Insulin-like growth factor binding protein-2 binding to extracellular matrix plays a critical role in neuroblastoma cell proliferation, migration, and invasion.

IGF binding proteins (IGFBPs) modulate IGF cellular bioavailability and may directly regulate tumor growth and invasion. We have previously shown that IGFBP-2 binds and localizes IGF-I to the pericellular matrix and have provided some evidence suggesting that the heparin binding domain (HBD) or the arginine-glycine-aspartic acid (RGD) integrin binding motif may be involved in these interactions. However, the precise mechanisms involved remain to be elucidated. We therefore mutated the HBD or RGD sequence of IGFBP-2 and investigated consequent effects on extracellular matrix (ECM) binding, IGF-induced proliferation, and migration of neuroblastoma cells. IGFBP-2 and its arginine-glycine-glutamic acid (RGE) mutant similarly bound ECM components, whereas binding of mutant HBD-IGFBP-2 to each of the ECM substrates was markedly reduced by 70-80% (P < 0.05). IGF-I (100 ng/ml) increased incorporation of 3H-thymidine in neuroblastoma SK-N-SHEP cells by approximately 30%, an effect blunted by exogenously added native or either mutant IGFBP-2. Overexpression of IGFBP-2 and its RGE mutant potently promoted SHEP cell proliferation (5-fold), whereas SHEP cell proliferation was negligible when HBD-IGFBP-2 was overexpressed. Addition or overexpression of IGFBP-2 and its RGE mutant potently (P < 0.05) enhanced SHEP cell migration/invasion through the ECM. However, overexpression of the HBD-IGFBP-2 mutant potently inhibited (50-60%) SHEP cell invasion through ECM. Thus, IGFBP-2, which binds to the ECM, enhances proliferation and metastatic behavior of neuroblastoma cells, functions that directly or indirectly use the HBD but not the integrin binding sequence. Our novel findings thus point to a key role for the HBD of IGFBP-2 in the control and regulation of neuroblastoma growth and invasion.

Polymorphisms in the Il12b gene affect structure and expression of IL-12 in NOD and other autoimmune-prone mouse strains.

Interleukin (Il)-12 is a heterodimeric cytokine composed of 35 and 40 kD chains that plays a key role in the induction of Th1 cells, a T cell subset involved in many autoimmune diseases. We report here the cDNA sequence encoding the IL-12 p40 subunit from the autoimmune-prone non-obese diabetic (NOD) mouse, which spontaneously develops type 1 diabetes. Compared with the C57BL/6 sequence, there are two base changes that lead to amino acid replacements. Other autoimmune-prone strains, but not the diabetes-resistant NOR strain, share the same allele as NOD. We found both trans- and cis- allele-dependent effects on levels of basal and induced IL-12p40 expression. Furthermore, we show that one of these changes results in a structural change in the p40 molecule, as evidenced by the failure of a monoclonal antibody to bind NOD IL-12. These findings have implications for the predisposition to autoimmune responses in NOD and other autoimmune-prone mouse strains.

Linkage disequilibrium of a type 1 diabetes susceptibility locus with a regulatory IL12B allele.

Type 1 diabetes (T1D; or insulin-dependent diabetes mellitus, IDDM) is an autoimmune disease with both genetic and environmental components. In addition to the human leukocyte antigen (HLA) complex, the single major genetic contributor of susceptibility, an unknown number of other unidentified genes are required to mediate disease. Although many loci conferring susceptibility to T1D have been mapped, their identification has proven problematic due to the complex nature of this disease. Our strategy for finding T1D susceptibility genes has been to test for human homologues of loci implicated in diabetes-prone NOD (non-obese diabetic) mice, together with application of biologically relevant stratification methods. We report here a new susceptibility locus, IDDM18, located near the interleukin-12 (IL-12)p40 gene, IL12B. Significant bias in transmission of IL12B alleles was observed in affected sibpairs and was confirmed in an independent cohort of simplex families. A single base change in the 3' UTR showed strong linkage disequilibrium with the T1D susceptibility locus. The IL12B 3' UTR alleles showed different levels of expression in cell lines. Variation in IL-12p40 production may influence T-cell responses crucial for either mediating or protecting against this and other autoimmune diseases.

Evidence for two distinct classes of high affinity growth hormone binding proteins in pregnant rat serum.

