ASHG 2010: Adaptation and natural selection

Selected abstracts from the 2010 meeting of the American Society of Human Genetics.

Signals of local adaptation in human populations. T. S. Simonson1, C. D. Huff1, Y. Zhang1, W. S. Watkins1, D. J. Witherspoon1, J. Xing1, S. R. Woodward2, L. B. Jorde1 1) Department of Human Genetics, University of Utah, Salt Lake City, UT; 2) Sorenson Molecular Genealogy Foundation, Salt Lake City, UT.

Humans have adapted to various environmental conditions. Genome-wide scans of positive selection have provided a first glimpse into locally adapted regions of the genome. It is possible that many more examples of positive natural selection exist in populations that have not yet been surveyed. In order to enhance the current understanding of population-specific adaptation, we performed genome-wide selection scans for 14 previously unstudied world-wide populations from Africa, Europe, Asia, and the Americas using the iHS statistic. Five of the seven selection candidates previously reported in the HapMap populations (Sabeti 2007) were identified in at least one additional population in our sample at the 0.02 significance level. Two of the previously reported regions were found in three or more of the populations we sampled: 1) the genic region containing PDE11A (phosphodiesterase 11A), first identified in Europe, Japan, and China, was also significant in Bolivia, Mongolia, Thailand, and Tibet; 2) the genic region containing SLC30A9 (solute carrier family member 9), first identified in Japan and China, was significant in India, Pakistan, and Iraq. In contrast to finding many HapMap selection candidates in our additional populations, more than two-thirds of our selection candidate regions were not significant at the 0.05 level in any of the HapMap populations. We find that positive selection candidates vary considerably among these populations, highlighting novel signals of natural selection. These results suggest that distinct human populations sampled here have differentially adapted to specific environments.


Genome-wide patterns of allele frequency variation estimated from low-coverage Illumina sequencing of 2,000 Danish individuals. K. E. Lohmueller1, A. Albrechtsen2, Y. Li3, S. Y. Kim1, T. Corneliusen2, N. Vinckenbosch1, G. Tian4,5, E. Huerta-Sanchez1, A. Feder1,6, T. Jiang3, I. Hellmann7, O. Pedersen8,9,10, J. Wang2,3, R. Nielsen1,2, and the LuCamp Consortium 1) Integrative Biology and Statistics, University of California, Berkeley, Berkeley, CA; 2) Department of Biology, University of Copenhagen, Copenhagen, Denmark; 3) Beijing Genomics Institute, Shenzhen, China; 4) Beijing Institute of Genomics, Chinese Academy of Science, Beijing, China; 5) The Graduate University of Chinese Academy of Sciences, Beijing, China; 6) University of Pennsylvania, Philadelphia, PA; 7) Department of Mathematics, University of Vienna, Vienna, Austria; 8) Hagedorn Research Institute, Gentofte, Denmark; 9) Faculty of Health Sciences, University of Aarhus, Aarhus, Denmark; 10) Institute of Biomedical Sciences, University of Copenhagen, Copenhagen, Denmark.

We analyze low-coverage genome-wide next-generation sequencing data from 2,000 Danish individuals. Individual genotypes cannot be directly inferred from such data. However, we have developed a new statistical method to reliably estimate population allele frequencies of single nucleotide polymorphisms (SNPs). Our method eliminates the difficulty of inferring individual genotypes data by using the read counts and quality scores in the sample directly, without first attempting to estimate genotypes. Application of our new methods to the data allows the first estimates of genome-wide patterns of SNP allele frequency variation obtained from re-sequencing data. By comparing allele frequencies in different categories of sites, we are able to quantify the amount of selection acting in various non-coding parts of the genome. Interestingly, we also document a significant negative correlation between minor allele frequency and recombination rate. SNPs in regions of low recombination tend to have lower frequencies than SNPs in regions of the genome with higher recombination rates. Simulations using realistic demographic and selective models show that pattern is expected if positive natural selection is common in the human genome. Our results have important implications for understanding how natural selection has shaped patterns of variation across the human genome.


