Sampling

The excavation of the Xiaohe cemetery began in 2002. The lowest layer of the cemetery, comprising a total of 41 graves of which 37 have human skeletal remains, was excavated by the Xinjiang Archaeological Institute and the Research Center for Chinese Frontier Archaeology of Jilin University from 2004 to 2005. After the appropriate recording, the skeletal remains of 30 well-preserved individuals, together with sandy soil, were packed in cardboard boxes and sent to the ancient DNA laboratory of Jilin University, where they were stored in a dry and cool environment. All the samples were collected by two highly skilled scientists in our group, equipped with gloves and facemasks. As a result of the saline and alkaline character of the sand, the dry air and good drainage, the human remains are in excellent conditions. One intact femur and two tooth samples for each set of human remains were selected for DNA analysis, except for samples 84 and 121, which have only tooth samples. The archaeological information regarding the samples used in this study is summarized in Additional File 1.

Contamination precautions and decontamination

Previous studies have shown that the recovery of authentic ancient human DNA is possible [4, 5] when strict precautions are taken to prevent contamination. The following is a summary of the measures taken to avoid contamination and to ensure authenticity in the present study.

(1) Bone powdering, DNA extraction and amplification were carried out in three separate rooms in our laboratory which is dedicated solely to ancient DNA studies; all staff wore laboratory coats, facemasks and gloves and strict cleaning procedures [frequent treatment with bleach and ultraviolet (UV) light] were applied.

(2) In order to prevent contamination, polymerase chain reaction (PCR) tubes, tips, microcentrifuge tubes and drills were sterilized by autoclaving. Some of the reagents were exposed to UV light at 254 nm for at least 30 min. Tip boxes were soaked in 10% sodium hypochlorite solution. Extraction and amplification blanks were included in every PCR assay in order to detect any potential contamination from sample processing or reagents even though they were guaranteed DNA-free by the manufacturers.

(3) Multiple extractions and amplifications from the same individual were undertaken at different times and from two different parts of the skeleton by three different laboratory members.

(4) Three samples were sent to Fudan University in Shanghai, China, for an independent repetition.

(5) Cloning analysis was performed with all the samples, in order to detect potential heterogeneity in the amplification products due to contamination, DNA damage or jumping PCR.

(6) Different length fragments were amplified. Since there is an inverse correlation between fragment length and amplification efficiency for ancient DNA, contamination from modern DNA could be identified by this assay.

(7) Ancient DNA from animal remains (goat or cattle) found at the same site was isolated and amplified using the same procedures as those used for the human ancient DNA, again providing a negative control for our study.

(8) The sex of some of the morphologically intact individuals was determined by amplifying the amelogenin (AMG) gene and the results were compared to that of the morphological examination in order to monitor any potential contamination in the extraction.

Sample preparation and DNA extraction

The bone and tooth samples were processed independently. A fragment of bone, about 3 cm long, was cut from the intact femur or tibia by sawing. In order to remove any possible surface contamination of the samples by external DNA, each bone was drilled three times with three different drills to remove a layer about 1-3 mm from the top after removing the external soil using brush, and then soaked in a 5% sodium hypochlorite solution for 10 min, rinsed with distilled water and absolute alcohol successively and UV-irradiated (254 nm) on all sides for at least 45 min in a clean room. The samples were then pulverized by Freezer Mill 6850 after immersion in liquid nitrogen. Intact teeth were handled differently: they were soaked in a 5% sodium hypochlorite solution for 15 min after a preliminary treatment in which they were wiped with 10% sodium hypochlorite and then rinsed with absolute alcohol. After that, all sides of the samples were exposed to UV radiation (254 nm) for a minimum of 20 min before the sample was ground to powder by Freezer Mill 6750. DNA was extracted by means of a silica-based protocol [10]. One extraction blank was included for every three ancient samples. In brief, 0.5-2 g tooth/bone powder was incubated about 20 h at 50°C in lysis solution (1 mL 10% SDS, 4 ml EDTA (pH 8; 0.5 M, Promega, WI, USA), 100 uL of 10 mg/ml proteinase K (Merck, Darmstadt, Germany). After centrifugation, the solution was subsequently concentrated with centricons (Millipore, MA, USA) up to about 100 μL volume, and then the extraction proceeded according to the handbook of the QIAquick DNA Purification Kit (Qiagen, Hilden, Germany).

DNA amplification, cloning, and sequencing

Data analysis

Sequence alignments were performed using the Clustalx1.8 software. Comparison of DNA sequence homology was performed with Blast from the National Centre for Biotechnology Information. Median networks were constructed by Network 4.5 using a reduction threshold and the different weights for SNPs loci but the same weight for all HVSI polymorphism loci. No statistical analysis was performed in this study due to the small sample size.

