Cerebral, cerebellar, and brain stem gangliosides in mice susceptible to audiogenic seizures

Abstract— The quantitative and qualitative distribution of gangliosides was investigated in the cerebrum, cerebellum and brain stem of audiogenic seizure resistant (C57BL/6J) and susceptible (DBA/2J) mice at 21 days of age. The concentration of gangliosides (μg/unit weight) was higher in the DBA cerebrum and brain stem, but lower in the DBA cerebellum compared to the concentration in C57 mice. In general, the brain water content was lower in DBA mice than in C57 mice. The distributions of a number of gangliosides were found to be different between the two strains and the differences were often in the same direction across the three brain regions. The most consistant and significant difference in ganglioside pattern observed between the strains was the higher concentration of G_M1 in all three regions of the DBA brain. These results suggest that DBA mice have a more heavily myelinated CNS than C57 mice. The relationship of these observations to inherent audiogenic seizure susceptibility is discussed.     

Gangliosides in various brain areas of three inbred strains of mice

The ganglioside patterns of cerebellum, cortex, pons-medulla, hypothalamus, hippocampus and caudate nucleus of three inbred strains of mice (C57BL/6J, DBA/2J and BALB/cJ) have been analysed. All brain areas contained both the simple and complex species of gangliosides. GD1a was the major ganglioside in cortex, hippocampus and caudate nucleus whereas GT1b was the major species in cerebellum, hypothalamus and pons-medulla. In hippocampus, the percentages of GT1b and GD1a were quite similar. Pons and medulla exhibited the highest levels of GM1 (which approaches the value of GT1b) and the lowest values of GD1a. A ganglioside, termed here GT1L, was located between GD1b and GT1b. This ganglioside, which was present in highest amounts in cerebellum disappeared after alkali treatment. Highly significant differences were observed in the amounts and patterns of gangliosides among brain areas of the three strains. Highly significant differences (p<0.001) were also found in the ganglioside distribution of various brain areas among the strains, especially for tri-and tetrasialogangliosides between Balb and DBA. A significant difference of GM1 was observed in the cerebellum when comparing DBA with the two other strains. It is likely that the differences might be related to their relative abundances in certain cell types and for defining synaptic circuits in brain areas of some strains.     

Genetic analysis of barrel field size in the first somatosensory area (SI) in inbred and recombinant inbred strains of mice

We measured the combined area of posterior medial barrel subfield (PMBSF) and anterior lateral barrel subfield (ALBSF) areas in four common inbred strains (C3H/HeJ, A /J, C57BL /6J, DBA/2J), B6D2F1, and ten recombinant inbred (RI) strains generated from C57BL/6J and DBA/2J progenitors (BXD) as an initial attempt to examine the genetic influences underlying natural variation in barrel field size in adult mice. These two subfields are associated with the representation of the whisker pad and sinus hairs on the contralateral face. Using cytochrome oxidase labeling to visualize the barrel field, we measured the size of the combined subfields in each mouse strain. We also measured body weight and brain weight in each strain. We report that DBA/2J mice have a larger combined PMBSF/ALBSF area (6.15 +/- 0.10 mm(2), n = 7) than C57BL /6J (5.48 +/- 0.13 mm(2), n = 10), C3H/HeJ (5.37 +/- 0.16 mm(2), n = 10), and A/J mice (5.04 +/- 0.09 mm(2), n = 15), despite the fact that DBA/2J mice have smaller average brain and body sizes. This finding may reflect dissociation between systems that control brain size with those that regulate barrel field area. In addition, BXD strains (average n = 4) and parental strains showed considerable and continuous variation in PMBSF/ALBSF area, suggesting that this trait is polygenic. Furthermore, brain, body, and cortex weights have heritable differences between inbred strains and among BXD strains. PMBSF/ALBSF pattern appears similar among inbred and BXD strains, suggesting that somatosensory patterning reflects a common plan of organization. This data is an important first step in the quantitative genetic analysis of the parcellation of neocortex into diverse cytoarchitectonic zones that vary widely within and between species, and in identifying the genetic factors underlying barrel field size using quantitative trait locus (QTL) analyses.     

