HVS-I polymorphism screening of ancient human mitochondrial DNA provides evidence for N9a discontinuity and East Asian haplogroups in the Neolithic Hungary

Abstract

Analysis of mitochondrial mutations in the HVS-I region is an effective method for ancient human populational studies. Discontinuous haplotype data between the first farmers and contemporary Europeans has been described before. Our contribution is based on a survey initiated on the Neolithic skeletons from Hungarian archaeological sites in the Alföld. This Lowland, the Hungarian Plain, is well excavated as an important region for spread of Neolithic culture from Near East and Balkans toward Central and Western Europe, started circa 8000 years ago. HVS-I sequences from nt15977 to nt16430 of 11 such specimens with sufficient mitochondrial DNA preservation among an extended Neolithic collection were analysed for polymorphisms, identifying 23 different ones. After assigning all single-nucleotide polymorphisms, a novel, N9a, N1a, C5, D1/G1a, M/R24 haplogroups were determined. On mitochondrial control mutations at nt16257 and nt16261, polymorphic PCRs were carried out to assess their distribution in remains. Neolithic data set was compared with contemporary Vác samples and references, resulting in higher frequency of N9a in Alföld as a remarkable genetic discontinuity. Our investigation is the first to study mutations form Neolithic of Hungary, resulting in an outcome of Far Eastern haplogroups in the Carpathian Basin. It is worth further investigation as a non-descendant theory, instead of a continuous population history, supporting genetic gaps between ancient and recent human populations.

Introduction

Sequence variations of mitochondrial DNA (mtDNA) have been the focus of many human population genetic studies with ancient DNA. A great number of investigations has analysed the Y chromosome genetic data and/or the hypervariable regions of mtDNA, latter with the advantage of higher copy number over genomic DNA, and extensive population-level variability. mtDNA is harboured in the mitochondrion, the particle inherited only maternally and not affected by remarkable recombination. Altogether, these features make ancient mtDNA a very useful tool for prehistoric human lineage tracking.¹ For it, most studied part of the 16 569-bp long human mtDNA is the displacement loop, called D-loop or control region, from position 16 024 to position 576 of the circular mtDNA molecule.² This element is involved in the regulation of transcription and replication with the three hypervariable sequences, called HVS-I, HVS-II and HVS-III. At population level, in comparison with the rest of the genome, these are extremely variable elements of mtDNA. Several approaches co-exist in the field, regarding the different scopes of the different studies available. In population genetics, HVS-I is taken to be from nucleotide 16 024 to 16 400 to encompass the phylogenetically important high proportion of sequence variation within a very small and easily sequencable stretch. At the rest of hypervariability, 44–340 is regarded to be HVS-II and 438–576 is regarded to be HVS-III.

With the usage of specific polymorphisms originated from these regions of mtDNA, groups of haplotypes (maternal lineages) or haplogroups were created.³ Assignments showed fairly a great level of difference when comparing those of different continents to each other (www.mitomap.org). Major European mtDNA lineages were H, V, T, J, U, K, R1, R2, I, N1, X, W.⁴ Neolithic aDNA fragments of C haplogroup were also successfully extracted and sequenced from a Siberian skeletal remain,⁵ as well as ancient mtDNA haplotypings of some Europeans from Neolithic gene pools were published. These studies showed haplotypes such as H, N1a, K, HV, T, V, J, U3 from Central Europe,⁶ and H, I1, J1c, W1, T2, U4 from Iberia,⁷ as well as H, J, T, U4/H1b, U5, V from Northern Europe.⁸ Most of them are supposed to be involved in the West-Eurasian clad of mtDNA variants, although in some cases, they have been present with markedly different frequencies to the ones identified in the contemporary gene pools (For example N1a in Central Europe⁶).

DNA fragments in archaeological remains are able to survive for tens of thousands of years, but surprisingly, almost complete degradation just within a few months can occur as well.⁹ Stepping over the limitations, out of the achievements in Molecular Anthropology, the circa, 40 000-year-old Neanderthal,¹⁰ was one of the first to gain ancient mtDNA sequences from. Furthermore, determination of the whole sequence of the Neanderthal human made it possible to generally compare evolutionary aspects of polymorphisms.¹¹ Regarding the prehistoric humans, on the basis of polymorphisms revealed via sequencing, very interesting results were found in connection with the non-descendant theory on the first farmers of Central Europe, dating back to the Neolithic times.⁶

The quality and quantity of authentic DNA molecules in an ancient sample depend on many physical and chemical factors, which are not easy to predict even if the taphonomic history of a sample is known. Samples are also threatened by contamination with contemporary DNA, which can be ruled out using appropriate laboratory cautions.¹² Hydrolytic damage to the bonds in the phosphate sugar backbone of the DNA molecule results in single strand nicks and irreversible cross-linking between the strands of double helix, causing insensitivity towards enzymatic modifications, such as PCR amplification by polymerases. Hydrolytic attack on the sugar can cause depurination and deamination.¹³ DNA modifications take place as a function of time and have a strong impact on the quality and integrity of ancient DNA, decreasing the efficiency of the PCR, as well as the size of the detectable PCR products.⁹ Authentic ancient DNA quantities are usually limited with an accompanying phenomena of further fragmentation and cross-linking,¹⁴ so that researchers are rarely able to amplify templates up to a few hundred bp in size.¹⁵ Ancient mutations by PCR screening can only be trapped, if replication of experiments, including sequencing and cloning, are available. Thus, DNA degradation presents several problems in prehistoric samples;¹⁶ however, the rest of the opportunities of aDNA has been proven to be powerful tools in population reconstruction, because morphometry of skeletal remains alone appeared insufficient to describe migration events of historic populations.

