Clin Microbiol Infect 2011; 17: 821–829
The direct detection of ancient Mycobacterium tuberculosis molecular biomarkers has profoundly changed our understanding of the disease in ancient and historical times. Initially, diagnosis was based on visual changes to skeletal human remains, supplemented by radiological examination. The introduction of biomolecular methods has enabled the specific identification of tuberculosis in human tissues, and has expanded our knowledge of the palaeopathological changes associated with the disease. We now realize that the incidence of past tuberculosis was greater than previously estimated, as M. tuberculosis biomarkers can be found in calcified and non-calcified tissues with non-specific or no visible pathological changes. Modern concepts of the origin and evolution of M. tuberculosis are informed by the detection of lineages of known location and date.
Early Morphological Studies
You have to know the past to understand the present Dr Carl Sagan (1934–1996)
Tuberculosis (TB) is a disease that was recognized in ancient times, primarily by the characteristic changes to the spine (kyphosis or gibbus) that result in Pott’s disease. The eminent Greek physician Hippocrates (460–370 BC) gives a clear description of kyphosis, stating that it can be the result of disease, can occur above or below the diaphragm, and is associated with hard tubercles in the lungs, and abscesses in the lumbar region . He also noted that patients had a poor prognosis if the spinal curvature was above the diaphragm and if the patient was a child or young adult. Many other ancient and historical texts contain recognizable descriptions of TB, where the disease is identified as phthisis, scrofula, King’s Evil, lupus vulgaris, consumption, etc. . As palaeopathology emerged from physical anthropology and forensic science as a distinct discipline, diagnostic criteria based on skeletal changes were agreed [3,4].
As part of an ongoing process, detailed morphological studies were performed on more recent historical skeletal collections with contemporaneous records of the individual cases, including age, sex, occupation, symptoms, and cause of death [5–7]. This led to the realization that some bony changes, such as periostitis (surface changes caused by new bone formation) on ribs, were significantly associated with individuals who had been diagnosed as having clinical TB. Other conditions linked to recognized TB changes include hypertrophic osteoarthropathy (HOA) [8,9] and serpens endocrania symmetrica—a morphological sign of respiratory distress and increased vascularization around the brain .
The use of palaeohistology  also aids the recognition of more subtle changes associated with TB in calcified and non-calcified tissues. Mummified remains can demonstrate signs of TB, such as granulomas in lung and other organs. Indeed, Allison et al.  used histological methods to rehydrate such tissue from a pre-Columbian Andean mummy (AD 200–800), and demonstrated bacteria within granulomas that were acid-fast by the Ziehl–Neelson stain, with the typical microscopic appearance of mycobacteria. Radiological imaging gives additional data, and the use of computerized tomography facilitates the three-dimensional examination of morphological features. Micro-computerized tomography is becoming used more frequently, although there are concerns about the increased radiation load, which may impact on the recovery of biomolecular markers such as ancient DNA (aDNA) (see below).
Early studies, based on visual appearance, radiology, and a limited number of microscopic and histological investigations, led to the conclusion that TB definitely occurred in ancient populations but that it was comparatively rare. The earliest cases recognized were from Neolithic times [12,13], so the hypothesis that TB was associated with animal domestication and that human TB originated from Mycobacterium bovis  became widely accepted. The significance of the finding of acid-fast bacilli within granulomas in pre-Columbian mummified tissue  was not generally appreciated, and several subsequent publications included the suggestion that Columbus and subsequent European contacts introduced TB to the Americas.
Early Biomolecular Studies
The use of cloning to detect DNA from the extinct quagga in 1984  opened up the prospect of direct genetic studies of archaeological material. However, this only became practicable after the development of PCR , which enables rapid amplification of specific genetic loci defined by oligonucleotide primers. There was an explosion of studies based on extinct animals, plants, and other remains, several of which are now suspect, owing to the lack of appropriate measures to prevent contamination with modern DNA. The early development of this field in archaeology and palaeontology had a strong influence on the initial criteria that emerged regarding the precautions required for the production of verifiable data based on aDNA [16,17]. These have since been modified  to take account of our increased understanding of the sources of contamination  and the stability of Mycobacterium tuberculosis complex (MTBC) aDNA  as compared with other biomolecules [21,22].