These studies have established the presence of two major classes of high affinity growth hormone binding proteins in pregnant rat serum, designated GHBPa and GHBPb, with apparent native Mr of 257 K and 98 K respectively. GHBPa, which has not been identified previously, exhibits a binding affinity (2-5 nM(-1)) that is up to 20-fold higher than GHBPb (0.2-0.8 nM(-1)) and is the least abundant form, being approximately 15-20% of total serum GH-binding capacity. Western immunoblot analysis revealed that each GHBP is composed of several immunoreactive proteins which were reactive with carboxy-terminal (RB1615) and/or N-terminal (MAb263) domain antibodies, suggesting the presence of GHBPs with and without the hydrophilic tail. Of importance is that GHBPa exhibited significantly higher Mr (78-182 K, +DTT) than that predicted by GHBP cloning, suggesting that they may be covalently bound to other non-GH-binding proteins or may be distinct entities. GHBPb, on the other hand, was composed of smaller Mr (43/48 K, +DTT) "hydrophilic" tail-containing proteins, some of which were disulphide linked to a larger complex of approximately 110 K. These novel findings challenge the current view of the mechanism for generation of the rat serum GHBP and raise the intriguing possibility that the two classes of GHBP may play distinct and important roles in GH physiology.

Guinea pig serum contains a specific high affinity growth hormone-binding protein with novel ligand specificity.

Previous workers have suggested that guinea pig serum does not contain a GH-binding protein (GHBP) or that it is defective. The current studies, however, have identified and characterized the presence of GH-binding activity in guinea pig serum using gel chromatography to separate bound and free hormone. The detection of GH-binding activity is critically reliant on the type of radioligand used to measure binding. Clear identification of GH-binding activity was demonstrated with [125I]ovine GH (oGH), but specific binding could not be measured with [125I]human GH. The novel specificity was also shared by guinea pig liver membrane GH receptor (GHR) and cytosol GHBP, suggesting structural similarity in the GH-binding domain between the GHR and soluble GHBPs. The binding of oGH was dependent on serum concentration (5 microl serum produced 16.03 +/- 0.5% specific binding; mean +/- SEM; n = 11) and incubation time (equilibrium was reached by approximately 6 h at 21 C) and was completely reversible (t(1/2), approximately 2 h). Scatchard analysis revealed linear plots with an affinity constant (Ka) of 0.59 +/- 0.09 x 10(9) M(-1) and a capacity of 23,181 +/- 4,474 fmol/ml serum. Similar association constants were obtained for liver membrane GHR (0.79 +/- 0.22 x 10(9) M(-1)) and cytosol GHBPs (0.99 +/- 0.15 x 10(9) M(-1)), but the capacity, when expressed as femtomoles per g tissue, was significantly increased (4-fold) in cytosol (4,303 +/- 505) over that in membranes (1,071 +/- 257). There was no sex difference in Ka or level of GHBP in guinea pig serum. Surprisingly, the level of GH-binding activity was very low to undetectable in pregnant guinea pig serum. Characterization of the native structure of guinea pig GHBPs has indicated the presence of several proteins that are structurally distinct. Although the distribution of GH-binding activity covered a large Mr range (approximately 70-350 kDa) the major form of the circulating GHBP identified by gel chromatography had an apparent native Mr of 150-170 kDa. Partially purified GHBP (approximate Mr, 170 kDa) was covalently cross-linked to [125I]oGH and subjected to nonreducing SDS-PAGE. Specific GHBP complexes of 158 and 85 kDa were detected, suggesting that the partially purified GHBP complex may be composed of a smaller GHBP associated noncovalently with a non-GH-binding protein. "Pore limit" native PAGE (cathodic and anodic) revealed the presence of specific GHBPs of 363, 158, 74, and 55 kDa, which cross-hybridized with the rat liver membrane GHR monoclonal antibody mAb 263 but not with the rat serum GHBP-specific mAb 4.3. Interestingly, although GH binding was undetectable in pregnant guinea pig serum, Western immunoblot analysis with mAb 263 demonstrated the presence of a major immunoreactive GHBP band of 105 kDa in addition to 158- and 55-kDa GHBPs. The data indicate that the GHBPs are immunologically related to the rat membrane GHR, but provide no evidence to support the presence of a hydrophilic tail sequence homologous to that in the rat GHBP. These studies have identified in guinea pig serum GHBPs that exhibit novel ligand specificity, structural heterogeneity, and an immunological relationship to the rat liver membrane GHR. The identification of serum GHBP and the novel ligand specificity, which is also expressed by the liver membrane GHR, argue against the view that the guinea pig has a defective GHBP.