Distorted population genetic principles in a super-exponentially growing population. A. G. Clark1, A. Keinan2 1) Dept Molec Biol & Gen, Cornell Univ, Ithaca, NY; 2) Biol Statistics and Computational Biology, Cornell Unv, Ithaca, NY.

The global human population has been growing super-exponentially over the last 20-30 centuries, so that the current census size is approaching one million times the historical effective population size. This extraordinary situation implies a massive departure from equilibrium, such that human populations will continue to accumulate genetic variability for many millennia to come until a new equilibrium is reached. This presents an interesting theoretical problem regarding the structure of that variation, and encouraged us to revisit all the principles of population genetics in the context of this extraordinary demography. While many principles still hold, recent super-exponential growth produces an unusual site frequency spectrum, with nearly normal counts of SNPs in different frequency classes other than the very rarest classes, which are highly inflated. But this effect is only seen in sample sizes in the thousands. Our logistic branching process modeling also predicts a distortion of linkage disequilibrium in rapidly growing populations. The amplitude of random genetic drift is greatly attenuated as the population grows, and the probabilities of fixation and of loss are likewise reduced. Last, natural selection operates more efficiently in a large population and as a result, super-exponential growth has the effect of activating natural selection on for many standing variants. This phenomenon, combined with the skewed allele frequency spectrum of new mutations, produces key predictions for the frequency and characteristics of variants that are implicated in complex disease. All of these features show how poorly the historical effective size represents many population genetic attributes in a rapidly expanding population.


Signatures of natural selection in the first pilot experiment of the 1000 Genomes Project. R. D. Hernandez1, J. L. Kelley2, S. C. Melton2, A. Auton3, G. McVean3,4, G. Sella5, M. Przeworski2, 1000 Genomes Project 1) University of California, San Francisco, Department of Bioengineering and Therapeutic Sciences, San Francisco, CA, 94158; 2) University of Chicago, Department of Human Genetics, Chicago, IL, 60637; 3) University of Oxford, Wellcome Trust Centre for Human Genetics, Oxford, OX3 7BN, United Kingdom; 4) University of Oxford, Department of Statistics, Oxford, OX1 3TG, United Kingdom; 5) The Hebrew University, Jerusalem, Department of Evolution, Systematics and Ecology, Jerusalem, JR 91904, Israel.

Nearly 200 genomes from four populations have been resequenced at low coverage for the first pilot experiment of the 1000 Genomes Project. These data avoid many of the ascertainment biases that have plagued previous large-scale human data sets, allowing long standing evolutionary questions to be resolved. We used these data to characterize genomic signatures of natural selection using patterns of genetic diversity. We show that neutral diversity increases with genetic distance from coding regions, suggesting that selection on coding regions distorts patterns of diversity in their vicinity. In particular, we report a clear footprint of natural selection on diversity patterns around human-specific amino acid substitutions. In addition, we identified novel regions of the genome with extreme differences in allele frequencies between population samples. While all three findings reflect the action of natural selection, it remains unclear to what extent they are explained by adaptive evolution or purifying selection, with recent reports offering conflicting conclusions in this regard. To disentangle the relative contributions of the two evolutionary forces, we ran extensive simulations of the human genome, incorporating information about functional annotations, fine-scale genetic maps, and realistic demographic models of all four populations.


Geographic approaches to rare SNPs in human populations. T. B. Mersha1, R. A. Wilke2 1) Cincinnati Children’s Hospital Medical Center, Department of Pediatrics, University of Cincinnati, Cincinnati, OH, USA; 2) Division of Clinical Pharmacology, Vanderbilt University Medical Center, Nashville, TN, USA.