TB was the first infectious disease to be successfully detected by PCR in bones  and mummified tissue , owing to a fortuitous combination of circumstances. There were agreed morphological diagnostic criteria, appropriate DNA extraction procedures had been developed, and clinical microbiologists had designed specific PCR primers. The driving force for molecular diagnosis of TB was the extremely slow growth rate of the causative organism, M. tuberculosis, and the other closely related members of the MTBC. Clinical treatment is the same for all members of the MTBC, so no attempt was made at the time to distinguish between its members. The first primers used were based on an MTBC-specific locus in the insertion sequence IS6110, and, because it is usually present in multiple copies within the bacterial cell, it was the preferred locus for detection of tubercle bacilli. These early studies confirmed TB in Byzantine Turkey, pre-European contact Borneo, and a pre-Columbian Andean mummy (1000 BP). They were quickly followed by reports of MTBC aDNA in ancient Egypt, China, America, and Europe, summarized in a series of reviews [18,25–29].
It soon became apparent that MTBC aDNA could be detected in material with non-specific pathology, such as ribs with periostitis  and calcified pleura . Specific MTBC mycolic acids, found in the bacterial cell wall, can provide independent confirmation of TB [31,32], which is of particular significance, as the method is based on HPLC and there is no amplification of the target molecules. Recently, another category of cell wall lipid biomarkers, the phthiocerol dimycocerosate waxes, have enabled the detection of MTBC in archaeological material by the use of negative ion chemical ionization gas chromatography mass spectrometry of their diagnostic mycocerosic acid components. Redman et al.  investigated a group of 49 individuals from the 1837–1936 Coimbra Identified Skeletal Collection (Portugal), half of whom had records giving TB as a cause of death. There was a 72% correlation between detection of mycocerosate acid biomarkers with individuals who were listed as likely to have died from TB.
Re-evaluation of TB Palaeopathology as Elucidated by Biomolecular Techniques
It had been observed by Baron et al.  that bone samples from sites with obvious pathology, and from the same bone but of apparently normal appearance, were positive for MTBC aDNA. Their specimens were from a historical collection of individuals given a diagnosis of TB at autopsy, and they concluded that the tubercle bacilli had been carried into the bone with the bloodstream. They speculated further that this raised the possibility that other infectious agents retrievable in blood should be detectable by PCR in the absence of lesions—a supposition now known to be correct .
A small study of skeletal remains from 15–17th-century Lithuania may have been the first to deliberately include bones from individuals with no bone pathology. Faerman et al.  examined pathological lesions vs. healthy tissue from the same skeleton, using four individuals. Three additional skeletal remains, chosen at random from the same site, were also examined, plus soil samples to test the specificity of the PCR primers. Compact bone and teeth were examined. All seven individuals contained MTBC aDNA. One soil sample was also positive, but further analysis showed that only a sub-specimen containing small pieces of skin and cartilage was positive. The authors concluded that they had obtained direct evidence for haematogenous spread of tubercle bacilli. They observed that many clinical forms of TB leave no specific traces on the skeleton, and suggested that the disease may be detectable even in individuals represented by only a single bone or tooth.
Zink et al.  examined the relationship between morphological bony changes and TB in ancient Egypt. They examined 41 bones: 37 from the necropolis of Thebes-West (2120–500 BC), and four from the necropolis of Abydos (3000 BC). Only three subjects had pathological changes typical of TB, whereas 17 showed pathological changes that were non-specific but were likely to have been caused by TB. The remaining 20 specimens were viewed as controls, and had no visible pathology. Of 30 specimens that contained human DNA, nine were positive for MTBC DNA: two of the three with typical pathology, five of 13 with non-specific pathology but that were probable cases of TB (including two from 3000 BC), and two of 14 with no pathological changes.
Mays et al.  noted that tuberculous lesions in the ribs arise by extension from spinal lesions, from haematogenous spread from some remote soft tissue focus, or by direct spread from disease in the lungs, pleura, or chest wall lymphatic system. In the clinical literature, lytic lesions, caused by dissemination of disease from remote soft tissue foci via the bloodstream, are believed to be the most frequent form of rib involvement in TB, and in such cases bony changes would not be expected. Mays et al. examined 14 ribs from a deserted mediaeval village in north-eastern England (10–16th century AD), seven with pathological surface pathology, and seven forming an age-matched and sex-matched control group, but with no visual or radiological sign of pathology. MTBC aDNA was detected in one rib with pathology and in two control ribs, indicating that the absence of visible pathology does not preclude detection of tuberculous biomarkers. The failure to detect MTBC aDNA in bones with pathology is most likely attributable to poor DNA preservation and/or localized distribution within the bone.
