Ancient genome of Empress Ashina reveals the Northeast Asian origin of Göktürk Khanate

The following supplementary material is available online for this article at http://onlinelibrary.wiley.com/doi/10.1111/jse.12938/suppinfo: Fig. S1. The geographical location of this individual and Records of Empress Wude. Fig. S2. Proportion of C > T and G > A substitutions in human DNA across DNA fragments in the Ashina individual. The red curve represented C > T substitutions that increased at the 5′-end and the blue curve represented G > T substitutions that increased at the 3′-end, as expected for authentic ancient DNA. Fig. S3. Detailed principal component analysis (PCA) of 1667 Eurasian exhibited the west–east Eurasian cline and north–south Eastern Eurasian cline. The Ashina individual fell within present-day and ancient Eastern Eurasians along PC2, and clustered with modern Tungusic and Mongolic speakers, ancient populations in Northeast Asia and eastern Mongolia Plateau, especially the populations previously referred to as "Ancient Northeast Asian" (ANA) that are genetically homogeneous hunter-gatherers from Northeast Asia ("Baikal_EN", 5200–4200 BCE, "Mongolia_N", 6000–4400 BCE, "AR_EN", 5500–5300 BCE, and "DevilsCave_N", 5700 BCE) and post-Iron Age East Steppe nomadic people including Xianbei, Rouran, Khitan and part of Mongols. Fig. S4. ADMIXTURE results for selected Eurasians for K = 2–6. A, Cross-validation (CV) errors for the ADMIXTURE analysis based on 1240 K data set. B, CV errors for the ADMIXTURE analysis based on HO data set. C, ADMIXTURE results for K = 2–6 based on 1240 K data set. After pruning for linkage disequilibrium, the number of SNPs included in this analysis was 871 858, the included modern populations were from HGDP. D, ADMIXTURE results included Turkic populations for K = 2–6 based on HO data set. The number of included SNPs was 279 737. The ADMIXTURE analyses base on 1240 K and HO data set both revealed two summarized ancestries in Ashina: one related to West Eurasian ancestry, the dominating one associated with East Eurasians including ANA and southern ancestry. The results of ADMIXTURE also presented the genetic discrepancy among Ashina, earlyMed_Turk and CentralSteppe_Turk and the genetic heterogeneity among Ashina and present-day Turkic-speaking populations. Fig. S5. Close genetic relationship between ancient and present-day Eurasian populations and geographically different individuals of the Türkic Khaganate. We presented top 60 outgroup-f3 signal for each different group of the Türkic Khaganate. Horizontal bars represent the point estimate ±3 standard error. The outgroup-f3 of Ashina showed the close genetic affinity with East Eurasians than West Eurasians, especially northern East Eurasian. earlyMed_Turk who sampled in Mongolia Plateau presented close genetic relationship with East Eurasians, especially Rouran.The CentralSteppe_Turk harbored similar genetic relationship with Eurasians. Fig. S6. f Statistics in forms of f4 (X, Mbuti; Ashina, pre-Iron Age Northeast Asian populations) and f4 (pre-Iron Age Northeast Asian populations, Mbuti; Ashina, X) test whether pre-Iron Age Northeast Asian populations make contribution to the formation of Ashina. Darker orange and blue squares showed sharing at |Z-score | > 6. Blue and orange squares show sharing at 3 < |Z-score | <6. Gray squares indicate significantly less sharing at the same threshold. A, f4 (X, Mbuti; Ashina, pre-Iron Age Northeast Asian populations) showed Ashina formed a genetic clade together with Neolithic hunter-gatherers in Northeast Asia including Amuer River, Mongolia Plateau and DevilsCave, Baikal Lake (Russia_Shamanka_Eneolithic). In addition, Ashina showed close genetic affinity with population related to Bronze Age SlabGrave and Ulaanzukh culture in Mongolia Plateauwho harbors dominating ANA ancestry. B, f4 (pre-Iron Age Northeast Asian populations, Mbuti; Ashina, X) further provided evidence that ancestry related to hunter-gathers in Northeast Asia dedicated to the gene pool of Ashina as reflected in significantly positive f4 with the exception of Sino-Tibetan populations. Fig. S7. The analysis of f4 (West Eurasian, Mbuti; Ashina, East Eurasian) detected whether gene influx from West Eurasian flowed into the gene pool of Ashina basing on 1240 K data set. A–L, West Eurasians included West Steppe nomadic populations from Bronze Age to Iron Age, Anatolian and Iranian farmers, East hunter-gatherers (EHG), ANE (MA1, Ancient North European) and Botai and the Bactria-Margiana Archaeological