Changes in biological pathways during 6,000 years of civilization in Europe

Abstract

The beginning of civilization was a turning point in human evolution. With increasing separation from the natural environment, mankind stimulated new adaptive reactions in response to new environmental factors. In this paper, we describe direct signs of these reactions in the European population during the past 6,000 years. By comparing whole-genome data between Late Neolithic/Bronze Age individuals and modern Europeans, we revealed biological pathways that are significantly differently enriched in non-synonymous SNPs in these two groups and which therefore could be shaped by cultural practices during the past six millennia. They include metabolic transformations, immune response, signal transduction, physical activity, sensory perception, reproduction, and cognitive functions. We demonstrated that these processes were influenced by different types of natural selection. We believe that our study opens new perspectives for more detailed investigations about when and how civilization has been modifying human genomes.

Full text

1
Article
Discoveries
Title: Changes in biological pathways during 6,000 years of civilization in Europe
Authors: Evgeny Chekalin1, Alexandr Rubanovich1, Tatiana V. Tatarinova1,2,3,4, Artem Kasianov1,5, Nicole
Bender6, Marina Chekalina1, Kaspar Staub6, Nikola Koepke6, Frank Rühli6, Sergey Bruskin1*, Irina
Morozova6*
Affiliations:
1Vavilov Institute of General Genetics, Russian Academy of Sciences, Moscow, Russia
2Department of Biology, University of La Verne, La Verne, CA, USA
3A. A. Kharkevich Institute for Information Transmission Problems, Moscow, Russia
4Department of Fundamental Biology and Biotechnology, Siberian Federal University, Krasnoyarsk, Russia
5Center for Data-Intensive Biomedicine and Biotechnology, Skolkovo Institute of Science and Technology,
Moscow, Russia
6Institute of Evolutionary Medicine, University of Zurich, Zurich, Switzerland
* Shared authorship
Corresponding authors: Irina Morozova irina.morozova@iem.uzh.ch, Sergey Bruskin brouskin@vigg.ru

2
Abstract
The beginning of civilization was a turning point in human evolution. With increasing separation
from the natural environment, mankind stimulated new adaptive reactions in response to new environmental
factors. In this paper, we describe direct signs of these reactions in the European population during the past
6,000 years. By comparing whole-genome data between Late Neolithic/Bronze Age individuals and modern
Europeans, we revealed biological pathways that are significantly differently enriched in non-synonymous
SNPs in these two groups and which therefore could be shaped by cultural practices during the past six
millennia. They include metabolic transformations, immune response, signal transduction, physical activity,
sensory perception, reproduction, and cognitive functions. We demonstrated that these processes were
influenced by different types of natural selection. We believe that our study opens new perspectives for more
detailed investigations about when and how civilization has been modifying human genomes.

3
Introduction
It is generally accepted that the term “civilization” refers to any complex society characterized by
urban development, social stratification, symbolic communication forms (typically represented by writing
systems), and a perceived separation from and domination over the natural environment (Adams 1966).
From an evolutionary point of view, civilization started when humans, instead of reacting to the
environment, began to actively shape it. Since the Neolithic transition, mankind has experienced a shift to
agriculture, domestication of animals and plants, sedentism, significant increase in population density, and
exposure to new pathogens; most of these effects have been self-imposed. Humans have been creating the
artificial environment separating them from nature. This new environment induces new responses to it.
At present, it is supposed that culturally derived selection pressures should be stronger than non-
cultural ones (Bersaglieri, et al. 2004; Ehrlich 2000; Feldman and Laland 1996; Laland 2008; Laland, et al.
2010; Richerson and Boyd 2005). The main reason for this is that using cultural practices led to drastic
population growth. As a result, the number of targets for mutations (both advantageous and disadvantageous)
in the population increased, as did the number of individuals for selection (Laland 2008). Paradoxically,
mutations accumulated in human genomes as a result of relaxed natural selection can also serve as targets for
selection in new environmental conditions. Moreover, new cultural practices typically spread more quickly
than genetic mutations, and the more individuals exhibiting the cultural trait, the greater the intensity of
selection (Boyd and Richerson 1985; Cochran and Harpending 2009; Hawks, et al. 2007; Kimura 1955;
Laland 2008).
Culturally derived selection leaves signs in the human genome. Some of these signs (like lactase
persistence) are quite evident (Allentoft, et al. 2015; Beja-Pereira, et al. 2003; Gamba, et al. 2014; Holden
and Mace 1997), while many others are still uncertain (Galvani and Slatkin 2003; Libert, et al. 1998; Sabeti,
et al. 2005; Stephens, et al. 1998). Revealing and analyzing these selection signatures is of high importance
not only for improving our understanding of connections between the human organism and the environment
but also for deepening our insight into mechanisms of emergence of so-called “diseases of civilization.”
The earliest stages of human civilization are the Late Neolithic Age and the Bronze Age. These were
the epochs that gave rise to our present lifestyle. The six thousand years between that period and modern
times encompass the greater part of human civilization events. In this paper, we study the genetic
consequences of these cultural events.
Many different approaches are used to reveal and analyze possible selection signals (Field, et al.
2016; Grossman, et al. 2013; Mathieson, et al. 2015; Quach, et al. 2009; Sabeti, et al. 2007; Tang, et al.
2007; Voight, et al. 2006; Williamson, et al. 2007). Most are based on modern human genome-wide data
and, therefore, represent indirect evidence of selection. Objective information can be obtained by direct
comparison of ancient and modern human genomes. The first steps in this direction were made relatively
recently; they became possible thanks to whole-genome next generation sequencing of ancient samples.
These studies have revealed selection signatures in SNPs associated with skin pigmentation, diet, and

4
immunity, as well as with some complex traits, i.e., human height (Allentoft, et al. 2015; Dannemann, et al.
2016; Fu, et al. 2016; Mathieson, et al. 2015; Olalde, et al. 2014).
Providing that natural selection should act through phenotypes, we assume that selection signals for
multigenic traits should be analyzed not only at the level of individual SNPs but also at the level of
biological pathways, where the influence of individual SNPs is aggregated into functional groups. This
approach has been previously used, for instance, to study selection signatures between human and
chimpanzee lineages (Somel, et al. 2013). In the present study we applied pathway analysis to low-covered
whole-genome ancient DNA sequence data. We compared data on European Late Neolithic/Bronze Age
individuals (Allentoft, et al. 2015; Gamba, et al. 2014; Haak, et al. 2015; Mathieson, et al. 2015) with those
from modern European individuals (http://www.internationalgenome.org/) supposedly of Bronze Age
ancestry and occupying the same geographical area as their ancestors. Our aims were: i) to reveal non-
synonymous SNPs in ancient and modern groups; ii) to associate these SNPs with biological pathways; iii) to
calculate the differences in pathway enrichment between the ancient and modern groups. The revealed
differences indicate the processes that we suppose have been shaped by introduction of human cultural
practices during the past 6,000 years.

5
Results
Compatibility of the data
We compared whole genome data from 150 ancient samples (Supplementary Table 1 and 2) dated
between 3,500 and 1,000 BCE (Allentoft, et al. 2015; Gamba, et al. 2014; Haak, et al. 2015; Mathieson, et al.
2015) (Fig. 1) with data on 305 modern Europeans genotyped in the framework of the 1,000 Genomes
Project (Genomes Project, et al. 2015). We analyzed 40,573 synonymous and 48,860 non-synonymous SNPs
from the Bronze Age group vs 72,558 synonymous and 96,710 non-synonymous SNPs from the modern
group using the pipeline shown in Figure 2.
To test whether there is any genetic continuity between the Bronze Age group and the modern group,
we applied two different approaches. First, Principal Components Analysis (PCA) demonstrated that the
ancient and modern European individuals are co-located within the same cluster and are separated from
modern individuals from other geographic regions (Africa, America, and Asia) (Fig. 3).
Second, to test whether the analyzed modern individuals possess genetic ancestry of the Bronze Age
individuals, we measured the proportion of the Bronze Age individuals in modern samples. Figure 4 shows
that the linear composition of Bronze Age ancestry in the modern individuals is relatively high and varies
from 20% to 90%.
Therefore, we can confidently consider the analyzed modern Europeans to be genetic descendants of
the Bronze Age Europeans; this fact gives us the basis for studying microevolution changes that occurred
during the past six millennia in Europe.
Comparison of ancient and modern data
Due to the low coverage of each position on the genome in the ancient data, consideration of
individual SNPs for direct comparison of ancient and modern data does not produce biologically or
statistically significant results, since variant call at each ancient genomic position has limited fidelity.
Therefore, we considered one ancient “merged genome” and one modern “merged genome”. The ancient
“merged genome” was assembled from compiling all SNPs of European Bronze Age individuals, while the
modern “merged genome” was assembled from all SNPs of modern European individuals. Grouping of the
SNPs into KEGG biochemical pathways (see Material and Methods) gave the additional robustness to the
calculations.
We assumed that during neutral evolution the same biological pathways in the ancient and in the
modern groups should accumulate mutations at the same rate, while under selection pressure the rate of
accumulation of mutations in the same pathways should be different. Therefore, we calculated two types of
enrichment scores for pathways: 1) differential synonymous SNP enrichment (DSSE) scores between ancient
and modern groups and 2) differential non-synonymous SNP enrichment (DNSE) scores for these groups
(see Materials and Methods). The enrichment score for each pathway was calculated as the deviation of the
fraction of ancient SNPs in the given pathway from the expected fraction of SNPs in the ancient “merged