A further example of the detection of TB in the absence of visible markers is given by a study of skin tissues from 12 pre-Columbian (AD 140–1240) Andean mummies, where Konomi et al.  demonstrated MTBC DNA in two individuals. In addition, the study by Redman et al.  reported that there were no visible signs of TB in 11 of the 49 skeletons that were positive for mycocerosates. Therefore, both MTBC aDNA and cell wall lipid biomarkers have demonstrated molecular traces of TB in the absence of specific palaeopathology.
When MTBC aDNA was detected in bones from individuals with no obvious palaeopathology at all [30,35,36], this initially attracted criticism and the assumption of laboratory contamination, until anthropologists became aware of the low proportion of skeletal TB in clinical infections from the recent pre-antibiotic era. It is estimated that bone TB occurs in only 3–5% of infections , and that the spine is involved in approximately 40% of these. Therefore, the earlier views of the incidence of TB based solely on bone morphology were necessarily severe underestimates. The detailed studies on the relationship of rib lesions with TB [5–7] had indicated that this might be the case, but until the use of specific aDNA and lipid biomarkers, this could not be proved (Table 1).
|Confirmation of palaeopathological changes associated with TB|
|Gain a better understanding of the majority of past TB infections that leave no visible signs of disease|
|To examine ancient TB in relation to diet, health, and social structures|
|Geographical range in the past in relation to human and pathogen lineages|
|Comparison of microbial pathogens from the past with those of today|
|Provision of real-time markers of genetic changes|
|Enable monitoring of changes in virulence over a longer time-scale|
Identification of Strains and Lineages
The development of molecular diagnostics and the increased understanding of the MTBC genome encouraged efforts to identify individual members of the MTBC in ancient material and to distinguish between strains and lineages. Methods are based mainly on copy number of repetitive sequences, deletion analysis, and single-nucleotide polymorphisms (SNPs). Sreevatsan et al.  described three SNP types based on katG463 CTG to CGG and subsequent gyrA95 ACC to AGC mutations, thereby proposing an evolutionary pathway for the tubercle bacilli.
Spoligotyping is based on the direct repeat (DR) region of the MTBC . PCR primers in the DR region amplify up to 43 unique spacer regions that lie between each DR locus. The spacers are visualized by dot blot hybridization on a membrane, giving a fingerprint, and patterns are stored on an international database. Thus, Taylor et al.  were able to demonstrate, using spoligotyping and SNPs, that bones from mediaeval London (AD 1350–1538) were infected with M. tuberculosis rather than M. bovis. Both human and cattle bones with periostitis from an earlier mediaeval site in northern England were examined for M. tuberculosis and M. bovis , but only M. tuberculosis was detected, and only in human remains. Indeed, to date, there is only one site where human archaeological cases of M. bovis have been identified, and this concerns Iron Age pastoralists (360 BC to AD 230) from southern Siberia [43,44].
Animal bones have proved a difficult subject for the study of past TB. There is one report of an Iroquoian dog from the 16th century, with the TB-associated but non-specific palaeopathological condition HOA , in which MTBC aDNA was found. A histological study of the HOA pathological lesions in these remains  indicated that the dog had experienced a long and most likely arduous period of illness. The animal was buried with purpose and lived within a longhouse environment , thus providing the perfect opportunity for the transmission of an interspecies (and intraspecies) TB infection.