Inheritance of Lysosomal Acid β-Galactosidase Activity and Gangliosides in Crosses of DBA/2J and Knockout Mice

GM1 gangliosidosis is a progressive neurodegenerative disease caused by deficiencies in lysosomal acid beta-galactosidase (beta-gal) and involves accumulation and storage of ganglioside GM1 and its asialo form (GA1) in brain and visceral tissues. Similar to the infantile/juvenile human disease forms, B6/129Sv beta-gal knockout (ko) mice express residual tissue beta-gal activity and significant elevations of brain GM1, GA1, and total gangliosides. Previous studies suggested that inbred DBA/2J (D2) mice may model a mild form of the human disease since total brain ganglioside and GM1 concentration is higher while beta-gal specific activity is lower (by 70-80%) in D2 mice than in inbred C57BL/6J (B6) mice and other mouse strains. A developmental genetic analysis was conducted to determine if the genes encoding beta-gal (Bgl) in the D2 and the ko mice were functionally allelic and if the reduced brain beta-gal activity in D2 mice could account for elevations in total brain gangliosides and GM1. Crosses were made between D2 mice homozygous for the Bgld allele (d/d), and either B6/129Sv mice heterozygous for the Bgl+ allele (+/-) or homozygous for the ko Bgl- allele (-/-) to generate d/+ and d/- mice. Specific beta-gal activity (nmol/mg protein/h) showed additive inheritance in brain, liver, and kidney at juvenile (21 days) and adult (255 days) ages with the d/- mice having only about 16% of the beta-gal activity as that in the +/+ mice. These results indicate that the Bgl genes in the D2 and the ko mice are noncomplementing functional alleles. However, the d/- mice did not express GA1 and had total brain ganglioside and GM1 concentrations similar to those in the d/+ and +/+ mice. These results suggest that the reduced brain beta-gal activity alone cannot account for the elevation of total brain gangliosides and GM1 in the D2 mice.     

Genetic analysis of barrel field size in the first somatosensory area (SI) in inbred and recombinant inbred strains of mice

We measured the combined area of posterior medial barrel subfield (PMBSF) and anterior lateral barrel subfield (ALBSF) areas in four common inbred strains (C3H/HeJ, A?/J, C57BL?/6J, DBA/2J), B6D2F1, and ten recombinant inbred (RI) strains generated from C57BL/6J and DBA/2J progenitors (BXD) as an initial attempt to examine the genetic influences underlying natural variation in barrel field size in adult mice. These two subfields are associated with the representation of the whisker pad and sinus hairs on the contralateral face. Using cytochrome oxidase labeling to visualize the barrel field, we measured the size of the combined subfields in each mouse strain. We also measured body weight and brain weight in each strain. We report that DBA/2J mice have a larger combined PMBSF/ALBSF area (6.15?±?0.10?mm²,?n?=?7) than C57BL?/6J (5.48?±?0.13?mm²,?n?=?10), C3H/HeJ (5.37?±?0.16?mm²,?n?=?10), and A/J mice (5.04?±?0.09?mm²,?n?=?15), despite the fact that DBA/2J mice have smaller average brain and body sizes. This finding may reflect dissociation between systems that control brain size with those that regulate barrel field area. In addition, BXD strains (average n?=?4) and parental strains showed considerable and continuous variation in PMBSF/ALBSF area, suggesting that this trait is polygenic. Furthermore, brain, body, and cortex weights have heritable differences between inbred strains and among BXD strains. PMBSF/ALBSF pattern appears similar among inbred and BXD strains, suggesting that somatosensory patterning reflects a common plan of organization. This data is an important first step in the quantitative genetic analysis of the parcellation of neocortex into diverse cytoarchitectonic zones that vary widely within and between species, and in identifying the genetic factors underlying barrel field size using quantitative trait locus (QTL) analyses.     

Ganglioside expression in tissues of mice lacking the tumor necrosis factor receptor 1

