Neuroscience Discovery Research, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana 46285-0438
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
phencyclidine; methamphetamine; lysergic acid; brain cortex; spinal cord; hippocampus; traumatic injury; microarray; GeneChip
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Mining data from microarrays is crucial for extracting biological value from microarray results. To be effective, this process should be performed at multiple hierarchical levels. However, most studies to date in the field of neuroscience have focused on the analysis of single transcripts or clusters of altered transcripts. Given that the expression and actions of individual genes and proteins are dependent on time (e.g., developmental stage) as well as location (e.g., cellular domain and/or cell type), the need for interpreting alterations in the expression of a particular transcript as part of a multidimensional and multifactorial equation is clear.
Based on the complexity of data resulting from transcript profiling studies, understanding functionally relevant gene expression patterns requires the analysis of large data sets generated by multiple laboratories. Although several reviews have discussed gene alterations associated with particular neuropathologies, few studies have reported a reanalysis of publicly available data sets (e.g., Refs. 9 and 24). Here we combine published data sets from five microarray studies to examine global changes in nerve tissue gene expression. Our approach extends the biological conclusions derived from each study. In addition, we propose an alternative level of microarray data analysis: the identification and interpretation of overlapping gene expression patterns.
![]() |
MATERIALS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Five studies that fulfilled these conditions were selected using the National Center for Biotechnology Information (NCBI) Entrez PubMed tool. The sequence lists from these published sources were classified into two major data set groups: acute drug-induced and trauma-induced changes in nerve tissue gene expression. The first group included acute drug treatments using: lysergic acid diethylamide (LSD) (21), methamphetamine (METH) (22), and phencyclidine (PCP; Kaiser S, Foltz LA, George CA, Kirkwood SC, Bemis KG, Lin X, Gelbert LM, and Nisenbaum LK, unpublished observations). The second group included controlled mechanical injury of either hippocampus (18) or spinal cord (28).
It was not within the aims of this report to evaluate the experimental design and the validity of animal models reported by other peer-reviewed studies. All the selected studies used young adult outbred rats (230325 g). Only one study used female instead of male rats and Long-Evans instead of Sprague-Dawley rats (18). Other primary similarities and differences between the selected studies are summarized in Table 1.
|
Overlapping genes were functionally categorized using Incytes LifeSeq Gold and Proteome Bioknowledge Library (Incyte, Palo Alto, CA) as well as published sources (journal articles + reviews + textbooks). When appropriate, alternative symbols for a given gene are noted in parentheses.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
Common Changes in Gene Expression Induced by Traumatic Injury of Hippocampus and Spinal Cord
Transcript levels for 28 genes were altered by controlled traumatic injury of both hippocampus (18) and spinal cord (28) (Fig. 1B and Table 3).
|
Genes downregulated in hippocampus and spinal cord after traumatic injury (10 of 28 = 36%; Table 3).
A decrease in expression levels was observed in transcripts encoding structural proteins (i.e., Mtap2) as well as genes involved in signal transduction (i.e., Camkk1, Camk2b), ion transport (i.e., Kcnd2, Slc24a2), transmitter release (i.e., SNAP-25A, SNAP-25B), and neurotransmission (i.e., Gria1, Gria3, mGluR3).
Genes altered in opposite directions in response to injury in hippocampus and spinal cord (2 of 28 = 7%; Table 3).
In response to injury, smallest neurofilament (Nfl) and K+ channel TWIK (Kcnk1) were upregulated in hippocampus and downregulated in spinal cord.
