The mammalian paired box (Pax) genes encode a family of transcription factors involved in embryogenesis. The murine and human Pax8 genes are expressed in developing and adult thyroid as well as in the developing secretory system and at the lower level in adult kidney. In the secretory system expression is localized to the induced, extensively differentiating parts that undergo a transition from mesenchyme to epithelium. The human PAX8 gene generates at least five different alternatively spliced transcripts encoding different PAX8 isoforms. These isoforms differ in their carboxy-terminal regions downstream of the paired domain that has been shown previously to be responsible for the DNA binding. The PAX8a isoform contains a 63 amino-acid serine-rich region that is absent in the isoform PAX8b whereas PAX8c reveals a novel 99-amino-acid proline-rich region. This proline-rich region arises due to an unusual reading-frame shift in the PAX8 transcript. RNAse protection and RT(reverse transcription)-PCR analysis show the expression of all three PAX8 transcripts in human thyroid, kidney and five Wilms' tumors. Band-shift assay indicates a greatly reduced binding affinity of the isoform PAX8c to a DNA sequence from the promoter of the thyroperoxidase gene compared to the binding of PAX8a and PAX8b to this sequence. Deletion analysis of murine PAX8a indicates that its activating domain residues at the carboxy terminus of the protein which is shared by isoforms PAX8a and PAX8b. In accordance with these data PAX8a and PAX8b activate transcription from a thyroglobulin promoter as well as from a cotransfected synthetic PAX8-specific promoter/chlorampericol acetyltransferase (CAT) reporter containing a Pax8-binding oligonucleotide in front of the basal herpes simplex virus thymidine kinase (HSV-TK) promoter (P11/12-TK-CAT). However if the basal HSV-TK promoter of this reporter is substituted by a minimal adenovirus E1b TATA element, PAX8a and PAX8b fail to activate transcription. Of the three chimaeric forms containing the GAL4 DNA-binding domain at the amino-terminal end fused to the corresponding carboxy-terminal regions of the PAX8 isoforms beginning immediately downstream of the paired domain only a GAL4-PAX8b fusion significantly activates transcription from a cotransfected GAL4-specific upstream-activating-sequence (UAS)-TK-CAT reporter. Substitution of the basal HSV-TK promoter in this reporter by the minimal E1b TATA element does not affect this activation. These results indicate that the PAX8 isoforms display different functional properties and may also function differently in vivo.(ABSTRACT TRUNCATED AT 400 WORDS)

The conserved structure of the transcription factors of the Pax gene family may reflect functional conservation. We have demonstrated that the human Pax8 transcription factor is organized in several functional domains and contains two regions responsible for its nuclear localization, in addition to an activating region at the carboxy terminus of the protein and an inhibitory region encoded by the exon 9 present only in a splice variant PAX8a. Regions of PAX8 determining the nuclear localization of the PAX8flacZ fusions contain short amino acid sequences similar to several described nuclear localization sites (NLS). These NLS were identified in the paired domain and between the octapeptide and the residual homeodomain, respectively. The activating domain is encoded by the exons 10 and 11 and its function is modulated by the adjacent domains encoded by the exons 9 and 12. The domain encoded by exon 9 significantly inhibits the function of the activating domain. Pax8 is expressed in thyroid cells and its product binds promoters of the thyroglobulin and thyroperoxidase genes through its paired domain. Thyroid cell growth and differentiation depend on thyrotropin which, by stimulating cAMP synthesis, activates the CAMP-dependent protein kinase A (PKA). We have investigated a link between thyrotropin stimulation and gene activation by Pax8. Stimulation of cAMP synthesis augments Pax8-specific transcription in thyroid cells, indicating that PKA is involved in Pax8 activation. Cotransfection of GAL4/ PAX8 fusions and the catalytic subunit of PKA in A126, a PKA-deficient derivative of the PC12 pheochromocytoma cell line, synergistically activates the GAL4-specific reporter, suggesting the activating domain of PAX8 is dependent upon the catalytic subunit of the PKA. We propose that this dependence is due to a hypothetical adaptor which forms a target for PKA and interacts with the activating domain of PAX% We show that PAX8 isolated from the thyroid cell line FTRL5 is a phosphoprotein in which phosphorylation is not dependant on cAMP pathway activation. Our results suggest that Pax8 is part of the cAMP signaling pathway and mediates thyrotropin-dependent gene activation in thyroid cells. Investigation of the PAX8 expression in a panel of Wilms' tumors shows a striking correlation between the expression of PAX8 and another transcription factor, WT1, indicating that these two genes may interact in vivo.

