No Evidence of Neandertal mtDNA Contribution to Early Modern Humans

1Despite intense research efforts, no consensus has been reached about the genetic relationship between early modern humans and archaic human forms such as the Neandertals. While supporters of “multiregional evolution” argue for genetic exchange or even continuity between archaic and modern humans, proponents of a “single African origin” of contemporary humans claim that negligible genetic interaction took place. Mitochondrial DNA (mtDNA) sequences from early modern humans would in principle be able to resolve the question of a contribution of Neandertal mtDNA to modern humans. However, human DNA is pervasive in palaeontological and archaeological remains as well as in most laboratory environments (e.g.,;; ). It is therefore currently impossible to differentiate contaminating modern DNA sequences from endogenous human DNA in human remains. Thus, although mtDNA sequences have been reported from remains of early modern humans, it is not possible to determine whether such DNA sequences indeed represent endogenous DNA sequences. A related problem is that if a Neandertal fossil yields modern human-like DNA sequences, those might be discarded as putative contaminations, even if they may be endogenous and represent evidence for a close genetic relationship or interbreeding between the two groups.

2To explore the genetic relationship between early modern humans and Neandertals in spite of these difficulties, we made use of the fact that the four Neandertal mtDNA sequences determined to date can easily be distinguished from those of modern humans. This allowed us to ask whether all well-preserved Neandertal remains contain Neandertal-like mtDNA and whether all well-preserved early modern human remains fail to contain such DNA sequences. Thus, we did not attempt to determine DNA sequences that are similar to present-day human mtDNA. Instead, we determined whether Neandertal-like mtDNA sequences were present or absent in well-preserved remains of Neandertals and of early modern humans.

3The preservation of endogenous DNA in fossils is correlated with the amount, composition, and chemical preservation of amino acids. We find that endogenous DNA can be amplified from Pleistocene remains when the amino acid content is more than 30,000 parts per million (ppm), the ratio of glycine to aspartic acid between two and ten, and the aspartic acid racemization (i.e., the stereoisomeric D/L ratio) less than 0.10 (;, 2000;; data not shown). We analyzed the amino acid preservation of 24 Neandertal and 40 early modern human fossils (Table S1). Several important Neandertal fossils, such as La Ferrassie and Krapina, as well as important modern human fossils, such as Veternica, proved to be too poorly preserved to be likely to allow DNA retrieval. Thus, further destructive sampling of these specimens was not considered justified. However, four Neandertal and five early modern human fossils fulfilled the above criteria for amino acid preservation and were thus expected to contain endogenous DNA (Figure 1; Table 1). These samples were geographically well distributed across Europe (Figure 2) and included remains whose morphology is typical of Neandertals (e.g., La Chapelle-aux-Saints) and of modern humans (La Madeleine, Cro-Magnon). They also included samples that have sometimes been considered “transitional” between Neandertals and modern humans, based on their morphological features: Vindija and Mladecˇ.

4If low amounts of DNA are preserved in a specimen, some extracts will fail to contain DNA molecules by chance. Therefore, except in the case of Mladecˇ 2, in which the amount of material available permitted only two extractions, we extracted each of the four Neandertal and the five early modern human samples three times. For each extraction, amplifications were performed using two primer pairs: (i) “hominoid primers” that amplify homologous mtDNA sequences from the previously determined Neandertals and contemporary modern humans, as well as African great apes; (ii) “Neandertal primers” that, under the conditions used, amplify only Neandertal mtDNAs even in the presence of a large excess of modern human DNA. Since authentic ancient DNA is typically highly degraded, both primer pairs were designed to amplify short mtDNA fragments (72 and 31 bp, respectively, excluding primers). In each of these fragments, two substitutions allow the discrimination of previously determined Neandertal mtDNA sequences from contemporary modern human sequences. The sensitivity of both primer pairs is similar, as shown by the fact that they are both able to amplify single template molecules as judged from nucleotide misincorporation patterns. In order to determine the nature of the DNA sequences amplified, each amplification product was cloned and approximately 30 clones were sequenced for each “hominoid product” and ten clones for each “Neandertal product.”

