Mouse brain gene expression changes after acute and chronic amphetamine

Authors


Address correspondence and reprint requests to Dr George R. Uhl, Box 5180, Baltimore, md 21224 , USA. E-mail: guhl@intra.nida.nih.gov

Abstract

Gene expression changes are candidate mechanisms to contribute to long-term consequences of psychostimulant use. We use microarrays to examine the expression of 6340 genes in brains of mice killed 5 or 20 h following 14 day, twice-daily treatments with saline (SS), saline followed by a single 7.5 mg/kg amphetamine dose (SA), or repeated 7.5 mg/kg amphetamine doses (AA) that produce sensitization but no clear-cut neuronal toxicities. Arrays display robust hybridization for about 3600 transcripts. One hundred and seventeen of these expressed transcripts are candidate positives for drug-related changes, displaying > 1.8-fold differences from SS control values in whole brains of either SA or AA mice. Five transcripts reveal altered expression in both AA and SA mice. SA mostly enhances expression while AA treatments largely reduce expression. Fourteen SA and four AA changes in whole brain mRNA were replicated by > 1.8-fold changes in independent microarray assessments of either cerebral cortical or brainstem mRNAs, with more changes identified in frontal than in entorhinal/parietal cortical samples. About one-quarter of these changes persist in initial studies of mice killed 20 h after the last amphetamine injection. Each of these genes, including transcription factor, cellular regulatory, structural and other gene family members, are candidates to contribute to brain adaptations to psychostimulants.

Abbreviations used
AA

saline followed by repeated 7.5 mg/kg amphetamine doses

HSP

heat shock protein

SA

a single 7.5 mg/kg amphetamine dose

SS

saline

Repeated exposures to moderate doses of amphetamine lead to neuroadaptations that include enhanced behavioral responses termed ‘sensitization’, reduced responses or ‘tolerance’, and addiction (Berke and Hyman 2000). Neuroadaptations could also contribute to consequences of longer-term amphetamine use that can include psychosis and mood disturbances. Although neither the exact brain sites nor the detailed molecular mechanisms for these phenomena are completely understood, they can all be elicited by amphetamine doses that are significantly lower than those required to produce frank neurotoxicity.

Changes in gene expression repertoires are among the most promising sites at which to search for molecular mechanisms that might underlie these behaviorally significant neuroadaptations. Several studies have documented altered expression of a number of genes in brains of animals exposed to psychostimulants (Hope et al. 1992; Persico et al. 1993; Persico et al. 1995; Robinson and Kolb 1997; Berke et al. 1998; Cadet et al. 2001; Freeman et al. 2001a,b; Jayanthi et al. 2001; Takaki et al. 2001; Thiriet et al. 2001; Ujike et al. 2002; Xie et al. 2002). These studies reveal a number of features, including apparent ‘tolerance’ to amphetamine effects on gene expression (Persico et al. 1993), but have examined only limited numbers of genes or effects of moderate or high, neurotoxic doses.

Microarrays allow parallel screening of the expression patterns and regulation of thousands of genes. Examination of many genes in any single experiment also provides opportunities for false positive results that can limit confidence in the results of any single series of experiments. We now report the candidate positive genes whose expression changes in brains of mice treated with a single injection of amphetamine after two weeks of saline injections (SA) or in animals treated with two weeks of amphetamine injections (AA), both compared to control mice treated with two weeks of saline injections (SS). Several of these candidate positive findings can be replicated in independent studies of mRNAs from several brain regions of separate groups of SS, SA and AA mice. These results, and initial data from mice killed 20 h after the last amphetamine injections, provide an initial estimate of the fraction of genes whose expression might be altered by such amphetamine doses across large brain regions. They provide a contrast with data recently reported using neurotoxic doses of this drug (e.g. Xie et al. 2002). These data document the types of genes regulated acutely or chronically by psychostimulants, the different patterns of gene regulation induced by acute versus chronic amphetamine and evidence for striking tolerance for many amphetamine-induced gene expression changes. Each of these results can inform our understanding of processes likely to underlie neuroadaptations induced by short- and long-term psychostimulants, including those that may have behavioral relevance.