The simultaneous study of genomic loci across numerous populations will facilitate a better understanding of the role of genetic variation in different geographic regions. We examined global allele frequencies across all HapMap populations at 3.7 million SNPs to search for loci under recent selection pressure, with a goal toward identifying variants that contribute to ecotypic phenotypic divergence. By examining private SNPs in HapMap samples from geographically separated populations, we demonstrate that 463 loci (mapping to 38 genes) were fixed (delta= 1). These private loci included four non-synonymous coding SNPs: rs4536103 (NEUROG3), rs1385699 (EDA2R), rs11946338 (ARHGAP24), and rs4422842 (CACNA1B), and regulatory variants in four additional genes (IER5L, NPNT, SESTD1 and EXOC6B) were undergoing recent positive selection. Over all, very few genes in the human genome had extreme allele frequency differences among populations, and the bulk of variations at the nucleotide levels are not visible at the phenotypic level. We explore pathway-oriented strategies to identify signaling mechanisms that may have influenced human adaptation to different environments. These fixed SNPs contribute to variability in several cellular processes. As the clinical community moves toward application of genome-wide SNP scanning in large practice-based cohorts, our pathway-oriented approach has relevance to pharmacogenomic data. An improved understanding of these pathways in genomic regions of fixed loci may help explain race/ethnicity-specific variability in treatment outcome as ancient genes are exposed to modern drugs.


Genome-wide scans of European population groups identify multiple loci with evidence for positive selection in Sardinians. R. Kosoy1, F. Macciardi2,3, F. Taddeo3, D. Cusi4, S. Lupoli5, H. Chen6, M. L. Ransom1, M. F. Seldin1 1) Dept Biological Chemistry, Univ California, Davis, Davis, CA; 2) Department of Psychiatry and Human Behavior, Univ California, Irvine, Irvine, CA; 3) Genomics and Bioinformatics Unit, Univ of Milan - Fondazione Filarete, Italy; 4) School of Medicine. Dept. of Sciences and Biomedical Technologies. University of Milan, Italy; 5) INSPE, San Raffaele Scientific Institute, Milan, Italy; 6) Department of Genetics, Harvard Medical School, Boston, MA.

Identification of genes undergoing natural selection in various human populations may help identify genetic factors that increase or decrease the risk for specific diseases within that population. We applied multiple methods to detect signatures of positive natural selection using genome-wide scans of eight different European populations. The methods included Fst as a measure of population differentiation and two algorithms (XPEHH, and iHS), which detect the presence of extended identical haplotypes, arising due to a partial or a complete sweep in a population. In addition, we also used XPCLR, a likelihood method for detecting selective sweeps by modeling the multilocus allele frequency differentiation between pairs of populations. These methods were applied to 475,574 SNPs genotyped in English, Irish, Norwegians, Northern Italians, South Italians, Sardinians, Ashkenazi Jews and Arabs (number of individuals in each group between 70 and 780) in a total of 2035 individuals. A number of loci with strong evidence for positive selection were identified in the Sardinian population by pair-wise analyses with four other South European populations (Southern Italians, Northern Italians, Ashkenazi Jewish, and Arabic groups) and three North European populations (English, Norwegians, and Irish). In particular, multiple SNPs in 2p22.1 and 2q37.8 regions showed positive results with XPEHH scores > 0.9 and Fst scores > 0.14 for pair wise comparisons and with clearly defined boundaries. The first of the regions spans 440kb, and contains a part of a single gene, SLC8A1 (solute carrier family 8), previously suggested as a candidate for hypertension. The second region is 220kb long, and contains six genes, HDLBP, SETP2, FARP2, STK25, BOK, and THAP, the first of which codes for high density lipoprotein binding protein, whose likely function is in cell sterol metabolism. The pair-wise differences were strongest between Sardinian and other South European populations, suggesting that the natural selection at these genomic regions is largely confined to Sardinian population, and is not a part of the South Europe versus North Europe difference.


Combining signatures of selection and disease-associated loci in three ethnic groups. X. Sim1, R. T. H. Ong1,2, W. T. Tay3, C. Suo1, T. Y. Wong3,4,5, E. S. Tai5,6, K. S. Chia1,5, Y. Y. Teo1,5,7,8 1) Centre for Molecular Epidemiology, National University of Singapore, Singapore, Singapore; 2) NUS Graduate School for Integrative Science and Engineering, National University of Singapore, Singapore; 3) Singapore Eye Research Institute, Singapore National Eye Centre, Singapore; 4) Centre for Eye Research Australia, University of Melbourne, Australia; 5) Department of Epidemiology and Public Health, National University of Singapore, Singapore; 6) Department of Medicine, National University of Singapore, Singapore; 7) Department of Statistics and Applied Probability, National University of Singapore, Singapore; 8) Genome Institute of Singapore, Agency for Science, Technology and Research, Singapore.