The arrival of a presumed Neolithic culture to Europe, accompanied by continent-wide and small-scale migrations, has been just the last steps in human prehistory.¹⁷ Two extreme hypotheses were proposed on the spread of farming during the Neolithic in Europe. The replacement hypothesis suggested that the extensive immigration by demic diffusion from the Near East enhanced the onset of agriculture.¹⁸ The other extreme model was based on a trade and cultural diffusion system, which would have left the gene pool of prehistoric Europe essentially autochthonous.¹⁹ There must have been clearly a spectrum of possible events between the two extremes,²⁰ caused by the consequences of differences among geographic regions, as analysed in Central Europe,⁶ in Iberia,⁷ as well as in Scandinavia.⁸ The Hungarian Plain in Central Europe, which is mainly the Alföld region of Hungary, was one of the major geographic regions in these processes.¹⁸ The radiocarbon data showed a very rapid spread of this culture between Hungary and northern Germany only within no more than 200 years.¹⁹ At that time, the first agriculturists to the Carpathian Basin arrived from the Balkan.^{20, 21} Archaeologists classified them to Körös-Čris-Starçevo culture (KSC)-occupied Western Transylvania and South Hungary circa 7500 years ago, according to radiocarbon data from many of these archaeological sites.²² The expansion of farming came to a halt in the eastern and southern Carpathian Basin for up to 1000 years. During this time, farming techniques are thought to have adapted to the Central European ecological and climatic conditions. Then a new Neolithic culture, the Alföld Linear Pottery culture (ALP) were developed north of the Hungarian Plain. It became widely distributed throughout Central-, Western- and Northern Europe. Regarding several archaeological evidences, Alföld had an important position in the spread of Neolithic, so the early Neolithic findings from here might as well reveal the key data of the process.

Study of ALP aDNA from Central Europe showed a higher frequency of the today rare N1a haplogroup in each region considered in that study, verifying genetic gaps between the first farmers and recent humans.⁶ This fact has been found not be explained by genetic drift alone, so the same research group also studied Mesolithic DNA from Northern Europe and detected a high frequency of U haplogroup among other common European haplotypes.²³ Further extended analysis of haplogroups from ancient DNA of European early Neolithic farmers revealed their Near Eastern affinities and origin.²⁴

Determination of ancient human genetic structure is still a great challenge in Molecular Anthropology, so in our step-by-step mitochondrial polymorphism studies, we concentrated on evidences from reproducible PCR amplicon sequence data from some part of the Neolithic and contemporary remain collection in Alföld. These were samples from the Central Alföld region (Szarvas, Szakmár), which were part of the KSC, as well as from the northern part of Hungarian Plain (Mezőkövesd, Aszód, Polgár), which were part of ALP, overlapping region of KSC and ALP, like Ecsegfalva. The remains with excellent DNA preservation from the circa 150-year-old mummified humans of a Vác crypt were also extracted, extending the comparative studies. To determine the haplogroup pattern of Alföld, we concentrated on the specific ancient polymorphisms identified to describe our novel genetic data from a European region never covered before in depth by similar studies.

Materials and methods

Human skeletal remains from Hungarian archaeological sites

Human remains examined in the present study were selected from the large-scale collection of Hungarian Natural History Museum, which stores ∼480 of them, excavated in Alföld, the Hungarian Plain. All of them were previously dated to the Neolithic Period by archaeologists, supported by their radiocarbon data.²² Our collection consists of 70 human remains of KSC culture from 15 sites and 411 remains of ALP culture from another ∼15 sites, as well as some of the best preserved Neolithic human remains stored in other Hungarian museums (see Supplementary Figure 1 for locations of excavations). Storage of bone material has been organised in separate individual boxes, in dedicated and supervised storage halls with suitable records available for researchers. These storage boxes contain complete long bones of upper and lower limbs and intact teeth placed in their original position in the maxilla or mandible. These helped us at proving their clear-cut relationship to the neighbouring bones. After a thorough examination of the physical features and preservation of these remains, skeletons failing the simple criteria mentioned above were ruled out. The thick compact layers of diaphyseal part from femurs and humeri were sampled for pulverised compact long-bone preparations, whereas for tooth-root powder preparations, the undivided roots of incisives and canines were extracted.

In our osteological collection, 33 Neolithic remains seemed to be the best preserved specimens for the criteria of our ancient DNA technology; that is why we started with these ones in the extraction procedures. Finally, the number of most suitable remains was limited to 11, providing a sufficient amount of reproducibly amplifiable DNA (Table 1). Although further 8 out of the 33 contained extractable DNA, but with an insufficient amount for the consecutive repetitions, they were ruled out off the final sequence analysis. The contemporary mummy samples were excavated in the city of Vác (North Hungary), from the crypt of the Dominician Church, where citizens of the catholic society of Vác were buried. They were naturally mummified bodies of humans with a known identity and a detailed-historical record, dated to the late 18th and 19th centuries. Experimental design and DNA extractions from the 10-mummified remains and 2 medieval ones were carried out in the same manner as from the Neolithic ones (Table 1).

Table 1 Summary of the counts for the skeletal remains subjected to the extractions and sequence analysis

Reagents and kits for molecular biology

Molecular biology grade reagents (also DNAase–RNAase free, if applicable) were prepared in small aliquots, exposed to UV irradiation wherever possible. Reagents were purchased as follows: Chloroform, as well as phenol/chloroform/isoamyl alcohol (25/24/1), ethanol, EDTA, pH 8.5 and N-lauryl sarcosine from Sigma-Aldrich, St Louis, MO, USA; agarose and proteinase K from Fermentas, Glenburnie, MD, USA; proteinase K also from Invitrogen, Carlsbad, CA, USA; 2 × PCR Mastermix and Accugene DNA-free Water from Fermentas; Hyper Ladder V and 6 × Loading Dye from Bioline, London, UK; E.Z.N.A. Cycle Pure Kit and E.Z.N.A. Gel extraction Kit from Omega-Biotek, Norcross, GA, USA; standard oligonucleotides were synthesised at HVD Life Sciences, Ebersberg, Germany; Amicon Ultra-4 from Millipore, Billerica, MA, USA; Eppe5333000018 Mastercycler, as well as Eppendorf PCR tubes and plates from Eppendorf, Hamburg, Germany).