However, the rarity of animals in the pathological record can no doubt be explained by the fact that most collections of animal bones are from domesticated animals used for food. These rarely live a sufficient time to develop recognizable lesions or to die a natural death. Once the animals are killed, the carcases will be butchered, cooked, and eaten. Bones will be used if possible, or discarded. An exception to this scenario was discovered in the Natural Trap Cave in Wyoming, USA. This cave is 30 m deep and presents an unavoidable hazard on a game trail, thereby providing an unbiased sample over >100 000 years of the many passing species. Owing to the depth of the cave, falls are invariably fatal, and the temperature remains constant at 4–5°C, which is known to be conducive for aDNA persistence. Palaeopathological changes were occasionally noted in bovids, and a metatarsal from an extinct long-horned bison (17 870 ± 230 BP) was found to contain MTBC aDNA . Further characterization was attempted, and consensus spoligotyping indicated an ancestral version of the MTBC DR region, rather than the typing patterns associated with M. bovis or most extant strains of M. tuberculosis.
The first indication of human TB infections caused by another member of the MTBC was in a population study from Thebes West in ancient Egypt . Spoligotyping indicated that M. tuberculosis was present in most of the infected individuals, but there appeared to be one case where Mycobacterium africanum was implicated. A follow-up study of bone and soft tissue samples from 118 individuals compared the incidence of TB over three time periods. This revealed several samples with an M. africanum spoligotype pattern from a Middle Kingdom tomb (2050–1650 BC), whereas samples from later periods were all typical of M. tuberculosis . There was no indication of M. bovis. The authors concluded that these findings did not support the theory that M. tuberculosis had evolved from M. bovis.
Palaeoepidemiology of TB
To estimate disease prevalence over a long time-span, the occurrence of TB in vertebral bones only was examined at Thebes-West in Ancient Egypt. Specimens were from three time periods: the pre-dynastic to early dynastic period (c. 3500–2650 BC), the Middle Kingdom to the Second Intermediate Period (c. 2100–1550 BC), and the New Kingdom to the Late Period (c. 1450–500 BC) . MTBC aDNA was found in 18 of the 50 individuals with detectable human aDNA throughout these time periods, indicating the long-term association of host and bacterial pathogen. There were more positives in individuals with typical TB pathology, but seven of 50 positive cases had bones of normal appearance.
A population study on a more recent, but exceptionally well-preserved, collection of naturally mummified or partially mummified remains was based on a collection of 263 individuals buried during the 18th century, in a church crypt that was sealed and only rediscovered in the early 1990s [51,52]. Contemporaneous civic and church records were available, so it was possible to determine TB prevalence according to age, sex, occupation, and family group. TB was assessed initially by visual and radiographic evidence of palaeopathology and contemporary records. Independent research groups determined the presence of MTBC aDNA, and found over 60% of individuals to be infected . Molecular typing showed that three members of a family group were infected with separate strains of M. tuberculosis. A mother was infected with a different SNP genotype  from her two daughters , thus suggesting that, in this high-incidence population, infection was acquired from the community, not within the family.
An intriguing aspect of palaeomicrobiology is the ability to detect co-infections by the use of specific biomarkers. Donoghue et al.  detected both MTBC and Mycobacterium leprae aDNA in skeletal material from individuals with obvious palaeopathology for lepromatous leprosy. This had been prompted by reports in the clinical literature of leprosy patients dying of TB, as M. tuberculosis is the more virulent pathogen in such patients. Other past populations with molecular evidence of two or more infectious diseases include the Early Christian Nubian population of Kulubnarti, where both TB and leishmaniasis were identified, and 15–16th-century ‘wet’ Korean mummies, in which many intestinal parasites, TB and hepatitis B virus have been detected .
Setting the Molecular Clock—direct Determination of Past MTBC Lineages
Brosch et al.  demonstrated that the SNP types identified by Sreevatsan et al.  occurred in a lineage of M. tuberculosis strains that had already lost M. tuberculosis specific deletion 1 (TbD1) (Fig. 1), and redefined them as markers of three principal genetic groups (PGGs). The results of Fletcher et al.  therefore demonstrated that M. tuberculosis aDNA amplified from naturally mummified Hungarians from the 18th and 19th centuries belonged to katG463/gyrA95 PGG2 and PGG3, and that the TbD1 deletion occurred in the lineage of M. tuberculosis before the 18th century. At the time of publication this was of interest, as it suggested that the dramatic increase in the number of TB cases later in the 18th century in Europe involved the TbD1-deleted M. tuberculosis strains that are now prevalent in the developed world. It also provided a further demonstration that TB was caused by M. tuberculosis and not by M. bovis.