Natural selection leaves an imprint on the genetic architecture of populations at the selected site and provides an insight to the evolutionary adaptations of the human genome. Often, the roles of environmental factors that shaped the selection or targets of the selection are not immediately clear. The study of positive selection therefore hinges on an understanding of the genotype-phenotype relationships and how these association signals could be rooted in regions of positive selection. Darwinian medicine has suggested that vulnerabilities to diseases are shaped by selection pressures and the process of selection has an important role in shaping the humans evolved defenses against infectious and immune diseases. Hence it is of great interest to combine evidences of selection and disease association to enhance the understanding of phenotype-genotype relationships in both chronic and infectious diseases. Different statistical methods have been developed to detect signatures of positive selection across the frequencies spectrum. While multiple populations are more useful to detect selection events that has risen to fixation in one population and not in another, a characteristic of recent signature of positive selection (at frequencies of 30% to 80%) is unusually extended haplotypes that rise in frequency rapidly as recombination has yet to break down the long haplotype while the selected variant is passed down preferentially. With the availability of genome-wide data on three ethnic groups in Singapore, Chinese, Malays and Asian Indians of 2500 individuals in each cohort and sophisticated statistical algorithms such as iHS and XP-EHH, we surveyed the genome for selection signatures and associations with common diseases such as Type II Diabetes and complex traits including lipids traits. Our findings suggest the overlapping of selection signatures and diseases-associated loci could elucidate differences in diseases susceptibility in different populations and the importance of environmental variations across populations.


Population differentiation as a test for soft sweeps on standing genetic variation. H. Chen1,2, N. Patterson2, D. Reich1,2 1) Dept Gen, Havard Med Sch, Boston, MA; 2) Broad Institute of MIT and Harvard, Boston MA.

Recent studies have suggested that adaptations to environmental change in human populations in the past 10,000 years may have occurred in part as a “soft sweep”, that is selection on standing genetic variation. A number of statistical methods have been developed for detecting selective sweeps on newly arising genetic variants but a concern is that almost all have low power for soft sweeps. This is because the population genetic model for selective sweeps underlying these methods assumes that causal mutants under selection started from a single copy (“hard sweeps”), and soft sweeps generate a genetic polymorphism pattern that is strikingly different from classical signatures of hard sweeps. We propose a method that can detect soft sweeps efficiently, which is based on analysis of the haplotype differentiation patterns across populations. We incorporate haplotype patterns from reference populations (which presumably did not experience selection) as an approximation of the haplotype structure around the selected standing variant before selection. We then use population genetic theory to explicitly model the effect of a soft sweep on the allele frequencies of the nearby SNPs (single nucleotide polymorphisms). This approach can capture the characteristic genetic polymorphism patterns caused by soft sweeps and it achieves much better power compared with existing methods. We apply our method to several genome-wide SNP array data sets, including comparisons of Africans to non-Africans, and also northern to southern Europeans. Our method assumes that the selected variants are genotyped in the data set and thus can be expected to work substantially better on resequencing data. In addition to providing a new method for scanning for soft sweeps, our work also demonstrates that investigating the impact of natural selection in multiple populations simultaneously can provide insights about selection that would not be accessible in any one population.


Evidence of selection in the pigmentation genes SLC24A5 in Europeans, East Africans, and Southwest and Central Asians and SLC45A2 in East Asians and Native Americans. M. P. Donnelly, W. C. Speed, A. J. Pakstis, J. R. Kidd, K. K. Kidd Department of Genetics, Yale University School of Medicine, New Haven, CT.