Differential effects of histamine on the N-methyl-D-aspartate channel in hippocampal slices and cultures.

The effect of histamine on N-methyl-D-aspartate currents was investigated in pyramidal neurons in the CA1 region of acute hippocampal slices from juvenile rats. The objective was to compare histamine effects in the slice with those previously reported in acutely dissociated and cultured hippocampal neurons. Micromolar concentrations of histamine had no effect on N-methyl-D-aspartate mediated excitatory postsynaptic currents in the slice, in contrast to the large enhancement seen in culture under identical conditions. However, millimolar concentrations of histamine blocked these currents both in the slice and in culture. Possible reasons for the lack of enhancement in the slice were explored as follows. (1) Histamine could not penetrate the slice or was already present at high concentrations inside the slice. This was tested by recording N-methyl-D-aspartate currents elicited in outside-out patches pulled from the somas of CA1 slice neurons. Histamine still had no effect in patches, whereas the corresponding experiment for cultured neurons showed robust enhancement. (2) Slices release an endogenous ligand that binds with high affinity to the histamine site on the N-methyl-D-aspartate receptor, blocking its activation. This was tested by superfusing cultures with supernatant from homogenized slice tissue. Histamine enhancement was maintained in these cultures. (3) CA1 slices and cultures express different N-methyl-D-aspartate receptor subtypes. The reverse transcription-polymerase chain reaction technique was used to examine the expression of messenger RNA encoding N-methyl-D-aspartate receptor subunits in the two systems. No difference was found in the whole-tissue expression of messenger RNA for the NR2A, 2B or 2C subunits or for the eight known splice variants of the NR1 subunit. It is hypothesized that the differential enhancing effect of histamine in slices and culture involves posttranslational modifications or other factors that modulate the N-methyl-D-aspartate receptor/ion channel according to its environment.

In vivo chromatin structure of the murine interleukin-5 gene region: a new intact cell system.

The study of chromatin involvement in the regulation of gene expression has traditionally required the isolation of nuclei. However, cell fractionation techniques are subject to redistribution of proteins during the isolation procedure, which prevents rigorous physiologically relevant analysis. To eliminate the need to isolate nuclei and to analyze chromatin structures in vivo in response to agents regulating murine interleukin-5 (IL-5) gene activation, we have established a novel lysolecithin permeabilized intact cell system for suspension cell types, in this case T cells. Nuclear integrity of permeabilized cells is demonstrated by nuclear transport assays using confocal laser scanning microscopy. Results are identical in unstimulated and stimulated T cells, indicating that the chromatin structure after activation is not the result of gross alterations in nuclear protein transport properties. Potential new IL-5 gene regulatory regions are identified by DNase I hypersensitivity mapping. Our lysolecithin permeabilized intact cell system is amenable to physiologically relevant analysis of responses to signaling pathways at the level of chromatin, nuclear protein translocation and possibly other cellular functions in a variety of suspension and adherent cell types.

Growth hormone receptor/binding protein: physiology and function.

Soluble truncated forms of the growth hormone receptor (GHR) are present in the circulation of many species and are also produced by many tissues/cell types. The major high-affinity forms of these GH-binding proteins (GHBP) are derived by alternative splicing of GHR mRNA in rodents, but probably by proteolytic cleavage in other species. Questions still remain with respect to the origins, native molecular form(s), physiology, and function of the GHBPs, however. The observation that GH induces dimerization of the soluble GHBP and membrane GHR, and that dimerization of GHR appears to be critical for GH bioactivity suggests that the presentation of GH to target cells, in an unbound form or as a monomeric or dimeric complex with GHBP, may have significant implications for the ability of GH to activate specific postreceptor signaling pathways (tyrosine kinase, protein kinase C, G-protein pathways) known to be utilized by GH for its diverse biological effects. This minireview addresses some of these aspects and highlights several new questions which have arisen as a result of recent advances in our understanding of the structure, function, and signaling mechanisms of the membrane bound GHR.

Developmental expression of the growth hormone receptor gene in rabbit tissues.