Extraction and purification of ancient/contemporary DNA from skeletal remains

The surface of the intact diaphyseal part of long bones and/or the teeth from maxilla/mandible were soaked in commercial bleach for 10 min, followed by an extensive rinse in DNA-free water. Subsequently, in case of teeth, UV irradiation (1.0 J cm⁻² UV-C light for 2 × 10 min) on both sides was applied, and the radices were cut off and sliced using a cleaned-dental saw. UV-irradiated diaphyseal long bone pieces were pre-treated by scraping their outer surfaces using a Microtool (MF Perfecta, type 9975-E, W&H Dentalwerk, Bürmoos Gmbh, Bürmoos, Austria) followed by a similar UV-exposure on their both sides to efficiently get rid of all possible outer DNA contamination. Osteological materials cut into small cubes (∼0.5 cm), as well as slices of tooth radix were pulverised in a spherical mineralogy mill (Mixer Mill, Retsch, Haan, Germany). The mill was washed with bleach and DNA-free water rigorously, with 60 min of UV exposure between each sample pulverisation, respectively. Pulverisation procedure finally resulted in 0.4–1.0 g of bone/tooth powder, suspended in extraction buffer contained 0.1 M EDTA, pH 8.5; 0.5% N-lauryl sarcosine and 0.2 mg ml⁻¹ proteinase K, digested at 37 °C for 12–36 h in a final volume of 3 ml. The same incubation time was applied in parallel experiments. It was usually 36 h, but shorter incubations for 20 h were also applied and worked well, when an extra 10% of the original amount of proteinase K was added after 12 h. From the suspensions, DNA was extracted after incubation by the standard method²⁵ using phenol/chloroform/isoamyl alcohol (25/24/1; pH 7.5–8.0) and molecular biology grade Chloroform. The upper aqueous phase of each mixture was washed with DNA-free water and concentrated on Microcon mini-columns according to the manufacturer's instructions. Extracted DNA was further purified with E.Z.N.A. cycle-pure kit to get rid of any excess of hypochlorite and solvents in the samples before amplification and subsequent experiments. Final elution was made with 50 μl DNA-free water.

All pre-PCR modifications, including sampling, were carried out in two rooms with spatial and temporal separation from the post-extraction modifications. One extraction laboratory room was fitted with a positive air pressure and overnight circulation of UV-exposed air, as well as an UV-exposure laminar hood, and the other laboratory was only with the latter. In both, workers used suitable protective clothing (full body suits, hairnets, filter-containing facemasks and double-gloves) and equipments to avoid modern DNA contamination. Extraction tools and pipettes, disposable filter tips, tubes and racks were dedicated to this type of work only and never moved out. Both of the laboratories had been free of any recent DNA work, never any DNA preparations were carried out before. Frequent surface cleaning with detergent was regularly followed by 10% natrium-hypochlorite treatment. Overnight UV irradiation of surfaces and tools was to carry out by laboratory workers only. Molecular biology grade reagents were DNAase–RNAase free (if applicable) and prepared in small aliquots, exposed to UV irradiation wherever possible. Extractions were carried out in the UV-irradiated laminar hoods, separated from the PCR set-up, which was due in a separated PCR room, in another UV-irradiated hood.

For regular survey of any intralaboratory DNA contamination, experimental blanks were always applied, at least one in each set of experiments. If extraction blanks gave positive amplifications in PCR, all the extractions were repeated before the new PCRs, with reagents containing DNA-free water from a new batch. If the no-template control gave positive amplifications, chemicals were regarded as contaminated and disposed. Then, freshly opened DNA-free water and reagents were used for the consecutive PCR reactions. Finally, DNA extractions were prepared in the room without UV light and in the one with UV light (see above) to gain the same post-extraction results. Results for individuals M98, M88 and Ecsegfalva 23A were replicated at the Institute of Anthropology in Mainz, Germany (Palaeogenetics groups). Results were identical except for one substitution (potential sequencing error), which did not alter the haplogroup assignment. Otherwise, all Neolithic pre-PCR experiments in this study were carried out by Zsuzsanna Guba, whose mtDNA was typed in Mainz (Joachim Burger's laboratory) and her HVS-I was found identical to revised Cambridge Reference Sequence (rCRS).

PCR amplification of human HVS-I fragments from the extracted ancient or contemporary mtDNA

Performance of the extracted ancient DNA as template in the subsequent PCR experiments was dependent on its integrity as detailed above. A primer-walking strategy (see Supplementary Figure 2), similarly, but not identically to some previous aDNA studies,^{6, 24} were designed. The forward primer of the downstream amplicon well overlapped with the reverse primer of another upstream amplicon for each combination of primers. It also helped to cope with the low yield of PCR from fragmented DNA and supported the expected sizes of amplicons of interest from 64 to 455 bp of the mitochondrial HVS-I for a batch of several different PCR reactions (usually 6–8). However, some of these combinations of primers did not work in amplifications for unknown reasons, but the efficient ones with sizes of equal or greater than 100 bp were used (Figure 1). All the oligonucleotides were manufactured by HVD Life Sciences, and used in PCR according to the positions corresponding to the 5′ and 3′ nucleotides numbered after rCRS²⁶ as follows: LAF 15 977–15 995, LBF 16 096–16 114, LAR 16 161–16 141, LCF 16 192–16 212, LBR 16 253–16 234, LDF 16 268–16 287, LCR 16 365–16347, LDR 16 430–16 411.

Figure 1

Representation of HVS-I PCR amplicons with correct sizes and yield from the extracted mtDNA of Szarvas 23/20 sample after resolved in 2% agarose gels. Determined amplicon sizes were compared with sizes calculated from nucleotide positions of revised Cambridge Reference Sequence²⁶ as follows: B1 185, B2 277, B3 389, B4 454, B7 270, B10 174, B11 239, B12 98 and B13 163 bp. Blank extraction without bone material was carried out alongside of each preparation and numbered from 1 to 8, next to the corresponding Szarvas 23/20 amplicon. bp, base pair; st, molecular weight DNA standard; NTC, no-template control as negative PCR control.