6
genome”. Therefore, when there are more SNPs in the ancient “merged genome”, compared to what is
expected, the enrichment score is positive; when there are less SNPs in the ancient “merged genome”,
compared to what is expected, the enrichment score is negative. Hence, a positive enrichment score indicates
higher pathway enrichment in the ancient group; a negative enrichment score indicates higher pathway
enrichment in the modern group.
Comparative analysis of DSSE scores revealed that none of the pathways show significant
differences in synonymous SNP enrichment between the ancient and the modern groups (Supplementary
Table 3). This corresponds to the hypothesis of neutral evolution for this type of mutations. At the same time,
comparison of DNSE scores revealed 15 pathways that were differentially enriched in non-synonimous SNPs
between the Bronze Age and modern European individuals (Fig. 5, Table 1, Supplementary Table 4 and 5).
We also normalized non-synonymous SNPs on synonymous SNPs. The results (Figure 6) showed that all p-
values of the synonymous test, as well as most p-values of the non-synonymous test are inside the area of
non-significant differences (shaded rectangle). At the same time, p-values of the non-synonymous test for 15
differently enriched pathways are outside the area of non-significant differences. This confirms the
significance of differences in these pathways between ancient and modern Europeans.
The significance of the differences of enrichment scores between the ancient and the modern groups
was assessed using the Bonferroni correction with p<0.01 (Table 1). Benjamin et al. (Benjamin, et al. 2017)
proposed to use the threshold of p<0.005 (see Material and Methods). We suggest that, among 15 revealed
pathways, the two pathways that did not pass this threshold (“Pentose and glucuronate interconversions” and
“PI3K-Akt signaling pathway”) should be interpreted with caution. We also excluded two of the pathways:
“Metabolic pathways” since this grouping is too general and “Ascorbate and aldarate metabolism” since it is
very reduced in humans and its functions are not unique (Ye and Doak 2009).
Therefore, we have identified the following pathways to be significantly different between the
Bronze Age and modern groups: “Pentose and glucuronate interconversions”, “Drug metabolism by
cytochrome P450”, “Chemical carcinogenesis”, “ABC transporters”, “Antigen processing and presentation”,
“Graft-versus-host disease”, “Autoimmune thyroid disease”, “Hypertrophic cardiomyopathy”, “Olfactory
transduction”, “Oocyte meiosis”, “Long-term potentiation”, and “Dopaminergic synapse”.
We also compared the distribution of regulatory SNPs between Bronze Age and modern individuals
(see Material and Methods). A proportion test revealed no difference in enrichment of the KEGG pathways
between the ancient and the modern groups. This result was expected since the functions of most of the
revealed SNPs in the regulatory regions are not yet known. Non-functional SNPs contribute to noise
interfering the detection of functional SNPs (the same situation could happen if we analysed synonymous
and non-synonymous SNPs together).

7
Verification of the results
As an alternative hypothesis, we considered the possibility that the obtained results can be explained
by insufficient sequence coverage of pathways in Bronze Age individuals. To test this hypothesis, we
performed the following computations. First, we calculated the Spearman’s correlation coefficient between
enrichment score and fraction of covered length (Supplementary Fig. 1). The coefficient of determination
was R2=0.1. This implies that a change in the coverage can explain only 10% of the variability of pathway
SNP enrichment, and cannot be the leading cause of the observed effect. Second, we analyzed the median
coverage and median length of genes in the pathways (Supplementary Fig. 2). With the rare exception (three
pathways), more than 50% of individual genes were covered in the studied pathways. The revealed enriched
pathways were clustered together with unenriched pathways. Third, we calculated average coverage per bp
per sample per pathway and total coverage per bp per pathway (Supplementary Fig. 3). In general, there is
no relationship between enrichment and coverage. The only exception is olfactory transduction pathway
(Supplementary Fig. 3A), whose average coverage per sample is a bit lower in comparison to other
pathways. However, the total coverage for this pathway (Supplementary Fig. 3B), though a bit lower in
comparison to most of other pathways, is not an outlier (there are two other pathways with the same
coverage which did not show any difference in enrichment between ancient and modern groups). For our
calculations, we used data from total, not average, coverage. Therefore, there is no relationship between
enrichment and the gene’s size or coverage.
The observed trend might also be the result of general inter-population differences between the two
groups. To test this hypothesis, we calculated inter-population differences between modern European groups
using the same pathway enrichment analysis (Supplementary Table 6). No difference in enrichment in any
pathway was revealed between present-day Europeans. Therefore, the observed differences between the
Bronze Age and modern groups are the results of microevolution changes during the past 6,000 years.

8
Discussion
Since the Late Neolithic, the European lifestyle has changed drastically. The main factors
determining the relationship between environment and the human body have undergone significant
alterations. For example, pre-agrarian and early agrarian populations were exposed to environmental
influences from a comparatively small geographical area (Gillings, et al. 2015). In contrast, modern
Europeans exist in a globalized world where global travel (and corresponding environmental exposures) as
well as different new types of food, clothes, and other consumables are common. Many new factors have
appeared, such as dietary changes, new pathogens, new medications, as well as high population density and
closer connections between distant groups of people. All of these new conditions inevitably provoke
responses from the human body.
In this paper, we studied how introduction of different cultural practices during the past 6,000 years
could shape human genomes. We traced the microevolution of modern Europeans back to their ancestors,
carriers of the Late Neolithic and Bronze Age cultures. We revealed twelve biological pathways that are
significantly different between the Bronze Age and modern groups. For most of them (except three) the
number of non-synonymous mutations is higher in the modern group than in the Bronze Age group, which
means the accumulation of mutations during the past 6,000 years. In the next paragraphs we attempt to
explain what civilization events during the past millennia could have caused the changes in these pathways.
We detected significant changes in a number of pathways responsible for metabolism. One of them,
Pentose and glucuronate interconversions, is associated with carbohydrate metabolism. In the human
organism, this pathway mainly describes the transformation of UDP-glucose, α-D-glucose-1-phosphate, and
D-xylose (Du, et al. 2016). We suggest that changes in this pathway are the consequences of dramatic diet
modifications arising with the introduction of agriculture, an important event that stimulated the Neolithic
transition and progressed during the Bronze Age. One of three substrates entering the pentose and
glucuronate interconversions pathway, UDP-glucose, comes from the galactose metabolism pathway (Du, et
al. 2016). The main source of galactose in the modern human diet is lactose from milk. Humans are the only
mammals who have the ability to utilize lactose in adulthood. This ability is provided by a single mutation in
an enhancer region of the lactase gene (LCT) whose product lactase, a participant in the galactose
metabolism pathway, breaks down lactose (Enattah, et al. 2008; Lewinsky, et al. 2005). It is believed that in
Europe the LCT mutation arose in the Bronze Age or somewhat earlier as a result of milking (Allentoft, et al.
2015; Gamba, et al. 2014; Holden and Mace 1997). In modern Europe the mutation frequency is up to 100%
(Gerbault, et al. 2011) indicating strong positive selection of this gene. Apparently, such a significant change
in the galactose metabolism pathway could strongly affect the product (UDP-glucose) yield, which in turn
could modify the next pathway, pentose and glucuronate interconversions. Other substrates for the pentose
and glucuronate interconversions pathway are α-D-glucose-1-phosphate, the product of glycolysis, and D-
xylose, entering from the starch and sucrose metabolism pathway (Du, et al. 2016). Glucose and starch dairy
intake has changed dramatically during the past 6,000 years: as a result of agriculture, the ratio of
carbohydrate-rich food, especially grain-based products, has increased significantly in the human diet
(Cordain, et al. 2005). This ratio further increased after the Industrial transition in the 18th-19th century after

