First insights into the metagenome of Egyptian mummies using next-generation sequencing
- First Online:
- Cite this article as:
- Khairat, R., Ball, M., Chang, C.H. et al. J Appl Genetics (2013) 54: 309. doi:10.1007/s13353-013-0145-1
- 1.3k Views
We applied, for the first time, next-generation sequencing (NGS) technology on Egyptian mummies. Seven NGS datasets obtained from five randomly selected Third Intermediate to Graeco-Roman Egyptian mummies (806 BC–124AD) and two unearthed pre-contact Bolivian lowland skeletons were generated and characterised. The datasets were contrasted to three recently published NGS datasets obtained from cold-climate regions, i.e. the Saqqaq, the Denisova hominid and the Alpine Iceman. Analysis was done using one million reads of each newly generated or published dataset. Blastn and megablast results were analysed using MEGAN software. Distinct NGS results were replicated by specific and sensitive polymerase chain reaction (PCR) protocols in ancient DNA dedicated laboratories. Here, we provide unambiguous identification of authentic DNA in Egyptian mummies. The NGS datasets showed variable contents of endogenous DNA harboured in tissues. Three of five mummies displayed a human DNA proportion comparable to the human read count of the Saqqaq permafrost-preserved specimen. Furthermore, a metagenomic signature unique to mummies was displayed. By applying a “bacterial fingerprint”, discrimination among mummies and other remains from warm areas outside Egypt was possible. Due to the absence of an adequate environment monitoring, a bacterial bloom was identified when analysing different biopsies from the same mummies taken after a lapse of time of 1.5 years. Plant kingdom representation in all mummy datasets was unique and could be partially associated with their use in embalming materials. Finally, NGS data showed the presence of Plasmodium falciparum and Toxoplasma gondii DNA sequences, indicating malaria and toxoplasmosis in these mummies. We demonstrate that endogenous ancient DNA can be extracted from mummies and serve as a proper template for the NGS technique, thus, opening new pathways of investigation for future genome sequencing of ancient Egyptian individuals.
KeywordsAncient DNADNA survivalEgyptian mummiesEmbalming materialMEGANMetagenomicsNext-generation sequencingPreservationTemperature
Ancient Egyptians believed in the afterlife. Death marked a transformation from the corporeal transitory life on earth into a spiritual permanent life in the afterworld. In their beliefs, the physical body had to be preserved and its integrity was crucial to continue the existence in the hereafter (Ikram 2003). The will to protect and preserve the integrity of the physical body into a more perfect and eternal form is probably the key point which gave rise to artificial mummification (David 1997; Wisseman 2001; Ikram 2003; Lynnerup 2007).
During the course of the Egyptian history, the embalming priests developed sophisticated methods of artificial mummification and, for many people, the word ‘mummy’ immediately calls forth images of corpses entombed in sarcophagi.
Until recently, it was assumed that mummification started during the Dynastic period. It was supposed that the idea of artificial mummification was given to the Egyptians by observing the sand naturally desiccated and perfectly preserved bodies dating back to the Predynastic period (5000–3000 BC) (Ikram 2003).
However, excavations carried out at the sites of Hieraconpolis and Adaima in the south of Egypt have shown that, during the Naqada II Culture (3500–3150 BC), the first attempts of artificial preservation of the bodies through the use of resins and bandages were performed (Ikram 2003).
These discoveries forced egyptologists and scientists to re-consider their ideas on the origin of artificial mummification, which now appears to have developed as early as the Gerzean Culture (Naqada II) and lasted until the Christian Era, with several diversifications in the mummification methods throughout time (Ikram 2003).
Since mummification usually results in excellent human and animal soft tissue preservation (Pääbo 1985a; David 1997; Lynnerup 2007; David 2008; Corthals et al. 2012), molecular studies on Egyptian mummies started during the 1980s of the last century (Pääbo 1985b) and were followed by further reports with different focus and research interests (Nerlich et al. 1997; Zink et al. 2000; Zink and Nerlich 2003; Zink et al. 2006; Nerlich et al. 2008; Nerlich and Lösch 2009; Zweifel et al. 2009; Woide et al. 2010; Donoghue et al. 2010; Hawass et al. 2010; Hekkala et al. 2011; Hawass et al. 2012; Kurushima et al. 2012). In the last decade, the development of minimally invasive techniques allowed to gain deeper biological knowledge of mummies without causing major damage.
The challenge to reveal more genetic information from ancient tissues without performing massive sampling has catalysed the application of next-generation sequencing (NGS) technologies to archaeological specimens (Pusch et al. 2000; Wisseman 2001). The generation of large volumes of sequence data is the primary advantage over conventional methods (Lambert and Millar 2006; Metzker 2010; Pareek et al. 2011). Recent scientific discoveries that resulted from the application of NGS highlight the striking impact of these massively parallel platforms on genetics (Mardis 2008). These new sequencing methods are particularly suited to ancient DNA analysis because the generated sequence fragments are up to 400 bp in length, a size comparable to that found in most degraded ancient genomes (Green et al. 2006; Poinar et al. 2006; Rasmussen et al. 2010; Reich et al. 2010; Keller et al. 2012).