Common Changes in Telencephalic Gene Expression Induced by Mechanical Injury and Acute Administration of PCP
Sixty-two transcripts (Fig. 1C and Table 4) were differentially expressed in response to mechanical injury of the hippocampus (18) and PCP administration (Kaiser et al., unpublished observations). Upregulated transcripts represented 79% of the total overlapping transcript pool (49 of 62; Table 4). Included in this subset were genes involved in cell cycle control (e.g., Kras2 and Ddit3), cholesterol metabolism (e.g., Mvd and Idi1), inflammation (Ptgs2), transcription/translation (e.g., Sox11 and Narp), signal transduction (e.g., Sgk and Ptpn16), as well as receptors (e.g., Il6r and Nr4a2), channels/transporters (e.g., Slc2a1 and nucleoporin p45), growth factors (e.g., Bdnf and Pthlh), and cytoskeletal components (e.g., Gfap and Homer1). Four transcripts were downregulated (6%; Table 4) by hippocampal injury and after PCP administration: Cox8a, Alcam, Argbp2, and mGluR3. Transcripts upregulated in hippocampus by mechanical injury but downregulated in cortex by PCP (n = 5, 8%; Table 4) included Pcna, PRG1, Slc16a1, Kv8.1, and Igfbp2. Acute PCP upregulated four additional cortical transcripts (6%; Table 4) that were identified as downregulated in hippocampus after mechanical injury: Sult1a1, Pde2a2, Camkk1, and Klf9.
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Modeling Psychosis: Cortical Effects of Acute LSD, PCP, and METH Administration
Hallucinogens (e.g., LSD), N-methyl-D-aspartate receptor (NMDA-R) antagonists (e.g., PCP), and psychostimulants (e.g., METH) exert their psychotomimetic effects by acting primarily at specific targets within brain serotonin (5-HT), glutamate, and dopamine pathways, respectively. Considering that the behavioral effects of these drugs develop as a result of alterations in multiple neurotransmitter systems, some degree of phenomenological overlap between models is expected.
Acute LSD administration triggers behavioral symptoms that resemble some of those observed in schizophrenia (4). These effects are thought to result from interactions with 5-HT2A, 5-HT2C, and 5-HT1A receptor subtypes (6, 29) as well as dopamine receptors (12, 31). In the PFC, stimulation of 5-HT2A receptors (e.g., by LSD) increases the release of glutamate (1). Interestingly, many of the psychotomimetic effects of NMDA-R antagonists result from increased glutamatergic activity at non-NMDA glutamate receptors (19). Support for common glutamate-related pathways involved in the effects triggered by psychedelic hallucinogens (e.g., LSD) and NMDA-R antagonists (e.g., PCP) comes from studies reporting reversal of both PCP- and LSD-induced effects by metabotropic glutamate agonists (1, 17, 20).
In this respect, our analysis provides additional support for common paths activated in PFC after both LSD and PCP acute administration. We found that all five transcripts reported to be upregulated in cortex by LSD (Ania3, Egr2, Sgk, Nr4a3, and Nfkbia) (21) are also upregulated after PCP administration (Kaiser et al., unpublished observations; Fig. 1A and Table 2). These apparently divergent transcriptional changes may result from or reflect alterations in the following: 1) glutamatergic and dopaminergic signaling (i.e., Ania3, Egr2, Nr4a3), 2) glucocorticoid levels (i.e., Egr2, Sgk, Nr4a3), 3) cytoskeletal architecture and synaptic plasticity (i.e., Ania3), and 4) intracellular signal-transduction (i.e., Nfkbia). Based on these findings, we hypothesize that the pathways and specific genes identified in this analysis may be altered in psychosis and schizophrenia.
In contrast to LSD or PCP, acute METH administration rarely produces psychotic episodes in healthy humans. However, acute METH administration still induces behavioral alterations (e.g., euphoria, decreased need for sleep) that mimic many of the symptoms of mania and psychotic-mania (10). Given that amphetamines increase catecholaminergic activity, METH-induced effects may result primary as a consequence of increased dopaminergic activity (27).