9
which industrially processed flour and sugar became commonly available (Adler, et al. 2013; Cordain, et al.
2005). Therefore, we suppose that changes in nutrient consumption and thus in the metabolism of substrates
for the pentose and glucuronate interconversions pathway have caused an accumulation of non-synonymous
mutations, which could modify this pathway.
Other metabolic pathways are associated with the transformation of xenobiotics. They include Drug
metabolism by cytochrome P450 and chemical carcinogenesis (two closely related pathways, Metabolism
of xenobiotics by cytochrome P450 and Drug metabolism by other enzymes, have passed only the
Benjamini-Hochberg correction and not the Bonferroni one (Supplementary Table 3 and 4). These pathways
are closely connected because they have partially overlapping mechanisms (Lang and Pelkonen 1999;
Oliveira, et al. 2007) (indeed, the chemical carcinogenesis pathway shares approximately 70% of genes with
the cytochrome P450 metabolic pathway (Supplementary Table 7)). During the past several
millennia, substantial changes in human lifestyle were accompanied by introduction of large amounts of
different xenobiotics (including new types of food, alcoholic beverages, and microbial toxins). Some of them
(such as medications, plant fertilizers, and food additives) are supposed to improve the quality of human life.
Others (such as heavy metals and other pollutants) are side effects of civilization activities. All of these
substances can shape human genomes by causing mutations (directly or indirectly) or by inducing natural
selection. Our results suggest that new environmental factors in the form of xenobiotics have induced
genomic responses via increasing gene variability and, as a result, modification of corresponding
pathways. Unfortunately, we can see not only this adaptation but also an increase in the number of mutations
in the chemical carcinogenesis pathway.
The ABC transporters pathway can be considered a part of the human metabolic system. Human
ABC transporter genes encode transmembrane pumps that transport various substrates (including amino
acids, lipids, proteins, inorganic ions, drugs, and other xenobiotics) against concentration gradients (Moitra
and Dean 2011; Pohl, et al. 2005; Stefkova, et al. 2004; Vasiliou, et al. 2009). Therefore, changes in the
quantity or quality of these substrates through diet modifications or introduction of xenobiotics could also
affect genes encoding these transport proteins. Interestingly, signals of positive selection were detected
earlier in some genes associated with transport of vitamins and cofactors (Tang, et al. 2007; Voight, et al.
2006). In aggregate, these data suggest that changes in lifestyle have induced genetic modifications in a
system for transport of nutrients and xenobiotics in the human body during the past several millennia.
Antigen processing and presentation is a very important part of the adaptive immune system, which
is evolutionarily young and very reactive to environmental factors. It is the first line of host immune
defense that recognizes and initiates immune responses to a broad range of alien agents. The major
histocompatibility complex (MHC) plays the most important role in this process. Due to a very specific
mechanism of antigen interaction, MHC proteins are highly diverse, and the genes encoding them (human
leukocyte antigen genes, HLA) are the fastest evolving genes in the human body (Blum, et al. 2013; Forni, et
al. 2014). Unsurprisingly, the antigen processing and presentation pathway has been shaped during the past
6,000 years. The introduction of farming, which led to exposure to a huge variety of new pathogens, as well
as other civilization factors such as urbanization (thus increasing population

10
density, insufficient sanitation, peri-domestic animals, etc.) and development of trading routes increasing the
probability of disease spread etc., has changed the pathogenic environment drastically. Major pandemics,
such as the plague in Europe, could have also played a very important role in the selection of immune
system genes (Barnes, et al. 2011; DeWitte 2014; Laayouni, et al. 2014). It is quite possible that modern
medicine has also been modifying the genetic mechanism of immune response. This issue still needs
extensive research.
Graft-versus-host disease is considered by clinicians to be a disorder but from the evolutionary point
of view it represents a powerful system of immune response to alien agents. This alloimmunity is
evolutionarily ancient and seems to be an “unavoidable consequence” of a natural mechanism of antigen
processing and presentation (Lakkis and Lechler 2013). Indeed, graft-versus-host disease shares 71% of
common genes with the antigen processing and presentation pathway (most of them are HLA-genes)
(Supplementary Table 7). Being an inseparable part of the human defense system, alloimmunity should
evolve together with it. Therefore, we suppose that all the above-mentioned factors that caused changes
in the antigen processing and presentation process should act similarly on the graft-versus-host disease
pathway.
Another pathway connected with antigen processing and presentation is connected to
autoimmunity. We revealed selection signals for autoimmune thyroid disease. It shares 37% of genes (all
of them belong to HLA group) with the antigen processing and presentation pathway (Supplementary Table
7). Therefore, the emergence of this autoimmune disease is probably a cost of the fast adapting antigen
processing and presentation system; however, we believe that there are additional environmental factors that
contributed to the intensive evolution of this particular disorder. Autoimmune thyroid disease is a syndrome
characterized by chronic inflammation of the thyroid. It is believed to be specific for H. sapiens (Aliesky, et
al. 2013) but it is unknown when this disease appeared in the human population. Currently, autoimmune
thyroiditis is quite common in the European population (Vanderpump 2011). It can probably be connected
with the increased carbohydrate uptake after introduction of agriculture which, in turn, has increased thyroid
hormone levels in the human body (Kopp 2004). Increased levels of thyroid hormone, especially in
combination with inappropriate iodine supply, cause several detrimental systemic disorders (Kopp 2004;
Motomura and Brent 1998). Therefore, we assume that the emergence of new non-synonymous mutations is
probably an organismal reaction to this new hormonal status. Hypothetically, this reaction could be a kind of
prevention mechanism, or, on the contrary, a consequence of thyroid hyperfunction.
We revealed significant changes in the PI3K-Akt signaling pathway. It is one of the universal
signaling pathways, which are active in most of the human body's cells. It is responsible for a variety of
fundamental processes, such as apoptosis, cellular growth, proliferation, cell survival, metabolism, and
others (De Santis, et al. 2017; Duronio 2008; Engelman, et al. 2006; Song, et al. 2005). This pathway was
shown to play an important role in immunity, cancer, and long-term potentiation (Chen, et al. 2017; Fresno
Vara, et al. 2004; Hou and Klann 2004; Pons-Tostivint, et al. 2017; Porta, et al. 2014; Sui, et al. 2008;
Weichhart and Saemann 2008). The PI3K-Akt signaling pathway is activated by different stimuli including
antigens, inflammation, environmental toxicants, and drugs (De Santis, et al. 2017; Duronio 2008;

11
Engelman, et al. 2006; Song, et al. 2005). Therefore, any of the factors described above (changes in diet,
pathogen environment, xenobiotics) could affect this pathway and stimulate accumulation of non-
synonymous mutations in it.
The hypertrophic cardiomyopathy (HCM) pathway also shows signals of selection during the past
6,000 years. HCM is an autosomal dominant disease, which is manifested as a functional impairment of the
heart. It occurs in approximately 0.2% of modern populations (Cirino and Ho 1993; Marian 2010). The
course of the disease is very often asymptomatic; however, in some cases, especially with intensive physical
activity, a sudden cardiac death can occur. For example, hypertrophic cardiomyopathy is the leading cause of
sudden cardiac death in young athletes (American College of Cardiology Foundation/American Heart
Association Task Force on Practice, et al. 2011; Barsheshet, et al. 2011). We suppose that the higher
prevalence of non-synonymous SNPs in the modern group in comparison to the ancient group can be a
consequence of the gradual change in European lifestyle from pre-technological agrarians to modern post-
industrial societies: a redistribution of physical load, as well as of balance between calorie uptake and
physical activity (Lightfoot 2013). Genetic monitoring and adequate therapy (American College of
Cardiology Foundation/American Heart Association Task Force on Practice, et al. 2011; Cirino and Ho
1993) probably also play a role in the accumulation of HCM-associated mutations in modern Europeans and,
therefore, one can expect an even higher frequency of these mutations in the future.
Olfactory transduction, the capacity to discriminate odors, shows the strongest signal of selection
(Table 1). As reported before, olfactory genes in primates have a tendency to pseudogenization(Gilad, et al.
2003b; Pierron, et al. 2013; Somel, et al. 2013). In humans, approximately 60-70% of olfactory genes are
pseudogenes; this probably reflects a decreasing need for olfactory perception in great apes and especially in
humans (Gilad, et al. 2003a; Rouquier, et al. 1998). Indeed, relaxed selection has been described for most
human olfactory genes (Gilad, et al. 2003a; Somel, et al. 2013) leading to fast accumulation of mutations in
these genes (Gilad, et al. 2003b; Miyata and Hayashida 1981). According to our results, this process has also
been taking place during recent human microevolution. Most likely, the process of pseudogenization of
olfactory genes is still ongoing. At the same time, we cannot exclude the possibility that introduction of new
cultural practices (new types of food, perfume, etc.) provides new directions for selection, at least for some
olfactory genes.
All the pathways described above have been accumulating non-synonymous mutations during the past
6,000 years. At the same time, we revealed three pathways with the opposite pattern: the modern group has
significantly fewer mutations than the ancient one. These pathways are described below.
We revealed significant changes in a pathway associated with oocyte meiosis. Oogenesis is the most
important part of female reproductive function. It determines the timing of puberty and menopause as well as
the effectiveness of reproduction. It has been shown that all these parameters are strongly influenced by
environmental factors (Gluckman and Hanson 2006b; Gold 2011; Henneberg and Saniotis 2013).
Retrospective analysis and direct data suggest significant fluctuations in menarche onset during the past
several millennia (Gillette and Folinsbee 2012; Gluckman and Hanson 2006a, b; Henneberg and Saniotis
2013); a tendency has been reported for later menopause in modern European women, which is probably

