Differentiated demographic histories and local adaptations between Sherpas and Tibetans

Abstract

Background: The genetic relationships reported by recent studies between Sherpas and Tibetans are controversial. To gain insights into the population history and the genetic basis of high-altitude adaptation of the two groups, we analyzed genome-wide data in 111 Sherpas (Tibet and Nepal) and 177 Tibetans (Tibet and Qinghai), together with available data from present-day human populations.

Results: Sherpas and Tibetans show considerable genetic differences and can be distinguished as two distinct groups, even though the divergence between them (~3200-11,300 years ago) is much later than that between Han Chinese and either of the two groups (~6200-16,000 years ago). Sub-population structures exist in both Sherpas and Tibetans, corresponding to geographical or linguistic groups. Differentiation of genetic variants between Sherpas and Tibetans associated with adaptation to either high-altitude or ultraviolet radiation were identified and validated by genotyping additional Sherpa and Tibetan samples.

Conclusions: Our analyses indicate that both Sherpas and Tibetans are admixed populations, but the findings do not support the previous hypothesis that Tibetans derive their ancestry from Sherpas and Han Chinese. Compared to Tibetans, Sherpas show higher levels of South Asian ancestry, while Tibetans show higher levels of East Asian and Central Asian/Siberian ancestry. We propose a new model to elucidate the differentiated demographic histories and local adaptations of Sherpas and Tibetans.

Keywords: Gene flow; High-altitude adaptation; Next-generation sequencing; Population history; Sherpa; Tibetan.

Fig. 1

Culturally defined regions of historical Tibet and geographical locations of Sherpa and Tibetan samples analyzed in this study. The culturally defined regions in historical Tibet are illustrated in different colors: red, Ü-Tsang (central Tibet); green, Kham (eastern Tibet); and purple, Amdo (northeastern Tibet). Dots with distinct colors represent subgroups classified according to the collected geographical locations: blue for two SHP subpopulations; and non-blue for regional Tibetans. The locations of Amdo and Kham regions make Tibetans there more easily influenced by cultures and genetics from East Asians or Central Asians/Siberians. The figure was modified from one obtained from Wikipedia (https://en.wikipedia.org/wiki/Kham)

Fig. 2

Principal component analysis (PCA) for SHP, TBN, and their subgroups. PCA of a SHP and TBN within the context of some East Asians, b SHP and TBN, c SHP subgroups, and d TBN subgroups. Subgroups are classified according to their geographic locations. Numbers in parentheses denote variance explained by each principal component (PC). Note that three outliers in d (one in TBN.Nyingchi and two in TBN.Shigatse) were removed when we drew the figure

Fig. 3

Panel 2 dataset-based results of genetic admixture when assuming K = 6. Each individual is represented by a single line broken into K = 6 colored segments, with lengths proportional to the K = 6 inferred clusters. Results for all SHP and TBN are further summarized and displayed in the two large pie charts in the center of the circle plot with component proportion denoted as percentage. Proportions of each genetic component for SHP and TBN subgroups are summarized in the small pie charts with their proportions listed below. TC Tibetan major component, SC SHP major component, EAC East Asian major component, SC1 Central Asian/Siberian major component 1, SC2 Central Asian/Siberian major component 1, SAC South Asian major component

Fig. 4

Evidence of gene flow between SHP and TBN subgroups. We performed f ₃ tests to detect gene flow events from the TBN subgroup to SHP subgroup (Additional file 1: Figure S23), SHP subgroup to TBN subgroup (Additional file 1: Figure S24), and within SHP subgroups (Additional file 1: Figure S25). The f ₃ statistics were significantly negative (with Z score ≤3) for: a f ₃(SHP.Zhangmu; TBN.Shigatse, X) when X was assumed as a South Asian population; b f ₃(TBN.Nyingchi; SHP.Khumbu, X) when X is an East Asian population; and c f ₃(SHP.Zhangmu; SHP.Khumbu, X) when X was South Asians and some Central Asians/Siberians. Results provide evidence for gene flow from South Asians and Nepalese Sherpas to Chinese Sherpas, and from East Asians and Nepalese Sherpas to Tibetans in Nyingchi. **Significantly negative value with Z scores ≤3; *score of 3 < Z ≤ 2. Highlander subgroups are highlighted with red fonts and blue arrows

Fig. 5

The historical effective population size (N _e) and divergence time between SHP and TBN. Estimates of a N _e and divergence time between b SHP.Zhangmu and others and c SHP.Khumbu and others using MSMC. The N _e was estimated using autosomal sequences of two genomes (four haplotypes) for each population. Divergence time between each pair of populations was evaluated using autosomal sequences of four genomes, i.e., two individuals for each population. An autosomal mutation rate (μ _Auto) with 1.25 × 10⁻⁸ per base-pair per generation and 25 years per generations (g) were used

Fig. 6

A proposed model of demographic history of SHP and TBN. A simplified model for the origins and evolutionary history of Tibetans and Sherpas based on the observations and estimations from this study. GF gene flow, MRCA most recent common ancestor. Dashed lines indicate gene flow events and arrows denote directions. MRCA1, MRCA2, and MRCA3 are based on Fig. 5b. We inferred GF1 from the treemix results (Additional file 1: Figures S31 and S32) and the observation that both SHP (mainly for Chinese Sherpa) and TBN contain an East Asian genetic component (EAC) (Fig. 3). GF2 was based on the excess EAC in TBN compared to SHP (Fig. 3; Additional file 1: Figure S27). Based on the f ₃ tests (Fig. 4b; Additional file 1: Figure S22) and the higher proportion of EAC in Kham and Amdo Tibetans (Fig. 3), we confirmed GF3. GF4 is based on Fig. 4b and Additional file 1: Figure S23 and the historical record that Sherpas migrated from the Kham region in eastern Tibet to Nepal within the last 300–400 years, possibly supporting the genetic contact between Khumbu Sherpas and Kham Tibetans. GF5 is based on the excess Sherpa genetic component in Ü-Tsang Tibetans compared to that in Kham and Amdo Tibetans (Fig. 3) and also on the results shown in Additional file 1: Figure S26. GF6 is based on Fig. 4a. The higher South Asian component in Chinese Sherpas compared to that in Nepalese Sherpas (Fig. 3) and the f ₃ statistics (Fig. 4a) validated the presence of GF7. Population substructures in both SHP and TBN are based on PCA (Fig. 2), ADMIXTURE (Fig. 3), F _ST (Additional file 1: Figures S4 and S5), outgroup f ₃ tests (Additional file 1: Figures S8–S10), and D statistics (Additional file 1: Figures S28–S30). Estimates of MRCA1, MRCA2, and MRCA3 are based on Fig. 5b and Additional file 1: S34

Fig. 7

Example of a putatively functional adaptive variant. A novel missense variant (chr17: 19645417) located in ALDH3A1 was selected as an example. a The derived allele frequency (DAF) of this SNP in SHP and TBN was estimated based on the Target-genotyping panel (Tables 1 and 2). b Median-joining network of ALDH3A1 showing a Sherpa-specific haplogroup. Haplotypes consisted of the missense variant and 30 randomly selected shared variants between SHP and non-SHP residing at the ALDH3A1 region with minor allele frequency (MAF) larger than 5%. The derived allele is specific to SHP in the SHP-specific haplogroup. c Positive selection signals of extended haplotype homozygosity (EHH) and Integrated Haplotype Score (iHS). Analyses in b and c are based on 55 imputed genomes of Zhangmu Sherpas. d Functional consequences of the missense variant