Egyptian mummies have never been subjected to NGS until now. Consequently, our primary aim was not to determine entire genomes here. Instead, our goal was to determine the degree of information that can be gained from mummified tissues when they are analysed by NGS technologies.
Therefore, five Egyptian mummies and two unearthed pre-contact Bolivian lowlands skeletons were selected and characterised. Subsequently, the datasets from those people who lived in warm climates were contrasted to three recently published NGS datasets obtained from cold-climate region specimens, i.e. the Saqqaq, the Denisova hominid and the Alpine Iceman (Rasmussen et al. 2010; Reich et al. 2010; Keller et al. 2012). The genomes of these three individuals have been established by two robust methods, the Solexa and the SOLiD NGS technologies (Liu et al. 2012).
Here, we present the data from a first characterisation of Egyptian mummies using these new DNA sequencing technologies and demonstrate that the DNA from mummies can serve as a proper template for the NGS method. Through an adequate number of runs, entire genome sequencing of ancient Egyptian individuals is likely to become standard in the not too distant future.
Materials and methods
Characterisation of samples under investigation, comprising nine samples from warm environments and three from cold environments. Abbreviations used: AD Anno Domini, BC Before Christ, BP before present, ND not determined, < cold-climate sample, > warm-climate sample
Average temperature 15 °C
Mummified tissue (muscle)
Third Intermediate Period
Mummified tissue (muscle)
Hellenistic Period/Ptolemaic Period
Mummified tissue (muscle)
Mummified tissue (muscle)
Late Period/Hellenistic-Ptolemaic Period
Mummified tissue (muscle)
Mummified tissue (muscle)
Third Intermediate to Roman Period
Mummified tissue (muscle)
Third Intermediate to Roman Period
Mineral soil burial, South America
Mineral soil burial, South America
ca. 717 AD
Cryotic soil, Western Greenland
4044 ± 31 BP
Rasmussen et al. (2010)
Denisova SL3003 human
Cave soil, Russia
Earlier Middle Palaeolithic
Reich et al. (2010)
ca. 5300 BP
Old Copper age
Keller et al. (2012)
In order to differentiate by climate between all samples used in this study, we applied the term “warm-climate samples” to the Egyptian mummies and the Bolivian skeletons. The three remaining datasets taken from the literature were defined as “cold-climate samples” and shall contrast the Alpine Iceman, the Saqqaq Palaeo-Eskimo and the Denisova hominid from the aforementioned individuals. Assessing the effects of further parameters, e.g. humidity, pH, salt concentration etc., is not a part of the present study and will be the topic of future research.
DNA extraction and contamination monitoring
We adopted the previously published criteria for ancient DNA authentication (Richards et al. 1995; Roberts and Ingham 2008; Hawass et al. 2010; Keller et al. 2012; Hawass et al. 2012). DNA extraction work was conducted in a dedicated facility, physically isolated from the polymerase chain reaction (PCR) technology, the library preparation steps and the areas for the post-PCR work. Work surfaces were frequently cleaned with DNase and irradiated with UV light; disposable plastic items were used whenever possible, non-disposal items were baked at 200 °C, washed with DNase and irradiated with UV light. If solutions were bought in pre-made, these were finally prepared on a clean bench, autoclaved and sterilised by filtration (0.25-μm syringe filter; Nalgene, Thermo Fisher Scientific, Waltham, MA, USA). All work was conducted while wearing appropriate protective garments. Contamination was monitored by the use of negative and blank extraction controls, which were processed along with each sample. Mitochondrial DNA typing of all laboratory working members was performed.
Sampling of tissue, DNA extraction and DNA purification were performed according to a protocol published previously (Scholz and Pusch 1997). A slightly modified version of the aforementioned extraction protocol was also applied. Here, the DNA was extracted from the DNA-containing phase according to the protocol of Scholz and Pusch (1997), but instead of subsequent manual purification steps, we applied the MagNA Pure Compact System (Roche Applied Science, Penzberg, Germany) for automated purification. The DNA samples were also examined by spiking reactions to test the effect of inhibition due to the presence of co-extracted substances (Pusch and Bachmann 2004).
PCR analysis, cloning and Sanger sequencing
Mitochondrial PCR amplification was accomplished using specific primers designed from the D-loop control region of the Canis lupus familiaris mitochondrial genome. The primers used were forward 5-TGCATACAATACTCACAAGCTTTATTT-3 and reverse 5-GACTACGAGACCAAATGCGTGT-3, and amplified a DNA segment between positions 16,572 and 16,672.
Specific primers were designed to detect the gene sequence of the mitochondrial NADH dehydrogenase subunit 1 (nad1) (positions 496–603) of the Pinaceae species. The primer sequences were as follows: forward 5-ATGTCGGTCGACGATGCCGC-3 and reverse 5-AGGTGCCCAGCGATTCCTTCA-3.