Total RNA from several adult (6-18-month-old) rabbit tissues was characterized using an oligonucleotide probe derived from the extracellular domain of the nucleotide sequence of the rabbit growth hormone receptor (GH-R) cDNA. Multiple GH-R mRNA species of approximately 4.6, approximately 3.3, 2.1 and approximately 1.4 kb were detected. The major 4.6 kb transcript was detectable in all tissues examined but with quite marked abundance differences. The highest level of expression was observed in liver followed closely by muscle. A qualitative assessment of insulin-like growth factor I (IGF-I) mRNA abundance was made in these same tissues. The data showed that the tissue abundance of GH-R mRNA was not necessarily parallel to that of IGF-I mRNA. The ontogeny of GH-R mRNA was studied in rabbit liver, muscle, heart and kidney. Low levels of GH-R mRNA were detectable in all fetal tissues studied except kidney which showed relatively high levels, suggesting that GH may play an important role in early kidney development. The overall developmental pattern of GH-R mRNA was similar in heart, muscle and liver, being low in fetal and early neonatal (day 3) periods and reaching maximal levels between 2 and 6 months. However, in kidney the pattern contrasted markedly. Relatively high levels of GH-R mRNA were observed in fetal and early neonatal (day 3) kidney with little change throughout development. The developmental pattern of IGF-I gene expression was not necessarily co-ordinately regulated with the ontogenic pattern of GH-R gene expression.(ABSTRACT TRUNCATED AT 250 WORDS)

Serum binding proteins for growth hormone: origins, regulation of gene expression and possible roles.

The above discussion highlights the heterogeneity of the family of GH receptors/GHBPs and their mRNAs. Considerable uncertainty still exists as to the interrelationships between the various forms, the specific mechanisms for their generation and their possible significance in terms of modulating GH action and receptor function. Collectively, the regulatory data indicate that while the mRNAs encoding the membrane and soluble GH receptors/GHBPs in the rat are expressed by the same broad distribution of tissues, they can be differentially regulated. Such regulated expression implies a functional basis for production of GHBP. The wide tissue distribution of GHBP mRNA also suggests a role for GHBP as a paracrine/autocrine effector molecule, perhaps in addition to an endocrine role. Additional studies, both in vitro and in vivo, perhaps utilizing highly purified recombinant GHBP, will be required to provide more definitive information as to the true physiological role(s) of the circulating GHBPs.

Does the serum binding protein for growth hormone have a functional role?

The presence in the serum of several animal species of highly specific binding proteins for growth hormone is now accepted. The major binding protein is a truncated, variant form of the target tissue GH receptor and is synthesized by all tissues expressing the full-length GH receptor. The GH receptor and GH binding protein are not co-ordinately expressed, being produced in variable ratios between tissues. Such independence of expression suggests that the synthesis of the GH binding protein is regulated and therefore may be biologically relevant. The GH binding protein has been shown to increase the half-life of circulating GH by decreasing the metabolic clearance rate and degradation rate; to limit the volume of distribution of bound GH to approximately twice the intravascular space; and to inhibit receptor binding and biological actions of GH in in vitro model systems. These observations suggest that circulating GH binding protein does have a functional role in modulating both the circulating concentrations of GH and its availability and effectiveness for target tissue interaction. Other possible, but as yet unproven, functions for the GH binding protein are briefly discussed. The ability to produce GH binding protein by recombinant means should allow additional, more definitive studies to be performed, thereby providing greater opportunity to define the true biological role(s) of the GH binding protein in the metabolism and function of GH.

Structural and functional characterization of the gamma 1 subunit of GABAA/benzodiazepine receptors.

The GABAA receptor gamma 1 subunit of human, rat and bovine origin was molecularly cloned and compared with the gamma 2 subunit in structure and function. Both gamma subunit variants share 74% sequence similarity and are prominently synthesized in often distinct areas of the central nervous system as documented by in situ hybridization. When co-expressed with alpha and beta subunits in Xenopus oocytes and mammalian cells, the gamma variants mediate the potentiation of GABA evoked currents by benzodiazepines and help generate high-affinity binding sites for these drugs. However, these sites show disparate pharmacological properties which, for receptors assembled from alpha 1, beta 1 and gamma 1 subunits, are characterized by the conspicuous loss in affinity for neutral antagonists (e.g. flumazenil) and negative modulators (e.g. DMCM). These findings reveal a pronounced effect of gamma subunit variants on GABAA/benzodiazepine receptor pharmacology.

Differential benzodiazepine pharmacology of mammalian recombinant GABAA receptors.