Complex (BMAC) of Central Asia. East Eurasian included reported population without gene flow from West Eurasian in previous studies. Different shapes represented different Z-scores of f4. The results showed the gene pool of Ashina was influenced by the eastward migration of West Steppe pastoralists and the population related to BMAC who was attested to migrated along Inner Asia Mountains Corridor into the Mongolia Plateau and influenced ancient Tianshan and Mongolian people (f4 > 0). The detected Anatolia farmers, EHG and Iranian farmers, Botai-related ancestry in Ashina might be mediated by the eastward migration of West Steppe pastoralists and BMAC, respectively. Fig. S8. The significantly genetic heterogeneity within the Türkic Khaganate. A, B, The genetic divergency in individuals of the Türkic Khaganate revealed by f4(X, Mbuti; Ashina, earlyMed_Turk/CentralSteppe_Turk) showing earlyMed_Turk/CentralSteppe_Turk harbored more allele with West Eurasian than Ashina. A, earlyMed_Turk. B, CentralSteppe_Turk. Fig. S9. The genetic relationship of Ashina with Iron-Age Tianshan nomadic populations. A–C, The results of f4 in form of f4 (X, Mbuti; Ashina, Kazakhstan_Kanju/Kazakhstan_Wusun/Kyrgzstan_TianshanHun) all showed Iron-Age Tianshan nomadic people harbored diverse genetic profile with Ashina with more genetic affinity with West Eurasian and West Siberian hunter-gatherers (WSHG). In contrast, Ashina shared more allele with East Eurasians than Iron-Age Tianshan nomadic populations. Different shapes represented different Z-score. Fig. S10. The genetic relationship of Ashina with Xiongnu populations in different periods. Jeong et al. ( 2020) split early Xiongnu into two subgroups, earlyXiongnu_west (SKT010, SKT001, SKT003, SKT009, SKT008, AST001) and earlyXiongnu_rest (JAG001, SKT002, SKT004, SKT005, SKT006, SKT012), based on their individual genetic modeling results. A, B, The results of f4 in form of f4 (X, Mbuti; Ashina, earlyXiongnu_rest/earlyXiongnu_west) showed Ashina shared more genetic affinity with East Eurasians than early Xiongnu. The genetic discrepancy with Ashina was different in earlyXiongnu_rest and earlyXiongnu_west, showing the genetic affinity with ancient populations in Northeast Asia and Mongolia Plateau was similar in earlyXiongnu_rest and Ashina (no significant Z-score). While that genetic profile did not exist in earlyXiongnu_west who harbored more West Eurasian affinity than Ashina. C–E, The results of f4 in form of f4 (X, Mbuti; Ashina, lateXiongnu/lateXiongnu_han/lateXiongnu_sarmatian) presented the difference of harboring genetic relationships with Asina among late Xiongnu: late Xiongnu shared similar genetic affinity to ancient Northeast Asians with Ashina (non-significant Z-score when X included Northeast Asian populations), lateXiongnu_han possessed semblable genetic profile with Ashina (non-significant Z-score, except for AR_EN), that contrasted with lateXiongnu_sarmatian who shared more genetic affinity with West Eurasians than Ashina (significantly negative Z-score when X was West Eurasian populations). Fig. S11. The genetic relationship of Ashina with contemporary East Steppe nomadic populations including Xianbei and Rouran. A–C, The result of f4 in form of f4(X, Mbuti; Ashina, AR_Xianbei_IA/Mongolia_Xianbei/Rouran) revealed the genetic homogeneity of Ashina with Xianbei and Rouran (no significant Z score). Fig. S12. The genetic relationship of Ashina with Mongolic and Tungusic East Steppe pastoralists after the Türkic Khanate. (A–C) The result of f4 in form of f4(X, Mbuti; Ashina, lateMed_Khitan/lateMed_Mongol/Russia_Heishui_Mohe_early_Medieval) demonstrated the discrepancy of genetic relationship with Ashina in Khitan and Mongol empire; Mongolic Khitan and Tungusic Heshui_Mohe had genetic similarity with Ashina, while Mongols in Mongol empir period showed genetic differentiation with Ashina. Fig. S13. The genetic relationship of Ashina with historically Turkic-speaking pastoralists of Central Steppe after the collapse of the Türkic Khaganate. A–E, The result of f4 in form of f4 (X, Mbuti; Ashina, Kimak/KaraKhanid/Uigur/Karluk/Kipchak) indicated the high genetic differentiation between Ashina and later Turkic-speaking pastoralists of Central Steppe including Kimak, KaraKhanid, Uigur, Karluk and Kipchak Khaganate. The Central Steppe pastoralists had evident genetic affinity with West Eurasians (significantly negative Z-score when X included West Eurasians), indicating the prominent ancestry from West Eurasian in the Central Steppe pastoralists, that