All PCR fragments were amplified in a volume of 25 μl containing 1× FastStart PCR Master Mix (Roche Applied Science, Penzberg, Germany), 20 pmol of each primer, 10 mM of each dNTP and aliquots of the extracted DNA. The cycling conditions using a GeneAmp® PCR System 9700 Thermal Cycler (Applied Biosystems, Foster City, CA, USA) were 94 °C for 5 min, followed by 45 cycles of 94 °C for 30 s, 60 °C for 30 s and 72 °C for 30 s, followed by a final extension of 10 min at 72 °C.
Alternatively, the same protocol for thermal cycling but with an annealing temperature of 62 °C was applied to amplify the DNA of the animal mummy, an annealing temperature of 58 °C was applied to amplify the Pinaceae species and a temperature of 57 °C was used to amplify cloned DNA fragments by colony PCR.
The cloning of PCR products was performed with the CloneJET™ PCR Cloning Kit (Fermentas, Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer’s protocol. The colony PCR products were cleaned by ExoSAP-IT (USB Corporation, Cleveland, OH, USA) and were used for the Sanger cycle-sequencing reactions with BigDye Terminator v3.1 chemistry (Applied Biosystems, Foster City, CA, USA). Samples were run on a 3130 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA).
Both methods, the Solexa (Rasmussen et al. 2010; Reich et al. 2010) and the SOLiD technologies (Keller et al. 2012), are considered to be robust in their application to ancient DNA. The Roche 454 protocol was not employed by us, since it is better suited for nucleic acids with larger fragmentation sizes.
Whole genome libraries were generated for the SOLiD 3 Plus System (Applied Biosystems, Foster City, CA, USA). Genomic DNA was end-repaired using 1 μl of the end-polishing enzyme 1 (10 U/μl) and 2 μl of the end-polishing enzyme 2 (10 U/μl), 4 μl dNTP mix (10 mM), as well as 20 μl 5× end-polishing buffer, in a total volume of 100 μl. Following incubation at room temperature for 30 min, DNA was purified using the SOLiD™ Library Column Purification Kit (Applied Biosystems, Foster City, CA, USA). SOLiD™ adaptors P1 (5-CCACTACGCCTCCGCTTTCCTCTCTATGGGCAGTCGGTGAT-3) and P2 (5-AGAGAATGAGGAACCCGGGGCAGTT-3), each at 2.5 μM, were ligated to purified DNA with 40 μl of 5× T4 ligase buffer and 10 μl T4 ligase (5 U/μl), in a total volume of 200 μl at room temperature for 15 min. After another purification step, the ligated DNA was eluted in 40 μl nuclease-free water. No additional size selection was carried out in order to avoid loss of material. DNA was then incubated with 380 μl Platinum HiFi PCR Amplification Mix (Life Technologies, Carlsbad, CA, USA) and 10 μl each of both library PCR primers (primer 1: 5-CCACTACGCCTCCGCTTTCCTCTCTATG-3 and primer 2: 5-CTGCCCCGGGTTCCTCATTCT-3) in order to repair the gap in the double-stranded DNA molecules introduced during adaptor ligation. The following conditions for thermal cycling were applied: hold at 72 °C for 20 min plus another hold at 95 °C for 5 min, followed by two cycles at 95 °C for 15 s, 62 °C for 15 s, 70 °C for 1 min, one cycle at 70 °C for 5 min and a final hold step at 4 °C. The PCR product was purified using the PureLink™ PCR Purification Kit (Invitrogen, Carlsbad, CA, USA). Eluted DNA was again cycled using 100 μl of 2× Phusion HF Master Mix (Finnzymes, Thermo Fisher Scientific, Waltham, MA, USA) and 8 μl each of both library PCR primers 1 and 2 in a total volume of 200 μl. This mixture was divided into four PCR tubes and cycled using the following conditions: 12 cycles at 95 °C for 15 s, 62 °C for 15 s, 70 °C for 1 min, followed by one hold step at 70 °C for 5 min. All libraries were purified as described above and stored at −20 °C until sequencing using the SOLiD™ 3 Plus System (Applied Biosystems, Foster City, CA, USA).