We compared gamma-aminobutyric acid (GABA)-activated currents and their modulation by benzodiazepines in cultured human cells transfected with complementary desoxyribonucleic acid (cDNA) encoding different GABAA receptor subunits. Flunitrazepam, a benzodiazepine agonist which potentiates GABA responses in both neurons and astrocytes was only effective in receptors containing the gamma 2 subunit (alpha 1 beta 1 gamma 2 and alpha 5 beta 1 gamma 2). The beta-carboline methyl-4-ethyl-6,7-dimethoxy-beta-carboline-3-carboxylate (DMCM) decreased GABA-activated currents in receptors composed of alpha 1 beta 1 gamma 1 and alpha 1 beta 1 gamma 2 subunits but increased GABA-activated currents in receptors containing the alpha 5 subunit (alpha 5 beta 1 gamma 1 and alpha 5 beta 1 gamma 2). These results strongly suggest that flunitrazepam and DMCM do not act on isosteric sites and that differences in the responsiveness of GABAA receptors to these compounds are based on different subunit compositions of GABAA receptors.

Functional properties of recombinant rat GABAA receptors depend upon subunit composition.

GABA-gated chloride channels were expressed in human embryonic kidney cells following transfection of cDNAs encoding the alpha 1, beta 2, and gamma 2 subunits of the rat GABAA receptor (GABAR). Functional properties were determined using patch-clamp techniques in the whole-cell and outside-out configurations. Large whole-cell currents were observed in cells expressing the alpha 1 beta 2, alpha 1 gamma 2, and alpha 1 beta 2 gamma 2 subunit combinations. The unique characteristics of GABAR channels consisting of these subunit combinations depended upon the presence or absence of beta 2 and gamma 3 subunits. GABA-activated currents in cells expressing GABARs with the beta 2 subunit desensitized faster and showed greater outward rectification, and the channels had a shorter mean open time than GABARs composed of alpha 1 gamma 2 subunits. When the gamma 2 subunit was present the resulting GABAR channels had a larger conductance. The slope of the concentration-response curve was significantly steeper for GABARs composed of alpha 1 beta 2 gamma 2 subunits compared with GABARs consisting of alpha 1 beta 2 or alpha 1 gamma 2 subunit combinations.

Sequence and expression of a novel GABAA receptor alpha subunit.

Cloned cDNA encoding the bovine alpha 4 subunit of the GABAA receptor has been isolated. The predicted 51 amino acid long mature protein contains an exceptionally long intracellular domain and shares 53-56% sequence similarity to the previously characterized alpha 1, alpha 2 and alpha 3 subunits. Co-expression of alpha 4 and beta 1 in Xenopus oocytes resulted in the formation of GABA-gated chloride channels with expected pharmacology, although no benzodiazepine potentiation was observed. Northern analysis indicates that a 4 kb alpha 4 mRNA is expressed in the calf cerebellum, cortex and hippocampus but is barely detectable in the rat brain.

Binding sites for growth hormone in rabbit placental cytosol.

A soluble GH-binding protein, which cross-reacted with a monoclonal antibody (Mab) to the rabbit liver membrane GH receptor, has been identified in cytosol preparations from both fetal and maternal portions of rabbit placenta. Structural studies using gel filtration chromatography and chemical covalent cross-linking techniques have shown that the GH-binding protein in fetal/maternal placental cytosol has a native mol wt of 104,000 and a denatured subunit of about 57,000 mol wt (with or without dithiothreitol). Very low levels of GH-specific [125I]human (h) GH binding were observed in membrane preparations from the corresponding placentae, even after desaturation of any endogenously bound hormone by 5 M MgCl2. No PRL-specific binding was observed in placental membranes or cytosols. Scatchard analysis of [125I]hGH binding to fetal and maternal placental cytosol revealed linear plots with Ka values of 6.1 +/- 1.1 nM-1 (fetal) and 5.31 +/- 0.63 nM-1 (maternal; mean +/- SEM; n = 5). The binding capacity of maternal placental cytosol, when expressed as femtomoles per mg protein (170 +/- 10.10) or femtomoles per g tissue (3245 +/- 123), was about 3-fold higher than that for fetal placental cytosol. Northern blot analysis of fetal and maternal placental mRNA probed with a GH receptor oligonucleotide probe revealed hybridization to a 4.4 to 4.7-kilobase and a 2.2-kilobase species in fetal placenta only. The level of GH-specific binding observed in fetal and maternal placental cytosol did not correlate directly with the level of mRNA expression. A GH-binding protein has also been shown to be present in fetal rabbit serum and is known to be structurally and immunologically related to the rabbit placental and liver cytosolic GH-binding proteins. Scatchard analysis of [125I]hGH binding to fetal serum GH-binding protein revealed a single class of high affinity sites with a Ka of 4.64 +/- 1.29 nM-1 and a capacity of 338 +/- 167 fmol/ml serum (mean +/- SEM; n = 4). Given the relative binding capacities and the demonstration of GH receptor mRNA in fetal placental cytosol, it is highly unlikely that contamination of fetal placental cytosol by fetal serum accounts for all of the placental binding capacity observed. However, no such definitive conclusion regarding contamination by maternal rabbit serum of maternal placental cytosol can be made. The presence of GH-binding proteins in placental cytosol has not been described previously, and these observations suggest that GH may have a role in placental metabolism.

