In recent times, genetic technology has progressed sufficiently to elucidate upon some of the questions normally preserved for archaeologists. One such question concerns the fate of a group of hominins that roamed Europe and East Asia for at least 250,000 years. During this time, this species adapted and endured some of the harshest environments on offer, all while showing signs of a unique culture. Only for them to suddenly disappear from the fossil record approximately 30,000 years before present (BP) (cf. Barton et al. 2007). So, what happened to our closest evolutionary relatives, the Neanderthals?
Much of the debate concerning Neanderthals is fiercely contested. Perhaps one of the more prominent issues regards the evolutionary status of Neanderthals in regards to modern humans (Homo sapiens). It is clear that during their time in existence, Neanderthals almost certainly overlapped both geographically and temporally with anatomically modern humans. Yet, what remains uncertain is whether or not Neanderthals contributed to our ancestry. This frequently elicits numerous responses under a varying degree of banners, such as gene-flow, hybridisation, adaptive introgression amongst other claims (Wolpoff, 2004). And furthermore, are Neanderthals a completely separate species (Homo neanderthalensis)? Or do we go one step further and say they are in fact a subspecies of us (Homo sapiens neanderthalensis)?
Lodged in-between these questions are many theories vying to explain the origins of modern humans. Broadly speaking, these accounts can be narrowed down into two categories: the out-of-Africa model and multiregional evolution model. The basic difference here rests on the two models’ separate accounts of human dispersal, interbreeding and species boundaries (Barton et al., 2007).
The out-of-Africa hypothesis is fairly true to its name, proposing Africa is the place where modern humans emerged and subsequently spread out from. In this model, humans displaced archaic hominin populations without any significant genetic contribution from the likes of the Neanderthals. Meanwhile, a multiregional model argues interbreeding did occur, and that these archaic hominins simply evolved into the modern human populations. As Barton et al. (2007) note: “[…] this model implies that the species ancestral to modern humans were not true biological species.” (pg. 740).
Importantly, these two hypotheses can be held up to scrutiny from a genetic perspective. If the out-of-Africa model is correct, there should be no evidence for genetic contribution from Neanderthals after humans left Africa and spread throughout the world. However, if there is a genetic overlap between Neanderthal and human populations, it is the multiregional model that earns the plaudits (cf. Hawks et al., 2007). With these questions and subject areas in mind, this essay will now discuss and draw conclusions from the current body of literature into Neanderthal DNA.
2. Control Region Studies
2.1 Initial approaches to Neanderthal mitochondrial DNA
Initial investigations (Krings et al., 1997) into the genetic world of Neanderthals came about through improved methods of ancient DNA (aDNA) recovery, particularly mitochondrial DNA (mtDNA).
Mitochondria are small structures in the cytoplasm of cells, whose primary function is to generate energy through the production of Adenosine Triphosphate (ATP) (Russell, 2002). Contained within a mitochondrion is its own circular strand of DNA, which is vital for purposes of analysis because it is found across all respiring organisms – and can therefore be subjected to methods of comparison (ibid). Furthermore, as most mtDNA is maternally inherited, researchers are able to trace matrilineal line relationships of modern humans right back to our most recent common ancestor – Mitochondrial Eve (ibid). This is done by first tracing the nucleotide base mutations, and then using the molecular clock technique (Motoo, 1968) to calculate the genetic drift. Here, an approximate divergence estimate is devised, which in the case of Mitochondrial Eve is ~150,000 BP (Cann et al., 1987).
Mitochondrial genetics is also applied to that of archaic remains, offering an insight into the genetic relatedness between extant individuals and their evolutionary antecedents. Of particular importance is that mtDNA is normally well preserved in fossils due to its relative abundance – important for researchers in need of high levels of sequence data (ibid). Also, because mtDNA neither checks for copying errors nor undergoes genetic recombination it has a higher rate of mutations occurring at a constant, traceable rate (ibid). To do gather and sequence these small samples of ancient mtDNA polymorphisms, researchers use a method known as polymerase chain reaction (PCR) (cf. Krings et al., 1997).