The time-scale for the emergence of distinct human lineages was dramatically extended by the findings of Hershkovitz et al. . Studies on the pre-Pottery Neolithic site of Atlit Yam in the eastern Mediterranean, dating from 9250 to 8150 BP, demonstrated that a woman and infant were infected with M. tuberculosis, based on morphological changes to bones, the detection of MTBC-specific cell wall lipid biomarkers, and the detection of aDNA. Bones were well preserved, as the skeletal remains had been buried in thick clay under the sea, and DNA preservation was also good, so some molecular characterization was possible. Several different target loci were detected, including that of the TbD1 region. This demonstrated that these two individuals were infected with M. tuberculosis from a TbD1-deleted lineage.
Phylogenomics of MTBC Strains
The individual members of the MTBC (excluding the strains with smooth colonies classified as Mycobacterium canettii) are 99.95% identical on the basis of nucleotide sequence. This gave rise to the suggestion that there had been an evolutionary bottleneck at the time of speciation , estimated by ancestral sequence inference to have occurred 15 000–20 000 years ago. However, there are clear differences in phenotype, host tropism, and pathogenicity, so comparative genomic studies were undertaken. Brosch et al.  analysed 20 variable regions situated around the genome of a representative and diverse set of 100 MTBC strains, obtained from different hosts and with a broad range of geographical origins. The major finding was that MTBC strains have undergone reductive evolution by the unidirectional loss of chromosomal sequences. Once a non-repetitive region has been deleted, it cannot be replaced, and the deletion is a biomarker of a clone derived from a single cell and all of its descendants . The application of deletion analysis thus demonstrates unambiguously that M. bovis has undergone numerous deletions relative to M. tuberculosis, and M. tuberculosis is therefore a more ancestral lineage. The TbD1 locus is absent from the majority of strains now prevalent in Europe and the Americas, but occurs in strains of different geographical origin and in MTBC lineages associated with animal hosts. Therefore, Brosch et al.  and subsequent authors identified the TbD1 deletion as a marker of importance in our understanding of the evolution of the MTBC. This is supported by total genome sequencing of M. tuberculosis  and M. bovis , which has shown that the M. bovis genome is slightly smaller (98.5%) and has undergone 11 deletions in comparison with M. tuberculosis.
In the search for the source of the MTBC, Gutierrrez et al.  re-examined M. canettii, which Brosch et al.  had concluded was a distinct outlier group with the most ancestral characteristics. Indeed, it became clear that different strains of M. canettii displayed more variation than is found in all the other species or ecotypes  in the MTBC. This led the authors to the hypothesis that the MTBC had emerged, via an evolutionary bottleneck, from a postulated species with the expected range of internal clonal variation, termed ‘Mycobacterium prototuberculosis’. Using synonymous sequence diversity to estimate the minimal age of the last common ancestor, they concluded that the bacteria responsible for tuberculosis have co-evolved with early humans since at least the time of early hominids, between 2.6 and 2.8 million years ago. There is continuing debate about this suggested scenario, but, meanwhile, a possible case of TB in Homo erectus dated to (490–510) ± 0.05 Ka (thousand years) was described in a partial skeleton discovered in western Turkey . The diagnosis was based on endocranial palaeopathology distinct from serpens endocrania symmetrica, termed leptomeningitis tuberculosa, believed to be associated with tuberculous meningitis, but critics believe that this diagnosis is premature and based on inadequate evidence . This remains an intriguing possibility that awaits the application of next-generation molecular methods for clarification.
Relationship of Tubercle Bacilli to Different Human Societies
The use of standardized methods of reporting skeletal collections, and the study of past populations in the context of their diet, living conditions, work patterns, and type of society, has facilitated the study of health and disease over time . Evidence of habitations, population density, animal domestication, tools and agriculture is supplied by traditional archaeological assessments. Carbon and nitrogen stable isotope analysis can give information on the type of diet eaten, e.g. terrestrial, marine, freshwater, meat, or dairy. Coprolite analysis can also provide data on nutrition and calorie intake, based on identification of bones, seeds, pollen, etc. in preserved faeces. Strontium and oxygen isotopes can give an indication of location and mobility during a lifetime. There may also be evidence of major climatic events, such as earthquakes and tsunamis changing sea levels, or the height of the annual flooding of the Nile .