For nt16257 and nt16261, polymorphic site-specific primers were designed, respectively. This PCR set-up allowed us to distinguish the different polymorphic sequences from each other within a single PCR experiment, similarly to ARMS-PCR (Amplification Refractory Mutational System),²⁷ but here, primers were designed with the substituted nucleotide at their 3′ end. In this single-nucleotide polymorphic site-dependent PCR (SNP-PCR), a positive amplification was a consequence of the presence of SNP. M1 primers were designed at position 16257 with C or A, respectively, whereas M2 primers designed at position 16261 with replacing all four bases one-by-one at the 3′ end, respectively. The mut1C/A oligos from position 16239/8 to 16257 or mut2C/G/A/T from position 16244/3 to 16261 were used in polymorphic PCR with reverse primers LCR and/or LDR. In multiplex PCRs, lower efficiency was experienced, so mainly monoplex PCRs were applied in SNP-PCR experiments. All the overlapping primer pairs and polymorphic site-specific primer pairs were adjusted to the same melting temperature (58 °C), so that the PCR reactions could be run at the same annealing temperature and no touch-down approach was required. The cycle conditions as usual consisted of an initial denaturation at 94 °C for 2 min, 40 cycles of 94 °C for 35 s, 53 °C for 35 s, 72 °C for 35 s, followed by a final adenylation at 60 °C for 10 min. Amplification reactions were set up using 2 × Fermentas MasterMix in a final volume of 20 μl with 0.25 μM primers and 1–3 μl of the purified extract. One skeletal extraction provided enough DNA material for a maximum of 20 PCR reactions.

Determination of DNA sequences of ancient or contemporary mtDNA amplicons

PCR reactions were kept −20 °C after amplification until dissolved in 6 × loading buffer before electrophoresis. The samples of mtDNA amplicons and DNA standard Hyper Ladder V were resolved on 2.0% agarose gels. This type of molecular weight (MW) standard provided an opportunity by the well-defined DNA amount in bands to judge the amount of amplicons in micrograms simply from the gel. The DNA fragments of sizes expected were isolated with disposable scalpels (one for each sample) and purified from their corresponding gel slices, using E.Z.N.A. gel extraction kit. An aliquot of 10–20 ng DNA from each sample were subjected to a single sequencing reaction. After estimation of the DNA contents, the material from a single purification of a given PCR product was used as a supply for 2–10 sequencing reactions, depending on the efficiency of the PCR. The available number of sequencing reactions was determined by an easy densitometry from gel photos applied for estimating DNA amount. Sequencing reactions, using dissolved and/or air-dried DNA samples, were carried out by HVD Life Sciences, according to the manufacturers instructions (www.mwgbiotech.com). Either as PCR fragments or as inserts, the sequencing reactions on amplified DNA were performed on both DNA strands, at least in duplicates, with the PCR primers described as before. The direct sequencing of amplicons instead of their direct subcloning allowed us to clear the sequence data from sequences of any contaminating human or non-human DNA present in a low amount. In this case, the authentic PCR product provided us with an excess amount of DNA over that of the fragments derived from non-authentic sources, so that the sequence of it could easily be detected as authentic or samples be barcoded as artifacts. However, starting with the direct subcloning step, even after sequencing of a series of subclones, there would be a probability for the low amount of contaminating DNA fragments being represented among sequenced subclones with a higher rate than expected from the stochastic, which regularly occurred in subcloning experiments, causing serious ambiguity in the final sequence. But PCR products were also subcloned in pCMV5 vector for safe sequencing or directly sequenced without subcloning in a series of repetitive experiments. In this way, when ambiguous sequencing errors occurred in the primary sequence, they were usually caused by the poor quality of the given sample of the ancient DNA and not by the traces of contaminating DNA fragments. Then independent extractions and sequencings were repeated. After analysing all the sequence data, further sequencing reactions were still being carried out if required to verify the safe consensus sequence. In this case, sequence errors and real mutations had finally been distinguished to complete the reconstruction of the HVS-I authentic sequences of a specimen of interest. The number of necessary repetitions were also recorded (Tables 1 and 2).

Table 2 Summary of the single-nucleotide polymorphisms identified, and the corresponding haplogroup predictions for the remains from this study

Sequence analysis of HVS-I regions of the amplified mtDNA to track polymorphisms

First, all the crude sequences were aligned to each other by SE Central Clone Manager Suit 7.0 (Science & Educational Software, Cary, NC, USA). Global-Ref alignments at the default parameters given in that software. First, sequences of non-related artefacts and those from samples of insufficient quality were eliminated, then sequencing errors were ruled out by their unreproducible dissimilarity upon the repetitive alignments. If the sequences were confirmed by repetitive checks and the lack of errors verified, authentic sequences were aligned in Global-Ref alignments again to rCRS to identify specific polymorphisms. Similarity of these HVS-I sequences were tracked at the default parameters in SE Central Clone Manager Suit 7.0., using exhaustive multi-way alignments of all sequences from nt16145 to nt16367, as well as progressive assembly of alignments with neighbour-joining phylogeny, according to the instructions given in the software. This region was completely sequenced for all the specimens included in the same phylogeny. In this way, sequence differences depended on the mutational sites only and the length of the sequences did not have any effect on this analysis.

Haplogroup assignment was made on the basis of the above region of HVS-I sequences by using the data of mtDNAmanager (http://mtmanager.yonsei.ac.kr) and the Genebase mtDNA Haplogroup Reference Guide (www.genebase.com) version 2.6 (April, 2008). Except for the one novel haplotype, the information obtained from the HVS-I sequences, together with the result of typing different diagnostic control SNPs, allows us to classify 10 additional Neolithic samples into their corresponding mtDNA haplogroups in the well-known mtDNA phylogeny, ruling out the possibility that mutations occurred only during PCR reactions.