5When amplified with the hominoid primers, all Neandertal and all early modern human remains yielded modern human DNA sequences (see Table 1). In addition, five cave bear teeth from Vindija, Croatia, and one from Gamssulzen, Austria, extracted in parallel with the hominid samples, all yielded human sequences. This confirms previous results in showing that most, if not all, ancient remains yield human DNA sequences when amplification conditions that allow single DNA molecules to be detected are used. For three Neandertal and all five modern human remains, several different mtDNA sequences were retrieved from individual extractions, and in the case of one Neandertal and one modern human, at least two of the sequences were also found in an independent extraction from the same specimen. Additionally, one of the cave bear teeth yielded a human sequence found in two independent extracts. Thus, the fact that a DNA sequence is found in two independent extracts is a necessary, but not sufficient, criterion of authenticity when human remains are analyzed. This implies that in the absence of further technical improvements, it is impossible to produce undisputable human mtDNA sequences from ancient human remains. In addition to DNA sequences identical to those previously amplified from present-day humans, the Neandertal bones Vi-77 and Vi-80 from Vindija yielded four out of 89 and 73 out of 85 mtDNA sequences, respectively, that were identical to previously determined Neandertal sequences. Thus, these two specimens contain a proportion of Neandertal-like mtDNA sequences (i.e., sequences that carry two substitutions that differentiate Neandertal mtDNA sequences from modern human mtDNA sequences as described above) that is high enough to detect using primers that amplify also modern human DNA.

6When amplified with Neandertal-specific primers, Neandertal-like mtDNA sequences were amplified from two independent extractions from all Neandertal fossils (see Table 1; Figure 3). For one of these, Vi-80 from Vindija, DNA preservation was sufficient to allow the retrieval of longer fragments and thus the reconstruction of 357 bp of the hypervariable region I (see Supporting Information section; Figure S1). This mtDNA sequence was identical to that retrieved from another bone from the same locality (Vi-75; ). In contrast to the Neandertal remains, none of the early modern human extracts yielded any amplification products with the Neandertal primers, although these remains are similar in chemical preservation to the Neandertal remains (see Figure 1).

7Thus, all Neandertal remains analyzed yielded mtDNA sequences that are not found in the human mtDNA gene pool today but are similar to those found in four previously published Neandertals (see Figure 3). This is compatible with results suggesting that the extent of Neandertal mtDNA diversity was similar to that of current humans and lower than that of the great apes. It is noteworthy that this result is not an artifact created by discarding “modern-like” mtDNA sequences amplified from Neandertals, since all Neandertal remains with good biomolecular preservation yield “Neandertal-like” mtDNA sequence. Furthermore, none of the five early modern humans yields “Neandertal-like” mtDNA sequences in spite of the fact that these remains are as well preserved in terms of amino acids as the Neandertal remains. Thus, we fail to detect any evidence of mtDNA gene flow from Neandertals to early modern humans or from early modern humans to Neandertals.

8However, a relevant question is what extent of gene flow between Neandertals and early modern humans the current data allow us to exclude. In this regard, it is of relevance that the five early modern humans analyzed lived much closer in time to the Neandertals than do contemporary individuals. The probability that mtDNA sequences potentially contributed to modern humans by Neandertals were lost by drift or swamped by continuous influx of modern human mtDNAs in the Neandertal gene pool is therefore much smaller than when contemporary humans are analyzed (e.g., ). In fact, the five early modern humans analyzed almost double the amount of information about the Upper Pleistocene mtDNA gene pool since, under a model of constant effective population size, all contemporary humans trace their mtDNA ancestors back to only four to seven mtDNA lineages 20,000 to 30,000 years ago (Figure 4A; Figure S2), while all other mtDNA sequences present in the gene pool at that time have been lost by random genetic drift. Since the probability is very low (p < 0.007) that one or more of the five early modern humans analyzed here are among these few ancestors of current humans, the five Upper Pleistocene individuals can be added to the ancestors of the current mtDNA gene pool to allow us to ask what extent of Neandertal mtDNA contribution to early modern humans can be statistically excluded using the coalescent. Under the model of a constant human effective population size of 10,000 over time (Figure 4A), any contribution of Neandertal mtDNA to modern humans 30,000 years ago larger than 25% can be excluded at the 5% level (Figure S3). A more realistic scenario may be that the spread of modern humans was accompanied by an increase in population size before and during their migration out of Africa and subsequent colonization of western Eurasia (see Figure 4B). In that case, the Neandertal contribution that can be excluded is smaller (i.e., less gene flow could have taken place), but that depends critically on when and how the expansion occurred. Finally, under the unlikely scenario that population size was constant during the migration out of Africa and colonization of Europe and expanded only after a putative merging with Neandertals, the Neandertal contribution could have been larger, but this also depends on the nature of the growth (see Figure 4C).