TB is a disease that can exist as a latent infection throughout the lifetime of a host, but that may be reactivated or re-acquired when host resistance is impaired. In addition to the extremes of life, individuals become susceptible to active disease when suffering from poor nutrition, other serious medical conditions, or severe mental or physical stress. Today, it is estimated by the WHO that one-third of the global population is infected with tubercle bacilli but that only 10% of individuals will develop disease in their lifetime (Fig. 2). The implication of the high proportion of latent infections is that there has been a prolonged period of co-evolution of host and pathogen. This is also indicated by the detection of a TbD1-deleted lineage in the early Neolithic site at Atlit Yam.
Evolutionary biologists have postulated that, in the long period of human evolution that pre-dated the Neolithic, tubercle bacilli occupied an unusual ecological niche in human biology. While populations were primarily small, mobile, and often family groups, the organisms could survive only by surviving throughout the lifetime of their host, with transmission occurring from adults with reactivated disease resulting from an age-related failing immune system, or from other severe disease or stress. Infants are more susceptible to acquisition of disease, and, because of their immature immune system, may have a high mortality with primary disseminated TB, thereby providing an opportunity for transmission soon after infection. Therefore, tubercle bacilli adopt a quasi-commensal relationship with their host in the majority of cases .
This scenario explains the development of the clonal relationship of the MTBC complex members with their host, where particular bacterial lineages are associated with human lineages, even in the modern world and in second-generation or third-generation inhabitants of ethnically diverse cities  (Fig. 3). However, when the human population density rises significantly, there is a greater opportunity for transmission to susceptible hosts, so bacterial virulence is likely to increase, owing to natural selection. This has now been manifested by the development of evolutionarily modern lineages of M. tuberculosis, which induce a lower level of early inflammatory response in the host and demonstrate faster progression to active disease . Therefore, reflecting the co-evolution of host and pathogen, it is unsurprising that length of urbanization is a predictor of host genetic resistance to TB .
At this time of rapid population growth and the emergence of new virulent M. tuberculosis lineages, it is especially important that we understand the history of TB and evolution of the host–pathogen relationship. This chronological review also allows us to follow the sequence of observations, emergence of concepts and changes in our perception of this relationship (Table 2). Palaeomicrobiology offers a unique opportunity to detect past lineages that may no longer be extant. It also enables us to calibrate the molecular clock when calculating time to most recent common ancestor by ancestral sequence inference. By providing direct evidence of specific biomolecular markers, palaeopathologists are able to make better assessments of the past prevalence of disease and to identify TB-associated skeletal changes with more confidence. In this regard, it is important to extend palaeomicrobiological studies around the world to obtain as comprehensive a picture as possible. The development of next-generation technology and additional biomarkers, with greater stability than DNA, should enable us to explore even further back into the past.
|Confirmation of diagnosis from palaeopathology|
|Additional palaeopathological indicators of TB authenticated|
|Evidence of haematogenous and direct spread of tubercle bacilli within the body|
|Understanding that Mycobacterium tuberculosis DNA is in bones with no pathology|
|Appreciation that former estimates of TB infection were far too low|
|Confirmation of geographical distribution in ancient times; for example, TB existed in Borneo before European contact, existed in pre-Columbian America, and was widespread in ancient Egypt and Rome|
|M. tuberculosis complex DNA detected in North American bison from the Pleistocene (17 500 years ago)|
|Confirmation of palaeopathological lesions on bovids indicates that M. tuberculosis in North America was spread from Asia by ungulates that crossed the Bering land bridge|
|Direct evidence that humans were infected with a TbD1-deleted lineage of M. tuberculosis 9000 years ago|
|Mycobacterium africanum and M. tuberculosis both shown to exist in ancient Egypt over 4000 years ago|
|Mycobacterium bovis only clearly identified in human remains from Iron Age pastoralists in southern Siberia|
|Co-infections identified in some human remains|
I greatly appreciate the contribution made by my collaborating colleagues and their groups around the world, the research staff and students who have worked with me since 1992, and the authorities who made this work possible. Recent funding for the Vác Mummy Project is acknowledged from the Hungarian Scientific Research Fund (OKTA) Grant No. 61155.
Conflicts of interest: nothing to declare.