No overlap in cortical gene expression patterns was found between LSD and METH data sets (Fig. 1A). Possible explanations for this observation include: 1) acute administration of LSD and METH primarily activate different neurotransmitter pathways, 2) few transcriptional changes were detected by both studies because of their conservative microarray data analysis approach, and 3) different evaluation time points were used in each study (LSD, 1.5 h; METH, 24 h). Consistent with the clinical observation that PCP models a greater range of schizophrenic symptoms than either LSD or METH (15), alterations induced by LSD and METH were also reported after PCP administration (Kaiser et al., unpublished observations). Interestingly, the only METH-induced change found to overlap with the pool of transcripts altered after PCP administration was the transcript encoding the circadian rhythm-related D-box binding protein (Dbp; Table 2).
In Dbp knockout mice (11), as well as in schizophrenia (3, 13, 26) and depression (5, 16), the duration of sleep time cycles and the consolidation of sleep episodes are reduced [i.e., decreased slow wave sleep (SWS)]. Since several lines of evidence suggest that enhanced dopamine activity reduces SWS (8, 14, 23), and PCP as well as METH are known to increase cortical dopaminergic activity, this raises the possibility that dopamine is capable of modulating cortical Dbp levels. Consistent with Dbp showing a strong circadian rhythm in cerebral cortex (32), it is not surprising to observe discrepancies in the direction of transcriptional change at different time points following drug administration (in PCP study, Dbp downregulated after 4 h; in METH study, Dbp upregulated after 24 h). Acute psychoactive drugs may also temporally phase-shift circadian rhythms (2, 30, 33) and/or exacerbate normal patterns of cortical Dbp expression. Based on the analyses of both the PCP and METH studies, we hypothesize that abnormalities in the dopaminergic system may contribute to the dysregulation of circadian rhythms reported in neuropsychiatric disease.
Traumatic Injury, Nerve Tissue, and Excitotoxicity
In nerve tissue, traumatic injury (18, 28) and acute ischemic-hypoxic episodes (25) trigger a series of cellular and molecular cascades that eventually result in increased neuronal hyperexcitability, irreversible cellular dysfunction, and cell death. Generalized membrane depolarization coupled to suppression of inhibitory synaptic mechanisms converge to enhance excitatory neurotransmitter release (e.g., glutamate) and its associated risk for excitotoxicity. Moreover, activation of second messenger pathways because of increased synaptic neurotransmission has acute and chronic functional effects on both neurons and nonneuronal cells (e.g., glial and blood-brain barrier cells). Enhanced second messenger levels are known to alter cellular physiology, cell survival, and cell morphology via several mechanisms, including the generation and metabolism of reactive oxygen species (ROS), activation of proteases, and release of neuropeptides and cytokines as well as of trophic factors. Second messengers can also alter gene expression and elicit long-lasting changes in synaptic and cellular function. In addition to genes related to nerve tissue dysfunction, expression of genes implicated in preserving and restoring function are also altered following injury (7).
Analysis of the data sets used to build the list of overlapping transcripts in Table 3 shows changes in gene expression as early as 3 h after hippocampus and/or spinal cord injury. Alterations in the expression of genes encoding proteins related to gene transcription (i.e., increased Junb, Nfkb1, Irf1, Egr1, Cebpg, c-fos) and inflammatory response (i.e., augmented cytokines Ccl3 and Cxcl2, Il1b, Selp, Hmox1), neurotransmitter control dysfunction (i.e., decreased presynaptic SNAP-25A and SNAP-25B), ionic imbalance and excitability regulation (i.e., Slc24a2, K+ voltage-gated channels: Kcnk1 and Kcnd2), as well as cytoskeletal (i.e., Nes, Mtap2, Nfl) and extracellular matrix (i.e., Icam1) reorganization are observed. By 1224 h after injury, in addition to these effects, endogenous attempts to repair and functionally stabilize the injured nerve tissue begin (i.e., alteration in Cpg21, Ccnl, Camkk1, Camk2b, Gria3, Gria1, Grm3, and Npy transcript levels). No transcripts associated with apoptotic mechanisms were identified as changed, probably because of the small number of cells expressing these transcripts and the relatively low sensitivity of the microarray approach (e.g., compared with real-time PCR). Interestingly, a recent study examining transcriptional changes induced by ischemia in cortex (25) also detected an alteration in a high percentage of the genes listed in Table 3 (22 of 28 transcripts = 79%; see Table 3, asterisks). This suggests that common mechanisms triggered after trauma may be related to ischemia and hypoxia rather than just the mechanical trauma per se. In addition to the expected injury-related phenomena (e.g., inflammation, altered neurotransmission, oxidative stress), this overlapping transcript list suggests that, independent of the nerve tissue affected (either peripheral or central), traumatic injury induces a common set of changes possibly in response to secondary ischemia/hypoxia. At the molecular level it also provides evidence for: 1) altered Ca2+ signaling (e.g., Camkk1, Camk2b), 2) impaired presynaptic terminals (e.g., SNAP-25A, SNAP-25B), 3) dysregulation of cellular excitability (e.g., Kcnd2, Kcnk1), and 4) possible alterations in pain perception (e.g., Npy).