Materials and methods

Twelve-week-old male C57BL/J6 mice were housed four to a cage with free access to food and water and maintained on a 12-h light/dark cycle (lights on at 07.00 h) for 2 weeks before initial injections. Mice in each of five treatment groups received intraperitoneal injections at 16.00 h on day one, at 08.00 h and 16.00 h on days 2–10, and between 08.00 and 09.00 h on day 11. Chronic amphetamine (AA) mice received 20 injections of 7.5 mg/kg amphetamine (d-amphetamine sulfate). Acute amphetamine (SA) mice received 19 injections of saline followed by a single 7.5 mg/kg amphetamine injection. Control (SS) mice received 20 injections of saline. Mice tolerated each of the amphetamine regimens without mortality or significant weight loss Mice were killed 5 h after their last injection.

For initial experiments assessing effects observed at a ‘withdrawal’ timepoint, chronic amphetamine/withdrawal (AA/W-20) mice received 19 injections of amphetamine followed by a single saline injection. These animals displayed differences from SS control mice that were compared with those from saline/amphetamine withdrawal (SA/W-20) mice, which received 18 saline injections followed by a single amphetamine injection at 16.00 h of day 10 and a single saline injection at 08.00 h of day 11. Mice in each of these two withdrawal groups were thus killed 20 h after their last amphetamine exposure.

Mice were killed by cervical dislocation, brains were quickly removed, immersed briefly in ice-cold saline, and whole brains or rapidly dissected brain regions were frozen on dry ice and stored at −70°C. Cortex samples were dissected from three blocks formed by rapid coronal razor blade cuts that divided the cerebral cortex into thirds and removed the cingulate cortex using a rapid razor blade cut that removed a 0.5-mm medial strip. Prefrontal and frontal cortex samples were then dissected from the underlying white matter from the rostral 1/3 block, while entorhinal and parietal cerebral cortical samples were dissected from the middle block. Brainstem was defined as the midbrain, pons and medulla. Dissection speed and reproducibility were emphasized for maximal RNA quality; smaller subcortical structures were not dissected. Total RNA was isolated using TRI Reagent (MRC, Inc., Cincinnati, OH, USA) as described (Sokolov 1998). PolyA + , mRNA enriched RNA was isolated from total RNA using Oligotex mRNA kits (Quagen Inc. Valencia, CA, USA). Poly A + RNA was used in hybridization experiments for total brain, while total RNA was used for cortical regions and brain stem. Biotinylated cRNA probes for hybridization were prepared according to the manufacturer. Hybridizations were carried out using Affymetrix Mu6500 microarrays. Data were initially analyzed using Affymetrix GeneChip Analysis Suite 3.3. ‘Present/absent’, ‘changed’, and ‘fold change’ values were generated for each gene in SA and AA samples in comparisons with SS samples from microarrrays hybridized, washed and scanned in parallel. Initial comparisons of data from AA/W-20 and SA/W-20 mice to data from SS mice were also performed. Subsequent analyses used Microsoft Excel and Access.

We attempted to balance the feasibility of these microarray studies with the need to assess experimental variability as follows. For primary analyses of whole brains comparing AA or SA to SS mice, poly A + RNA from brains of 3–5 mice was pooled. Biotin-labeled hybridization probe prepared from mRNA extracted from each pool of brains was hybridized to one or two separate microarrays. Results reported here are from three replicates of such experiments.

We averaged the results from three to four independent replicate hybridization experiments comparing whole brain mRNAs from three different groups of mice subjected to SA, AA and SS regimens. We defined genes with candidate changes in whole brain gene expression as genes for which three criteria were satisfied: (i) Hybridization was sufficiently robust that the gene's expression was rated as ‘present’ by Affymetrix criteria in at least one of the SS, SA or AA groups of mice; (ii) hybridization levels were rated as ‘changed’ by Affymetrix criteria in at least one hybridization experiment; and (iii) the mean differences between SS and either SA or AA mice averaged > 1.8-fold in the 3–4 replicate comparisons.

For analyses of brain regions, RNA from frontal cortex, entorhinal/parietal cortex and brainstems of pools of four animals for each condition was used to prepare biotin labeled hybridization probe which was hybridized to two or three microarrays. These experiments were replicated for each of two pools for cerebral cortical mRNAs. Results from ‘candidate positive’ genes in whole brain experiments were compared with those from brain regions. Whole-brain gene expression changes were ‘confirmed’ by ≥ 1.8-fold changes in brain regional samples, such results seek both generalization and replication.