Mutation frequencies (F) were calculated upon the data set described above in this study, where the number of extracted remains (not the number of extractions) was regarded as total number (N). The number of cases of a single substitution (n) was divided by N to obtain mutation frequency as a ratio: F=n/N. For comparison, mutational frequencies from a contemporary study²⁹ were also calculated in the same way with the help of Table 1 in that publication. The significant differences between the Neolithic and the contemporary frequency values were determined by a pairwise χ²-test.²⁸

Results

The present study has been the first to analyse the ancient DNA content of the Neolithic archaeological collection excavated in Hungary, so this comprehensive study was aimed to supply an initial survey of reliability as well. Thus, the state of this archaeological material proved to be crucial in the extraction procedures that the remains were always selected with care regarding their condition, and the code that never the complete material can be destroyed by repetitive sampling, appreciating the protected state of stored skeletons as national treasures. Finally, 11 Neolithic and 12 historic/contemporary remains were repetitively analysed for their reproducibly extractable DNA content. The strict criterias of aDNA sampling were successfully applied in these cases (Table 1). Unlike in Neolithic case studies, in contemporary DNA samples, all the fragments of interest produced acceptable yield for PCR and subsequent sequence analysis.

Primer combinations of our PCR strategy were systematically used to test the integrity of templates of interest. Not all combinations of synthesised primers performed successfully in PCR, but amplicon B1, B2, B3, B4, B7, B10, B11, B12 and B13 proved to be always amplifiable. PCR amplicons from the extractions were than resolved on and purified from 2% agarose gels. Szarvas 23/20 tooth-root powder sample is a representation of the most successful PCR experiments from the Neolithic aDNA of interest. All the amplicons listed as before produced clearly visible DNA bands, but the longest ones (B3, B4) somehow with a lower yield (Figure 1). All but B4 were isolated from this experiment for repetitive sequencing, and similar DNA bands were used for every remains analysed later on. As the outcome of this pilot study on Szarvas 23/20 individual, the entire HVS-I and some extra sequences have been constructed from the overlapping data set as reported before.³⁰ Three mitochondrial control mutations were verified in the Szarvas 23/20 sequence: C16223T, C16257A and C16261T (Figure 2). These polymorphisms were detected for the first time from the Neolithic archaeological remains, so they represent novel findings to our knowledge.

Figure 2

Summary of sequence alignments of reconstructed HVS-I sequences from Neolithic/historic/contemporary skeletal remains of interest. Sequence polymorphisms within HVS-I determined by comparisons with the revised Cambridge Reference Sequence are given for orientation, with nucleotide numbering of the human mtDNA above the representative alignment. mitCRS, revised Cambridge Reference Sequence;²⁶ Neolithic sequence of Csongrad 20, Csongrád Bokrospuszta grave 20; Ecsegfalva 23A, Ecsegfalva 23A specimen;⁶ Kisköre 15, Kisköre grave 15; Mkövesd 25, Mko?vesd Mocsolyás grave 25; Szakmár 8, Szakmár-Kisülés grave 8; Szarvas 23_20, Szarvas 23 grave 20; Szarvas 23_22, Szarvas 23 grave 22; Vörs 52, Vörs Máriaasszonysziget object 52; Dobó ruszka, sequence of 16th century Dobóruszka; sequence of 19th century samples of Mummies from Crypt of Vác Hungary: M98 b1 M98 t1, M88, M129; Aszód, Aszód specimen; Folyás 111, Folyás grave number 111; Szegvár 25, Szegvár Tûzköves grave number 25; Zalavár23, sequence of 9th century Zalavár grave number 23.

SNP-PCR for control mutation sites 16257 and 16261 (Figure 3) were applied as polymorphic PCR on further Neolithic aDNA extracts. Szarvas 23/20 samples produced polymorphic PCR amplicons 16257A and 16261T as expected upon their previously determined nucleotide sequence. Neolithic DNA samples without any previous sequence data were also included, and among them, Szakmár Kisülés 8 appeared with the same double mutational pattern (Figure 3 and Supplementary Table 1), whereas Szarvas 23/22 showed only 16261T polymorphism, but not 16257A, derived from the same archaeological site as Szarvas 23/20. However, contemporary M98 mummy-DNA extract and other Neolithic ones such as Csanytelek 6, Kisköre 15 and Szegvár 25 produced non-polymorphic (16257C and 16261C) pattern except for Csongrád 20, but only 16257A is shown for the latter in Figure 3, and 16261T was not checked by PCR because of a low DNA yield, but detected by sequencing only. Sometimes SNP-PCR resulted in bands in both version of nucleotide-dependent PCRs, clearly indicating the presence of contaminating DNA, but it is not possible to distinguish between the contamination and the authentic DNA bands in this case. Instead, PCRs from a new batch of samples with single version clear-cut bands on gels were reproducibly provided always from new extractions. The usefulness of SNP-PCR instead of sequencing to prevent us from working on the contaminating DNA was always demonstrated in such situations.

Figure 3

Results of SNP-PCR screening for control polymorphisms C16257A and C16261T in the ancient mtDNA samples. Alternative amplifications were performed for the 126 or 127-bp long PCR products for site 16257, whereas 186 or 187-bp long ones for site 16261, respectively. All amplicons were identical, except for the SNP-site nucleotide as given in this scheme. For 16257 site, not only LCR, but also LDR oligonucleotide was used for amplification providing further amplicons with sizes of 191 or 192 bp (Szarvas, Szakmár). All the monoplex PCR amplicons were resolved in 2% agarose gels. Archaeological samples were as follows: Neolithic remains in this study were as Szarvas 23/20, Szarvas 23/22, Szakmár-Kisülés 8, Csanytelek 6, Kisköre 15, Csongrád 20, Szegvár 25, and recent mummified human specimens as M98 t1 of mummy No. 98 Crypt of Vác. Abbreviations: bp, base pair; st, molecular weight DNA standard; Blank, corresponding extraction without bone material; NTC, no-template control as negative PCR control; B13, positive PCR control with primers LDF and LDR; M1C or M1A, DNA fragments amplified by forward primers for control polymorphisms C16257A applying C or A as a last nucleotide, at the 3′ end of each primer, respectively, at the substitution site 16257. M2C, M2A, M2G or M2T, DNA fragments amplified by forward primers for control polymorphisms C16261T applying C, A, G or T as a last nucleotide, at the 3′ end of each primer, respectively, at the substitution site 16261.