Two clones were isolated in a three-hybrid screen of a rat fetal brain P5 cDNA library with an intronic splicing enhancer of the amyloid precursor protein (APP) gene as RNA bait. These clones represent the rat homologues of the previously described genes CUG-binding protein (CUG-BP) and Siah-binding protein (Siah-BP). Both interact in a sequence-specific manner with the RNA bait used for library screening as well as with the CUG repeat. In contrast, no interactions were observed in the three-hybrid assay with other baits tested. In two-hybrid assays, Siah-BP interacts with U2AF65 as well as with itself. EWS, an RGG-type RNA-binding protein associated with Ewing sarcoma, was identified as an interacting partner for the CUG-BP homologue in a twohybrid assay for protein±protein interactions performed with various factors involved in RNA metabolism. Splicing assays performed by RT-PCR from cells cotransfected with certain cDNAs and an APP minigene, used as a reporter, indicate exclusion of exon 8 if the CUG-BP homologue is present. We conclude that clone AF169013 and its counterpart in human CUG-BP could be the trans-acting factors that interact with the splicing enhancer downstream of exon 8, and in this way influence alternative splicing of the APP minigene.Keywords: alternative splicing; amyloid precursor protein (APP); CUG-binding protein (CUG-BP); three-hybrid system.Alternative usage of splice sites requires cis-acting elements, such as 5 H and 3 H consensus sequences for the binding of the basic splicing machinery, as well as gene-specific elements for binding of regulatory trans-acting factors [1,2]. Consequently, specific exon or intron sequences capable of strongly stimulating or repressing splice-site usage have been designated splicing enhancers or silencers, respectively. Although several proteins have been identified in Drosophila that act in this way [3±6], the regulatory molecules that direct alternative splicing in vertebrates remain elusive. Furthemore, reports describing such causative interactions in vertebrates have been from simple biochemical analysis and not genetic functional assays. One of the most extensively clarified cases reported so far describes biochemical interactions within the c-src gene. Analysis of mouse c-src gene indicates that the neuronal inclusion of the N1 exon requires an intronic splicing-enhancer sequence between 117 and 1142 nucleotides downstream of the exon [7]. The central, most conserved, portion of this enhancer sequence (nucleotides 38±70) is called the downstream control sequence [8], which binds to a complex of regulatory proteins that is important in allowing 18-nucleotide exon N1 splicing. A protein containing four K homology RNA-binding domains is shown to be the key component in the co-operative assembly of the multiprotein complex at the splicing enhancer [9]. The latest report demonstrates by genetic and biochemical methods that Nova-1 regulates in a sequencespecific manner alternative splicing of two inhibitory receptor pre-mRNAs, glycine receptor a2 and...

Recent evidence indicates a crucial role for paired box genes in mouse and human embryogenesis. The murine Pax8 gene encodes a sequence-specific transcription factor and is expressed in the developing secretory system as well as in the developing and adult thyroid. This restricted expression pattern suggested involvement of the Pax8 gene in the morphogenesis of the above organs and prompted us to investigate the PAX8 gene in humans. In this report, we describe the isolation and characterization of PAX8 cDNAs from a human adult kidney cDNA library. An open reading frame of 450 amino acids contains the 128 amino acid paired domain at its amino-terminal end. The predicted human and mouse Pax8 proteins show 97.8% conservation and are identical in their paired domains. Two independent cDNA clones reveal differential splicing of the PAX8 transcripts resulting in the removal of a 63 amino acid serine-rich region from the carboxy end of the predicted Pax8 protein. The truncated Pax8 protein becomes more similar to the predicted murine Pax2 protein, that is also expressed during kidney development and lacks the serine rich region. RNAse protection analysis shows the presence of both PAX8 transcripts in human thyroid, kidney and five Wilms’ tumors. No truncated Pax8 transcripts could be detected in mouse kidney. In situ hybridization to sections of human embryonic and fetal kidney showed expression of PAX8 in condensed mesenchyme, comma-shaped and S-shaped bodies. In contrast, PAX2 expression was present mainly in the very early stages of differentiation, in the induced, condensing mesenchyme. This restricted expression pattern suggests a specific role for both genes during glomeruli maturation.(ABSTRACT TRUNCATED AT 250 WORDS)

The aim of this study was to investigate whether mutations in hepatocyte nuclear factor (HNF)-4gamma, a transcription factor homologous to HNF-4alpha, contribute to the etiology of early-onset type 2 diabetes. Linkage between diabetes and two polymorphic markers at the HNF-4gamma locus (D8S286 and D8S548) was evaluated in 32 multigenerational families with early-onset autosomal-dominant type 2 diabetes unlinked to known maturity-onset diabetes of the young genes. Total logarithm of odds (LOD) scores were strongly negative (-50.3 at D8S286 and -46.2 at D8S548), but linkage could not be excluded in 15 families having LOD scores >-2.0. To screen these pedigrees for HNF-4gamma mutations, the gene structure was defined. Because reverse transcriptase-polymerase chain reaction experiments indicated that the first 1,674 bp of the published cDNA sequence (3,248 bp) were a cloning artifact, the correct cDNA sequence was determined by 5' rapid amplification of cDNA ends (RACE) and primer extension assay. Based on the new cDNA sequence (1,731 bp), 11 exons were found. After screening the 5' flanking region and all coding exons for mutations, we identified several polymorphisms, one of which affected the amino acid sequence (M190I). However, no mutations segregating with diabetes could be found in these families. We conclude that genetic variability in the HNF-4gamma gene is unlikely to play a major role in the etiology of early-onset autosomal-dominant type 2 diabetes.