For initial studies of ‘withdrawal’ conditions in AA/W-20 and SA/W-20 mice, RNA was pooled from whole brains of four mice and hybridization of two preparations of biotin labeled hybridization probes was repeated to two sets of microarrays. Differences from SS controls found in mice subjected to AA/W-20 regimens was compared to those from mice with SA/W-20 regimens.

Confirmation of candidate positive results was also sought by comparison to data from the literature, results of reverse transcriptase/polymerase chain reaction (RT-PCR) mRNA assays, and data from western blot protein quantitation studies. RT-PCR assays were carried out using parallel amplification of endogenously expressed β-actin as an internal standard. Standard curves were constructed for each RT-PCR assay. Quantitation used measurements taken at least 3–4 cycles before the end of the exponential phase of PCR reactions. The PCR cycle consisted of: 30 s at 94°C, 30 s at 60°C, 30 s at 72°C followed by 1 min at 72°C. TSC22 was assessed in 24 PCR cycles using sequences 5′-CTGCAGCCTACTCCTTGCTTC-3′ and 5′-CCAGTTACCGGAAACAACATC-3′, CREM in 28 cycles using 5′-CAGAAGAAGCAACTCGCAAGC-3′ and 5′-CCTGCCCCGATTAGAGTTCAC-3′; synaptotagmin V: with 24 cycles using 5′-GCCCAAGCCCAAGTCCATC-3′ and 5′-TGCGGAGAACCGATCAAAGT-3′, DUSP7 with 22 cycles of: 5′-CGTTCAGTGACAGTTACTGTAGC-3′ and 5′-CACTGGGTGCGTGGTTGTC-3′; and SERCA2: with 32 PCR cycles using 5′-AAGCTACCGAGACTGCTCTCAC-3′ and 5′-CTGTTGCCGCACAGCTCAC-3′. Western blotting for heat shock protein (HSP)-27 was carried out using 10% PAAG gels loaded with 20 µg of protein per lane, polyclonal anti-HSP-27 antisera (Upstate Biotechnology, Lake Placid, NY, USA) and amplified alkaline phosphatase (Bio-Rad Laboratories, Hercules, CA, USA). HSP amounts were within the linear range of this assay. Total protein staining with Coomassie blue confirmed equal protein loading.

Results

Initial analyses

‘Candidate’ transcripts for amphetamine regulation in whole brain mRNAs

Preliminary gel analyses revealed similar mRNA size distributions in samples from SS, SA and AA mice (data not shown). Approximately 3300 out of the 6300 transcripts assessed reached Affymetrix criteria ‘present’ in at least one of three replicate experiments in total brain. One hundred and seventeen transcripts reached Affymetrix criteria for ‘present’ and ‘changed’ as well as displaying ≥ 1.8-fold differences in hybridization intensity/average difference values comparing data from SS control mice to either SA or AA data (ftp://137.187.144.38/). Sixty-three and 59 transcripts reached these criteria in comparisons of SA versus SS and AA versus SS mRNAs, respectively. These are ‘candidate’ amphetamine-regulated genes regulated by acute and chronic amphetamine. Five transcripts reveal altered expression in both AA and SA mice.

SA results differed from AA data. SA gene expression changes were largely up-regulation; 38 transcripts were up-regulated and 25 down-regulated. AA gene expression changes were largely down-regulation; 18 transcripts were up-regulated and 41 down-regulated (ftp://137.187.144.38/). Five candidate genes shared altered expression in both SA and AA mice (Table 1). Fifty-eight of the 63 genes whose expression was modulated by a single amphetamine exposure thus appeared to display ‘tolerance’ after chronic exposures. Fifty-four of the 59 genes regulated by chronic exposures were not affected by the single amphetamine administrations to SA mice.