12
connected with lifestyle and overall quality of life(Gold 2011). Several hypotheses discuss the fluctuations in
the number of childbirths: alterations in nursing and dietary habits in early agriculturalists might have caused
a shortening of birth intervals (Gluckman and Hanson 2006a, b; Gold 2011; Hewlett and Lamb 2005; Kolata
1974), and subsequent introduction of artificial childbirth reduction (including induced abortions and, later,
contraception) decreased the number of pregnancies. In turn, all these changes affected the number of
menstrual cycles during a woman’s life. Overall, it is expected that changes in the duration of the
reproductive period and in the number of maturing oocytes might affect oocyte meiosis. The decrease in the
number of non-synonymous mutations in the oocyte meiosis pathway during the past six millennia probably
implies that despite all environmental changes, in Europeans there was a tendency to keep the organism's
homeostasis in such an important process as reproduction. The other possibility can be a shift in the mutation
spectrum in order to adapt to the new environmental conditions.
Two more pathways (long-term potentiation and dopaminergic synapse) for which the number of
non-synonymous substitutions in the modern group is significantly less than in the ancient group are
associated with cognitive functions, especially memory and learning. Information is probably the most
rapidly changing factor of our environment. During the past millennia, ways of information presentation and
perception have been completely altered. Six thousand years ago information was being accumulated from a
relatively small geographic area and changed relatively slowly. With the evolution of transport and
transmission techniques information capacity has expanded globally, and the quantity and quality of data to
process have been markedly enlarged. Moreover, the main cognitive tasks in Europe have also dramatically
changed during this time period (e.g., tool-making vs car driving). This presumably affects such information
perception systems as learning capability and memory. However, mutations in cognitive function genes can
lead to detrimental consequences (indeed, mutations in genes in long-term potentiation and dopaminergic
synapse pathways can cause schizophrenia, obsessive-compulsive disorder, Parkinson’s disease, drug
addiction, and many other neurological and neuropsychiatric disorders (Bibb 2005; Centonze, et al. 2005;
Kauer and Malenka 2007)). These deleterious mutations should be eliminated through strong selection, both
directly and indirectly via sexual selection connected with behavioral reactions. Data on molecular evolution
of the human brain are still controversial, but most researchers suggest that coding regions of most human
brain genes are subjects of negative selection (Duret and Mouchiroud 2000; Hill and Walsh 2005; Huang, et
al. 2013; Miyata, et al. 1994; Tuller, et al. 2008). Our results agree with this suggestion; at the same time, the
observed trend can indicate directional changes as a response to the modified cognitive tasks.
In summary, we have revealed selection signatures in functional processes responsible for metabolic
transformations, immune responses including protection against pathogens, alloimmune and autoimmune
reactions, signal transduction, physical activity, sensory perception, reproduction, and cognitive functions.
Interestingly, different environmental factors have induced different types of natural selection. An increase in
the number of non-synonymous mutations in modern humans can indicate signs of either positive or relaxed
selection, whereas a decrease suggests negative or, on the contrary, strong positive selection. For the
identification of the exact type of selection an additional analysis is required.

13
The weakness of our approach is that it is impossible to identify selection signals caused by
modifications in a single gene (like, for example, it was done in the work of Mathieson et al., 2015). Instead,
it is possible to reveal multiple modifications in pathways that are the result of many weak signals
undetectable by using other methods. Therefore, we believe that our results complement existing data on
recent selection in the European population. Based on our results, we suppose that the most important
civilization events that have affected adaptive reactions are changes in diet and the pathogenic environment,
the introduction of xenobiotics, modifications in lifestyle and in the information background. To our
knowledge, our work is the first evidence for natural selection on the functional level. Our results show that
even during a relatively short period of time, the human genome can be significantly shaped by selection if
the selection is induced by man.
Our results raise a number of questions, namely, when did selection began to influence the revealed
processes? How their subsequent evolution was affected? To address these issues, further analyses on
previous (Early/Middle Neolithic, Paleolithic) and intermediate (Iron Age, Middle Ages) time periods should
be performed. We are convinced that, with the emergence of new data, we will better understand how deeply
and how rapidly biochemical and metabolic pathways can be affected by cultural and social changes.

14
Material and Methods
Ancient data preparation
We used published data from 159 European samples dated 3500-1000 BCE (Allentoft, et al. 2015;
Gamba, et al. 2014; Haak, et al. 2015; Mathieson, et al. 2015). The focus of our investigation was the Bronze
Age; however, since the borders between different archaeological cultures and time periods are blurred, we
also used samples attributed to the Late Neolithic (Supplementary Table 1). Selected individuals probably
spoke Indo-European family languages that currently prevail in Europe (Haak, et al. 2015). Most of the Late
Neolithic/Bronze Age individuals have been previously shown to be genetically related to Yamnaya culture
(Allentoft, et al. 2015; Haak, et al. 2015; Lazaridis, et al. 2014) and to most modern European ethnic groups
(Allentoft, et al. 2015; Haak, et al. 2015).
According to the authors (Allentoft, et al. 2015; Haak, et al. 2015), genomic reads successfully
passed quality controls on mitochondrial and bacterial DNA contamination. To ensure authenticity and
remove batch effects, we used a Bayesian approach implemented in mapDamage 2.0 (Jonsson, et al. 2013).
We trimmed the past two nucleotides from each sequence; we further restricted our analyses to sites with
base quality ≥ 20. To achieve statistical significance of the results we implemented the pipeline described in
Figure 2 and briefly outlined below. SNPs were called independently in every sample, filtered by mapping
quality (Q > 30) and SNP quality (QUAL > 20); if possible alleles were supported by the same number of
sequence reads, we selected an allele at random. We set the allele to ‘no call’ if the position was not covered
by sequence reads. Genotypes for samples were called using the ‘call’ command of bcftools (samtools,
bcftools) (Li, et al. 2009) and filtered for quality score (QUAL ≥ 20) and the coverage was required to be at
least three per sample.
We calculated the density and number of non-synonymous SNPs per sample (Supplementary Table
6). We excluded eleven samples from analysis based on 1) absence of genotypes in every position, 2) out-
grouping during PCA analysis shown in the original publication (Supplementary Table 2), or 3) due to
enormous SNPs numbers compared to other samples. For this, we computed the proportions of SNPs in the
samples through all the 305 pathways in the KEGG database. To filter the samples, we calculated the
proportion of SNPs in every sample for every pathway and then acquired the kernel density distribution for
median proportions of SNPs per sample per pathway. The samples which were outside the 99th percentile
were rejected from further analysis. The 99th percentile for the median proportion of SNPs per pathway was
7.0%, while the samples RISE98, RISE00, and RISE423 had average relative proportions of SNPs per
pathway of 7.1, 7.0 and 15.1 % respectively. Therefore, these samples were rejected from further analysis.
The final Bronze Age subset consisted of 150 samples (Supplementary Table 2).
The resulting SNPs were annotated with the ANNOVAR (Wang, et al. 2010) tool using the hg19
human genome annotation and the refGene database
(http://varianttools.sourceforge.net/Annotation/RefGene). Synonymous and non-synonymous SNPs were
pooled into two separate single datasets, resulting in a collection of 40,573 synonymous SNPs and 48,860

15
non-synonymous SNPs, respectively. Next, we calculated the numbers of synonymous and non-synonymous
SNPs per KEGG pathway.
Modern data preparation
Modern data were obtained from the latest release of the “1000 Genomes Project” database
(Genomes Project) (http://www.internationalgenome.org). We selected data only for European populations
with Indo-European roots. Originally, the European subset includes British, Finnish, Spanish, Italians, and
Utah residents with Northern and Western European ancestry. First, we excluded the Utah residents:
although they have European ancestry, the past several centuries they have been living in different
geographical and cultural conditions, having different lifestyle, different diet etc.(Willett, et al. 2006). Next,
we excluded Finnish, since their population history is different from other European populations (Lao, et al.
2008). The Modern dataset contained 305 individuals: 91 from the British population in England and
Scotland, 107 from the Iberian peninsula (Spain) and 107 individuals from Toscani (Italy). SNPs were
functionally annotated with the ANNOVAR tool (Wang, et al. 2010) using the hg19 human genome
annotation and the refGene database.
Depth files correction
Due to poor data sequence coverage, even after aggregation of sequence reads from all the Bronze
Age samples, complete genome coverage had not been achieved. Prior to calculating the distribution of non-
synonymous SNPs in the Bronze Age and modern Europeans, to avoid artificially high enrichment scores,
we restricted our analysis of modern Europeans to genomic positions covered by the Bronze Age sequence
reads. Samtools (Li, et al. 2009) was used for coverage calculation, then the results were filtered to keep
coverage above 3 and mapping quality above 30 (Q > 30). We generated a list of covered bases and used this
list to select those SNPs in the modern human subsets that are covered in the Bronze Age samples. After this
filtering, the modern subsets contained 72,558 synonymous and 96,710 non-synonymous SNPs.
KEGG annotation and preparation for enrichment analysis
Distribution of SNPs in genes
A combined lists of 1) synonymous SNPs and 2) non-synonymous SNPs from the Bronze Age
individuals and present-day Europeans was mapped onto 305 KEGG pathways (Du, et al. 2016), and counts
of SNPs per pathway were computed. To minimize the false-positive rate, we included only pathways
containing more than five genes with SNPs and with sum covered pathway length more or equal to 50% in
aggregated ancient data. (Table 1, Supplementary Table 2).
Enrichment analysis
To analyze differences in numbers of SNPs per pathway between the Bronze Age and present-day
individuals, we calculated 1) differential synonymous SNP enrichment (DSSE) scores and 2) differential