Skin pigmentation is one of the most recognizable human phenotypes and tends to vary on a latitudinal cline, even within Europe. The derived alleles of missense SNPs in SLC24A5 (rs1426654) and SLC45A2 (rs16891982) have both been implicated in light skin pigmentation among Europeans. We have collected data for these markers from 4474 individuals in 107 population samples. The derived alleles of both SNPs were observed at high frequencies throughout Europe, though the derived allele of rs16891982 is found at lower frequencies in Southern Europe. The derived allele of rs1426654 was also found at moderate to high frequencies 2 to 100% in East Africa, Southwest Asia, and Central Asia, whereas the derived allele of rs16891982 was seen at frequencies of 0 to 58% in these populations. At SLC24A5 a single allele of a 13-SNP (including rs1426654) haplotype covering ~146 kb accounts for ~95% of the chromosomes in Europe. At SLC45A2, we saw no significant LD around rs16891982. Using the REHH test, we found strong evidence of selection for the derived allele of rs1426654 in Europe as well as East Africa, Southwest Asia, and Central Asia where it had not previously been seen and were able to confirm the evidence of selection in East Africa using nHS. We saw no or very weak evidence of selection for rs1689192 using REHH or nHS among European and nearby populations. We did find previously unidentified evidence of selection in East Asians and Native Americans at a SNP about 15 kb upstream of SLC45A2. The allele showing evidence of selection at this SNP is found at high to moderate frequencies throughout the world except in Europe where it is virtually absent. Given its location it is likely part of or in LD with an upstream regulatory element upon which selection is/was acting. Taken together these results suggest several conclusions about the evolution of skin pigmentation in humans. First two SNPs shown to play a role in pigmentation in a region of the world can show different distribution patterns Second, it suggests that light skin among Europeans evolved both by means of natural selection and neutral factors. Finally, the evidence of selection in the upstream region of SLC45A2 in East Asians and Native Americans, suggests that though light skin pigmentation likely evolved separately in regions where light skin pigmentation is predominant, the evolution at some genes may have occurred independently through different variants. Funded in part by NIH Grant GM57672.


The Genetic Basis and Evolutionary History of PTC Bitter Taste Perception in Africa. M. C. Campbell1, A. Ranciaro1, A. Froment2, D. Zinshteyn3, J.-M. Bodo4, D. Drayna5, P. Breslin6,7, S. A. Tishkoff1,3 1) Dept Gen, Univ Pennsylvania Sch Med, Philadelphia, PA; 2) Ecoanthropology and Ethnobiology Musée de l’Homme, Paris, France; 3) Department of Biology, University of Pennsylvania, Philadelphia, PA; 4) Ministry of Scientific Research and Innovation, Yaoundé, Cameroon; 5) National Institute on Deafness and Other Communication Disorders, Rockville, MD; 6) Monell Chemical Senses Center, Philadelphia, PA; 7) Department of Nutritional Sciences, Rutgers University, New Brunswick, NJ.

Bitter taste perception is an important dietary adaptation which may have evolved to prevent the accidental ingestion of harmful plant toxins. The ability to taste the bitter synthetic compound phenylthiocarbamide (PTC) is a highly variable trait in humans and is also correlated with the ability to detect naturally bitter substances in food. A large proportion of the phenotypic variance in PTC taste sensitivity has been attributed to genetic variability at TAS2R38, a bitter taste receptor gene, located on chromosome 7. In order to better understand the genetic basis and evolutionary history of PTC taste perception, we examined sequence variation within a 2975 bp region encompassing the 1002 bp coding exon of the TAS2R38 gene, in 588 individuals from more than 50 ethnically diverse populations from East and West Central Africa, as well as in a comparative sample of 132 non-Africans from different geographic regions. We also examined genotype-phenotype associations in a large subset of Africans, totaling 463 individuals, from the above sampled populations. Our analyses uncovered higher levels of nucleotide and amino acid haplotype variability, including an excess of rare non-synonymous polymorphisms, in Africans relative to non-Africans. The estimated time to the most recent common ancestor (TMRCA) of genetic variation within the TAS2R38 coding region is ~2.3 million years, consistent with the pattern expected under ancient balancing selection. The diverse amino acid haplotypes present in African populations were also associated with a broader range of PTC taste sensitivity than is typically observed outside of Africa. In addition, we found that several rare non-synonymous polymorphisms significantly modify PTC sensitivity in African populations, demonstrating the effect of single amino acid substitutions on bitter taste perception. This research provides further information regarding the genetic basis of phenotypic variation in Africa, and serves as a model system for understanding the influence of natural selection, as well as common and rare amino acid variation, on variable traits in humans.