Table 1.  “Confirmed” amphetamine-regulated transcripts
GeneDescriptionConfirmationGenBank#SAAASA/W
WB
AA/W
WB
WBFE/pBStWBFE/pBSt
  1. TGF, tumor growth factor; NGF, nerve growth factor. Bold numbers: Transcripts which reach Affymetrix criteria for ‘present’, ‘changed’, and > 1.8 change, and have confirmed changes. Normal font numbers: Transcripts changed ≥1.8-fold (not reaching Affymetrix criteria for both ‘present’ and ‘changed’) and have confirmed changes. These changes are not counted as ‘confirmations’ in the text and are provided here for illustration only. ‘NA’: gene identification from BLAST search of probe sequences provided by Affymetrix (personal communication) when the Genebank ID, provided by Affymetrix was no longer valid. Changes are confirmed by one or more of the following criteria: (i) changed in the whole brain and confirmed in brain regions; (ii) changed in a brain region and confirmed in another brain region; (iii) changed in SA and confirmed in AA mice; (iv) changed in whole brain and confirmed by RT/PCR or western analyses; (v) changed in whole brain and confirmed by duplicate probe sets; (vi) changed in whole brain or a region and confirmed by literature. Genes confirmed by literature included CHOP10 (Berke et al. 1998), Fos B (Persico et al. 1993; Persico et al. 1995; Berke et al. 1998; Ujike 2001), C/EBP (Cadet et al. 2001), c-JUN (Hope et al. 1992; Persico et al. 1993; Berke et al. 1998; Cadet et al. 2001), Pax8 (Cadet et al. 2001), TAC1 (Lindefors 1992; Berke et al. 1998), Camk2d (Ujike 2001), Scg2 (Kuzmin and Johansson 1999), actin 1 (Maeno et al. 2000). WB: Whole brain; F: frontal cortex; E/p: Entorhinal/parietal cortex; BSt: brainstem.