Finally, the sequencing strategy produced HVS-I sequences between nt16145 and nt16367, and they were successfully constructed in case of 17 remains, including 11 Neolithic ones (Figure 2). Samples from the same remain with sequence alterations, suspecting exogenous contamination, were ruled out, because in this case, the authentic sequence could not be identified safely, except for the prior application of SNP-PCR. Because this region of HVS-I represented only a part of the whole hypervariable region of mtDNA, the summary of the identified haplogroups and their polymorphisms (Table 2) could have only been concluded if the mutations characteristic for the haplogroup of interest were present within this region, and no other mutations were required for assigning the different ones. Upon these facts, a novel and four East Asian (N9a, D1/G1a1, C5, M/R24) haplogroups were identified in this study for the Neolithic. The two Körös individuals (KSC) bearing the N9a polymorphisms at nt 16257 with the parallel presence of 16261 mutation were derived from two different localities, Szarvas and Szakmár. Interestingly, the nearby Ecsegfalva 23A ALP specimen from the subsequent culture had N1a haplogroup, and its different polymorphic pattern has been detected before the present study.⁶ One ALP specimen (Csongrad 20) also carried the characteristic N9a mutations (Figure 2, Table 2). For mummy98, different polymorphisms were reproducibly found in the tooth and the bone samples of N1b and HV4a haplogroups, respectively. This could be caused by heteroplasmy or simply by DNA traces from the excavations, whereas other reasons were also possible. Parallel aDNA samples have always been applied for teeth and bones, but never similar discrepancies were detected.

Subsequently, contemporary mummy samples were screened by SNP-PCR for the presence of polymorphisms characteristic for the N9a haplogroup. We examined 8 out of 10 Vác-mummy specimens, namely mummy no. 22, 48, 50, 77, 127, 129, 164 and 179, of which the mummy no. 179 carried the characteristic co-occurrence of 16257A and 16261T as our first non-Neolithic N9a haplogroup assignment (Figure 4). The presence of these control mutations was also confirmed by sequencing B10 PCR amplicons of mummy179, providing only partial sequences (so it is only given in Table 2, but not in Figure 2 and in Figure 5).

Figure 4

Results of SNP-PCR screening for control polymorphisms C16257A and C16261T in the contemporary Vác mummy specimens. Archaeological samples were as follows: mummified human specimens from the Crypt of Vác No. 22, 48, 50, 77, 127, 129, 164 and 179. Abbreviations: bp, base pair; st, molecular weight DNA standard; −ve, extraction without bone material; NTC, no-template control as negative PCR control; +ve, positive PCR control with primers LDF and LDR; M1A, DNA fragments amplified by SNP-PCR only for polymorphic version, choosing the A-nucleotide version of forward primer M1 applied at the substitution site 16257. M2T, DNA fragments amplified by SNP-PCR only for polymorphic version, choosing the T-nucleotide version of forward primer M2 applied at the substitution site 16261.

Figure 5

Comparison of HVS-I sequences by neighbour-joining phylogeny. Dendrograms were constructed with a distance-based tree-building method for the relationship analysis among Neolithic, contemporary, Mesolithic and evolutionary sequences. (a) This dendrogram based on the alignment of Figure 2 between nt16145 and nt16367 of the revised Cambridge Reference Sequence (rCRS),²⁶ for abbreviations of ancient specimens refer to legend of Figure 2. Haplogroup prediction of ancient samples are given with letters in bold after their names. mitCRS, rCRS; Derenburg, Derenburg 1 specimen from Germany;⁶ Stone Age Eulau, Eulau sequence from the Later Stone Age Corded Ware Culture, Germany;³¹ Guinea, recent New Guinea sequence;³² North Indian, recent North Indian sequence;³³ Philippines, recent Philippine sequence;³⁴ Siberian, Siberian specimen;⁵ Iberian, Iberian specimen.⁷ (b) In this dendrogram for European Neolithic mtDNA sequences, the overlapping regions between nt16055 and nt16378 were considered for alignment. Catalonia sequences⁷ (1, 6, 8, 11, 12, 14, 21, 22, 23) from Spain and Siberia specimen,⁵ Szarvas 23/20 from this study, Derenburg 1, 3 from Germany and Flomborn1, Halberstadt2, Unterwiederstedt5, Ecsegfalva 23A⁶ were compared. (c) This dendrogram depicted to show relationship of Szarvas 23/20 specimens and other prehistoric mitochondrial sequences to those of older human forms. Mesolithic Lebyazhinka, Kretuonas (from Baltic Region) and Ostorf sequences,²³ Tyrolean iceman³⁵ from Austria and Neanderthalensis mtDNA sequences,¹¹ corresponding the overlapping regions between nt15977 and nt16430 of rCRS.

Using the HVS-I sequences determined in this study, a comparative analysis, which depended on the mutations detected in them, was performed. In this way, the consensus sequences compared were different only at mutational sites, which were characteristic for the determined sequence of the specimen of interest. This approach was set to determine a relationship between the contemporary sequences derived from the references and the Neolithic data set. In this neighbour-joining phylogeny (Figure 5), non-polymorphic Neolithic sequences within haplogroup H are grouped together with HV4a, H12, M/R24 and U3 ones as a Western Eurasian set. Neolithic sequences with determined mutations of 16257A and 16261T, including Szarvas 23/20 composed an N9a group, separated from N1a ones. Sequence of C5 haplogroup showed the tightest relationship with a Siberian, whereas the sequence with 16223T, 16325C, 16362C with some South Asian. Interestingly, the novel haplotype of Szarvas 23/22 showed a definite outlying position (Figure 5).

Only the Neolithic sequence of Szarvas 23/20, as a complete Neolithic HVS-I, was applied for further phylogenetic analysis. In this comparison, it somehow appeared separated from Catalonian, Siberian and Western European Neolithic samples, but closer to the Tyrolean iceman or Mesolithic sequences (Ostorf, Kretuonas, Lebyzhanska) than to the Neanderthalensis as expected (Figure 5).