contrasted with Ashina who harbored dominating ancestry from Northeast Asian. Fig. S14. The genetic relationship of Ashina with Central Steppe pastoralists in Medieval period. A–C, The result of f4 in form of f4(X, Mbuti; Ashina, Kazakhstan_His/Kazakhstan_GoldenHordeAsian/CentralSteppe_Medieval_Nomad) showed diverse genetic relationship with Ashina among Central Steppe pastoralists in Medieval period. Kazakhstan_GoldenHordeAsian formed a genetic clade together with Ashina, which indicating the similar genetic profile between them. However, Kazakhstan_His and CentralSteppe_Medieval_Nomad showed closer genetic affinity with West Eurasians than Ashina. Fig. S15. Genetic heterogeneity of the Türkic Khaganate. Heatmap of P-value of pairwise qpWave among post-Iron Age Central/East Steppe pastoralists. "++" represented values greater than 0.05, and "+" represented values <0.05 and >0.01. Computations were based on the outgroup set (Mbuti.DG + Onge.DG + Russia_MA1_HG.SG + Russia_Kostenki14.SG + Iran_GanjDareh_N + Kazakhstan_Eneolithic_Botai.SG + Russia_Sintashta_MLBA.SG + AR19K + Mongolia_N_North+UpperMid_YR_LN). Fig. S16. The genetic relationship with Ashina of present-day Altaic-speaking populations. A, We conducted f4(X, Mbuti; Ashina, Tungusic/Mongolic/Turkic) to investigate the genetic relationship of present-day Altaic populations with Ashina, and found that Mongolic and Tungusic speakers formed genetic clade with Ashina, and influenced by additional gene flow from millet farmers in Yellow River or West Eurasian in some groups (negative Z-score). By contrast, Turkic populations showed extremely disparate genetic profile with Ashina with the exception of Yakut, Dolgan, Tuvinian and Salar who presented some extent of genetic affinity with Ashina. B, We further performed f4 (Mbuti, Ashina; Tungusic/Mongolic, Turkic) to provide robust evidence of supporting the close genetic affinity with Ashina in Tungusic and Mongolic populations rather than in Turkic populations. Tungusic and Mongolic speakers did have closer genetic relationship with Ashina than Turkic populations (Z < -3). "-"represented −6 < Z-score < −3; "--" represented Z-score < −6; "+" represented 3 < Z-score < 6; "++" represented Z-score >6. C, The result of f4 (Mbuti, Ashina; Tungusic, Mongolic) further revealed Tungusic population shared more alleles with Ashina than Mongolic populations as reflected in significantly negative f4 values with the exception of Eavek_FastEast. Fig. S17. The genetic affinity between Turkic-speaking populations and ancient populations related with the diffusion of Turkic language. Heatmap of P-value of pairwise qpWave among Turkic-speaking populations and ancient populations related with the diffusion of Turkic languages. The genetic relationships between Turkic-speaking population and ancient populations associated with the diffusion of Turkic language revealed the significantly genetic differentiation of Ashina/earlyMed_Turk/earlyXiongnu/lateXiongnu/earlyMed_Uigur and Turkic-speaking populations (P < 0.01), CentralSteppe_Turk formed pairwise clade with Uzbek and Dolgan, the formed pairwise clades existed in other pairwise combination, including pairwise combination of Kimak with Tatar_Siberian, Uyghur, Karakalpak, Nogai_Astrakhan, Nogai_Stavropol and Uzbek, pairwise combination between lateXiongnu to Karakalpak (P > 0.05). "++" represented values >0.05, and "+" represented values <0.05 and >0.01. Computations were based on the outgroup set (Mbuti.DG + Onge.DG + Russia_MA1_HG.SG + Russia_Kostenki14.SG + Iran_GanjDareh_N + Anatolia_N + CHG + AR19K + Mongolia_N_North+UpperMid_YR_LN). Fig. S18. The potential East Eurasian ancestral source of Turkic population. We performed f4(X, Mbuti; Pop, Turkic) to further determine whether other East Eurasians are more suited to explain the East Eurasian ancestry of Turkic than Ashina. X was chosen from representative populations including Chuanyun (the Southern East Asian), YR_LN (millet farmers in Yellow River), Russia_Sintashta_MLBA and Botai. Pop included other potential East Eurasian ancestral source of Turkic-speakers. We observed that Ashina was more suited to approximate East Eurasian ancestry of Turkic populations where Z-scores of f4 (X, Mbuti; Ashina, Turkic) tended to approach to zero. Supplementary Document. Section 1 Archaeological Site and Sample Description of the Xiaoling Mausoleum. Section 2 Material and Methods. Section 3 Ancient Data Analyses and Genetic Characterization of Ancient Individual. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.