Transcription factors
 TSC22TGFβ regulated leucine zipper1,2,3,4AA0507331.91.92.02.41.9
 DBPD-Box binding protein1,2U29762−2.12.81.91.8
 Hes5Hairy/enhancer of split5 CNS homolog1D321322.12.3−2.5
 CHOP10C/EBP or LAP heterodimer6X670831.81.82.0−2.22.3
 FosBAP1 binding1,2,6X148972.52.92.1
 CREMCREB modulator1,2,4M602853.42.37.4  1.9−3.0
 C/EBPβCCAAT/enhancer binding protein β6X626002.01.91.8
 C/EBPδCCAAT/enhancer binding protein δ6X618001.8
 c-JUNAP1 binding6J041151.8−2.02.0
 Pax8Developmental regulator6X574871.9
 Foxg1Developmental regulator3U367602.12.2
Cellular stress/molecular chaperones
 HSP 27Heat shock protein 27 kDa1,2,4,5W189503.73.72.23.1
 HSP 27Heat shock protein 27 kDa4W080577.62.94.010.0
 HSP 47Serine protease inhibitor homolog (J6)1,2D129073.13.52.92.8
 HSP 105βHeat shock protein 105 kDa β1L404063.11.82.3
 HSP 25Heat shock protein 25 kDa5AA0234582.12.0
 HSP 25Heat shock protein 25 kDa5AA0150576.92.32.83.3
 HSP 90αHeat shock protein HSP-90 α3,5NA2.02.0−2.0
 HSP 90αHeat shock protein HSP-90 α5NA2.2
 MIDA1Binds to helix-loop-helix3D637841.92.1
Signaling pathways
 Ifnb2Interferon β, type 21,2J004242.23.32.5
 TAC1β-Preprotachykinin A3,6D175842.12.0
 AGTAngiotensinogen5AA106347– 2.8−2.0
 AGTAngiotensinogen5W131362.4−5.7−3.0
 AGTAngiotensinogen5W174731.8
 AGTAngiotensinogen5W136322.3−2.02.42.3−2.8−1.8
 PrlProlactin2X044183.13.7−3.4−2.926.020.0
 GhSomatotropin, growth hormone2,3Z466638.75.6−2.55.8
 ROCK-2Rho-associated protein kinase1U585131.81.83.2−2.7
 Mnat1Protein kinase, ménage a trois 13U352493.0−3.21.8
 DUSP7Dual specificity phosphatase 74W784431.8−2.1
 LckBP1Substrate for tyrosine kinase p56lck3X84797−7.76.9
 Camk2dCAM kinase II, δ chain6W096642.4−1.8
 FGFR-1FGF receptor3AA0602592.5−4.63.12.8
 SERCA2Ca2+-transporting ATPase1,3AA1534841.92.92.52.44.72.9
 YEL031WP-type cation translocating ATPase1AA015291−2.02.32.2−2.2−1.9
 Atp1a2Na+/K+-transporting ATPase α12,3AA1179733.01.9 1.8−2.1−2.2−5.7
 InhbaInhibin β-A, mesoderm formation2X696194.010.02.3−1.8
 Amfr7TM autocrine motility factor receptor3AA0601872.01.9
 Ptgs2/Cox2Cyclooxygenase 22M882422.42.4
 Adcy9Adenylyl cyclase 93U306021.81.8
Synaptic function
 Tubb4Tubulin β4 chain3W65827−2.02.0
 SytvSynaptotagmin V4W301332.11.8−2.9
 Scg2Secretogranin II6AC0838872.3−3.0−3.9−2.6−6.7
 SNCAα-Synuclein, specific for synapses3AA108571−1.92.1−2.43.6
 actin 1Actin6AA1095272.2−4.7
Cell cycle
 Ccnd3Cyclin 31,2U438441.81.92.2
 Sirt2Histone deacetylase3AA1055363.12.5
 DNAtopo IIIDNA topoisomerase III3W827932.12.2
Protein synthesis/degradation
 EPRSMultifunctional aa-tRNA synthetase1,2,5AA1044772.22.72.32.0
 EPRSMultifunctional aa-tRNA synthetase5AA0622372.8
 EPRSMultifunctional aa-tRNA synthetase5AA1702232.12.0
 METRSMethionyl-tRNA synthetase3AA0603682.7−2.22.4
 EF-2Elongation factor5NA2.0−2.7
 EF-2Elongation factor5AA1052941.8
 CTSDCathepsin D2X528862.22.52.5
 Ddx19RNA helicase from the DEAD box3AA146329−5.2−5.9−1.8
Others
 PC4/TIS7Interferon- and NGF-inducible protein1,2W296692.32.62.42.1
 xlr3aB-cell surface antigen 3b1,2U026002.22.22.91.8
 Kpna2Karyopherin α 2, nuclear import3D557202.12.9
 cwsp1Cell wall structural protein 11,3AA1250972.72.02.5
 BTG1B-cell translocation gene 13L168461.81.8
 PBEFPre-B cell enhancing factor3AA0601671.92.0
 PC-TPPhosphatidylcholine transfer protein3AA0890972.3−4.9−2.25.5−4.0−4.9
 CD34Hematopoietic stem cell antigen1AA064307−2.0−3.42.22.8−4.3−1.9
 MP4Proline-rich protein5X584383.2−1.82.7
 MP4Proline-rich protein5X584382.42.15.3
 cP-450IIIACytochrome P-450IIIA3X604522.41.81.8
 PTACoagulation factor XI3AA0665312.02.1

‘Candidate’ genes from brain regional expression patterns

Initial analyses of the gene expression changes manifest in brain regions were also undertaken using the same criteria of presence, change and ≥ 1.8-fold change that were applied to the whole brain data. More genes were candidates to be regulated by acute or chronic amphetamine in frontal (n = 125) than entorhinal/parietal cortical samples (n = 44; 12 overlapping with data from frontal cortex), consistent with the more abundant frontal cortical innervation by monoamine systems that are impacted by amphetamine. AA regimens were more effective than SA in altering transcript expression in frontal cortex (AA/SA = 78/57), while opposite results were noted in entorhinal/parietal cortex (12/37). Few transcripts altered in cortical samples were among the 63 genes changed > 1.8-fold in brain stem, 32 with SA, 38 with AA and seven with both regimens (see ftp://137.187.144.38/ for the complete list of genes changed in the regions or whole brain).

‘Candidate’ genes from analyses of ‘withdrawal’ timepoints

Initial analyses of gene expression in pools of mice killed 20 h after their last amphetamine injection revealed that 27%, 32 out of the 117 candidate gene expression changes found five hours alter injection, persisted at 20 h (ftp://137.187.144.38/). More transcripts displayed persistent changes after AA (22 out of 59) than after SA regimens (10 out of 63), as anticipated.