Screening the NCBI nucleotide database provided further information about mitochondrial control polymorphisms C16223T, C16257A and C16261T. In our BLASTN search, the Szarvas 23/20 sequence query resulted in recent Vietnamese polymorphisms (VN-79, data not shown) that are identical to these control polymorphisms, but none from Europe, as it was expected for base substitutions with dominantly Asiatic distribution.

Determination of the mutational frequencies of the Neolithic substitutions found in this study was carried out to compare them with those of a contemporary Slav population²⁹ as an example. This was an available sample of contemporary Slav population of Checks representing the modern population from this part of Central Europe, being geographically only a short distance from the sites of investigation and finding not exactly in the same region as Alföld. The recent data of that study were used, with an extension up to hundreds of individuals. The χ²-test was applied pair wise to estimate the level of significance between the two sets of data (Table 3). Frequencies of control mutations assigned to N9a haplogroup showed a significant difference between the Neolithic and the contemporary samples as a sign of a clear genetic discontinuity. In our Neolithic study, C16223T, C16257A and C16261T were the most common substitutions, whereas in the contemporary Slavs, most common ones were found at sites 16126, 16193, 16223, 16294, 16311, so C16223T was the only one substitution common in both the Neolithic and contemporary Czechs data set. However, its frequencies in the two data set were also significantly different (Table 3). Interestingly, a slight significance between the Neolithic and contemporary frequencies of some ‘neighbouring’ mutations, such as 16324, 16325 and 16327 could be seen as well. Like in case of the ones of N9a, these substitutions were characteristic also for the Asian haplogroup assignments of M/R24, D1/G1a and C5, respectively (Table 2).

Table 3 Summary of the comparative study of mutational frequencies

Assigned Neolithic haplogroups were depicted on the map of the Carpathian Basin to determine their geographical distribution in Alföld (Figure 6). This showed a quite even occurrence of East Asian haplogroups in both KSC (N9a, C5) and ALP (N9a, D1/G1a, M/R24). Although other ALP haplogroups appeared on the northern part, N9a was localised in the southern part of Alföld in both KSC and ALP, which corresponds to the central region of the former Neolithic occupation in the one-time Hungarian Plain.

Figure 6

Geographic distribution and cultural classification (in brackets) of the assigned Neolithic haplogroups of the present study. Alföld region is marked by a dashed line (circa 50 000 km²). The area of the map corresponds to the territory of Hungary and border territories of the neighbouring countries localised mainly in the Carpathian Basin. There is a distance of circa 80 km between the from north-to-south oriented part of River Danube and that of River Tisza, visible as grey curved lines in the middle of Alföld. For cultural classification and designation of the remains refer to Table 2.

Discussion

Ancient mtDNA analysis of Central-European ALP people⁶ has been published, followed by that of the Mesolithic²³ humans, and very recently, an extended comparative study²⁴ of a German ALP site, too. The predominant mtDNA haplotypes³⁶ described in these ancient samples were also found in recent Europeans, but at considerably different frequencies, which seemed unlikely to be caused by simple demographic processes. Only one ALP remain from Alföld was involved in these studies, the Ecsegfalva 23A specimen with N1a haplogroup.⁶ However, more Neolithic findings in Hungary were proved to have excellent DNA preservation, making the analysis of nuclear DNA possible (Szarvas specimens).³⁷ It seemed from these studies that tracking of a direct lineage can not be made between the first farmers and the recent Europeans as it was suggested by the demic diffusion model. Population demographic simulations also referred to other significant population exchange/movements, as the mtDNA-haplotype diversity in the Neolithic was different with a discontinuity from the recent one.

However, the presence of East Asian haplogroups in the first place, the N9a, very rare in Central-Europeans,²⁹ was fairly unexpected. Scientific/anthropological interpretation of this phenomenon is not an easy task at all. What seemed to be sure was that the Neolithic and present mtDNA pattern was somehow markedly different in this study, as well as in other previous studies, comparing the frequencies of HVS-I mutations. These were probably caused by random drift and/or bottleneck, which could have been more common in human prehistory than expected. In this case, the rare variants in the contemporary might have been more common a few thousand years ago. The extension of the sampling onto even more Neolithic remains is expected to provide further important data to help at the explanation of this phenomenon. In the first place, the direct environment and the surrounding remains of the ones carrying N9a mutations were to be examined. Several immigrations from Asia to Europe or vice versa are known from Archaeology, as well as from written sources, in the historic times. Hungarian people are known to be arrived from Central-Asia into the Carpathian Basin in the 10th century AD, but a representative mtDNA survey of the current Hungarian population is still lacking, except for the one with Hungarian nationality from Secklerland in Transylvania (Romania). It is also noteworthy that previous nuclear DNA analyses showed no significant differences in the Hungarian's genetic composition as compared with the neighbouring countries. Instead of them, the closest representative recent sample of the Czechs was used as a comparative sample.²⁹ For East Eurasian haplogroups, low frequencies in the East European samples have been experienced in those studies as a phenomenon of different discountinuities.

However, in our case study, Alföld's Neolithic populations were characterised by the occurrences of some rare East Asian haplogroups. They were detected both in the KSC (N9a, C5, D1/G1a) and the archaeologically consecutive ALP (N9a, M/R24). This timely occurrence of N9a is to be subjected to further examination for finding a possible genetic connection for the subsequent cultures of Central Europe, and their comparison with a more extended Neolithic case study would be of a great importance.