Whole genome libraries were also applied to the Genome Analyzer IIx (GAIIx) (Illumina, San Diego, CA, USA). The recommended protocol of the New England Biolabs NEBNext™ DNA sample prep kit was used (New England Biolabs GmbH, Ipswich, MA, USA). About 50 ng/μl of total DNA was end-repaired using 1 μl of DNA Poly I (Klenow LF) and 5 μl T4 DNA polymerase, 4 μl dNTP mix (10 mM) and 10 μl of 10× end-polishing buffer, in a total volume of 100 μl. Following incubation at 20 °C for 30 min, the DNA was purified using Agencourt AMPure XP (Beckman Coulter Genomics, Brea, CA, USA), A-tailed and ligated to Solexa adaptors with the sequences 5-GATCGGAAGAGCTCGTATGCCGTCTTCTGCTTG-3 and 5-ACACTCTTTCCCTACACGACGCTCTTCCGATCT-3. In a total volume of 50 μl, these adaptors were ligated to purified DNA using 25 μl of 2× T4 ligase buffer and 5 μl Quick T4 ligase by incubation at room temperature for 15 min. After another round of purification using Agencourt AMPure XP (Beckman Coulter Genomics, Brea, CA, USA), the ligated DNA was eluted in 10 μl nuclease-free water. Eluted DNA was cycled using 25 μl of 2× Phusion HF Master Mix (Finnzymes, Thermo Fisher Scientific, Waltham, MA, USA) and 1 μl each of both library PCR primers 1 and 2 in a total volume of 50 μl. This mixture was cycled using the following protocol: 12–18 cycles at 95 °C for 30 s, 65 °C for 1 min, 70 °C for 30 s, followed by 5 min at 70 °C and a final hold at 4 °C. PCR products were again purified using Agencourt AMPure XP (Beckman Coulter Genomics, Brea, CA, USA) and applied to the NGS sequencer model Genome Analyzer IIx.
Furthermore, for a detailed metagenomic analysis of two mummy DNAs which had been thoroughly pre-characterised by standard PCRs, they were pooled and additionally used for library preparation according to the recommended Illumina protocol and subjected to sequencing using the Solexa platform (Illumina, San Diego, CA, USA) (Table 1).
Starting with the 2-ng/μl dilution of the fragment library preparation, a 60-pg/μl dilution was applied to the emulsion PCR to perform single-molecule amplification of the SOLiD 3 libraries using the EZBead™ system (Applied Biosystems, Foster City, CA, USA). After a 3′-end modification of the DNA bound to magnetic beads, the templated beads were deposited onto chemically modified slides and loaded into the flow cells of a SOLiD™ 3 Plus system (Applied Biosystems, Foster City, CA, USA). Fifty base pairs of DNA fragments were sequenced using the SOLiD™ TOP Sequencing Kit (Applied Biosystems, Foster City, CA, USA).
According to the Illumina protocol, Solexa libraries were diluted to a final concentration of 6 pM to be ready for cluster amplification using the TruSeq PE Cluster Kit v.2. Subsequently, the Solexa samples were sequenced according to the manufacturer’s recommendations using the TruSeq SBS Kit v.5 GA on a Genome Analyzer model GAIIx (Illumina, San Diego, CA, USA). The read length using this NGS system was 75 bp/100 bp paired end.
NGS data and metagenomic analyses
NGS reads obtained from the sequencing were mapped against the human genome assembly hg19 with Bowtie software (Langmead et al. 2009) or the Burrows–Wheeler Aligner (BWA) (Li and Durbin 2009) on the Galaxy server. Visualisation of the mitochondrial output file was accomplished with the output file generated by the BWA software. The BAM files obtained after the mapping steps were processed and handled using the Integrative Genomics Viewer (IGV).
Intraspecies comparison was done using one million reads from each of our datasets and from the three previously published datasets specific for cold environments (Saqqaq: Rasmussen et al. 2010; Denisova: Reich et al. 2010; Alpine Iceman: Keller et al. 2012).
Blast alignments were accomplished using either the Blastn or megablast algorithms against the NCBI nucleotide collection with seed length 33. Mapping and Blast analysis was done on the Galaxy main server (http://main.g2.bx.psu.edu/root), the Tübingen Galaxy server (https://galaxy.informatik.uni-tuebingen.de/galaxy-local/), the Vienna server at canis.csb.univie.ac.at or felis.csb.univie.ac.at, and the public GALAXY@WUR server at http://galaxy.wur.nl:8080/galaxy/root. Blast results were analysed using MEGAN software (Huson et al. 2007).
Two closely related protocols for DNA extraction were applied to seven tissue biopsies from five randomly selected Egyptian human mummies, an Egyptian dog mummy sample and two bone biopsies from pre-contact South-American skeletons (Table 1).
Since it was expected that the metagenomes of Egyptian mummies also contain plant DNA originating from the embalming “recipes”, a further line of experiments was conducted for the authentication of results.
Following the purification steps, the libraries were cloned into Fermentas CloneJET™ PCR vectors (Fig. 4b) and the obtained clones were sequenced with the Sanger method. The analysis of the sequences showed the proper composition of NGS amplicons, with the presence of the two library adaptors and a short cloned fragment in between the two. The size range of the small-scale plasmid library was mainly between 200 bp and 400 bp (Fig. 4b).