Natural selection and sex-specific demography shape patterns of genetic variation across the genome. J. D. Wall1, A. E. Woerner2, F. L. Mendez3, J. C. Watkins4, M. P. Cox2, M. F. Hammer2, 3 1) Inst Human Gen, Univ California, San Francisco, San Francisco, CA; 2) ARL Division of Biotechnology, University of Arizona, Tucson, AZ; 3) Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ; 4) Department of Mathematics, University of Arizona, Tucson, AZ.

An analysis of six complete European genomes reveals that the ratio of X-linked to autosomal diversity deviates from the expected value of three-quarters. However, the direction of this deviation depends on whether a particular sequence is close to or far from the nearest gene. This pattern may be explained by stronger locally acting selection on X-linked versus autosomal genes, combined with larger effective population sizes for females than for males.


Evidence of natural selection at genetic regions associated with HIV-1 control is geographically restricted. Y. C. Klimentidis1, B. Aissani2, S. Shrestha2 1) Section on Statistical Genetics, Department of Biostatistics, University of Alabama at Birmingham, Birmingham, AL; 2) Department of Epidemiology, University of Alabama at Birmingham, Birmingham, AL.

HIV susceptibility and pathogenicity exhibit both inter-individual and inter-group variability. The etiology of inter-group variability is still poorly understood, and could be partly linked to genetic differences between groups. These genetic differences may be traceable to different regimes of natural selection in the 60,000 years since human radiation out of Africa. Here, we examine population differentiation and haplotype patterns at several loci identified through genome-wide association studies on HIV-1 -control, as determined by viral-load setpoint in Caucasian and African-American populations. We use the genome-wide SNP dataset on the Human Genetic Diversity Panel of 53 world-wide populations to compare measures of FST and extended haplotype homozygosity (EHH) at these candidate regions to the rest of the genome. We find that the Europe-Middle East pair-wise FST in the associated regions is elevated compared to the rest of the genome, while the sub-Saharan Africa-Middle East pair-wise FST is very low, suggesting that genetic differentiation (diversifying/positive selection) occurred outside of sub-Saharan Africa, while balancing or purifying selection occurred in sub-Saharan Africa. We also find greater EHH, indicative of recent positive selection at these associated regions, among all population sub-groups except for sub-Saharan Africans and Native Americans. These findings corroborate findings from other studies suggesting recent evolutionary change at immunity-related regions among Europeans, and shed light on the potential genetic and evolutionary origin of population differences in HIV-1 control.


Natural Selection at High Altitude: Andean Patterns of Adaptation to an Extreme Environment. A. Bigham1, M. D. Shriver2, L. G. Moore3, F. Leon-Valerde4, E. J. Parra5, T. D. Brutsaert6 1) Pediatrics, Univ Washington, Seattle, WA; 2) Department of Anthropology, Pennsylvania State University, University Park, Pennsylvania, USA; 3) Departments of Public Health Sciences, Anthropology and Obstetrics-Gynecology, Graduate School of Arts and Sciences, Wake Forest University, Winston-Salem, North Carolina, USA; 4) Departamento de Ciencias Biológicas y Fisiológicas, Universidad Peruana Cayetano Heredia, Peru; 5) Department of Anthropology, University of Toronto, Mississauga, Mississauga, Ontario, Canada; 6) Department of Exercise Science, Syracuse University, Syracuse, New York, USA.

