins that interact with topoisomerase I. BMC Genomics 2002;3:1.
26. Pisani DF, Cabane C, Derijard B, et al. The topoisomerase 1-interacting protein BTBD1 is essential for muscle cell differentiation. Cell Death Differ 2004;11:1157–65.
27. Kawamata T, Taniguchi T, Mukai H, et al. A protein kinase, PKN, accumulates in Alzheimer neurofibrillary tangles and associates endoplasmic reticulum-derived vesicles and phosphorylates tau protein. J Neurosci 1998;18:7402–10.
28. Shibata H, Oda H, Mukai H, et al. Interaction of PKN with neurone-specific base helix-loop-helix transcription factor, NDRF/NeuroD2. Brain Res Mol Brain Res 1999;74:126–34.
29. Wand G, Levine M, Zweifel L, et al. The cAMP-protein kinase A signal transduction pathway modulates ethanol consumption and sedative effects of ethanol. J Neurosci 2001;21:5297–303.
30. Ehringer MA, Thompson J, Conroy O, et al. High-throughput sequence identification of gene coding variants within alcohol-related QTLs. Mamm Genome 2001;12:657–63.
1 SAM identifies genes with statistically significant changes in expression by assimilating a set of gene-specific t-tests. Each gene is assigned a score on the basis of its change in gene expression relative to the standard deviation of repeated measurements for that gene. Genes with scores more than a threshold are deemed potentially significant (17). Cited Here...
2 To estimate the FDR, nonsense genes are identified by analyzing permutations of the measurements. The threshold can be adjusted to identify smaller or larger sets of genes, and FDRs are calculated for each set (17). Cited Here...
© 2005 International Anesthesia Research Society