A first analysis of the plant/herbals content in our mummies indicated that Ricinus communis was one of the main components found in mummies 1 and 4. Populus spec. sequences were most frequent in mummy 2, while Pinus spec. is highlighted in the pooled mummy sample DMG56. Linum, Olea, Prunus, Abies, Allium and Lotus sequences were identified to a lesser extent in a number of mummies. Although mummy 2 (382–234 BC) and mummy 4 (358–204 BC) originate from similar times, we note differences in their resin composition with regard to the Populus spec. sequences. While mummy 2 showed a total of 1,074 Populus DNA reads, there is not a single read count in mummy 4 for the taxon Populus. None of these plant DNAs could be identified in the metagenomes of the three cold-climate specimens.
Rasmussen et al. (2010)
Reich et al. (2010)
Keller et al. (2012)
The archea were even less represented in all samples from both cold and warm climates (0–0.051 %), except for the South-American skeletons, which showed an approximately 10-fold increased proportion, both about 0.4 % (Fig. 6).
The eukaryotic reads were abundant in the cold-climate samples (Saqqaq: 99.7 %, Iceman: 98.7 % and Denisova: 89.6 %), whereas they strongly varied in our mummy datasets, ranging between 6.9 % and 96 % (Table 2, Fig. 6). According to the percentage of eukaryotic reads, values were widely scattered from higher percentages in Egyptian mummy samples 1a, 2a and 3 (80–96 %) to very low percentages (6.86 % in mummy sample 4).
More differences could be pinpointed when analysing different biopsies taken from the same mummy; this concept is exemplified at its best in sample 1a and 1b taken from mummy 1 and in samples 2a and 2b taken from mummy 2 (Table 2, Fig. 6).
It is highly likely that the discrepancy seen in the bacterial content between the first (sample 1a from mummy 1 and 2a from mummy 2) and second samplings (sample 1b from mummy 1 and 2b from mummy 2) is consistent with a bacterial bloom. This was mainly due to the absence of an adequate environment monitoring in the new repository of the Institute of Pre- and Protohistory where the heads have been recently relocated.
The ratio of assigned reads to the total number of reads spans a wide range within the datasets of Egyptian mummies. It is moderately high in, for example, mummy samples 2b, 4 and 5 (10.8 %, 12.5 % and 13 %, respectively), but resembles the ratios observed in the Denisova (26.3 %) and the Iceman (10.8 %) cold-climate samples (Table 2).
The relative proportion of the Actinobacteria is the main difference highlighted in the “bacterial fingerprint”. On this basis, two major groups of samples could be defined. One group is represented by the Egyptian mummies and the glacier mummy Ötzi the Iceman, which display a relative amount of ≤10 % Actinobacteria in the group of bacteria.
The second group is the so-called “non-mummy group”, which includes the two South-American skeletons, the Denisova hominid and the Saqqaq Paleo-Eskimo (Fig. 7). Here, the relative amount of Actinobacteria is largely increased (≥28 %). Moreover, within the mummy group, an increased percentage of Firmicutes can also be noted.
Conversely, Proteobacteria show a very divergent range through all the samples; they span from moderately low in the Iceman mummy (5.94 %) to high in the Egyptian mummy sample 4 (78.44 %). This implies that Proteobacteria cannot be used as a fingerprint for the unequivocal identification of mummies.
Despite a general variability, the Egyptian mummies and the Iceman show a more homogenous metagenomic picture and can be easily contrasted both to the warm-climate lowland South-American samples and to the cold-climate Saqqaq and Denisova specimens.
Actinobacteria are very common in soil where they decompose organic materials. Firmicutes often build endospores and can survive in extreme conditions for a long time. Our data show that no correlation between the environmental temperature and the “bacterial fingerprint” of a given specimen exists.
Comparisons among the three previously published datasets of the Iceman, Denisova hominid and the Saqqaq (Rasmussen et al. 2010; Reich et al. 2010; Keller et al. 2012) and the new datasets from the Egyptian mummies were performed using the Blast results of a subset of one million reads from each dataset. This was in order to obtain the percentage of reads assigned to human DNA in comparison to the total number of assigned reads.
The Ötzi glacier mummy resides in position number one on this ranking (81 % human content), followed by the Denisova hominid (70 %) and the Saqqaq (68 %). Egyptian mummy samples 1a, 2a and 3 showed a similarly high human content ranging from 54 % to 64 %, thus, resembling the value obtained for the Saqqaq permafrost-preserved hair. Less abundant human DNA is observed in the remaining six samples represented by the South-American skeletons and the remaining four Egyptian mummy biopsies.
The coverage of HVR2 helps to detect a number of diagnostic mutations, which are considered to be authentic, since all of these single-nucleotide polymorphisms (SNPs) are absent in the mitochondrial DNA sequences of our laboratory members. Diagnostic base deviations with a good segment coverage point to the mitochondrial haplogroup I2, but further analysis is required in order to consolidate this tentative result.
Mummification reached its apex in the 21st Dynasty (1064–940 BC), when the embalmers started manipulating the flesh and “turned the prepared body into a more perfect image of itself” (Ikram 2003). The embalming priests used sophisticated “recipes” to prepare the bodies for the afterworld (Buckley and Evershed 2001; Wisseman 2001; Buckley et al. 2004).