16
non-synonymous SNP enrichment (DNSE) scores. The calculations for DSSE and DNSE were performed in
a same way; below, we describe the calculations for DNSE.
First, we calculated the number of non-synonymous SNPs in both the ancient and modern groups.
We assume that during neutral evolution similar pathways accumulate non-synonymous SNPs at the same
rate, and during enrichment analysis such pathways would fit a normal distribution, while pathways that are
affected by evolutionary pressure would be outliers from this distribution (Supplementary Fig. 4). If K = 305
is the total number of studied pathways, and i = 1,…, K, number the pathways, in Bronze Age and modern
samples, ni and mi denote the number of non-synonymous SNPs per ith pathway. The expected (equilibrium)
fraction of non-synonymous SNPs in ancient data is given by p = n/ (n + m), where p is the fraction of
ancient non-synonymous SNPs in the whole analyzed subset, n is the amount of ancient non-synonymous
SNPs in KEGG pathways, m is the amount of modern non-synonymous SNPs in KEGG pathways. The
fraction pi of ancient non-synonymous SNPs in the ith KEGG pathway is pi = ni/ (ni + mi), where ni is the
amount of ancient non-synonymous SNPs in the ith pathway, mi is amount of modern non-synonymous SNPs
in the ith pathway. From acquired numbers enrichment DNSE scores were computed for every pathway with
continuity correction (Fleiss, et al. 2003):
After computing the DNSE scores (distributed normally, Shapiro-Wilk test p-value > 0.01), we
calculated p-values using Bonferroni and Benjamini-Hochberg corrections and identified the differentially
enriched pathways. The pathways were considered to be differentially enriched if absolute value of the
DNSE score > 4, and the adjusted p-value < 0.01. However, in 2017 in Nature Human Behavior (Benjamin,
et al. 2017) the manuscript “Redefine statistical significance” was published, where it was proposed to
decrease the p-value threshold from 0.01 to 0.005. We implemented the proposed threshold on our data to
avoid further false-positive enrichment signals resulting in alternative lists of enriched pathways.
In order to normalize nonsynonymous SNPs on synonymous SNPs, we performed the following
procedure. Bonferroni-adjusted p-values were log-transformed (base 10) and multiplied by the sign of the
DNSE statistic, so that positive scores correspond to enrichment in modern groups and negative scores to
enrichment in ancient groups, respectively. As a condition of significance, we required the following: the p-
value of the nonsynonymous test was below the p-value of the synonymous test for each pathway. In
addition, it was required for the Bonferroni-corrected p-value to be below 0.01. For each pathway, the p-
value of the synonymous test was above the p-value of the corresponding nonsynonymous test.

17
Validation of the method
To validate our method, we compared it with the method implemented by Somel et al., 2013. We
calculated DNSE scores between chimpanzee and their ancestors (combined genomes from different species
of primates; data from Prado-Martinez et el., 2013, https://www.nature.com/articles/nature12228). Our
results confirmed the conclusions of Somel with colleagues: olfactory transduction pathway demonstrated
the signature of relaxed selection in chimpanzee (enriched in comparison with primates; it is the only
pathway enriched in chimpanzee) (Supplementary Table 8). As in Somel et al., 2013, proteasome pathway
did not demonstrate any signs of selection (no pathway enrichment) (see Somel et al., Figure 2 and our
manuscript, Supplementary Table 8). We also revealed several pathways which are enriched in primates in
comparison to chimpanzee. This can indicate possible negative or strong positive selection in chimpanzee.
However, this suggestion requires additional thorough analysis which is outside the scope of our paper.
Comparison of regulatory SNPs distribution
To compare regulatory SNPs in Bronze Age and modern individuals, we extracted 10,000
experimentally validated promoters and 5'-UTRs from the DBTSS database (https://dbtss.hgc.jp). Sequences
[TSS-1000, TSS+1000] were extracted and MATCH software with TRASNFAC database
(https://www.ncbi.nlm.nih.gov/pubmed/12824369, with parameters set to minimize false positive matches)
was applied to identify putative transcription factor binding sites (TFBS) in those sequences. A total of
61,451,840 putative TFBS were identified in these regions. The TRANSFAC database is highly degenerate
with different entries having the same or similar matrices, therefore producing overlapping predictions on a
genome. Such overlapping putative TBFS were merged into 88,513 contiguous regulatory sequences.
Furthermore, we removed those regions that were not fully covered by ancient DNA sequences, leaving us
with 31,036 regulatory fragments. A proportion test was performed similarly to the calculation of enrichment
scores in coding regions.
PCA and reAdmix
The principal component analysis (PCA) was carried out in R using the ADMIXTURE vectors for
Ancient and European/Worldwide modern individuals. The ADMIXTURE software implements a model-
based Bayesian approach that uses a block-relaxation algorithm to compute a matrix of ancestral population
fractions in each individual (Q files) and infer allele frequencies for each ancestral population (P files)
(Alexander and Lange 2011; Alexander, et al. 2009). We applied ADMIXTURE in unsupervised mode to the
combined dataset of modern and ancient individuals. We varied the number of components between K = 6
and K = 17, recording the value of cross-validation (CV) error and picked K = 7 for the PCA analysis as a
sufficient number of components to distinguish subpopulations from each other.

18
The PCA analysis was performed using the R package princomp with centering and scaling
parameters and then visualized using the first two components cumulatively corresponding to 60% of the
variance among worldwide modern and ancient individuals.
Additional ancient samples for reAdmix (Kozlov, et al. 2015) analyses were obtained from
(Allentoft, et al. 2015; Gamba, et al. 2014; Haak, et al. 2015; Mathieson, et al. 2015). This dataset was
combined with the modern European samples from the 1000 Genomes database. The resulting dataset
contained (1) the Bronze age subset used in this study (n = 150), (2) Early Neolithic data (n = 32), and (3)
Western European hunter-gatherers (n = 12). A reference dataset was assembled from all ancient individuals,
and modern individuals were represented as a linear combination of ancient ones using the reAdmix
algorithm. Each modern population was represented as
Modern Population = w1BA + w2EN + w3WHG + ɛ,
Where BA is “Bronze Age”, EN is “Early Neolithic”, WHG is “Western Europe hunter-gatherers” data and ɛ
is an unassigned part, where coefficients were determined using the differential evolution algorithm. Modern
individuals from 1000 Genomes (British population (n = 6), Toscani (n = 6) and Iberian (n = 5)) were
clustered within self-reported ethnic groups based on similarity of their admixture vectors, and the self-
reported identity was validated using leave-one out procedure and Euclidian distance to the reference
population. Average contributions of ancient genomes to modern individuals were computed for each cluster
of modern individuals.
Data access
SFTP access with ANNOVAR annotated vcf files for ancient man and filtered nonsynonymous and
synonymous files from modern samples:
ip 85.89.112.202
port 2203
username: bronze_man
password: bronze_man
Acknowledgments
We are grateful to Prof. Maciej Henneberg (University of Adelaide, Australia), members of the
Institute of Evolutionary Medicine (University of Zurich, Switzerland), and members of System Biology and
Computational Genetics Seminar (Vavilov Institute of General Genetics, Russia) for valuable comments and
fruitful discussion of the manuscript.
This work was supported by a Mäxi foundation grant (Zurich, Switzerland; awarded to F.R) and by the NSF
Division of Environmental Biology (1456634 to T.T.).