This study presents a comparative analysis of gangliosides from lymphoid (spleen and thymus) and other tissues (brain, liver, lung, muscle) of C57BL/6 mice homozygous (-/-) and heterozygous (+/-) for the tumor necrosis factor receptor 1 (TNFRp55). Quantitative and qualitative differences in the expression of the lipid-bound N-acetylneuraminic (Neu5Ac) and N-glycolylneuraminic acid (Neu5Gc) and of various ganglioside biosynthesis pathways were detected between the tissues of the TNFRp55 -/- and the control TNFRp55 +/- mice. Sialic acid profiles showed a strong decrease in the absolute amount of sialic acids (Neu5Ac + Neu5Gc) in the lungs and thymus of homozygous (1.41 and 0.3 ng/mg wet weight, respectively) compared with control heterozygous animals (7.18 and 2.05 ng/mg wet weight, respectively). Considerable differences of Neu5Ac/Neu5Gc ratios in the lungs, muscle, spleen, and thymus were also detected. The gangliosides GM3(Neu5Ac) and GM3(Neu5Gc) were the dominant gangliosides in the lungs of the control animals, whereas the knockout mice almost completely lacked these structures in this organ. Reduced expression of GM1b-type gangliosides (GM1b and GalNAc-GM1b) was also found in the lungs, spleen, and thymus of the TNFRp55 knockout mice. On the other hand, neolacto-series gangliosides were more abundant in the lungs, brain, and muscle of the knockout mice, whereas their expression in the liver, spleen, and thymus was similar in both groups of animals. This study provides in vivo evidence that TNF signaling via the TNFRp55 is involved in the acquisition of a distinct ganglioside assembly in different mouse organs. TNFRp55 signaling seems to be especially important for the activation of the GM1b-type ganglioside biosynthetic pathway that is a unique characteristic of the mouse lymphoid tissues.     

Differences in liver ganglioside patterns in various inbred strains of mice

The ganglioside patterns in the liver of different inbred and hybrid strains of mice were investigated. The inbred strains were Balb/cAnNCr1BR, C57BL/6NCr1BR, DBA/2NCr1BR. C3H/HeNCr1BR; the hybrid strain was the Swiss albino. The following major gangliosides were found to be present in mouse liver: GM3-NeuAc; GM3-NeuGl, GM2 [a mixture of one species carrying N-acetylneuraminic acid (NeuAc) and one carrying N-glycollylneuraminic acid (NeuGl)], GM1 and GD1a-(NeuAc,NeuGl). The qualitative and quantitative patterns of liver gangliosides were markedly different in the various inbred strains of mice; in Balb/cAnNCr1BR strain, ganglioside GM2 was preponderant (99.2% of total ganglioside content); in C57BL/6NCr1BR, the major ganglioside was GM2 (90.4%), followed by GM3-NeuAc (5.6%) and GM3-NeuGl (4.0%); in DBA/2NCr1BR, GM2 accounted for 77.1%, GD1a-(NeuAc,NeuGl) 18.9% and GM1 3.1% of gangliosides; in C3H/HeNCr1BR, GM2 constituted 50.6%, GM1 22.8% and GD1a-(NeuAc,NeuGl) 22.1%. In the hybrid Swiss albino mice, liver ganglioside composition markedly varied from one animal to another, GM3-NeuGl, GM2 and GD1a-(NeuAc,NeuGl) being the predominant gangliosides in the various cases.     

Comparison of lipids in total brain tissue from five mouse genotypes.

Brain tissue from adult male and female mice of the C57BL/6J, C57BL/6J-AW-J, BALB/cJ, SJL/J, and DBA/2J genotypes was examined for brain weight, total protein, total lipid, cholesterol, phospholipid, plasmalogen, sulfatide, nonganglioside-glycolipid sphingosine, and ganglioside N-acetyl neuraminic acid, fatty acid, and sphingosine. No significant differences were found between sexes for any of these constituents. When compared to the overall average obtained for other animals, the DBA/2J, C57BL/6J-AW-J, and BALB/cJ mice contained lower quantities of plasmalogen and sulfatide compared to the overall averages obtained for the other genotypes. In addition, the sterol content in DBA/2J mice was significantly higher than the overall average value obtained for the other animals.     

Biochemical Study of Heterosis for Brain Myelin Content in Mice

Heterosis (hybrid vigor) for brain myelin content has been examined in detail in (C57BL/6J x DBA/2J)F1 hybrid mice at 17 days of age. The amount of myelin isolated from the F1 hybrid brain is greater than that isolated from either parental strain. In addition, the total protein content in the myelin of the three genotypes showed the following trend: F1 greater than DBA greater than C57. However, no discernible differences in myelin protein compositions could be detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Analysis of the whole brain for several myelin-associated constituents such as GM1 ganglioside, 2',3'-cyclic nucleotide 3'-phosphohydrolase (CNPase), 5'-nucleotidase, and carbonic anhydrase indicated that heterosis exists for these components. No heterosis was found for such nonmyelin constituents as gangliosides GD1a, GT, GQ, RNA, DNA, and choline acetyltransferase. A developmental study of the whole brain CNPase indicated that the heterotic effect was greatest during the most active period of myelination (17-30 days). We conclude that the heterotic effect is specific for myelin content and is probably the result of an accelerated myelin synthesis. The heterotic effect should have great potential as a new model for studying aspects of myelinogenesis.     

