The liver is an important target organ of thyroid hormone. However, only a limited number of hepatic target genes have been identified, and little is known about the pattern of their regulation by thyroid hormone. We used a quantitative fluorescent cDNA microarray to identify novel hepatic genes regulated by thyroid hormone. Fluorescent-labeled cDNA prepared from hepatic RNA of T3-treated and hypothyroid mice was hybridized to a cDNA microarray, representing 2225 different mouse genes, followed by computer analysis to compare relative changes in gene expression. Fifty five genes, 45 not previously known to be thyroid hormone-responsive genes, were found to be regulated by thyroid hormone. Among them, 14 were positively regulated by thyroid hormone, and unexpectedly, 41 were negatively regulated. The expression of 8 of these genes was confirmed by Northern blot analyses. Thyroid hormone affected gene expression for a diverse range of cellular pathways and functions, including gluconeogenesis, lipogenesis, insulin signaling, adenylate cyclase signaling, cell proliferation, and apoptosis. This is the first application of the microarray technique to study hormonal regulation of gene expression in vivo and should prove to be a powerful tool for future studies of hormone and drug action.

No abstract

Thrombospondin 1 (TSP1) transcription is stimulated by glucose, resulting in increased TGF-␤ activation and matrix protein synthesis. We previously showed that inducible expression of the catalytic domain of cGMPdependent protein kinase (PKG) inhibits glucose-regulated TSP1 transcription and transforming growth factor (TGF)-␤ activity in stably transfected rat mesangial cells (RMCs(tr/cd)). However, the molecular mechanisms by which PKG represses glucose-regulated TSP1 transcription are unknown. Using a luciferase-promoter deletion assay, we now identify a single region of the human TSP1 promoter (؊1172 to ؊878, relative to the transcription start site) that is responsive to glucose. Further characterization of this region identified an 18-bp sequence that specifically binds nuclear proteins from mesangial cells. Moreover, binding is significantly enhanced by high glucose treatment and is reduced by increased PKG activity. Gel mobility shift and supershift assays show that the nuclear proteins binding to the 18-bp sequence are USF1 and -2. USF1 and USF2 bound to the endogenous TSP1 promoter using a chromatin immunoprecipitation assay. Glucose stimulates nuclear USF2 protein accumulation through protein kinase C, p38 MAPK, and extracellular signal-regulated kinase pathways. Increased PKG activity down-regulates USF2 protein levels and its DNA binding activity under high glucose conditions, resulting in inhibition of glucose-induced TSP1 transcription and TGF-␤ activity. Overexpression of USF2 reversed the inhibitory effect of PKG on glucose-induced TSP1 gene transcription and TGF-␤ activity. Taken together these data present the first evidence that USF2 mediates glucose-induced TSP1 expression and TSP1-dependent TGF-␤ bioactivity in mesangial cells, suggesting that USF2 is an important transcriptional regulator of diabetic complications.

The role of retinoic acid receptors (RARs) in intercellular regulation of cell growth was assessed by targeting a dominant-negative RARα mutant (dnRARα) to differentiated suprabasal cells of mouse epidermis. dnRARα lacks transcriptional activation but not DNAbinding and receptor dimerization functions. Analysis of transgenic mice revealed that dnRARα dose-dependently impaired induction of basal cell proliferation and epidermal hyperplasia by all-trans RA (tRA). dnRARα formed heterodimers with endogenous retinoid X receptor-α (RXRα) over RA response elements in competition with remaining endogenous RARγ-RXRα heterodimers, and dose-dependently impaired retinoiddependent gene transcription. To identify genes regulated by retinoid receptors and involved in cell growth control, we analyzed the retinoid effects on expression of the epidermal growth factor (EGF) receptor, EGF, transforming growth factor-α, heparin-binding EGFlike growth factor (HB-EGF) and amphiregulin genes. In normal epidermis, tRA rapidly and selectively induced expression of HB-EGF but not the others. This induction occurred exclusively in suprabasal cells. In transgenic epidermis, dnRARα dose-dependently inhibited tRA induction of suprabasal HB-EGF and subsequent basal cell hyperproliferation. Together, our observations suggest that retinoid receptor heterodimers located in differentiated suprabasal cells mediate retinoid induction of HB-EGF, which in turn stimulates basal cell growth via intercellular signaling. These events may underlie retinoid action in epidermal regeneration during wound healing.