Confirmatory analyses

Candidate transcripts from whole brain studies replicated in analyses of brain regional mRNAs

We tested whether several ‘candidate’ amphetamine-regulated genes identified in whole-brain studies were also regulated in brain regional samples dissected from different mice treated with the same AA, SA and SS regimens. Eighteen of the 117 candidate transcripts were confirmed, displaying > 1.8-fold changes in at least one of the three brain regions examined (Table 1). Fourteen of these genes were changed after SA, and four genes after AA. In the brain regional experiments, 0.7–2.0% of the transcripts displayed ≥ 1.8-fold differences between SS and either SA or AA mice, thus chance would have provided less than one confirmation rather than the 18 observed here. Further, three of the confirmed genes, PC4/TIS7, TSC22, and tRNA synthetase, displayed ≥ 1.8-fold changes in each of the three brain regions examined in SA mice (Table 1). On average fewer than 1/10000 transcripts would have experienced such reproducible changes by chance.

Genes with expression changes detected by several probe sets on the arrays

Microarrays used here contain several multiple probe sets with oligonucleotides complementary to the same mRNAs. Altered expression of HSP25, HSP27, HSP90, multifunctional aa-tRNA synthetase, angiotensinogen, proline-rich protein MP4 and elongation factor EF-2 were each confirmed by consistent results from several independent probe sets targeted to different regions of their mRNAs. Several cases produced different results from different probe sets, results that could reflect false positives, alternative mRNA splicing or the different specificities and cross-hybridization possible with different probe sets.

Confirmation by RT-PCR and western assays

We selected a sample of six ‘candidate’ genes from whole brain studies for confirmation at the mRNA level using RT-PCR techniques. Five of these were confirmed, including TSC22, CREM, synaptotagmin V, SERCA2 and DUSP7 (Fig. 1). Altered PP1 expression could not be confirmed. Western approaches also confirmed up-regulation of HSP27 at the protein level (Fig. 2).

Figure 1.

RT-PCR confirmation of altered expression of a sample of five genes in whole brain. Levels of the mRNAs were measured as described in Materials and methods. Data are mean ± SEM. Mice were treated with a single injection of amphetamine after two weeks of saline injections (SA), with 2 weeks of amphetamine injections (AA), or with two weeks of saline injections (SS). 5–7 mice per group were used for CREM, TSC22, DUSP7, SERCA2A, and 3–4 mice per group were used for STGV. **p < 0.0001 (one-tailed t-test) *p < 0.01 (one-tailed t-test).

Figure 2.

Western blot analysis of HSP27 protein expression in whole brain. (a) Representative western blot of HSP27. Quantification of HSP27 protein expression. (b) Data are means ± SEM.

Concordance with prior literature

Changes in a number of the ‘candidate’ genes were supported by concordance with prior results reported in the literature, although administration paradigms varied between these studies and our current approaches. Expression changes confirmed in this manner include changes in CHOP10, FosB, TSC22, c-JUN, C/EBP, Pax8, β-preprotachykinin (TAC1), CAMK2, secretogranin II and actin 1 (Table 1).

These genes whose expression changes were confirmed by at least one of these approaches fall into several gene families, as follows:

(a) Transcription factors: Single exposure to amphetamine regulates expression of 11 transcription factor genes (Table 1). These include FosB, CHOP10, c-JUN, C/EBP and Pax8 which have been previously shown to respond to acute dopaminergic stimulation (Hope et al. 1992; Persico et al. 1995; Berke et al. 1998; Cadet et al. 2001; Ujike 2001). Interestingly, chop-10 can heterodimerize with A/EBP and C/EBP gene products, also up-regulated by SA treatments (Table 1). HES-5, DBP, PC4, FOSB, CREM, TSC22 display evidence for tolerance and change in whole brain and brain regional samples, confirming the widespread nature of psychostimulant effects.

(b) Chaperone/cell stress genes: Six cell-stress genes are induced by SA treatments, but only MIDA1 remains up-regulated after AA regimens. Confirmation of the up-regulation of HSP25, HSP27 and HSP90-alpha by each of two sets of oligonucleotides and of HSP27 protein in western analyses adds to confidence in their identification. Elucidation of changes in HSP27, HSP47 and HSP105 in whole brain and brain regions in SA but not in AA mice points again to the breadth of amphetamine-induced brain changes.