N9a is a rare haplogroup in the one-time and contemporary populations of Europe as well. Neither N9a, nor any of the other East Asian haplogroups have been found in any ALP sites^{6, 24} with more western location. In a very important study from a German site (Derenburg), as an example, Neolithic samples proved to have N1a and other haplogroups with near-Eastern genetic roots, and no N9a was detected there.²⁴ However, the different mitochondrial control polymorphisms of the N9a have been examined by other groups in recent populations from different geographic regions world-wide, and several distinct patterns were found. Interestingly, control region polymorphism C16261T was detected in mtDNA of ancient bone samples of Pacific Islanders,³⁸ in recent samples of Basques,³⁹ Greenland Eskimos,⁴⁰ Australian aboriginal populations,⁴¹ as well as in Indians⁴² and Malaysians.⁴³ Control region polymorphism C16257A was distinctly detected in Japanese⁴⁴ and Chinese people.⁴⁵ C16223T polymorphism was found in Africa, where it was widespread. In the literature, a distribution of 7% in Europe and 65% in Mongolia and 91% in Africa was described for this polymorphism.²

Presence of the N9a haplogroup has also been shown by an extended study of the Siberian Kurgan people,⁴⁶ with only one specimen of N9a assignement out of 26-examined remains dated between 3800–1600 BP. Upon evidences from mtDNA, Y-chromosome and pigment gene polymorphisms, these samples of South-Siberian people were supposed to have European genetic relations. The connection between East Asian haplogroups and the Neolithic population could be further revealed by aDNA analysis of more near-Eastern/Anatolian Neolithic human remains.

Stepping beyond Genography, control mutations assigned to N9a have also been checked in connection with diseases such as mitochondrial encephalomyopathies⁴⁷ and Parkinson's disease. In connection with the latter, the C16257A and C16261T were mentioned.⁴⁸ Interestingly, C16257A was also examined in mitochondrial myopathy and encephalopathy.⁴⁹ If a connection to any of these diseases can be verified for N9a polymorphisms, assessing their contribution, very interesting research on their putative human diseases are to be carried out.

In the present study on the ancient DNA of the Hungarian Neolithic samples, the remarkable genetic discontinuity between our small Neolithic mitochondrial gene pool and a contemporary one was mainly realized in the more frequent occurrence of the East Asian N9a and was accompanied by N1a, D1/G1a1, C5, M/R24 and H haplogroups. C16223T, C16257A and C16261T were the most common mutations, the latter two were assigned only to N9a haplogroup. In contrary, in a contemporary Slav sample,²⁹ the set of the most common substitutions was completely different, the ones at sites 16126, 16189, 16193, 16223, 16294, 16311, and finally C16223T as well. The East Asian ones were found both in the KSC and ALP times of different Hungarian Neolithic remains handled separately in a series of experiments. This data also suggested a discontinuity, which was modelling differences within a circa-8000-year timespan. Seeing the surprisingly common occurrence of N9a haplogroup, SNP-PCR primers were then designed to amplify their mutations from mummy samples. Although sample sizes did not allow us to compare directly these populations (Neolithic Alföld and almost contemporary Vác) with each other, challenging up to a number of 50 samples for each would give even more information of haplogroup composition. In that case, stepping forward over χ²-test, genetic diversity analysis could be carried out on the extended data set according to a described statistical approach⁵⁰ available for the population genetics. In our opinion, the substitution ‘hotspots’ around nt16324/5/7 would have been worth investigating in depth, with newly designed SNP-PCR primers in a larger scale for their frequencies. But for an extended N9a frequency analysis, up to 100 samples by SNP-PCR and completed HVS-I sequencing would be rewardable. In that case, N9a frequency might be decreased both in the Neolithic and the contemporary (Vác and/or Czechs, Hungarians). People from the same settlement of interest could not be expected as a good reference for statistical analysis, because of the known population changes in history of Central Europe. However, sampling of recent Hungarian population would be necessary, but for addressing this problem, a different experimental strategy is required. In fact, some of the discrepancies have been ruled out by the lucky opportunity of an almost recent 19th century mummy collection. These people lived together in a well-documented manner, which can be verified from clerical registry.

Using SNP-PCR set-up and real-time PCR, efficiently and quickly screening a high number of Neolithic remains, the expected polymorphisms could be detected in larger samples and compared with an extended database search for references to gain frequencies previously described. This might help at producing higher sample sizes for aDNA studies, which would be seeked for a proper Genography of the Neolithic Central Europe. The χ²-statistics²⁸ and/or SPSS multivariate statistics on more polymorphic HVS-I sites of different East Asian haplogroups can be carried out on this extended data set. Expected significant changes in the distribution of mutations can provide a further basis to track more discontinuities and descendant lineages in more details.

Conclusions

In the work presented here, ancient DNA investigation on a previously uninterpreted archaeological collection of the Neolithic Hungary (circa 8000–6000 BP) has been successfully conducted. Mitochondrial HVS-I sequences were completed in case of 11 human remains and 29 different ancient SNPs were identified. In case of the 2 medieval remains and the 10 Vác mummies analysed, further 13 SNPs were described by sequencing and SNP-PCR.

For the well-preserved Szarvas 23/20 remains, the entire HVS-I sequence has been reconstructed, proving the presence of control mutations for nt16223, nt16257 and nt16261. Those supported the first characterisation of Neolithic N9a haplogroup, which was known to be connected mainly to East Asia, in the contemporary human populations.

Polymorphic PCR strategy has also been carried out as a screening approach to track the presence of nt16257A and nt16261T mutations in other Neolithic and contemporary remains. SNP-PCR experiments supported by sequence analysis verified further three specimens (Szakmár, Csongrád and mummy179) assigned to the East Asian N9a haplogroup. Final alignment of determined sequences resulted in even more haplotypings, one N1a, as well as a novel haplogroup with unknown origin and three other Asian (C5, D1/G1a, M/R24).

Phylogenetic studies of Neolithic sequences showed the expected relations to the ones described previously, as well as some sequence differences comparing with other groups of Neolithic people from other region of Europe, like Catalonia, Germany, also with the Mesolithic in the Baltic or with the Neanderthal.

A comparison of the frequencies of detected substitutions characteristic for N9a haplogroup, in a pairwise χ²-test between the Carpathian Neolithic from Alföld and a contemporary Czech population, verified significancy for the differences. This can be regarded as an indication of some genetic discontinuity, also described in previous Neolithic aDNA studies of Europe. These results also highlighted a surprisingly common occurrence of N9a both in KSC and ALP culture.

Geographical distribution of N9a haplogroup, as well as that of other East Asian ones, was depicted and they were found to be located mainly in the middle of Alföld, the Hungarian Plain, which was a main Neolithic region both for KSC and ALP culture. The Frequency studies and small-scale Genography underlined the specific features of our Neolithic sampling.