19
References
Adams RM. 1966. The evolution of urban society: early Mesopotamia and prehispanic Mexico. Chicago,:
Aldine Pub. Co.
Adler CJ, Dobney K, Weyrich LS, Kaidonis J, Walker AW, Haak W, Bradshaw CJ, Townsend G, Soltysiak
A, Alt KW, et al. 2013. Sequencing ancient calcified dental plaque shows changes in oral microbiota with
dietary shifts of the Neolithic and Industrial revolutions. Nat Genet 45: 450-455, 455e451. doi:
10.1038/ng.2536
Alexander DH, Lange K 2011. Enhancements to the ADMIXTURE algorithm for individual ancestry
estimation. BMC Bioinformatics 12: 246. doi: 10.1186/1471-2105-12-246
Alexander DH, Novembre J, Lange K 2009. Fast model-based estimation of ancestry in unrelated
individuals. Genome Res 19: 1655-1664. doi: 10.1101/gr.094052.109
Aliesky H, Courtney CL, Rapoport B, McLachlan SM 2013. Thyroid autoantibodies are rare in nonhuman
great apes and hypothyroidism cannot be attributed to thyroid autoimmunity. Endocrinology 154: 4896-
4907. doi: 10.1210/en.2013-1717
Allentoft ME, Sikora M, Sjogren KG, Rasmussen S, Rasmussen M, Stenderup J, Damgaard PB, Schroeder
H, Ahlstrom T, Vinner L, et al. 2015. Population genomics of Bronze Age Eurasia. Nature 522: 167-172.
doi: 10.1038/nature14507
American College of Cardiology Foundation/American Heart Association Task Force on Practice G,
American Association for Thoracic S, American Society of E, American Society of Nuclear C, Heart Failure
Society of A, Heart Rhythm S, Society for Cardiovascular A, Interventions, Society of Thoracic S, Gersh BJ,
et al. 2011. 2011 ACCF/AHA guideline for the diagnosis and treatment of hypertrophic cardiomyopathy:
executive summary: a report of the American College of Cardiology Foundation/American Heart
Association Task Force on Practice Guidelines. J Thorac Cardiovasc Surg 142: 1303-1338. doi:
10.1016/j.jtcvs.2011.10.019
Barnes I, Duda A, Pybus OG, Thomas MG 2011. Ancient urbanization predicts genetic resistance to
tuberculosis. Evolution 65: 842-848. doi: 10.1111/j.1558-5646.2010.01132.x
Barsheshet A, Brenyo A, Moss AJ, Goldenberg I 2011. Genetics of sudden cardiac death. Curr Cardiol Rep
13: 364-376. doi: 10.1007/s11886-011-0209-y
Beja-Pereira A, Luikart G, England PR, Bradley DG, Jann OC, Bertorelle G, Chamberlain AT, Nunes TP,
Metodiev S, Ferrand N, et al. 2003. Gene-culture coevolution between cattle milk protein genes and human
lactase genes. Nat Genet 35: 311-313. doi: 10.1038/ng1263
Benjamin DJ, Berger JO, Johannesson M, Nosek BA, Wagenmakers EJ, Berk R, Bollen KA, Brembs B,
Brown L, Camerer C, et al. 2017. Redefine statistical significance. Nature Human Behaviour: 1. doi:
10.1038/s41562-017-0189-z

20
Bersaglieri T, Sabeti PC, Patterson N, Vanderploeg T, Schaffner SF, Drake JA, Rhodes M, Reich DE,
Hirschhorn JN 2004. Genetic signatures of strong recent positive selection at the lactase gene. Am J Hum
Genet 74: 1111-1120. doi: 10.1086/421051
Bibb JA 2005. Decoding dopamine signaling. Cell 122: 153-155. doi: 10.1016/j.cell.2005.07.011
Blum JS, Wearsch PA, Cresswell P 2013. Pathways of antigen processing. Annu Rev Immunol 31: 443-473.
doi: 10.1146/annurev-immunol-032712-095910
Boyd R, Richerson PJ. 1985. Culture and the evolutionary process. Chicago: University of Chicago Press.
Centonze D, Siracusano A, Calabresi P, Bernardi G 2005. Long-term potentiation and memory processes in
the psychological works of Sigmund Freud and in the formation of neuropsychiatric symptoms.
Neuroscience 130: 559-565. doi: 10.1016/j.neuroscience.2004.09.032
Chen PH, Yao H, Huang LJ 2017. Cytokine Receptor Endocytosis: New Kinase Activity-Dependent and -
Independent Roles of PI3K. Front Endocrinol (Lausanne) 8: 78. doi: 10.3389/fendo.2017.00078
Cirino AL, Ho C. 1993. Hypertrophic Cardiomyopathy Overview. In: Adam MP, Ardinger HH, Pagon RA,
Wallace SE, Bean LJH, Mefford HC, Stephens K, Amemiya A, Ledbetter N, editors. GeneReviews(R).
Seattle (WA).
Cochran G, Harpending H. 2009. The 10,000 year explosion : how civilization accelerated human evolution.
New York: Basic Books.
Cordain L, Eaton SB, Sebastian A, Mann N, Lindeberg S, Watkins BA, O'Keefe JH, Brand-Miller J 2005.
Origins and evolution of the Western diet: health implications for the 21st century. Am J Clin Nutr 81: 341-
354.
Dannemann M, Andres AM, Kelso J 2016. Introgression of Neandertal- and Denisovan-like Haplotypes
Contributes to Adaptive Variation in Human Toll-like Receptors. Am J Hum Genet 98: 22-33. doi:
10.1016/j.ajhg.2015.11.015
De Santis MC, Sala V, Martini M, Ferrero GB, Hirsch E 2017. PI3K Signaling in Tissue Hyper-
Proliferation: From Overgrowth Syndromes to Kidney Cysts. Cancers (Basel) 9. doi:
10.3390/cancers9040030
DeWitte SN 2014. Mortality risk and survival in the aftermath of the medieval Black Death. PLoS One 9:
e96513. doi: 10.1371/journal.pone.0096513
Du J, Li M, Yuan Z, Guo M, Song J, Xie X, Chen Y 2016. A decision analysis model for KEGG pathway
analysis. BMC Bioinformatics 17: 407. doi: 10.1186/s12859-016-1285-1
Duret L, Mouchiroud D 2000. Determinants of substitution rates in mammalian genes: expression pattern
affects selection intensity but not mutation rate. Mol Biol Evol 17: 68-74.
Duronio V 2008. The life of a cell: apoptosis regulation by the PI3K/PKB pathway. Biochem J 415: 333-
344. doi: 10.1042/BJ20081056
Ehrlich PR. 2000. Human natures : genes, cultures, and the human prospect. Washington, D.C.: Island Press
for Shearwater Books.
Enattah NS, Jensen TG, Nielsen M, Lewinski R, Kuokkanen M, Rasinpera H, El-Shanti H, Seo JK,
Alifrangis M, Khalil IF, et al. 2008. Independent introduction of two lactase-persistence alleles into human

21
populations reflects different history of adaptation to milk culture. Am J Hum Genet 82: 57-72. doi:
10.1016/j.ajhg.2007.09.012
Engelman JA, Luo J, Cantley LC 2006. The evolution of phosphatidylinositol 3-kinases as regulators of
growth and metabolism. Nat Rev Genet 7: 606-619. doi: 10.1038/nrg1879
Feldman MW, Laland KN 1996. Gene-culture coevolutionary theory. Trends Ecol Evol 11: 453-457.
Field Y, Boyle EA, Telis N, Gao Z, Gaulton KJ, Golan D, Yengo L, Rocheleau G, Froguel P, McCarthy MI,
et al. 2016. Detection of human adaptation during the past 2000 years. Science 354: 760-764. doi:
10.1126/science.aag0776
Fleiss JL, Levin B, Paik MC. 2003. Statistical methods for rates and proportions. Hoboken, N.J.: J. Wiley.
Forni D, Cagliani R, Tresoldi C, Pozzoli U, De Gioia L, Filippi G, Riva S, Menozzi G, Colleoni M, Biasin
M, et al. 2014. An evolutionary analysis of antigen processing and presentation across different timescales
reveals pervasive selection. PLoS Genet 10: e1004189. doi: 10.1371/journal.pgen.1004189
Fresno Vara JA, Casado E, de Castro J, Cejas P, Belda-Iniesta C, Gonzalez-Baron M 2004. PI3K/Akt
signalling pathway and cancer. Cancer Treat Rev 30: 193-204. doi: 10.1016/j.ctrv.2003.07.007
Fu Q, Posth C, Hajdinjak M, Petr M, Mallick S, Fernandes D, Furtwangler A, Haak W, Meyer M, Mittnik A,
et al. 2016. The genetic history of Ice Age Europe. Nature 534: 200-205. doi: 10.1038/nature17993
Galvani AP, Slatkin M 2003. Evaluating plague and smallpox as historical selective pressures for the CCR5-
Delta 32 HIV-resistance allele. Proc Natl Acad Sci U S A 100: 15276-15279. doi: 10.1073/pnas.2435085100
Gamba C, Jones ER, Teasdale MD, McLaughlin RL, Gonzalez-Fortes G, Mattiangeli V, Domboroczki L,
Kovari I, Pap I, Anders A, et al. 2014. Genome flux and stasis in a five millennium transect of European
prehistory. Nat Commun 5: 5257. doi: 10.1038/ncomms6257
Genomes Project C, Auton A, Brooks LD, Durbin RM, Garrison EP, Kang HM, Korbel JO, Marchini JL,
McCarthy S, McVean GA, et al. 2015. A global reference for human genetic variation. Nature 526: 68-74.
doi: 10.1038/nature15393
Gerbault P, Liebert A, Itan Y, Powell A, Currat M, Burger J, Swallow DM, Thomas MG 2011. Evolution of
lactase persistence: an example of human niche construction. Philos Trans R Soc Lond B Biol Sci 366: 863-
877. doi: 10.1098/rstb.2010.0268
Gilad Y, Bustamante CD, Lancet D, Paabo S 2003a. Natural selection on the olfactory receptor gene family
in humans and chimpanzees. Am J Hum Genet 73: 489-501. doi: 10.1086/378132
Gilad Y, Man O, Paabo S, Lancet D 2003b. Human specific loss of olfactory receptor genes. Proc Natl Acad
Sci U S A 100: 3324-3327. doi: 10.1073/pnas.0535697100
Gillette MT, Folinsbee KE 2012. Early menarche as an alternative reproductive tactic in human females: an
evolutionary approach to reproductive health issues. Evol Psychol 10: 830-841.
Gillings MR, Paulsen IT, Tetu SG 2015. Ecology and Evolution of the Human Microbiota: Fire, Farming
and Antibiotics. Genes (Basel) 6: 841-857. doi: 10.3390/genes6030841
Gluckman PD, Hanson MA 2006a. Changing times: the evolution of puberty. Mol Cell Endocrinol 254-255:
26-31. doi: 10.1016/j.mce.2006.04.005