Authenticity of results

A total of 23 reproducible mtDNA fragments (360 bp) were obtained from 30 individual sets of the Xiaohe human remains, after discarding seven samples due to failed amplification or irreproducible results. Three of the 23 DNA fragments were similar to the DNA of two people involved in the study and were also removed from the study, even though they yielded consistent results through three or more independent extractions.

The remains of 20 individuals were selected for cloning analysis. Eight to twelve clones from two independent amplifications of the 20 individuals were selected for automated DNA sequencing. Although a few positions were different from direct sequencing of the PCR products, which could be due to random Taq misincorporation or DNA damage, the consensus sequence from cloning was consistent (see Additional File 2). In order to validate the results generated in our laboratory, human remains from three Xiaohe individuals (100, 119, and 127) were sent to the ancient DNA laboratory of Fudan University for further analysis and identical results were obtained.

The following facts further confirm the authenticity of our results: (1) an inverse correlation between the size of the PCR amplicons and the amplification efficiency for the Xiaohe samples (209 bp > 235 bp > 399 bp) was found; (2) molecular sexual identification results were in accordance with morphological sex assignments; (3) HVR-I sequences match well with the key coding region SNPs according to the well-defined mtDNA phylogenetic tree [16, 17]; and (4) parallel studies were also performed on Xiaohe animal samples, and no human DNA was detected. Collectively, 20 mtDNA fragments from 30 human remains were obtained that were inferred to be authentic (Table 3). The sequences have been submitted to GenBank with accession numbers FJ719792-FJ719811. This high success rate suggests that the DNA from the Xiaohe human remains is well preserved, as would be expected in samples originated from dry environmental conditions.

MtDNA haplogroup profile and distribution

The 20 mtDNA fragments containing eight different haplotypes that can be further assigned to five haplogroups, which belong to subhaplogroups of macrohaplogroups M and N (Figure 2).

The dominant haplogroup in the Xiaohe people was the East Eurasian lineage C, shared by 14 Xiaohe individuals who were associated with two different mtDNA haplotypes (S1 and S2). According to the coding region 11969 G to A, all lineage C found in the Xiaohe people was further changed to subhaplogroup C4, which had D-loop group-specific polymorphisms at nucleotide positions (np) 16298 (T to C) and 16327 (C to T) [18]. Interestingly, the haplotype S1 shared by these 10 of the 14 Xiaohe individuals did not have the 16223T, a mutation had been found in the majority of modern lineage C populations. This haplotype was found merely in modern Evenks and Udegeys of southeastern Siberia but the frequency is low [18]. It is interesting to note that our study found a mutation, unique to the Xiaohe people but rare among lineage C, 16309 G (see Additional File 3). To the best of our knowledge, the lineage C with 16309 G was observed only in three people from modern central Asia to date, who also possess another two mutations in HVRI region [19].

Besides the East Eurasian lineage, two West Eurasian mtDNA haplogroups H and K were found among the Xiaohe people. H lineage is the most common mtDNA haplogroup in West Eurasia [20], but haplogroup H with a 16260T was shared by only nine modern people in Genbank, including one Italian, one German, one Hungarian, one Portuguese, one Icelander and four English people. Haplogroup K, a western Eurasian-specific haplogroup, is mainly distributed in Europe, central Asia, and Iran [20, 21]. However, haplogroup K with 16134T, found in the Xiaohe people, has not been found in modern people to our knowledge.

Among the Xiaohe people, three sequences with the unique HVRI motif 16189-16192-16311 formed a subcluster (Figure 2) and were not shared by modern people. They are identified as macrohaplogroup R through sequencing the PCR amplicons at np10400 and np12705 in the coding region. The np12308, np14766, np10031, np4917, np3970, and 9 bp deletion, which are the diagnostic sites for the main subhaplogroups of R, were further examined [15]. The results showed that they are related neither to the West Eurasian haplogroups UK, TJ, HV, R11 and R1, nor to the East Eurasian haplogroups B and F. So we designated them as haplogroup R* temporarily. Another sequence with motif 16223-16304, shared by some people from East Asia, India, and Europe, was assigned to haplogroup M*.

Y chromosome haplogroup profiling and distribution

Fifteen individuals' AMG amplicons were obtained from the 20 Xiaohe individuals (whose mtDNA was successfully amplified), among which seven individuals were identified as male and eight as female. The Y chromosome haplogroup of the seven males were all assigned to haplogroup R1a1a through screening the Y-SNPs at M89, M9, M45, M173 and M198 successively. Haplogroup R1a1a is widely distributed in Eurasia: it is mainly found in Eastern Europe, Central Asia, South Asia, Siberia, ancient Siberia, but rare in East Asia [22–24].