Genetic influences on the timing of puberty in mice

Genetic influences on the timing of three pubertal events--vaginal opening, first vaginal cornification, and onset of cyclicity--were studied in C57BL/6J, DBA/2J, and C3H/HeJ mice and in two F1 hybrid strains (B6D2F1 and B6C3HF1). Marked genotypic differences were found. Among inbred strains, differences in the onset of vaginal opening and first vaginal cornification (C3H less than DBA less than C57) did not parallel those for the onset of cyclicity (C3H much greater than DBA = C57). Compared to parental strains, F1 hybrid strains were intermediate for times of vaginal opening and first vaginal cornification, consistent with the model in which the genetic effects on the timing of these events are additive. By contrast, onset of cyclicity occurred significantly earlier in the F1 hybrids than in their parent strains, indicating heterosis for one or more genes specifying this event. Body weights also differed among the genotypes from weaning onward, but these differences were only partially correlated with the differences in the timing of the pubertal events. Thus, genetic influences other than those affecting body weight contribute to the differential timing of pubertal events in these mouse strains. These results reveal marked genetic variation in the timing of puberty, and indicate that the set of genes specifying the timing of vaginal opening and first vaginal cornification differs from those specifying the onset of cyclicity.     

Brain Adrenoceptor Binding Sites in Mice Susceptible (DBA/2J) and Resistant (C57 Bl/6) to Audiogenic Seizures

We have examined if the age-related susceptibility of DBA/2J mice to audiogenic seizures is the result of an abnormality in the number or sensitivity of brain adrenoceptors. The binding of alpha 1-, alpha 2-, and beta-adrenoceptor ligands to membranes prepared from whole brain or regions of brain of DBA/2J mice was measured at various ages, corresponding to the periods before, during, and after the maximal sensitivity to audiogenic seizures. For comparison, we have studied concurrently age-matched C57 Bl/6 mice, a strain resistant to audiogenic seizures at all ages. There was no difference in the binding of alpha 2- or beta-adrenoceptor ligands to whole brain membranes between the two strains of mice at any age. The maximal number of alpha 1-adrenoceptor binding sites was lower in whole brains of DBA/2J mice than of C57 Bl/6 mice at all ages studied except 13-15 days of age. The differences were small (maximally 17%) but were statistically significant at 21-23 days of age, the time of maximal sensitivity of DBA/2J mice to audiogenic seizures. No difference between the two strains was found in the number or affinity of alpha 1- or alpha 2-adrenoceptor binding sites at any age in any of the brain regions studied. The age-related susceptibility of DBA/2J mice to audiogenic seizures is not the result of an abnormality in number or sensitivity of alpha 2- or beta-adrenoceptor binding sites, but a reduced number of alpha 1-adrenoceptor binding sites may be involved.     

Genetic influences on oestrous cyclicity in mice: evidence that cycle length and frequency are differentially regulated.

Studies in C57BL/6J, DBA/2J and C3H/HeJ mice and in two F1 hybrid strains (B6D2F1 and B6C3HF1) 2-5 months old revealed marked genotypic differences among inbred strains. C57 mice had three times as many regular (3-6 days) cycles as DBA and C3H mice, due largely to fewer pseudopregnant-like (7-14 day) cycles. C57 had longer regular cycles than DBA and C3H mice. Although the frequencies of regular cycles of DBA and C3H mice were similar, the cycles of C3H mice were shorter than those of DBA mice. The results indicated that the genetic determinants of the frequency of regular cycles differ from those specifying cycle length. Frequency of regular cycles of F1 hybrids was either intermediate between the parent strains (B6D2F1) or similar to the C57 strain (B6C3HF1), suggesting that regular cycle frequency shows additive genetic variation in the former crosses, but mostly dominant variance in the latter background. Regular cycles were either shorter than in both parent strains (B6D2F1) or similar to one of them (B6C3HF1), indicating heterosis and dominance for genes specifying short cycles. Although the lack of reciprocal crosses meant that maternal effects and possible genomic imprinting effects could not be assessed, these results reveal marked genetic influences on cycle length and frequency and suggest that some of the genes specifying these two traits differ.     