22
Gluckman PD, Hanson MA 2006b. Evolution, development and timing of puberty. Trends Endocrinol Metab
17: 7-12. doi: 10.1016/j.tem.2005.11.006
Gold EB 2011. The timing of the age at which natural menopause occurs. Obstet Gynecol Clin North Am 38:
425-440. doi: 10.1016/j.ogc.2011.05.002
Grossman SR, Andersen KG, Shlyakhter I, Tabrizi S, Winnicki S, Yen A, Park DJ, Griesemer D, Karlsson
EK, Wong SH, et al. 2013. Identifying recent adaptations in large-scale genomic data. Cell 152: 703-713.
doi: 10.1016/j.cell.2013.01.035
Haak W, Lazaridis I, Patterson N, Rohland N, Mallick S, Llamas B, Brandt G, Nordenfelt S, Harney E,
Stewardson K, et al. 2015. Massive migration from the steppe was a source for Indo-European languages in
Europe. Nature 522: 207-211. doi: 10.1038/nature14317
Hawks J, Wang ET, Cochran GM, Harpending HC, Moyzis RK 2007. Recent acceleration of human
adaptive evolution. Proc Natl Acad Sci U S A 104: 20753-20758. doi: 10.1073/pnas.0707650104
Henneberg M, Saniotis A. 2013. The future mismatch between biological and social development of youths.
Hewlett BS, Lamb ME. 2005. Hunter-gatherer childhoods : evolutionary, developmental, & cultural
perspectives. New Brunswick, N.J.: Aldine Transaction.
Hill RS, Walsh CA 2005. Molecular insights into human brain evolution. Nature 437: 64-67. doi:
10.1038/nature04103
Holden C, Mace R 1997. Phylogenetic analysis of the evolution of lactose digestion in adults. Hum Biol 69:
605-628.
Hou L, Klann E 2004. Activation of the phosphoinositide 3-kinase-Akt-mammalian target of rapamycin
signaling pathway is required for metabotropic glutamate receptor-dependent long-term depression. J
Neurosci 24: 6352-6361. doi: 10.1523/JNEUROSCI.0995-04.2004
Huang Y, Xie C, Ye AY, Li CY, Gao G, Wei L 2013. Recent adaptive events in human brain revealed by
meta-analysis of positively selected genes. PLoS One 8: e61280. doi: 10.1371/journal.pone.0061280
Jonsson H, Ginolhac A, Schubert M, Johnson PL, Orlando L 2013. mapDamage2.0: fast approximate
Bayesian estimates of ancient DNA damage parameters. Bioinformatics 29: 1682-1684. doi:
10.1093/bioinformatics/btt193
Kauer JA, Malenka RC 2007. Synaptic plasticity and addiction. Nat Rev Neurosci 8: 844-858. doi:
10.1038/nrn2234
Kimura M 1955. Stochastic processes and distribution of gene frequencies under natural selection. Cold
Spring Harb Symp Quant Biol 20: 33-53.
Kolata GB 1974. !Kung hunter-gatherers: feminism, diet, and birth control. Science 185: 932-934. doi:
10.1126/science.185.4155.932
Kopp W 2004. Nutrition, evolution and thyroid hormone levels - a link to iodine deficiency disorders? Med
Hypotheses 62: 871-875. doi: 10.1016/j.mehy.2004.02.033
Kozlov K, Chebotarev D, Hassan M, Triska M, Triska P, Flegontov P, Tatarinova TV 2015. Differential
Evolution approach to detect recent admixture. BMC Genomics 16: S9. doi: 10.1186/1471-2164-16-s8-s9

23
Laayouni H, Oosting M, Luisi P, Ioana M, Alonso S, Ricano-Ponce I, Trynka G, Zhernakova A, Plantinga
TS, Cheng SC, et al. 2014. Convergent evolution in European and Rroma populations reveals pressure
exerted by plague on Toll-like receptors. Proc Natl Acad Sci U S A 111: 2668-2673. doi:
10.1073/pnas.1317723111
Lakkis FG, Lechler RI 2013. Origin and biology of the allogeneic response. Cold Spring Harb Perspect Med
3. doi: 10.1101/cshperspect.a014993
Laland KN 2008. Exploring gene-culture interactions: insights from handedness, sexual selection and niche-
construction case studies. Philos Trans R Soc Lond B Biol Sci 363: 3577-3589. doi: 10.1098/rstb.2008.0132
Laland KN, Odling-Smee J, Myles S 2010. How culture shaped the human genome: bringing genetics and
the human sciences together. Nat Rev Genet 11: 137-148. doi: 10.1038/nrg2734
Lang M, Pelkonen O 1999. Metabolism of xenobiotics and chemical carcinogenesis. IARC Sci Publ: 13-22.
Lao O, Lu TT, Nothnagel M, Junge O, Freitag-Wolf S, Caliebe A, Balascakova M, Bertranpetit J, Bindoff
LA, Comas D, et al. 2008. Correlation between genetic and geographic structure in Europe. Curr Biol 18:
1241-1248. doi: 10.1016/j.cub.2008.07.049
Lazaridis I, Patterson N, Mittnik A, Renaud G, Mallick S, Kirsanow K, Sudmant PH, Schraiber JG,
Castellano S, Lipson M, et al. 2014. Ancient human genomes suggest three ancestral populations for present-
day Europeans. Nature 513: 409-413. doi: 10.1038/nature13673
Lewinsky RH, Jensen TG, Moller J, Stensballe A, Olsen J, Troelsen JT 2005. T-13910 DNA variant
associated with lactase persistence interacts with Oct-1 and stimulates lactase promoter activity in vitro. Hum
Mol Genet 14: 3945-3953. doi: 10.1093/hmg/ddi418
Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R, Genome
Project Data Processing S 2009. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25:
2078-2079. doi: 10.1093/bioinformatics/btp352
Libert F, Cochaux P, Beckman G, Samson M, Aksenova M, Cao A, Czeizel A, Claustres M, de la Rua C,
Ferrari M, et al. 1998. The deltaccr5 mutation conferring protection against HIV-1 in Caucasian populations
has a single and recent origin in Northeastern Europe. Hum Mol Genet 7: 399-406.
Lightfoot JT 2013. Why control activity? Evolutionary selection pressures affecting the development of
physical activity genetic and biological regulation. Biomed Res Int 2013: 821678. doi: 10.1155/2013/821678
Marian AJ 2010. Hypertrophic cardiomyopathy: from genetics to treatment. Eur J Clin Invest 40: 360-369.
Mathieson I, Lazaridis I, Rohland N, Mallick S, Patterson N, Roodenberg SA, Harney E, Stewardson K,
Fernandes D, Novak M, et al. 2015. Genome-wide patterns of selection in 230 ancient Eurasians. Nature
528: 499-503. doi: 10.1038/nature16152
Miyata T, Hayashida H 1981. Extraordinarily high evolutionary rate of pseudogenes: evidence for the
presence of selective pressure against changes between synonymous codons. Proc Natl Acad Sci U S A 78:
5739-5743.
Miyata T, Kuma K, Iwabe N, Nikoh N 1994. A possible link between molecular evolution and tissue
evolution demonstrated by tissue specific genes. Jpn J Genet 69: 473-480.