However, apart from rare exceptions in which some steps of the mummification procedure are reproduced (i.e. the coffin of Djedbastiuefankh, Pelizaeus Museum, Hildesheim—Late Period; the Rhind Magical Papyrus—ca. 200 BC; three papyri in the Cairo, Durham Oriental and Louvre Museums—around 1st century AD), “the Egyptians are curiously silent about the modes of mummification in both their written and figurative sources” (Ikram 2003).
The lack of direct evidence was filled with information derived from numerous written sources. The Ionian Greek writer Herodotus (5th century AD) provided the earliest written accounts on mummification (Book II of “The History”). Coupled and augmented with the records of Diodorus Siculus (1st century BC) and further completed by the writings of Porphyry (3rd century AD), these sources have long provided the basis of current knowledge on mummification techniques (Wisseman 2001; Ikram 2003).
The methods of mummification, defined as the deliberate act of preservation of the body after death, varied and diversified during the different periods of the Egyptian history (Wisseman 2001; Ikram 2003; David 2008; Jeziorska 2008).
The increasingly sophisticated biogeochemical and molecular techniques, as well as experimental mummification, allowed scientists to gain further information about how mummification was practised (Pääbo 1985a; Brier and Wade 1995; Brier and Wade 1997; Zimmerman et al. 1998; Barraco et al. 1977; Brier and Wade 1997; Wisseman 2001; Buckley and Evershed 2001; Aufderheide 2003; Kaup et al. 2003; Ikram 2003; Ikram 2005; Buckley et al. 2004; Metcalfe and Freemont 2012).
Due to mummies’ excellent tissue preservation, studies aiming to target both DNA and proteins harboured in mummified tissues were successfully carried out (Pääbo 1985b; Nerlich et al. 1997; Zink et al. 2000; Zink et al. 2001; Zink and Nerlich 2003; Zink et al. 2006; Kaup et al. 2003; Koller et al. 2003; Bianucci et al. 2008; Nerlich et al. 2008; Woide et al. 2010; Hawass et al. 2010; Donoghue et al. 2010; Hekkala et al. 2011; Corthals et al. 2012; Hawass et al. 2012; Kurushima et al. 2012).
The general feasibility of PCR-based DNA studies using different ancient samples originating from warm climates was evidenced (Lassen et al. 1994; Poinar et al. 1996; Fox 1997; Pusch et al. 2003; Poinar et al. 2003; Zink and Nerlich 2003; Zink and Nerlich 2005; Gilbert et al. 2008; Campos et al. 2012; Kurushima et al. 2012). Similarly, ancient human DNA from Egyptian mummies was retrieved with conventional molecular genetic methodology (e.g. Pääbo 1985b; Krings et al. 1999; Rutherford 2008; Hawass et al. 2010; Hawass et al. 2012).
We have tested, for the first time, the feasibility and fidelity of the NGS methods when applied to mummified Egyptian tissues. Since there is a common fear for contaminant modern human DNA in ancient specimens, a thorough contamination monitoring and authentication procedure was applied.
All chemicals and consumables were controlled as recommended by the ancient DNA guidelines (Richards et al. 1995; Roberts and Ingham 2008; Hawass et al. 2010; Keller et al. 2012; Hawass et al. 2012). Extraction blanks and negative controls were used along with the samples for PCRs and library preparation, which showed negative results.
In order to monitor for authentic DNA sequences in Egyptian mummified tissues, a muscle biopsy from an embalmed Egyptian dog was also tested by PCR. The results revealed the presence of authentic Canis lupus ancient DNA and the absence of human contaminant DNA.
Furthermore, we determined the presence of herbal DNA from plants, some of which might have been components of the embalming “recipes”. The DNA results for Pinus, Picea and Ricinus were confirmed by standard PCRs with subsequent Sanger sequencing and served as a further proof of authenticity (Pinus confirmation is shown in Fig. 3).
The positive results for authentic dog and herbal/plant DNAs are much less susceptible to contamination than human DNA and encouraged us to proceed with the library creation employing ancient Egyptian tissue of human mummies dating from 806 BC to 124 AD.
A first survey of the NGS-generated metagenomes of the randomly selected Egyptian mummies indicated further proof of authenticity by observing the signatures of two protozoan pathogens, namely, Plasmodium falciparum and Toxoplasma gondii (Fig. 5a).
The pathogen representation was reproducible among different runs originating from the same mummy and is specific to the warm-climate specimens. Furthermore, the presence of Plasmodium pathogens was confirmed by the highly specific PCR technology coupled with subsequent Sanger sequencing.
Four blood-parasite species belonging to the genus Plasmodium (P. falciparum, P. vivax, P. malariae, P. ovale) are responsible for different forms of human malaria. Parasites are transmitted to humans through the bite of female Anopheles mosquitoes. Plasmodium falciparum causes the most dangerous and severe form of malaria, termed malaria tropica. Previous reports have shown the presence of falciparum malaria in mummies from ancient Egypt (Nerlich et al. 2008; Bianucci et al. 2008; Hawass et al. 2010).