Comparison of lipids in total brain tissue from five mouse genotypes

Brain tissue from adult male and female mice of the C57BL/6J, C57BL/6J-A^w—J, BALB/cJ, SJL/J, and DBA/2J genotypes was examined for brain weight, total protein, total lipid, cholesterol, phospholipid, plasmalogen, sulfatide, nonganglioside—glycolipid sphingosine, and ganglioside N-acetyl neuraminic acid, fatty acid, and sphingosine. No significant differences were found between sexes for any of these constituents. When compared to the overall average obtained for other animals, the DBA/2J, C57BL/6J-A^w−J, and BALB/cJ mice contained lower quantities of nonganglioside—glycolipid sphingosine. The DBA/2J and C57BL/6J-A^w−J mice also contained lower quantities of plasmalogen and sulfatide compared to the overall averages obtained for the other genotypes. In addition, the sterol content in DBA/2J mice was significantly higher than the overall average value obtained for the other animals.     

Analysis of behavior using genetical genomics in mice as a model: from alcohol preferences to gene expression differences.

Most familial behavioral phenotypes result from the complex interaction of multiple genes. Studies of such phenotypes involving human subjects are often inconclusive owing to complexity of causation and experimental limitations. Studies of animal models argue for the use of established genetic strains as a powerful tool for genetic dissection of behavioral disorders and have led to the identification of rare genes and genetic mechanisms implicated in such phenotypes. We have used microarrays to study global gene expression in adult brains of four genetic strains of mice (C57BL/6J, DBA/2J, A/J, and BALB/c). Our results demonstrate that different strains show expression differences for a number of genes in the brain, and that closely related strains have similar patterns of gene expression as compared with distantly related strains. In addition, among the 24 000 genes and ESTs on the microarray, 77 showed at least a 1.5-fold increase in the brains of C57BL/6J mice as compared with those of DBA/2J mice. These genes fall into such functional categories as gene regulation, metabolism, cell signaling, neurotransmitter transport, and DNA/RNA binding. The importance of these findings as a novel genetic resource and their use and application in the genetic analysis of complex behavioral phenotypes, susceptibilities, and responses to drugs and chemicals are discussed.     

NaCl taste thresholds in 13 inbred mouse strains

Molecular mechanisms of salty taste in mammals are not completely understood. We use genetic approaches to study these mechanisms. Previously, we developed a high-throughput procedure to measure NaCl taste thresholds, which involves conditioning mice to avoid LiCl and then examining avoidance of NaCl solutions presented in 48-h 2-bottle preference tests. Using this procedure, we measured NaCl taste thresholds of mice from 13 genealogically divergent inbred stains: 129P3/J, A/J, BALB/cByJ, C3H/HeJ, C57BL/6ByJ, C57BL/6J, CBA/J, CE/J, DBA/2J, FVB/NJ, NZB/BlNJ, PWK/PhJ, and SJL/J. We found substantial strain variation in NaCl taste thresholds: mice from the A/J and 129P3/J strains had high thresholds (were less sensitive), whereas mice from the BALB/cByJ, C57BL/6J, C57BL/6ByJ, CE/J, DBA/2J, NZB/BINJ, and SJL/J had low thresholds (were more sensitive). NaCl taste thresholds measured in this study did not significantly correlate with NaCl preferences or amiloride sensitivity of chorda tympani nerve responses to NaCl determined in the same strains in other studies. To examine whether strain differences in NaCl taste thresholds could have been affected by variation in learning ability or sensitivity to toxic effects of LiCl, we used the same method to measure citric acid taste thresholds in 4 inbred strains with large differences in NaCl taste thresholds but similar acid sensitivity in preference tests (129P3/J, A/J, C57BL/6J, and DBA/2J). Citric acid taste thresholds were similar in these 4 strains. This suggests that our technique measures taste quality-specific thresholds that are likely to represent differences in peripheral taste responsiveness. The strain differences in NaCl taste sensitivity found in this study provide a basis for genetic analysis of this phenotype.     

Genetically determined differences in ethanol sensitivity influenced by body temperature during intoxication

The present study investigated the importance of body temperature during intoxication in mediating differences between five inbred strains of mice (C57BL/6J; BALB/cJ; DBA/2J; A/HeJ; 129/J) in their acute sensitivity to the hypnotic effects of ethanol. Mice exposed to 22 degrees C after ethanol injection became hypothermic and exhibited statistically significant differences between strains in rectal temperatures at the return of the righting reflex (RORR), duration of loss of the righting reflex (LORR), and blood and brain ethanol concentrations at RORR. Exposure to 34 degrees C after injection offset ethanol-hypothermia and markedly reduced strain-related differences in rectal temperatures and blood and brain ethanol concentrations at RORR. Brain ethanol concentrations at RORR were significantly lower in C57, BALB, DBA and A/He mice exposed to 34 degrees C compared to mice exposed to 22 degrees C during intoxication suggesting that offsetting hypothermia increased ethanol sensitivity in these strains. Taken with previous in vitro studies, these results suggest that genetically determined differences in acute sensitivity to the behavioral effects of ethanol reflect differences in body temperature during intoxication as well as differences in sensitivity to the initial actions of ethanol at the cellular level.     