24
Moitra K, Dean M 2011. Evolution of ABC transporters by gene duplication and their role in human disease.
Biol Chem 392: 29-37. doi: 10.1515/BC.2011.006
Motomura K, Brent GA 1998. Mechanisms of thyroid hormone action. Implications for the clinical
manifestation of thyrotoxicosis. Endocrinol Metab Clin North Am 27: 1-23.
Olalde I, Allentoft ME, Sanchez-Quinto F, Santpere G, Chiang CW, DeGiorgio M, Prado-Martinez J,
Rodriguez JA, Rasmussen S, Quilez J, et al. 2014. Derived immune and ancestral pigmentation alleles in a
7,000-year-old Mesolithic European. Nature 507: 225-228. doi: 10.1038/nature12960
Oliveira PA, Colaco A, Chaves R, Guedes-Pinto H, De-La-Cruz PL, Lopes C 2007. Chemical
carcinogenesis. An Acad Bras Cienc 79: 593-616.
Pierron D, Cortes NG, Letellier T, Grossman LI 2013. Current relaxation of selection on the human genome:
tolerance of deleterious mutations on olfactory receptors. Mol Phylogenet Evol 66: 558-564. doi:
10.1016/j.ympev.2012.07.032
Pohl A, Devaux PF, Herrmann A 2005. Function of prokaryotic and eukaryotic ABC proteins in lipid
transport. Biochim Biophys Acta 1733: 29-52. doi: 10.1016/j.bbalip.2004.12.007
Pons-Tostivint E, Thibault B, Guillermet-Guibert J 2017. Targeting PI3K Signaling in Combination Cancer
Therapy. Trends Cancer 3: 454-469. doi: 10.1016/j.trecan.2017.04.002
Porta C, Paglino C, Mosca A 2014. Targeting PI3K/Akt/mTOR Signaling in Cancer. Front Oncol 4: 64. doi:
10.3389/fonc.2014.00064
Quach H, Barreiro LB, Laval G, Zidane N, Patin E, Kidd KK, Kidd JR, Bouchier C, Veuille M, Antoniewski
C, et al. 2009. Signatures of purifying and local positive selection in human miRNAs. Am J Hum Genet 84:
316-327. doi: 10.1016/j.ajhg.2009.01.022
Richerson PJ, Boyd R. 2005. Not by genes alone : how culture transformed human evolution. Chicago:
University of Chicago Press.
Rouquier S, Taviaux S, Trask BJ, Brand-Arpon V, van den Engh G, Demaille J, Giorgi D 1998. Distribution
of olfactory receptor genes in the human genome. Nat Genet 18: 243-250. doi: 10.1038/ng0398-243
Sabeti PC, Varilly P, Fry B, Lohmueller J, Hostetter E, Cotsapas C, Xie X, Byrne EH, McCarroll SA, Gaudet
R, et al. 2007. Genome-wide detection and characterization of positive selection in human populations.
Nature 449: 913-918. doi: 10.1038/nature06250
Sabeti PC, Walsh E, Schaffner SF, Varilly P, Fry B, Hutcheson HB, Cullen M, Mikkelsen TS, Roy J,
Patterson N, et al. 2005. The case for selection at CCR5-Delta32. PLoS Biol 3: e378. doi:
10.1371/journal.pbio.0030378
Somel M, Wilson Sayres MA, Jordan G, Huerta-Sanchez E, Fumagalli M, Ferrer-Admetlla A, Nielsen R
2013. A scan for human-specific relaxation of negative selection reveals unexpected polymorphism in
proteasome genes. Mol Biol Evol 30: 1808-1815. doi: 10.1093/molbev/mst098
Song G, Ouyang G, Bao S 2005. The activation of Akt/PKB signaling pathway and cell survival. J Cell Mol
Med 9: 59-71.
Stefkova J, Poledne R, Hubacek JA 2004. ATP-binding cassette (ABC) transporters in human metabolism
and diseases. Physiol Res 53: 235-243.

25
Stephens JC, Reich DE, Goldstein DB, Shin HD, Smith MW, Carrington M, Winkler C, Huttley GA,
Allikmets R, Schriml L, et al. 1998. Dating the origin of the CCR5-Delta32 AIDS-resistance allele by the
coalescence of haplotypes. Am J Hum Genet 62: 1507-1515. doi: 10.1086/301867
Sui L, Wang J, Li BM 2008. Role of the phosphoinositide 3-kinase-Akt-mammalian target of the rapamycin
signaling pathway in long-term potentiation and trace fear conditioning memory in rat medial prefrontal
cortex. Learn Mem 15: 762-776. doi: 10.1101/lm.1067808
Tang K, Thornton KR, Stoneking M 2007. A new approach for using genome scans to detect recent positive
selection in the human genome. PLoS Biol 5: e171. doi: 10.1371/journal.pbio.0050171
Tuller T, Kupiec M, Ruppin E 2008. Evolutionary rate and gene expression across different brain regions.
Genome Biol 9: R142. doi: 10.1186/gb-2008-9-9-r142
Vanderpump MP 2011. The epidemiology of thyroid disease. Br Med Bull 99: 39-51. doi:
10.1093/bmb/ldr030
Vasiliou V, Vasiliou K, Nebert DW 2009. Human ATP-binding cassette (ABC) transporter family. Hum
Genomics 3: 281-290.
Voight BF, Kudaravalli S, Wen X, Pritchard JK 2006. A map of recent positive selection in the human
genome. PLoS Biol 4: e72. doi: 10.1371/journal.pbio.0040072
Wang K, Li M, Hakonarson H 2010. ANNOVAR: functional annotation of genetic variants from high-
throughput sequencing data. Nucleic Acids Res 38: e164. doi: 10.1093/nar/gkq603
Weichhart T, Saemann MD 2008. The PI3K/Akt/mTOR pathway in innate immune cells: emerging
therapeutic applications. Ann Rheum Dis 67 Suppl 3: iii70-74. doi: 10.1136/ard.2008.098459
Willett K, Jiang R, Lenart E, Spiegelman D, Willett W 2006. Comparison of bioelectrical impedance and
BMI in predicting obesity-related medical conditions. Obesity (Silver Spring) 14: 480-490. doi:
10.1038/oby.2006.63
Williamson SH, Hubisz MJ, Clark AG, Payseur BA, Bustamante CD, Nielsen R 2007. Localizing recent
adaptive evolution in the human genome. PLoS Genet 3: e90. doi: 10.1371/journal.pgen.0030090
Ye Y, Doak TG 2009. A parsimony approach to biological pathway reconstruction/inference for genomes
and metagenomes. PLoS Comput Biol 5: e1000465. doi: 10.1371/journal.pcbi.1000465

26
Table 1. Biological pathways differently enriched in ancient and modern groups
Note. Differential SNP enrichment of KEGG pathways between ancient and modern individuals. Positive
DNSE values correspond to pathways that have more SNPs in genomes of ancient individuals, while
negative DNSE values correspond to pathways that have more SNPs in genomes of modern Europeans.
Pathway
ID
Pathway name
Ancient
SNPs
count
Modern
SNPs
count
DNSE
score
p.value
p.value
adjusted
Bonferroni
Enriched
Bonferroni
0.01
threshold
Enriched
Bonferroni
0.005
threshold
hsa00040
Pentose and glucuronate
interconversions
26
117
-4.29
1.75×10-05
5.05×10-03
Modern
No
hsa00053
Ascorbate and aldarate metabolism
20
99
-4.20
2.69×10-05
7.77×10-03
Modern
No
hsa00982
Drug metabolism - cytochrome P450
84
259
-4.38
1.20×10-05
3.48×10-03
Modern
Modern
hsa01100
Metabolic pathways
2512
4982
-4.69
2.76×10-06
7.98×10-04
Modern
Modern
hsa02010
ABC transporters
209
533
-4.44
8.93×10-06
2.58×10-03
Modern
Modern
hsa04114
Oocyte meiosis
298
337
5.58
2.41×10-08
6.97×10-06
Ancient
Ancient
hsa04151
PI3K-Akt signaling pathway
969
2019
-4.18
2.94×10-05
8.50×10-03
Modern
No
hsa04612
Antigen processing and presentation
232
578
-4.37
1.27×10-05
3.66×10-03
Modern
Modern
hsa04720
Long-term potentiation
202
215
5.13
2.91×10-07
8.41×10-05
Ancient
Ancient
hsa04728
Dopaminergic synapse
294
365
4.45
8.61×10-06
2.49×10-03
Ancient
Ancient
hsa04740
Olfactory transduction
756
1817
-7.09
1.35×10-12
3.89×10-10
Modern
Modern
hsa05204
Chemical carcinogenesis
101
305
-4.62
3.86×10-06
1.12×10-03
Modern
Modern
hsa05320
Autoimmune thyroid disease
181
482
-4.65
3.25×10-06
9.38×10-04
Modern
Modern
hsa05332
Graft-versus-host disease
166
444
-4.50
6.71×10-06
1.94×10-03
Modern
Modern
hsa05410
Hypertrophic cardiomyopathy (HCM)
347
843
-4.95
7.48×10-07
2.16×10-04
Modern
Modern

Figure 1. Location of ancient samples analyzed in the study. Data from (Gamba et al. 2014; Allentoft
et al. 2015; Haak et al. 2015; Mathieson et al. 2015) (for details, see Supp. Table 1 and 2).

Figure 2. Principal pipeline of the study. Additional parameters and tool versions are listed in
Methods.

Figure 3. Principal component analysis. A: world groups; B: European groups.

Figure 4. Proportion of ancient genomes in modern European individuals.

Figure 5. Differential SNP enrichment of KEGG pathways between ancient and modern individuals.
Positive DNSE values correspond to pathways that have more SNPs in genomes of ancient individuals, while
negative DNSE values correspond to pathways that have more SNPs in genomes of modern Europeans.

Figure 6. Relationship between the p-values for synonymous and non-synonymous SNPs in the
studied pathways.
1 – Oocyte meiosis
2 – Long-term potentiation
3 – Dopaminergic synapse
4 – Pentose and glucuronate interconversions
5 – Drug metabolism - cytochrome P450
6 – Chemical carcinogenesis
7 – ABC transporters
8 – Antigen processing and presentation
9 – Graft-versus-host disease
10 – PI3K-Akt signalling pathway
11 – Ascorbate and aldarate metabolism
12 – Metabolic pathways
13 – Autoimmune thyroid disease
14 – Hypertrophic cardiomyopathy (HCM)
15 – Olfactory transduction