Differences in the developmental patterns of somatotrophic and lactogenic receptors in rabbit liver cytosol.

These studies have examined the ontogeny of specific GH- and PRL-binding proteins in rabbit liver cytosol across the fetal, early neonatal, and adult periods. The precise hormonal specificity of binding of the somatotropic/lactogenic ligand, [125I]human GH, was determined by displacement with monoclonal antibodies against either rabbit mammary gland PRL receptors or liver GH receptors. The developmental changes were evaluated by gel filtration chromatography, which allowed a measure of both the extent of binding and the molecular size (Mr). Scatchard analysis indicated that fetal liver cytosol contained mostly high affinity PRL receptors (Ka, 13.78 +/- 0.98 nM-1; capacity, 127 +/- 24.0 fmol/g tissue; mean +/- SEM; n = 5) and was low in GH-specific binding. This contrasts markedly with adult liver cytosol, which is a rich source of GH receptors (Ka, 2.44 nM-1; capacity, 20,765 fmol/g tissue), but is low in PRL-specific binding. Despite the deficiency of GH receptors in fetal liver, an abundant receptor-like GH-binding protein was present in fetal serum. Of several fetal tissue cytosols examined, only the placenta contained significant GH-specific binding; no tissues other than liver contained PRL-specific binding. After parturition PRL receptors in liver cytosol increased and predominated up until day 3 (Ka, 27.97 +/- 6.27 nM-1; capacity, 495 +/- 95.09 fmol/g tissue; n = 7), a period during which GH-specific binding was increasing only slightly. By day 6 a striking switch-over occurred, and GH receptors, which had a 4-fold lower affinity, became predominant (Ka, 6.89 +/- 0.63 nM-1; capacity, 600 +/- 47 fmol/g tissue; n = 5), while PRL-specific binding fell dramatically to levels observed in adult liver cytosol. After day 6 GH receptors increased steadily, reaching the high levels observed in adulthood by 2 months (Ka, 3.95 nM-1; capacity, 16,300 fmol/g tissue; n = 2), while PRL-specific binding appeared to change little. Structural and immunological analyses of the cytosolic GH and PRL receptors in the fetal/early neonatal period revealed similarities with the adult liver cytosolic GH receptor and mammary gland cytosolic PRL receptor, respectively. The increase in GH receptors on day 6 was clearly illustrated by cross-linking studies which showed the emergence of a GH-specific binding protein structurally distinct from PRL-specific binding proteins. These studies have demonstrated that in the rabbit major changes occur in the GH/PRL receptor profile during the early period of growth and suggest that important developmental changes occur in the requirement for GH and/or PRL action during this period.

GABAA receptor beta subunit heterogeneity: functional expression of cloned cDNAs.

Cloned cDNAs encoding two new beta subunits of the rat and bovine GABAA receptor have been isolated using a degenerate oligonucleotide probe based on a highly conserved peptide sequence in the second transmembrane domain of GABAA receptor subunits. The beta 2 and beta 3 subunits share approximately 72% sequence identity with the previously characterized beta 1 polypeptide. Northern analysis showed that both beta 2 and beta 3 mRNAs are more abundant in the brain than beta 1 mRNA. All three beta subunit encoding cDNAs were also identified in a library constructed from adrenal medulla RNA. Each beta subunit, when co-expressed in Xenopus oocytes with an alpha subunit, forms functional GABAA receptors. These results, together with the known alpha subunit heterogeneity, suggest that a variety of related but functionally distinct GABAA receptor subtypes are generated by different subunit combinations.