Little is known about the overall patterns of thyroid hormone (Th)-mediated gene regulation by the main Th receptor (Tr) isoforms, Tr-α and Tr-β, in vivo. We used 48 complementary DNA microarrays to examine hepatic gene expression profiles of wildtype and Thra and Thrb knockout mice under different Th conditions: no treatment, treatment with 3,3',5-triiodothyronine (T 3 ), Th-deprivation using propylthiouracil (PTU), and treatment with a combination of PTU and T 3 . Hierarchical clustering analyses showed that positively regulated genes fit into three main expression patterns. In addition, only a subpopulation of target genes repressed basal transcription in the absence of ligand. Interestingly, Thra and Thrb knockout mice showed similar gene expression patterns to wild-type mice, suggesting that these isoforms co-regulate most hepatic target genes. Differences in the gene expression patterns of Thra/Thrb double-knockout mice and Th-deprived wild-type mice show that absence of receptor and of hormone can have different effects. This large-scale study of hormonal regulation reveals the functions of Th and of Tr isoforms in the regulation of gene expression patterns. EMBO reports 4, 581-587 (2003) doi:10.1038/sj.embor.embor862 In the absence of Th, Trs recruit co-repressor complexes (for example, nuclear corepressor (NCoR) and silencing mediator of Tr and retinoic acid receptor (Smrt)) that lead to the histone deacetylation of local chromatin in the promoter region, and decrease the basal transcription of positively regulated target genes (Burke & Baniahmad, 2000). In the presence of Th, Trs recruit co-activator complexes that contain the p160 steroidreceptor co-activator (SRC) family members and other cofactors that have histone acetyltransferase activity, as well as other complexes that can associate with the RNA polymerase II complex (Yen, 2001;Zhang & Lazar, 2000). The overall effect of these ligand-dependent interactions is to promote the transcriptional activation of positively regulated target genes. Trs also negatively regulate target genes, but the molecular mechanism by which they do this is poorly understood (Yen, 2001).Recent studies have shown that Thra and Thrb knockout mice have different phenotypes, and have also suggested that different Tr isoforms may have distinct roles (Flamant & Samarut, 2003;Forrest & Vennstrom, 2000). Two groups have generated Thra knockout mice that have distinct phenotypes (Flamant & Samarut, 2003;Forrest & Vennstrom, 2000). This was due to the targeting of different loci for homologous recombination, which resulted in the loss of different combinations of Tr-α isoforms. Thra knockout mice that lack all Tr-α isoforms (Thra 0/0 ), including short Tr-α isoforms that are generated by internal transcription initiation, are viable and fertile, and have a mild phenotype (Gauthier et al., 2001;Macchia et al., 2001).

To determine whether 9-cis retinoic acid receptors (RXRs) regulate the biological activity of all-trans retinoic acid (tRA) and its receptors (RARs) in skin, we have targeted a dominant-negative RXRot (dnRXRoL) lacking transactivation function AF-2 to differentiated suprabasal keratinocytes in the epidermis of transgenic mice. Driven by the suprabasal-specific keratin-10 gene promoter, expression of dnRXRa severely reduced the ability of RAR-selective ligands tRA and CD367 to induce epidermal mRNA levels of the CRABPII, CRBPI, and CRBPII genes, which contain RA-responsive elements (RAREs) DR1 and/or DR2. It also reduced gene-specific, synergistic induction of CRBPI mRNA by a combination of CD367 and RXR-selective SRl1237. Like endogenous RXR~, dnRXR~ in epidermal nuclear extracts from the transgenic mice competitively formed heterodimers with endogenous RARy on RAREs, suggesting that dnRXR~ impairs retinoid signaling by competing with endogenous RAR~,-RXRoL heterodimers. Histologically, the epidermis of dnRXR~ mice showed no detectable developmental abnormalities. Surprisingly, in adult animals, the suprabasal expression of dnRXRoL significantly reduced the ability of topically applied tRA to stimulate proliferation of undifferentiated keratinocytes in the basal layer of epidermis. RXR-selective ligands alone had no detectable effects on both normal and transgenic mouse epidermis. Accordingly, we suggest that in vivo: (1) in suprabasal keratinocytes, retinoids regulate gene transcription via RAR-RXR heterodimers in which RAR confers a predominant ligand response, whereas RXR AF-2 is required for liganded RAR AF-2 to efficiently trans-activate target genes, and (2) this suprabasal RXR-assisted mechanism indirectly regulates proliferation of basal keratinocytes likely via intercellular signaling.[Key Words: Retinoids; 9-cis retinoic acid receptor; transgenic mouse; epidermal keratinocytes; gene transcription; cell proliferation] Received September 4, 1996; revised version accepted November 11, 1996.Vitamin A (retinol) is an important regulator of epithelial cell homeostasis. All-trans retinoic acid (tRA) is the major biologically active metabolite of vitamin A. Clinically, tRA and related synthetic retinoids are widely used in the therapy of skin disorders such as cystic acne, psoriasis, photoaging skin, and certain epithelial malignancies (for review, see Peck and Di Giovanna 1994). These effects are thought to be mediated by members of two nuclear receptor gene families, retinoic acid receptor (RAR) genes and 9-cis retinoic acid (9cRA) receptor (RXR) genes. Each of the two families comprises three ZCorresponding author.