Telencephalic Changes in Gene Expression Induced by Mechanical Injury and Psychotomimetics
As discussed in the previous sections, both traumatic injury and acute PCP administration induce the central release of excitatory neurotransmitters such as glutamate and dopamine. Therefore, enhanced levels of these excitotoxic neurotransmitters (in particular glutamate) may be responsible, directly or indirectly, for many of the transcriptional changes that occur in response to both traumatic injury and acute PCP (Table 4). Interestingly, only a few of these transcripts were also altered in response to cortical ischemia (25). Compared with the higher transcript overlap observed after traumatic injury of hippocampus and spinal cord, this observation suggests that little or no ischemia occurs in the cortex following acute PCP administration.
Most common alterations in gene expression induced after trauma and PCP administration (Table 4) are consistent with the induction of survival mechanisms aimed to reduce cytotoxicity and increase repair and structural reorganization. Alterations in these genes may reflect various factors including the induction of local inflammatory and stress-related events, alteration in glutamatergic and dopaminergic activity, Ca2+ signaling, changes in glucose and lipid metabolism, as well as involvement of brain nonneuronal cell types (e.g., astrocytes) and the blood-brain barrier.
Finally, we would like to speculate on the practical usefulness of this information. These two mechanistically different events (trauma and drug administration) are known to induce brain cell death. Considering that they trigger a common gene expression pattern in different brain areas, we propose that such a pattern could be relevant to the in vivo evaluation of treatments for injury (either mechanical or pharmacological) as well as early brain toxic effects of drug candidates. To both confirm and expand the conclusions derived from our analysis, followup studies will be necessary to identify and localize alterations in gene expression at the transcriptional, protein, and functional levels.
CONCLUSIONS
In this paper, we present evidence supporting the analysis of overlapping transcriptional changes for the identification of common mechanisms triggered by diverse events in different nerve tissues. From a practical point of view, defining overlapping patterns of gene expression allows a better understanding of the early stages in gene expression changes that are key in predictive biology. Applying this approach, for example, in toxicology, could lead to the detection of early molecular events (i.e., a common gene expression pattern) that constitute a biomarker of toxic cellular mechanisms, such as might be activated in response to induced glutamate excitotoxicity. Our analysis demonstrates the importance of interpreting transcript profiling data within the context of common pathways and mechanisms rather than solely specific to a given tissue or stimulus. Finally, we also highlight the significance of combining data derived from multiple microarray studies that examine diverse biological events for a broader interpretation of data from a particular microarray study.
![]() |
ACKNOWLEDGMENTS |
---|
![]() |
FOOTNOTES |
---|
Address for reprint requests and other correspondence: L. K. Nisenbaum, Neuroscience Discovery Research, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, IN 46285-0438 (E-mail: l.nisenbaum{at}lilly.com).
10.1152/physiolgenomics.00125.2003.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|