First developed by Kary Mullis (Russell, 2002), PCR employs a single primer, essentially a short sequence of DNA, and uses this as a template to produce numerous copies in a process known as amplification. Ignoring specific details involved in amplification, the basic output of this process are products called amplimers; and it is these sequences of DNA molecules that, when compared with contemporary samples, can “enable us to make evolutionary comparisons between ancient forebears and present-day descendents.” (ibid, pg. 191 – 192).
In the case of Neanderthals, PCR has largely been used to yield data pertaining to two control regions of mtDNA – hypervariable region 1 (HVRI) and hypervariable region 2 (HVRII) (Krings et al., 2000). As their name suggests, these non-coding regions are highly polymorphic (Russell, 2002) and can therefore employ the aforementioned molecular clock technique.
2.2 Neanderthal mtDNA control regions: what can it tell us?
One such study occurred in 1997, when a research team at the University of Munich (Krings et al.) were able to extract a 378-letter section of mtDNA from a Neanderthal-type specimen found in 1856. At the time, the team were only able to examine a 105 base pair (bp) segment of HVR1; yet, despite these limitations, they did manage to offer the first extensive analyses between human and Neanderthal mtDNA sequences.
In their paper, Krings and colleagues detail how they compared the Neanderthal HVR1 sequence against 994 extant human mitochondrial lineages in an effort to determine genetic relatedness. They found that on average modern human sequences differ by 8.0 ± 3.1 substitutions, whereas Neanderthals differ from modern humans by 27 ± 2.2 substitutions (ibid). Differences between specific human groups (such as Africans or Europeans) were also compared to clarify the range of modern human diversity.
Next, phylogenetic analyses of 16 chimpanzee lineages, 986 human lineages and the Neanderthal sequence were performed to ‘[…] agree with the pairwise comparisons of sequence difference in placing the Neanderthal mtDNA sequence outside the variation of modern human mtDNA.’ (ibid, pg. 25). Additionally, they proposed an estimated date of divergence of Neanderthal and human mtDNA between 550,000–690,000 BP, providing significant supportive evidence for palaeontological and archaeological records that place the split between 250,000–300,000 years ago. The rationale behind this is explained by Krings et al. in which the ‘divergence of genes is expected to predate the divergence of populations’ (ibid, pg. 27).
As there is a perceived divergence between Neanderthal and human mtDNA lineages, it can also be argued that very little, perhaps none whatsoever, admixture took place when the two populations temporally and geographically overlapped. Using data from the phylogenetic analyses, Krings et al. demonstrate the Neanderthal sequence to statistically correlate as an ‘outgroup’ – a genetic grouping, for the purposes discussed, outside the modern human mtDNA lineage –, and not contributing to modern human mtDNA variation (ibid).
Since the 1997 paper, there have been several further studies examining control regions of Neanderthal mtDNA (Krings et al., 1999; Krings et al., 2000; Ovchinnikov et al., 2000; Schmitz et al., 2002; Knight, 2003; Serre et al., 2004; Lalueza-Fox et al., 2005; Caramelli et al., 2006; Orlando et al., 2006). On the whole, these papers all lend confirmatory evidence for the Neanderthal-Human ancestral mtDNA divergence date provided by Krings et al. (1997). They also clarify some issues over the limited sample data, through sequencing mtDNA from several Neanderthal specimens found in different locations (Serre et al., 2004; Orlando et al., 2006). In particular, larger sample data reaffirmed original notions of Neanderthals being a genetic outgroup (Serre et al., 2004).
One new piece of speculation brought forth from some of these later studies (Krings et al., 2000) is that the diversity of Neanderthal mtDNA appears to be restricted. Diversity is important as it helps predict the initial population size out of which the general population sprung. In the case of Neanderthals, it is suggested (ibid) they might parallel humans whose “[…] low genetic diversity seen in both mtDNA and nuclear DNA sequences is likely to be the result of a rapid population expansion from a population of small size” (ibid, pg. 145). Though what this implies is not altogether clear, – except for suggestions of a genetic bottleneck taking place prior to Neanderthal-human ancestral split–, as we are not sure what exactly caused such a significant and rapid population expansion in humans.