Toxoplasma gondii is the other protozoan species that we identified. The definitive host of Toxoplasma gondii is the cat. Cats were cult animals in ancient Egypt and an abundance of cats was mummified during the Late and Graeco-Roman periods (Ikram 2005).
Recent genetic evidence supports the notion that ancient Egyptians used domesticated cats, Felis silvestris catus, for votive mummies and imply that taming of the cats occurred prior to or during Predynastic and Early Dynastic Periods (Kurushima et al. 2012). Hence, frequent contacts among domesticated cats and humans can explain the presence of this specific pathogen in Late and Graeco-Roman human mummies.
In addition to the human, bacteria, viruses and fungi DNAs, a fair amount of plant DNA in the metagenomes of our Egyptian mummies was identified.
Herodotus’ accounts mention myrrh, cassia, palm wine, “cedar oil” and “gum” as the main plant components of the embalming “recipes”. Previous research (Buckley and Evershed 2001) showed that, even if salt natron was widely used as a desiccant, due to the warm environmental conditions inside the tombs, the bodies would have decomposed without the application of specific organic substances.
Chemical analyses carried out on 13 Egyptian mummies dating from the mid-Dynastic period (ca. 1900 years BC) to the late Roman Period (AD 395) suggested that unsaturated plant oils and animal fats were the key components in mummification. Subsequently, more exotic substances were mixed up to this base and applied either on the bodies or on the bandages (Buckley and Evershed 2001). The peculiar properties of the unsaturated oils and fats allowed them to polymerise spontaneously. The polymerisation would have, in turn, produced a “highly cross-linked aliphatic network, which would have stabilized otherwise fragile tissues and/or wrappings against degradation by producing a physico-chemical barrier that impedes the activities of microorganisms” (Buckley and Evershed 2001).
Buckley and Evershed (2001) showed that beeswax and coniferous resins were used in the embalming procedures and that their use increased in importance over time, being more enhanced in later periods. Those components were found both on the bodies and on the wrappings. Components diagnostic for Pistacia resin were also found in a Ptolemaic female mummy.
In a subsequent study, some components diagnostic for Pistacia exudates were also chemically characterised in three further but undated mummies (Nicholson et al. 2011).
Some of the distinctive plant taxa seen in the mummy MEGAN output coincide with certain ingredients reported to have been used in the “embalming” recipes (Serpico and White 2000).
Castor (Ricinus communis), linseed (Linum usitatissimum), olive (Olea europaea L.) and almond (Prunus dulcis), whose oils could be used in the moisturising of the body in the embalming process (Serpico and White 2000), were identified in trace amounts in our mummies.
Fir (Abies cilicica) and pine (Pinus spp.) (Fig. 5b), which are considered to be the main components of the embalming resins (Serpico and White 2000), were well represented in our mummy datasets. While pine was identified in five mummies, fir was present only in one mummy. Finally, other ingredients, which might have been used to influence the smell of a mummified body, like populus (Populus euphratica), garlic (Allium sativum) and lotus (Nymphaea lotus), were also identified. Additional comparisons with forthcoming herbal and plant DNA datasets obtained from the NGS analysis of other mummified tissues from Ancient Egypt are pending. This will show whether the data are due to extraneous DNA from scattered pollen or if it is endogenous and authentic to the mummy as a part of the applied embalming substances. If the latter holds true, then we will undoubtedly learn more about the mysterious mummification process from ancient Egyptian times.
For further metagenomic analyses, two warm-climate specimens were processed along with the Egyptian mummies. The skeletons S1000 and S2000 were unearthed from the Llanos de Moxos eastern plain (Bolivian Amazonia), a vast seasonally savannah region. The climate is characterised by two marked seasons: the rainy season from October to April, when precipitation can reach 500 mm per month, and the dry season, with monthly precipitations of less than 50 mm. Temperatures are high throughout the year (Lombardo and Prümers 2010).
For comparison purposes, three previously published datasets of cold environment samples (i.e. the Denisova hominid, the Saqqaq and the Ötztal Iceman) were included.
The Blast results of all NGS datasets showed distinct differences among the metagenomes of the cold-climate samples compared to the Egyptian mummies and to the two samples from South America.
All cold-climate specimens yielded a high proportion of Eukaryota DNA within the total DNA (89.6–99.7 %). One of the analysed samples, the Egyptian mummy 1a, reached a top value of 96 %, therefore, displaying a similarity to the cold-climate specimens (Fig. 6).
We speculate that the abundance of Actinobacteria DNA in the Saqqaq, the Denisova hominid and in the two Bolivian skeletons is due to the overall contact with soil. The soil-living Actinobacteria are present in cold as well as in hot or arid regions. Therefore, this fingerprint is independent of the environmental temperature parameter.