Activity and electrophoretic profiles of brain aldehyde dehydrogenases in mice genetically selected for their ethanol preference

The activities and electrophoretic profiles of NAD⁺-aldehyde dehydrogenases (EC 1.2.1.3, ALDHs) in the different brain regions from the mice of inbred C57BL 6J strain (alcohol preferring) and DBA/2 sirain (alcohol avoicling) were studied.

• 1.1. Kinetic studies on ALDH of the different brain regions revealed the existence of at least two ALDHs in both strains.
• 2.2. The ALDH activity in the cerebellum, which was measured spectrophotometrically with 0.1 and 5mM acetaldehyde as the substrates, was the highest of the three regions and it was about 2-fold higher than that in the cortex in both strains.
• 3.3. The ALDH activities of mice of DBA/2 strain were higher than those ol mice of C57BL/6J strain in all the regions. Only in the cortex the F₁ hybrids (intermediate alcohol preferring) between C57BL/6J and DBA/2 strains had intermediate ALDH activity between their parental strains.
• 4.4. Electrophoretic analysis of cortex ALDH revealed that strain differences were observed in the isozyme pattern between pH 7.2 and 7.8 with cither acetaldehyde or propionaldehyde as substrate. These findings suggest that the ALDH activity in the cerebral cortex could be related to alcohol preference.

     

Gamma aminobutyric acid in different strains of mice effect of ethanol

Four strains of mice (C57BL/6J, DBA/2J, LS and SS), which differ in their voluntary intake and/or neural sensitivity to ethanol, were examined for the contents of γ-aminobutyric acid (GABA) in various brain regions. The effect of an acute dose of ethanol (4 g/kg) on the contents of this amino acid neurotransmitter in these mice were studied. The objective is to determine if the differences in sensitivity to ethanol may be reflected in differences in brain GABA contents in these animals.Results indicate that a previously documented ethanol-induced elevation of GABA in the whole mouse brain is a phenomenon observed in a variety of brain regions. No correlation was observed between GABA contents in the brain and neural sensitivity to ethanol.     

Complex control of GABA(A) receptor subunit mRNA expression: variation, covariation, and genetic regulation

GABA type-A receptors are essential for fast inhibitory neurotransmission and are critical in brain function. Surprisingly, expression of receptor subunits is highly variable among individuals, but the cause and impact of this fluctuation remains unknown. We have studied sources of variation for all 19 receptor subunits using massive expression data sets collected across multiple brain regions and platforms in mice and humans. Expression of Gabra1, Gabra2, Gabrb2, Gabrb3, and Gabrg2 is highly variable and heritable among the large cohort of BXD strains derived from crosses of fully sequenced parents--C57BL/6J and DBA/2J. Genetic control of these subunits is complex and highly dependent on tissue and mRNA region. Remarkably, this high variation is generally not linked to phenotypic differences. The single exception is Gabrb3, a locus that is linked to anxiety. We identified upstream genetic loci that influence subunit expression, including three unlinked regions of chromosome 5 that modulate the expression of nine subunits in hippocampus, and that are also associated with multiple phenotypes. Candidate genes within these loci include, Naaa, Nos1, and Zkscan1. We confirmed a high level of coexpression for subunits comprising the major channel--Gabra1, Gabrb2, and Gabrg2--and identified conserved members of this expression network in mice and humans. Gucy1a3, Gucy1b3, and Lis1 are novel and conserved associates of multiple subunits that are involved in inhibitory signaling. Finally, proximal and distal regions of the 3' UTRs of single subunits have remarkably independent expression patterns in both species. However, corresponding regions of different subunits often show congruent genetic control and coexpression (proximal-to-proximal or distal-to-distal), even in the absence of sequence homology. Our findings identify novel sources of variation that modulate subunit expression and highlight the extraordinary capacity of biological networks to buffer 4-100 fold differences in mRNA levels.