9It is noteworthy that under the model of constant population size, about 50 early modern human remains would need to be studied to exclude a Neandertal mtDNA contribution of 10%. To exclude a 5% contribution, one would need to study more early modern human remains than have been discovered to date. Thus, definitive knowledge of the extent of a putative contribution of Neandertals to the modern human gene pool will not be possible, although extensive studies of variation in the current human gene pool may clarify this question. It is, however, worthwhile to note that samples considered as anatomically “transitional” between modern humans and Neandertals, such as Vindija and Mladecˇ, analyzed here, fail to show any evidence of mtDNA admixture between the two groups. Thus, while it cannot be excluded that Neandertals contributed variants at some genetic loci to contemporary humans, no positive evidence of any such contribution has yet been detected.

10About 10 mg of bone were removed from each specimen and analyzed as in ) with minor modifications. In brief, proteins are hydrolyzed and amino acids labeled with o-phtaldialdehyde/N-acetyl-L-cysteine and analyzed by high performance liquid chromatography (Shimadzu, Kyoto, Japan) under conditions that separate the different amino acids as well as their stereoisomers. Eight amino acids are analyzed and their respective concentration measured: D- and L-alanine, glycine, D- and L-aspartic acid, serine, glutamic acid, valine, D- and L-leucine, and isoleucine.

11DNA extractions were performed in a laboratory dedicated to ancient DNA work. In this laboratory, positive air pressure is maintained with filtered air at all times, and all areas and equipment are treated with UV light when the laboratory is not used. A maximum of six bone or teeth samples were processed together with two blank extractions. Neandertal samples were always processed together with early modern human samples or cave bear samples. For each extraction, the samples were ground and between 30 mg and 120 mg of bone powder was extracted as in ). mtDNA sequences were amplified by polymerase chain reaction (PCR) using 5 μl of extract and 60 cycles. In addition, a minimum of four blank PCRs were performed together with each amplification from extracts. The “Neandertal-specific” amplification was carried out using the primers NL16230/NH16262 and an annealing temperature of 60°C. We consider it highly unlikely that the Neandertal-specific mtDNA fragments represent contaminations from other Neandertals, given that none of the extracts of modern humans or cave bears processed in parallel with the Neandertal remains yielded such products. The “hominoid” amplification was performed with the primers L16022/H16095 and an annealing temperature of 54°C. PCR products were cloned into Escherichia coli using the TOPO TA cloning kit (Invitrogen, Leek, The Netherlands), and ten or 30 clones of each amplification were sequenced on a ABI 3700 (Applied Biosystems, Foster City, California, United States).

12Given that previous analyses of mtDNA sequences have rejected a model of complete panmixia between Neandertals and early modern humans, we focused on the estimation of the level of admixture between Neandertals and early modern humans that can be excluded. For this purpose, we considered a population of early modern humans that merged at Tm with a (genetically different) population of Neandertal individuals (see Figure 4) from which point the fused population was panmictic. The probability of picking K individuals by chance in the merged population that all carry a modern human mtDNA sequence is (1 − c)K, where c represents the Neandertal genetic contribution to the merged population. If none of n mtDNA sequences sampled in the merged population is Neandertal-like, we can exclude (at the 5% level) contributions that give a probability smaller than 0.05 of observing only modern human sequences, i.e., (1 − c)K < 0.05. The number of ancestors of n samples at the time t is represented by a probability distribution, An(t). Thus, the probability of observing only one kind of sequences in n samples becomes:

13where K vary from 1 to n. For a population of constant size over time, Pr(An(t) = K) has been derived in ). We estimated the number of ancestors of n samples at time t as the expected value of An(t), E(An(t)), according to this model and calculate the probability of observing only human sequences for different values of c.

14The entire hypervariable region I sequence was determined from this specimen using amplifications and clones given in Figure S1. Its sequence is identical to the sequence previously determined from individual Vi-75 from Vindija. We could exclude cross-contamination from the old extract to this bone because different primers were used and some of the fragments of mtDNA amplified from Vi-80 were longer than those used to determine the sequence of Vi-75. Morphological analyses do not exclude that these two fragmentary bones (Vi-75 and Vi-80) may come from a single individual. Carbon-14 accelerator mass spectrometry dating, conducted in the Ångstrom Laboratory (Uppsala University, Sweden), yielded a date for Vi-80 of 38,310 ± 2,130 BP (before present). Since Vi-75 has been previously dated to over 42,000 BP, the possibility exists that the dates overlap since 42,000 BP is within two standard deviations of the Vi-80 date. Therefore, the bone labeled Vi-80 that yields the new mtDNA sequence could either be (i) a fragment of the same skeleton (individual) that was already successfully extracted, (ii) a bone from another individual maternally related to the first individual amplified, or (iii) another unrelated individual having by chance the same mtDNA sequence, which is not unlikely given the apparently low mtDNA diversity of Neandertals.