Evolution of an enhancer

Previous work has identified several noncoding regions of the human genome that have evolved particularly rapidly. Shyam Prabhakar and colleagues now report an initial functional dissection of the most rapidly evolving element, HACNS1, which has accumulated 16 human-specific changes in the 6 million years since the human-chimpanzee split (Science 321, 1346–1350; 2008). The authors tested human HACNS1, as well as its orthologs from chimpanzee and rhesus macaque, in a transgenic mouse enhancer assay. They found that the human element functions as a developmental enhancer, driving reporter gene expression in the anterior limb bud, pharyngeal arches, ear and eye of mid-gestation (E11.5) embryos. Reproducible expression driven by the human element extended into the distal-most region of the anterior limb bud, whereas that driven by the chimp and rhesus orthologs did not. At a later stage (E13.5), HACNS1 promotes reporter gene expression in the anterior developing forelimb, particularly in the shoulder and anterior junction of the forearm and handplate, and in some cases in the future anterior-most digit of the forelimb. The authors suggest that the human-specific adaptive evolution of HACNS1 may have contributed to uniquely human aspects of digit and limb patterning, possibly by influencing expression of the flanking genes CENTG2 and GBX2. AP

Dissecting species-specific expression

It is known that binding sites and targeted genes for highly conserved transcription factors vary between human and mouse. Now, Duncan Odom and colleagues show that genetic sequence, rather than differences in transcription factors, epigenetic machinery or cellular environment, play a primary role in directing species-specific transcription programs (Science advance online publication 11 September 2008; doi:10.1126/science.1160930). The authors investigated binding of the HNF1α, HNF4α and HNF6 transcription factors in hepatocytes from an aneuploid mouse strain, Tc1, which carries human chromosome 21. None of the genes encoding these transcription factors is located in chromosome 21, so all of the binding events can be attributed to mouse proteins. Almost all transcription factor binding events normally present on human chromosome 21 in human hepatocytes are present on human chromosome 21 in Tc1 hepatocytes. Conversely, transcription factor binding on the orthologous mouse chromosome 16 is largely identical between normal mouse hepatocytes and Tc1 hepatocytes. Similarly, species-specific enrichment of H3K4me3 histone marks and species-specific gene expression patterns were conserved on human chromosome 21 and mouse chromosome 16 in Tc1 hepatocytes. These findings demonstrate the important role that DNA sequence differences have in determining patterns of species-specific expression. EN

Pathways in cancer genomes

The Cancer Genome Atlas (TCGA) reports the first pilot phase of the project (Nature advance online publication 4 September 2008; doi:10.1038/nature07385), with an integrative analysis of genetic alterations in 206 glioblastoma samples surveyed for copy number, gene expression and epigenetic changes. TCGA also examined half the samples for mutations in 600 genes previously implicated in cancer. In related studies, Bert Vogelstein and colleagues report efforts from their private cancer genome studies on glioblastoma (Science advance online publication 4 Sept 2008; doi:10.1126/science.1164368) and pancreatic cancer (Science advance online publication 4 Sept 2008; doi: 10.1126/science.1164382). In both studies, Vogelstein and colleagues sequenced 20,661 protein-coding genes and examined copy number changes and gene expression. The glioblastoma and pancreatic tumor studies examined 22 and 24 tumor samples, respectively, and found an average of 60 altered genes per glioblastoma tumor and 63 per pancreatic tumor. These papers showed that although mutations varied between tumors, only a small number of specific pathways were disrupted in a large proportion of the tumors. This suggests that targeting these pathways, rather than the function of single genes, may be a useful approach for treatment development. OB

The POPRES resource

Pharmacogenetic applications of case-control association studies require a sufficient number of well-matched controls representing individuals outside of the northern European populations that are typically sampled. Matthew Nelson and colleagues describe the ongoing development of the Population Reference Sample (POPRES), which is a publicly available resource of genotypes that can be used for pharmacogenetic and other studies (Am. J. Hum. Genet. 83, 347–358; 2008). The POPRES collection comprises genotypes of 5,886 individuals drawn from ten ongoing studies, which include African Americans, Japanese, Taiwanese, northern Europeans, Indian Asians, Australians and Mexicans. They were genotyped using the Affymetrix 500K platform, although the authors suggest that selected subsets will be genotyped with other panels, including Illumina 550K and 1M. Principal-component analysis provides a high degree of discrimination among individuals from different continents, suggesting that the genotypes are of high quality. In a small study of 21 individuals with abacavir-associated hypersensitivity reaction, the use of POPRES controls selected by four different methods in a genome-wide association study results in the consistent identification of the known susceptibility region in the MHC. All of the data from POPRES will be available from the NCBI database of Genotypes and Phenotypes, pending acceptance of a standard Data Use Certification. AP

Rearrangement hot spot on 1q21.1

Evan Eichler and colleagues previously identified a 1.35-Mb region on 1q21.1 as a likely target of recurrent rearrangement and reported a single individual with a de novo deletion of this region from a screen of 290 individuals with idiopathic mental retardation (Nat. Genet. 38, 1038–1042, 2006). Eichler and colleagues (N. Engl. J. Med. advance online publication 10 September 2008; doi:110.1056/NEJMoa0805384) have now screened 5,218 individuals with mental retardation, autism or congenital anomalies and identified 25 individuals with overlapping 1q21.1 deletions. The deletions—which in some cases arose de novo and in others were inherited from mildly affected or unaffected parents—were associated with variable phenotypes, including mild-to-moderate mental retardation, dysmorphic features and cardiac defects. Notably, the same 1q21.1 deletion was recently found to be associated with schizophrenia (Nature 455, 232–236, 2008; Nature 455, 237–241, 2008). The authors also reported nine individuals with the reciprocal duplication, whose phenotypes included mental retardation or autism spectrum disorder and mild dysmorphic features. Collectively, these studies reveal marked heterogeneity in the phenotypes associated with recurrent 1q21.1 rearrangements and highlight an emerging class of disorders arising from local hot spots of genomic instability. KV

Written by Orli Bahcall, Emily Niemitz, Alan Packer & Kyle Vogan