The liver is an important target organ of thyroid hormone. However, only a limited number of hepatic target genes have been identified, and little is known about the pattern of their regulation by thyroid hormone. We used a quantitative fluorescent cDNA microarray to identify novel hepatic genes regulated by thyroid hormone. Fluorescent-labeled cDNA prepared from hepatic RNA of T3-treated and hypothyroid mice was hybridized to a cDNA microarray, representing 2225 different mouse genes, followed by computer analysis to compare relative changes in gene expression. Fifty five genes, 45 not previously known to be thyroid hormone-responsive genes, were found to be regulated by thyroid hormone. Among them, 14 were positively regulated by thyroid hormone, and unexpectedly, 41 were negatively regulated. The expression of 8 of these genes was confirmed by Northern blot analyses. Thyroid hormone affected gene expression for a diverse range of cellular pathways and functions, including gluconeogenesis, lipogenesis, insulin signaling, adenylate cyclase signaling, cell proliferation, and apoptosis. This is the first application of the microarray technique to study hormonal regulation of gene expression in vivo and should prove to be a powerful tool for future studies of hormone and drug action.

Unliganded thyroid hormone receptors (TRs) interact with corepressors and repress basal transcription of target genes in cotransfection and in vitro studies. Currently, little is known about the function of corepressors in vivo. We thus used a mouse albumin promoter to generate several transgenic mouse lines that overexpressed a dominant negative mutant corepressor, NCoRi, in liver. The transgenic mice had normal liver weight, appearance, and minimal changes in enzyme activity. To study the effects of NCoRi on transcription of hepatic target genes, we examined T 3 -regulated gene expression of hypo-and hyperthyroid transgenic mice. In hypothyroid mice, hepatic expression of Spot 14, Bcl-3, glucose 6-phosphatase, and 5-deiodinase mRNA was higher in transgenic mice than littermate controls whereas these genes were induced to similar levels in T 3 -treated mice. Derepression was not observed for malic enzyme mRNA expression in hypothyroid mice. Thus, NCoRi selectively blocked basal transcription of several thyroid hormone-responsive genes but had no effect on ligand-mediated transcription. Additionally, compensatory increases in endogenous SMRT and NCoR mRNA were observed in hypothyroid transgenic mice. Interestingly, hepatocyte proliferation as detected by BrdUrd incorporation was increased in transgenic mice. The gene profile in transgenic mouse livers was studied by cDNA microarray, and several genes related to cell proliferation were induced. In summary, our studies show that NCoR plays important roles in mediating basal repression by TRs and may prevent cellular proliferation in vivo.Thyroid hormone receptors (TRs) 1 and retinoic acid receptors (RARs) can repress basal transcription in the absence of ligand and activate transcription upon ligand binding, in positively regulated target genes. TRs and RARs mediate basal repression through interactions with corepressors, such as NCoR (nuclear receptor corepressor) and SMRT (silencing mediator for retinoid and thyroid hormone receptors, Refs. 1, 2). NCoR and SMRT are both 270 kDa in size and their overall amino acid identity is 43% (3, 4). These corepressors have two nuclear hormone receptor interaction domains in the carboxyl terminus and three transferable repression domains in the amino terminus (5, 6). NCoR binds to Sin3, which in turn, recruits histone deacetylases (7-11). The formation of the NCoR⅐Sin3⅐HDAC complex by TR leads to hypoacetylation of local histones resulting in conformational changes in the nucleosome structure and decreased access of enhancers and components of the basal transcriptional machinery to the promoter region and transcriptional start site (7-10). In contrast, transcriptional activation occurs via ligand-dependent recruitment of p160 co-activators by TR, which then forms complexes that contain histone acetyltransferase activity (12). These complexes may then exchange with other complexes that share components with the RNA pol II transcriptional initiation complex (13,14).The NCoR⅐Sin3⅐HDAC complex not only mediates transcriptional...