It is clear that control regions do not make up the entire mtDNA genome (Green et al., 2008), and as such fail to tell the whole story about Neanderthal mtDNA evolution (Wolpoff et al., 2004). In the majority of these studies, focus is placed on HVRI sequences, with only two (Krings et al., 1999; Krings et al., 2000) managing to describe HVRII sequences. Such limitations surrounding mtDNA data is far from conclusive.
For instance, Gutierréz et al. (2002) incorporate both the HVRI and HVRII sequences into their dataset and come to startlingly different conclusions. They find ten of the African sequences form an outgroup to other Africans, non-Africans and two Neanderthals. On the basis of this data, some Neanderthals are more related to humans than ten Africans. Such discrepancies may support claims for Neanderthals as a subspecies of Homo sapiens (Wolpoff et al., 2004). Or perhaps it is simply just a case of the data being incomplete. With these doubts looming over mtDNA, it appears many unsolved questions remain over the Neanderthals’ genetic heritage.
3. Beyond control regions: A complete mitochondrial genome
3.1 A genomics approach
As previously established: simply examining control regions of mtDNA is not a forgone conclusion in respect to maternal divergence estimates (cf. Gutierréz et al., 2002). These inherent flaws mean that conclusions based on Neanderthal mtDNA are hard to reconcile with insights from other disciplines. One solution is to sequence a Neanderthal mtDNA genome. But before we delve into this, I will briefly describe genomics – and how it differs from genetics.
Arriving at a distinction between genomics and genetics is relatively straightforward: the latter deals specifically with small, single gene regions, while the former examines a much larger cross-section of gene regions, more commonly described as a genome (Russell, 2002). In particular, genomics has brought forth a greater understanding of the interactions between genes (epistasis), their role in producing multiple phenotypic traits (pleiotropy) and between-species comparisons (DNA-DNA Hybridisation) (ibid).
Especially relevant to much of this essay’s enquiry is the latter point, of between-species comparisons, where a potential Neanderthal genome can be cross-referenced with that of humans and chimpanzees (Mikkelsen et al., 2005). This is made possible through improved sequencing techniques, which for the most part are highly technical, but simply put: these methods improve upon PCR by keeping more sequences in tact during amplification (Green et al., 2006). Therefore, more of the genome can be read without fears of preferentially amplifying contaminated aspects of a sequence, while expanding the focus to larger sequences of variation. Such is the basis of the 454 group’s sequencing of an entire Neanderthal mitochondrial genome (Green et al., 2008).
3.2 Reinforcing control regions
For the most part the Neanderthal mitochondrial genome reinforces the original conclusions arrived at by the control region studies. It places the divergence date of the human and Neanderthal mtDNA lineages at ~660,000 BP; significantly overlapping with initial HVR1 studies (Krings et al., 2000). Subsequently, it resolves some challenges put forward by Gutierréz et al. (2002). In particular, by using mtDNA from 53 modern humans around the globe the authors show Neanderthal mtDNA “[…] falls outside the variation of extant human mtDNA variation.” (Green et al., 2008, pg. 422). Expanding upon this, it is estimated that Neanderthal mtDNA variation to be three times as different as human-human variation (ibid).
The team also open up a new avenue of investigation by sequencing protein-coding genes in mtDNA. Formerly, these were unavailable for analysis as control regions are strictly non-coding – and it is here where genomics really demonstrates its advantages. Specifically, they compare the ratio of synonymous to nonsynonymous changes in 13 protein-coding genes, with additional aims of elucidating the role purifying selection plays (Green et al., 2008). Interestingly there are more nonsynonymous changes on the Neanderthal lineage, indicative of “[…] relaxed efficacy of purifying selection in Neanderthals.” (Clark, 2008, pg. 389). This inefficiency in removing deleterious polymorphisms is accounted for by Neanderthals having a smaller effective population size to that of modern humans (ibid), lending support to a restricted Neanderthal mtDNA genepool (Krings et al., 2000).
An alternative explanation is to invoke positive selection, as it would also accelerate nonsynonymous changes (Clark, 2008). This leads to us to Green et al.’s (2008) second discovery on the human form of COX2: four nonsynonymous amino acid substitutions since separation from Neanderthals.