Conversely, Actinobacteria are scarcely represented in all warm and cold-climate mummies (i.e. the Egyptian mummies and the Iceman), since they were not buried in soil. The Iceman was embedded within glaciers and ice-preserved until its discovery in 1991. The Egyptian mummies were embalmed, bandaged and buried in sarcophagi placed inside tombs. Hence, the bacterial fingerprint allows discrimination between mummified tissues and non-mummified tissues independently of the parameter temperature.
The human DNA content in the analysed Egyptian mummies was variable. Human DNA was abundant in the mummy extracts 1a, 2a and 3. Lower amounts were present in extracts 1b, 2b, 4 and 5 (Fig. 8).
Since two different biopsies from specific mummies were processed, a sample-dependent effect with regard to the amount of human DNA was identified. For example, samples 1a and 1b taken from mummy 1 and samples 2a and 2b taken from mummy 2 differ in the human DNA content as well as in the individual metagenomic composition (Figs. 6 and 8). The first sampling was performed prior to the unfortunate bacterial bloom that took place in the entire repository of the Institute of Pre- and Protohistory (Tübingen, Germany), and included the biopsies termed 1a, 2a and 3 (Fig. 6). Consequently, an excess of bacterial DNA is noted in all the other samples taken from mummies whose preservation condition had macroscopically worsened due to inappropriate environmental conditions within the mummy collection.
Finally, three of five randomly selected Egyptian mummies showed a representation of the human DNA content comparable to the recently published ancient genomes from cold-climate environments like the Saqqaq (Rasmussen et al. 2010), the Denisova hominid (Reich et al. 2010) and the Alpine Iceman (Keller et al. 2012) (Fig. 8). These data show that temperature is not the decisive parameter influencing human and animal DNA survival in archaeological samples.
The mummy datasets were additionally mapped against the hg19 reference genome and showed low absolute numbers of mitochondrial reads in relation to the total number of human reads (0 to 1 %).
This low amount of mitochondrial reads serves as further proof for authenticity. In the case of modern human DNA contamination, significantly higher amounts of extraneous DNA in the observed metagenomes should have been observed. Furthermore, all NGS-generated mitochondrial data were cross-checked against the haplotypes of our laboratory staff. Identical SNP patterns were never detected.
Even on the small-scale level of NGS sequencing for an initial characterisation of our samples and displaying only a mitochondrial read count of 1 %, we showed that one Egyptian mummy yielded a well-covered HVR2 region of the mitochondrium and gave an indication for haplogroup I2 (Fig. 9). Moreover, it is believed that haplogroup I2 has its phylogenetic origin in the Near East/West Asia (Derenko et al. 2007; Saunier et al. 2009; van Oven and Kayser 2009; Palanichamy et al. 2010; Terreros et al. 2011; Fernandes et al. 2012; Behar et al. 2012).
On the basis of all the data obtained so far, the possibility of retrieving authentic DNA from Egyptian mummies has been highlighted.
We conclude that warm temperature does not cause a complete degradation of nucleic acids and that the mummification appears to be a benefit with regard to DNA survival and preservation. Moreover, we emphasise that the human DNA content can reach sufficient amounts to obtain a good coverage not only for the mitochondrial genome, but also for the entire nuclear genome, when sufficient NGS runs are applied.
The analysis of next-generation sequencing (NGS)-generated mummy metagenomes reveals different DNAs. This variety of DNA can provide new insights in both the retrospective diagnosis of clinically relevant pathogens (e.g. bacteria, viruses and protozoa) and the chemical characterisation of the ingredients used during the embalming process. Furthermore, it can be used in the characterisation of the mitochondrial and nuclear haplotypes and, in turn, may give clues about the individuals’ phylogeny or even kinship. Hence, modern NGS technology appears to be a powerful aid to mummy research and strengthens the new discipline of Molecular Egyptology (Hawass et al. 2010).
To sum up, a good proportion of eukaryotic DNA in a specimen does not automatically imply that it has to come from a cold environment. On the other hand, a distinct difference within the warm-climate specimens can be pinpointed: DNA from mummified tissues can be differentiated from DNA from unearthed skeletons originating from warm-climate regions. Furthermore, a high bacterial content in a specimen does not speak against a good recovery of an endogenous human genome.
The bacterial fingerprint cannot predict experimental NGS success or failure in the specimen under investigation, but it can be used for sample typing.
We show that a “bacterial fingerprint” can be applied for the identification of mummified tissue signature as mummies independently from the climate in which they have been preserved. On the basis of this “bacterial fingerprint”, both the Egyptian mummies and the glacier mummy of the Alpine Iceman can be grouped together, even though they experienced very different temperatures over thousands of years.
We are grateful to Heiko Prümers, Iris Trautmann and Andreas Keller for their technical support and for providing sample/library access. Financial support was provided by the Graduiertenförderung Tübingen to R.K., M.B., C.-C.H.C. and C.M.P. Further support was obtained from the DAAD-GERLS program.
Conflict of interest
The authors declare that they have no conflict of interest.