High-altitude hypoxia, or the decrease in oxygen levels caused by lowered barometric pressure, challenges the ability of humans to live and reproduce. Human physiological responses to high-altitude have been extensively documented among long term high-altitude residents (i.e. Andeans and Tibetans). However, additional research is necessary to understand the genetic basis for the observed physiological traits. To this end, we performed a genome scan to identify selection nominated candidate genes or gene regions for high-altitude adaptation using the Affymetrix Inc. Genome-Wide Human SNP Array 6.0. We scanned across each chromosome to discern genomic regions with previously unknown function with respect to altitude phenotypes and examined groups of genes functioning in oxygen metabolism and sensing, such as the hypoxia inducible transcription factor (HIF) pathway, for evidence of directional selection. Candidate genes or gene regions were identified by comparing the patterns of variation between Andeans (n=49) and two low-altitude populations, Mesoamericans (n=39) and East Asians (n=90), using four tests for natural selection. The tests included the locus-specific branch length (LSBL), the log of the ratio of heterozygosities (lnRH), Tajima’s D, and a long-range haplotype test. Several genes including NOS2A, PRKAA1, EGLN1, and TNF as well as regions on chromosomes 12 and 15 showed significant evidence of natural selection. From these top candidates, ten single nucleotide polymorphisms (SNPs) were selected for further study and genotyped in a sample of n=141 Peruvian Quechua with associated resting and exercise phenotypic data. The subject participants included Q-SL subjects (Quechua sea-level) who were lifelong sea level residents transiently exposed to hypobaric hypoxia and Q-HA subjects (Quechua at high-altitude) who were lifelong residents of high-altitude. Significant associations were found for arterial oxygen saturation (SaO2) and genotype at PRKAA1 and the candidate region on chromosome 15 (repeated measures ANOVA, p<0.05 after FDR correction). These markers, along with the group effect (i.e. born high or born low), display relatively large effects on SaO2 with each explaining ~6-8% of the variance in this phenotype. Furthermore, the genotypic effects on SaO2 are broadly similar between the Q-SL and Q-HA groups. The results provide key insights into the patterns of genetic adaptation to high altitude in Andean populations. Natural Selection at High Altitude: Andean Patterns of Adaptation to an Extreme Environment. A. Bigham1, M. D. Shriver2, L. G. Moore3, F. Leon-Valerde4, E. J. Parra5, T. D. Brutsaert6 1) Pediatrics, Univ Washington, Seattle, WA; 2) Department of Anthropology, Pennsylvania State University, University Park, Pennsylvania, USA; 3) Departments of Public Health Sciences, Anthropology and Obstetrics-Gynecology, Graduate School of Arts and Sciences, Wake Forest University, Winston-Salem, North Carolina, USA; 4) Departamento de Ciencias Biológicas y Fisiológicas, Universidad Peruana Cayetano Heredia, Peru; 5) Department of Anthropology, University of Toronto, Mississauga, Mississauga, Ontario, Canada; 6) Department of Exercise Science, Syracuse University, Syracuse, New York, USA.

High-altitude hypoxia, or the decrease in oxygen levels caused by lowered barometric pressure, challenges the ability of humans to live and reproduce. Human physiological responses to high-altitude have been extensively documented among long term high-altitude residents (i.e. Andeans and Tibetans). However, additional research is necessary to understand the genetic basis for the observed physiological traits. To this end, we performed a genome scan to identify selection nominated candidate genes or gene regions for high-altitude adaptation using the Affymetrix Inc. Genome-Wide Human SNP Array 6.0. We scanned across each chromosome to discern genomic regions with previously unknown function with respect to altitude phenotypes and examined groups of genes functioning in oxygen metabolism and sensing, such as the hypoxia inducible transcription factor (HIF) pathway, for evidence of directional selection. Candidate genes or gene regions were identified by comparing the patterns of variation between Andeans (n=49) and two low-altitude populations, Mesoamericans (n=39) and East Asians (n=90), using four tests for natural selection. The tests included the locus-specific branch length (LSBL), the log of the ratio of heterozygosities (lnRH), Tajima’s D, and a long-range haplotype test. Several genes including NOS2A, PRKAA1, EGLN1, and TNF as well as regions on chromosomes 12 and 15 showed significant evidence of natural selection. From these top candidates, ten single nucleotide polymorphisms (SNPs) were selected for further study and genotyped in a sample of n=141 Peruvian Quechua with associated resting and exercise phenotypic data. The subject participants included Q-SL subjects (Quechua sea-level) who were lifelong sea level residents transiently exposed to hypobaric hypoxia and Q-HA subjects (Quechua at high-altitude) who were lifelong residents of high-altitude. Significant associations were found for arterial oxygen saturation (SaO2) and genotype at PRKAA1 and the candidate region on chromosome 15 (repeated measures ANOVA, p<0.05 after FDR correction). These markers, along with the group effect (i.e. born high or born low), display relatively large effects on SaO2 with each explaining ~6-8% of the variance in this phenotype. Furthermore, the genotypic effects on SaO2 are broadly similar between the Q-SL and Q-HA groups. The results provide key insights into the patterns of genetic adaptation to high altitude in Andean populations.