The high polymorphic rate of mtDNA, and its ability to accrue deleterious mutations, suggests genetic drift and purifying selection are the only processes involved in shaping mtDNA sequence variation (Stewart et al., 2008). COX2 could challenge this notion if found to display significant functional effects, implying a role for positive selection (Kelley and Swanson, 2008). If positive selection is found to be acting in mtDNA, then it throws into doubt some hypotheses about Neanderthal-human admixture for one very important reason: the absence of Neanderthal mtDNA in the human lineage would be expected (Cyran and Kimmel, 2005).
Until the role of COX2 can be determined, it is probably best to revert to the default position, where the amino acid substitutions “[…] may represent either a minor adaptive advantage, perhaps of regulatory relevance, or have no significant functional consequences for mitochondrial function.” (Green et al., 2008, pg. 423).
3.4 The limitations of mtDNA
Given all the contamination problems involved in sequencing (ibid); the destructive process of amplification (Noonan et al., 2006); making large-scale conclusions from a limited specimen pool (Wolpoff, 2004); and not to forget, the various assumptions made on theoretical underpinnings (such as the absence of positive selection (ibid) and clock-like rate of molecular evolution (Strauss, 1999)), then there are still fundamental drawbacks of using mtDNA.
It is fairly obvious mtDNA cannot reveal all when discussing gene flow between populations, especially those as closely related as humans and Neanderthals. Aside from being inherited through a non-Mendelian process, excluding any insight into paternal lineages, human mtDNA only encodes for a small portion of the genome (approximately 16,000 base pairs) when compared to the 3 billion base pairs found in sexually transmitted nuclear DNA (Russell, 2002).
So, while construing mtDNA as advantageous in producing divergence estimates, it cannot, by itself at least, offer much more insight into addressing the issue of human-Neanderthal admixture. To shed light on these issues, nuclear DNA is needed which, until recently, had only been found in one Neanderthal remnant capable of sequencing: a 38,000 year old specimen recovered from Vindija, Croatia (Noonan et al., 2006).
 Anatomically modern humans will subsequently be referred to as modern humans or just humans.
 It has been suggested that mtDNA can also be inherited through paternal lineages (Awadalla et al., 1999). Though these claims have been challenged (Elson et al., 2001), it is outside the scope of this essay to discuss such issues and it will be assumed that mtDNA is inherited maternally.
 Nucleotide base mutations are also referred to in this essay as Single Nucleotide Polymorphisms (SNPs)
 The term outgroup can also apply to other monophyletic clades within taxonomy, for example: Chordates and Echinoderms are more closely related than Mollusks, therefore Mollusks are considered to be an outgroup of the Chordate-Echinoderm clade. See http://en.wikipedia.org/wiki/Outgroup for more information.
 The authors from whom the above quote was taken suggest the use of complex language could have been an important factor in humanity’s rapid population expansion, which as you might guess has further implications in suggesting Neanderthals might have had language – a topic touched upon later in this essay (see FOXP2).
 Though as we shall see, some sequencing techniques do not even require amplification (Noonan et al., 2006).
 Silent amino acid changes (Kelley and Swanson, 2008).
 Amino acid altering (ibid)
 Purifying selection is just another term for negative selection, which is when selection is made against any deleterious mutations (Barton et al., 2007).
 The full, tongue twisting title: COX gene encoding subunit 2 of cytochrome c oxidase.
 Genetic drift shows how random mutations can become fixed in a population without having any adaptive value, and is based on the variance in the fitness of an individual to produce offspring (Barton et al., 2007).
 Positive selection “favours one allele over another or favours increased values of a quantitative trait.” (Barton et al., 2007, pg. 476).
Blum MG, & Rosenberg NA (2007). Estimating the number of ancestral lineages using a maximum-likelihood method based on rejection sampling. Genetics, 176 (3), 1741-57 PMID: 17435232
Green, R., Malaspinas, A., Krause, J., Briggs, A., Johnson, P., Uhler, C., Meyer, M., Good, J., Maricic, T., & Stenzel, U. (2008). A Complete Neandertal Mitochondrial Genome Sequence Determined by High-Throughput Sequencing Cell, 134 (3), 416-426 DOI: 10.1016/j.cell.2008.06.021
John Hawks’ weblog is an indispensable resource for Neanderthal genetics: http://johnhawks.net/weblog/reviews/neandertal_dna