(c) Genes encoding signaling pathway proteins: Eighteen genes related to signaling pathways were regulated by amphetamine. Virtually all changes are down-regulation, present mostly after AA treatments. Many of the changes persist in animals killed after 20 h. Altered regulation of these genes suggests plausible ways in which AA treatments could have widespread effects on cellular information trafficking. Most of these genes show only modest response to SA treatments, suggesting requirements for engaging sensitization-like mechanisms or for reaching the elevated amphetamine levels only available after cumulative dosing. Up regulation of CAMK2d by cocaine has been reported (Ujike 2001).

Communication between cells can also be affected. AA, but not SA treatments, down-regulate angiotensinogen (Table 1). Both SA and AA up-regulate expression of the substance p precursor β-preprotachykinin A gene, previously reported to be modulated by dopaminergic stimulation (Lindefors 1992; Berke et al. 1998).

(d) Synaptic function:  Features of neuronal cell biology can change in surprising fashions, as has been previously reported for modulation of secretogranin by cocaine (Kuzmin and Johansson 1999) and down-regulation of actin after amphetamine (Maeno et al. 2000).

(e) Other amphetamine regulated genes: Three genes are involved in differentiation/development/cell cycle. Cyclin D3 displays marked up-regulation by SA and apparent tolerance after AA (Table 1). SA but not AA markedly up-regulates multifunctional aa tRNA synthetase, while AA but not SA down-regulate elongation factor E2.

Discussion

Microarray data from animals killed following pharmacological manipulations can be discussed in terms of the strengths and the limitations of the results obtained, the classes of data generated, and the sorts of insights that could come from exploration of the gene expression changes described here.

The approaches described here demonstrate evidence for substantial sensitivity and specificity. Almost half of the genes whose sequences are represented on the microarrays produced hybridization signals sufficient to achieve Affymetrix criteria for ‘presence’ in whole brain mRNAs. These array results document that more than 3% of the expressed genes display reproducible, sizable changes in expression after either acute or chronic amphetamine exposures in whole brain samples. Most of a subsample of these genes also display changed expression when studied by RT-PCR or western blot techniques. Several of the gene expression changes were identified independently by several different sets of oligonucleotide probes on the microarrays. Many of the alterations noted in whole brain samples were also found in analyses of brain regional samples. However, these approaches would be likely to be insensitive for identifying changes that were restricted to small brain regions if other larger brain regions also expressed in a drug-insensitive fashion.

Amphetamine changes a remarkable amount of gene expression in the brain. Many gene expression changes are sufficiently large and/or widespread that they are identified in studies of mRNAs extracted from whole brains. These gene expression patterns differ substantially from those noted in brains of rodents subjected to very high, 40–45 mg/kg neurotoxic doses of methamphetamine, although up-regulation of c-JUN, FosB and Pax8 are also noted following these high doses (Cadet et al. 2001; Xie et al. 2002). We have also recently documented altered whole-brain expression of many genes in mice with knockouts of the monoamine transporters that amphetamine blocks (G.R. Uhl and Q.-R. Liu, unpublished data).

These patterns of gene expression change also demonstrate selectivity. Gene expression changes noted in mice killed after SA regimens differ substantially from those noted after AA treatments. Most SA changes are up-regulation, and most of these acutely up-regulated genes are not persistently up-regulated in AA mice. Such changes thus appear to display ‘tolerance’ that we and others have previously reported for psychostimulant-induced changes in transcription factor gene expression levels (Hope et al. 1992; Persico et al. 1993). Conversely, most AA gene regulation is down-regulation of genes whose expression was not significantly affected by SA treatments. It could be argued that, as chronic amphetamine induces tolerance to most of the gene up-regulation produced by acute amphetamine administration, the relevance of acutely regulated genes to addiction-related neuroadaptations might be questionable. Conversely, however, the acute transcription factor changes could also play transient roles in the cascades of events necessary to produce chronic changes more relevant to the maintenance of the sensitized, and perhaps even the addicted state.

