Csaba Finta

A Statistical View of Genome Transcription

Recent analysis in gene expression points to the fact that our perception of genes as well-defined DNA segments within a vast excess of nonfunctional sequences may not necessarily be true. It has been known for some time that a gene may have variable transcriptional start sites, termination sites, and use a different set of exons. The most abundant combination of these three components (transcriptional start, termination, and exon selection) is generally considered as the major, canonical product of the gene whereas other combinations are regarded as variants. However, even alternative transcriptional start and termination may not set the outer borders of a gene, as evidence is accumulating on intergenic transcription (Ashe et al. 1997; Gribnau et al. 2000) and intergenic splicing (Magrangeas et al. 1998; Kowalski et al. 1999; Moore et al. 1999; Finta and Zaphiropoulos 2000a; Millar et al. 2000; Thomson et al. 2000). Moreover, analysis of transcriptome sequences clearly reveals that the majority of unique transcripts is present at levels lower than one copy per cell (Velculescu et al. 1999). These rare transcripts may contain genomic sequences that are flanked by the conserved GT and AG dinucleotides but distinct from what is considered as the typical exons of a gene (Hirosawa et al. 2000; Croft et al. 2000). In fact, a large variety of such sequences, present within or outside the borders of a canonical gene, may be found in these transcripts, including exons from neighboring genes (Fig. 1). A significant number of these rare RNA molecules has open reading frames and consequently can code for proteins. Attention has recently focused on the results of genomic sequencing projects, that is, the finding that the human genome encodes only 30,000 to 40,000 genes (International Human Genome Sequencing Consortium 2001; Venter et al. 2001), not much more than the 19,000 genes of C. elegans (the C. elegans Sequencing Consortium 1998). The fact that the complexity of an organism is not accurately reflected in the number of its genes may therefore signify the importance of the content of its unique RNA repertoire. Collectively these findings suggest that the concept of gene transcription may not suffice to include all variations in expressed genomic sequences. Thus, we propose a more general term, “genome transcription,” which incorporates RNA molecules containing sequences outside the borders of canonical genes. In this scenario, genes and gene transcription may be viewed as statistical peaks within a genome-wide pattern of expression of the genetic information. Accordingly, defined genes only represent DNA segments that are characterized by levels of transcriptional start, termination, and exon selection that are above an arbitrary threshold. Moreover, expressed sequences surrounding these peaks and encompassing noncanonical exon combinations (sequences that are often considered the result of incomplete or aberrant splicing) may not necessarily be nonfunctional. Rather, these would represent a reservoir of potential function. Intergenic sequences may be capCorrespondence to: C. Finta; email: csaba.finta@cnt.ki.se; or: P.G. Zaphiropoulos; email: peter.zaphiropoulos@cnt.ki.se J Mol Evol (2001) 53:160–162 DOI: 10.1007/s002390010204

A possible role of mouse Fused (STK36) in Hedgehog signaling and Gli transcription factor regulation

The segment polarity gene Fused (Fu) encodes a putative serine–threonine kinase Fu, which has been shown to play a key role in the Hedgehog signaling pathway of Drosophila. Human FU (hFU) has been shown to enhance the activity of Gli transcription factors, targets of the signaling pathway. However, Fu−/− mice do not show aberrant embryonic development indicating that mouse Fu (mFu) is dispensable for Hedgehog signaling until birth. In order to investigate if there are important differences between hFU and mFu, we cloned the cDNA, analyzed expression and tested the ability of mFu to regulate Gli proteins. Of the tested tissues only brain and testis showed significant expression. However, in transient overexpression analyses mFu was able to enhance Gli induced transcription in a manner similar to hFU. Thus, we turned to RNAi in order to test if mFu would be important for Hedgehog signaling after all. In one cell line with reduced mFu expression the Hedgehog signaling was severely hampered, indicating that mFu may have a role in Hedgehog signaling and Gli regulation in some cellular situations.

SCF (Fbxl17) ubiquitylation of Sufu regulates Hedgehog signaling and medulloblastoma development

Skp1‐Cul1‐F‐box protein (SCF) ubiquitin ligases direct cell survival decisions by controlling protein ubiquitylation and degradation. Sufu (Suppressor of fused) is a central regulator of Hh (Hedgehog) signaling and acts as a tumor suppressor by maintaining the Gli (Glioma‐associated oncogene homolog) transcription factors inactive. Although Sufu has a pivotal role in Hh signaling, the players involved in controlling Sufu levels and their role in tumor growth are unknown. Here, we show that Fbxl17 (F‐box and leucine‐rich repeat protein 17) targets Sufu for proteolysis in the nucleus. The ubiquitylation of Sufu, mediated by Fbxl17, allows the release of Gli1 from Sufu for proper Hh signal transduction. Depletion of Fbxl17 leads to defective Hh signaling associated with an impaired cancer cell proliferation and medulloblastoma tumor growth. Furthermore, we identify a mutation in Sufu, occurring in medulloblastoma of patients with Gorlin syndrome, which increases Sufu turnover through Fbxl17‐mediated polyubiquitylation and leads to a sustained Hh signaling activation. In summary, our findings reveal Fbxl17 as a novel regulator of Hh pathway and highlight the perturbation of the Fbxl17–Sufu axis in the pathogenesis of medulloblastoma.

Novel Mechanism of Action on Hedgehog Signaling by a Suppressor of Fused Carboxy Terminal Variant

The Suppressor of Fused (SUFU) protein plays an essential role in the Hedgehog (HH) signaling pathway, by regulation of the GLI transcription factors. Two major isoforms of human SUFU are known, a full-length (SUFU-FL) and a carboxy-terminal truncated (SUFU- ΔC) variant. Even though SUFU- ΔC is expressed at an equivalent level as SUFU-FL in certain tissues, the function of SUFU-ΔC and its impact on HH signal transduction is still unclear. In two cell lines from rhabdomyosarcoma, a tumor type associated with deregulated HH signaling, SUFU-ΔC mRNA was expressed at comparable levels as SUFU-FL mRNA, but at the protein level only low amounts of SUFU-ΔC were detectable. Heterologous expression provided support to the notion that the SUFU-ΔC protein is less stable compared to SUFU-FL. Despite this, biochemical analysis revealed that SUFU-ΔC could repress GLI2 and GLI1ΔN, but not GLI1FL, transcriptional activity to the same extent as SUFU-FL. Moreover, under conditions of activated HH signaling SUFU-ΔC was more effective than SUFU-FL in inhibiting GLI1ΔN. Importantly, co-expression with GLI1FL indicated that SUFU-ΔC but not SUFU-FL reduced the protein levels of GLI1FL. Additionally, confocal microscopy revealed a co-localization of GLI1FL with SUFU-ΔC but not SUFU-FL in aggregate structures. Moreover, specific siRNA mediated knock-down of SUFU-ΔC resulted in up-regulation of the protein levels of GLI1FL and the HH signaling target genes PTCH1 and HHIP. Our results are therefore suggesting the presence of novel regulatory controls in the HH signaling pathway, which are elicited by the distinct mechanism of action of the two alternative spliced SUFU proteins.