2 comments:

TGGP said...

Interesting that a light-skin allele was found at "moderate to high frequencies in East Africa". Isn't that where "ink black" tribes are supposed to live? Speaking of that region and another paper below, I recall a while back a paper discussing how Andeans, Tibetans and Ethiopean highlanders adapted to their low-oxygen environments. Andeans and Tibetans had different adaptations, but it was never noted how the Ethiopeans got by other than that they don't feature the same distinct adaptations as the others.

n/a said...

TGGP,

There is an abstract dealing with Ethiopia, as well:

Adaptations to high altitude in Ethiopia. G. Alkorta-Aranburu1, D. Witonsky1, C. Beall2, A. Di Rienzo1 1) Human Genetics , University of Chicago, Chicago, IL; 2) Anthropology Department, Case Western Reserve University, Cleveland, OH.

High-altitude (HA) resident human populations differ biologically from lowlanders; however, the biological bases of their differences are not well understood. The possibility of genetic adaptation to HA is supported by the high heritability of hemoglobin (Hb) concentration in Andean and Tibetan highlanders; a locus associated with higher oxygen saturation in Tibetans and positively selected haplotypes of EGLN1, PPARA and EPAS1 associated with lower Hb concentration in Tibetans. Epigenetic adaptation can also be hypothesized since invasive cancer clones show both genetic and epigenetic modifications due to hypoxia and nutrient deprivation.
To learn more about human adaptations to HA, we extended genomic analysis to two Ethiopian ethnic groups: the Amhara and the Oromo with, respectively, thousands of years and 500 years of residence at HA. Amhara populations at 3500-4000 and 1500 meters, and Oromo populations at 4000 and 1500 meters were sampled. HA samples show lower oxygen saturation and higher Hb concentration; but when ethnic groups are compared, Oromos show greater altitudinal differences. Then, methylation levels at ~27000 CpG sites in 17 HA and 17 low-altitude (LA) Oromos and Amharas were collected using Illumina’s HumanMethylation27 BeadChips. While methylation levels differ significantly between Amharas and Oromos, no evidence for epigenetic differences between HA and LA within the same ethnic group is observed. Next, genome-wide genotype data were collected in 48 HA and 48 LA Amharas, and 24 HA and 24 LA Oromos using Illumina’s 650Y or 1M SNP arrays. When allele frequency differences between HA and LA Ethiopians are compared to expectations, the Amharas and, especially, the Oromos show a significant excess of high-FST values. Interestingly, some of the genes with high-FST SNPs are good candidates for hypoxia driven adaptation based on their biological function; e.g., the Parkinson’s disease gene SEMA5A that promotes angiogenesis by increasing endothelial cell proliferation, migration, and decreasing apoptosis. In addition, these data allow us to test whether the variants and genes associated with Hb concentration and/or signals of adaptation to HA in Tibetans also play a role in Ethiopian adaptations to HA. We replicate in the Oromo one EPAS1 SNP significantly associated with Hb concentration in two Tibetan cohorts.