We use data from several sources to confirm candidate gene expression changes. While these approaches provide reasonable assurance against large numbers of false positive results, they do not share the same sensitivity. False-negative results are thus likely among the candidate genes whose changes are not confirmed in the brain regions sampled here. Failure to confirm might even be likely for genes with uneven expression patterns or changed expression in regions not sampled in the present brain regional studies.

Five genes are regulated by both acute and chronic amphetamine. These include the neuropeptide β-preprotachykinin A gene that produces the classically amphetamine regulated substance P and substance K neuropeptides (Kraft et al. 2001), MIDA1, and structural proteins that include tubulin, cell wall structural protein 1, and karyopherrin α2. Indeed, although thinking about adaptations to psychostimulants has classically emphasized adaptations in neurotransmission molecules, more recent depictions have also identified altered cell structures and neuronal connectivities, providing good fits with several of the results documented here.

We have attempted to group the genes whose expression patterns change into functional families. We are aware that the limits of our understanding of the functions of the products of many of these genes make ‘gene family’ assignments somewhat arbitrary (ftp://137.187.144.38/). However, the classes used here represent a starting point for analyses that can be refined as we improve our understanding of gene classes.

A number of genes that encode transcription factors are predominantly up-regulated in SA mice. When coupled to the prominent alterations in expression of ‘heat shock’/chaperone proteins in such animals, these changes form a whole-brain response that could alter the expression of classes of ‘downstream’ genes. Some of these changes could occur in the neurons that express the transporters at which amphetamine acts primarily; others could take place in neurons influenced by more distant consequences of amphetamine administration. It is conceivable, for example, that altered amphetamine-induced thermoregulation could contribute to some of these gene expression alterations (Miller et al. 1991; Lu and Das 1993).

Genes that encode major regulators of intracellular signaling functions are regulated to substantial degrees. Cell cycle genes could be expressed in dividing glia, or in relation to other features of largely postmitotic neurons. The striking changes in cyclin and histone deacetylase genes could conceivably be consistent with intracellular regulation in neurons.

Differences in cellular signaling could readily follow differential expression of molecules such as secretogranin II or preprotachykinin. Changes in synaptic proteins that participate in both pre- and postsynaptic neurotransmission mechanisms or altered expression of neuronal structural proteins could add to altered functional connectivity.

Many of the gene expression changes noted in these studies are not as easy to categorize, including the reproducible changes in multifunctional aa-tRNA synthetase. Such data does provide impetus for future biological investigations that could implicate such changes in toxic or adaptive responses to amphetamine. Not all gene expression changes need to be functionally significant, however. It is also conceivable that the transcription factor changes identified here could alter expression of ‘innocent bystander’ genes in the cells that are directly or indirectly influenced by amphetamines. Further work will help us distinguish between epiphenomena and functional changes.

Broad patterns of remarkable gene up-regulation by acute amphetamine, down-regulation by chronic amphetamine, and persistence of several of the changes as much as a day after the last amphetamine dose provide a framework on which to build a more fine grained picture that studies gene expression changes in smaller, more physiologically relevant brain regions or cell groups. These more anatomically aggressive approaches should keep in mind the greater chances for mRNA degradation during prolonged dissections of many small brain regions that may produce a trade off between anatomical and quantitative accuracies, however. Arrays that examine most of the murine genes, rather than the sample studied here, will extend these sorts of analyses. Studies of animals which have self-administered drugs, in comparison to the present data from mice that have received drugs passively, will refine the picture. Gene network models may help to understand some of these patterns of coordinate gene regulation. However, even these current initial data provide a rich picture that appears to provide multiple clues to interesting biochemical consequences of amphetamine administration.

Many of these profound changes in gene expression could even be correlated with some of the behaviorally relevant consequences of acute and chronic amphetamine administration, including sensitization to behavioral effects of amphetamine, addictive and psychotomimetic properties of amphetamine. We have previously shown that a G protein beta subunit identified as up-regulated by amphetamine in subtracted/differential display paradigms could be blocked by oligonucleotide administration intracerebroventricularly, resulting in a striking and reversible blockade of amphetamine sensitization (Wang et al. 1997). The large number of gene expression changes reported both here and in other studies of psychostimulant-induced gene expression provide a wealth of opportunities to examine the behavioral phenotypes conferred by these altered patterns of gene expression.

Acknowledgement

We thank NIDA-IRP for financial support.

Ancillary