Genome-Wide Comparison of Medieval and Modern Mycobacterium leprae

See allHide authors and affiliations

Science  12 Jul 2013:
Vol. 341, Issue 6142, pp. 179-183
DOI: 10.1126/science.1238286

Leprosy: Ancient and Modern

In medieval Europe, leprosy was greatly feared: Sufferers had to wear bells and were shunned and kept isolated from society. Although leprosy largely disappeared from Europe in the 16th century, elsewhere in the world almost a quarter of a million cases are still reported annually, despite the availability of effective drugs. Schuenemann et al. (p. 179, published online 13 June; see the 14 June News story by Gibbons, p. 1278) probed the origins of leprosy bacilli by using a genomic capture-based approach on DNA obtained from skeletal remains from the 10th to 14th centuries. Because the unique mycolic acids of this mycobacterium protect its DNA, for one Danish sample over 100-fold, coverage of the genome was possible. Sequencing suggests a link between the middle-eastern and medieval European strains, which falls in line with social historical expectations that the returning expeditionary forces of antiquity originally spread the pathogen. Subsequently, Europeans took the bacterium westward to the Americas. Overall, ancient and modern strains remain remarkably similar, with no apparent loss of virulence genes, indicating it was most probably improvements in social conditions that led to leprosy's demise in Europe.


Leprosy was endemic in Europe until the Middle Ages. Using DNA array capture, we have obtained genome sequences of Mycobacterium leprae from skeletons of five medieval leprosy cases from the United Kingdom, Sweden, and Denmark. In one case, the DNA was so well preserved that full de novo assembly of the ancient bacterial genome could be achieved through shotgun sequencing alone. The ancient M. leprae sequences were compared with those of 11 modern strains, representing diverse genotypes and geographic origins. The comparisons revealed remarkable genomic conservation during the past 1000 years, a European origin for leprosy in the Americas, and the presence of an M. leprae genotype in medieval Europe now commonly associated with the Middle East. The exceptional preservation of M. leprae biomarkers, both DNA and mycolic acids, in ancient skeletons has major implications for palaeomicrobiology and human pathogen evolution.

Leprosy, which results from infection with the unculturable pathogen Mycobacterium leprae, was common in Europe until the 16th century, when it essentially disappeared. In contrast, disease prevalence has remained high in the developing world. During the past 20 years, elimination efforts using multidrug therapy have been largely successful, leading to the perception that leprosy is no longer a global health threat despite an annual incidence of over 225,000 cases worldwide (1). To understand the evolution and phylogeography of the leprosy bacillus and to investigate the disappearance of leprosy from Europe, we have used DNA capture techniques and high-throughput sequencing (HTS) to obtain near-complete genome sequences of M. leprae from 11th- to 14th-century skeletal remains and from recent biopsies of leprosy patients.

The 3.3 Mb genome of M. leprae has undergone massive gene decay (2), and half of the coding potential has been lost, as reflected by the presence of ~1300 pseudogenes. This reductive evolution offers an explanation for the long generation time—~14 days in humans—and our inability to culture the bacillus in vitro (2). The phylogeny of modern M. leprae has been investigated by using genome sequencing and a combination of variable number tandem repeat and single-nucleotide polymorphism (SNP) typing, defining four major branches (35). However, very little is known about the strains responsible for leprosy in the past. Leprosy is among the few infections that contribute to skeletal changes; hence, attempts to trace its course through history have been made in an archaeological context (6). Molecular biomarkers have permitted detection of M. leprae in skeletal remains from many different time periods and geographical locations (7, 8). Studying ancient DNA (aDNA) by using a capture approach has the potential to provide new insight into pathogen evolution, as recently illustrated through full-genome reconstruction of Yersinia pestis from skeletal remains (9).

Extracts of bone and teeth from 22 medieval skeletons (Fig. 1, table S1, and supplementary materials notes 1 and 2) with osteological lesions suggestive of leprosy—from cemeteries in Denmark (n = 8 extracts) (10), Sweden (n = 8 extracts) (7), and the United Kingdom (n = 6 extracts) (11)—were screened for the presence of M. leprae DNA by using a bead capture approach of three genomic loci (12). Human mitochondrial DNA (mtDNA) fragments were enriched simultaneously in order to evaluate the characteristic nucleotide misincorporation patterns expected of ancient human (13) and pathogen DNA (14) so as to confirm their authentic ancient origin (supplementary materials note 3). Although DNA damage patterns corresponding to a medieval origin were found in the human mtDNA (15), the captured M. leprae DNA revealed much less damage (tables S1 and S2 and figs. S1 to S3), an observation that was entirely unexpected. We therefore relied on traditional authenticity criteria including blank controls, independent replication (16), and other biomarkers, such as mycolic acids (supplementary materials notes 4, 6, and 10; figs. S4 and S5; and tables S4 to S7). Five skeletal samples (3077 from Sweden, Jorgen_625 and Refshale_16 from Denmark, and SK8 and SK14 from the United Kingdom) (table S2) fulfilled our initial criterion for progression to HTS, performed before and after DNA repair (supplementary materials notes 2 and 3). Four samples (3077, Refshale_16, SK8, and SK14) yielded insufficient DNA (table S8) and hence were enriched for M. leprae by using DNA array capture (supplementary materials note 5). Unexpectedly, no enrichment for Jorgen_625 was necessary because an astonishing 40% of the reads mapped to M. leprae (table S8). It was thus possible to do a de novo assembly providing over 100-fold genomic coverage (table S9) and 169 contigs, separated by gaps corresponding to repetitive regions (Fig. 2A), that aligned perfectly with the modern M. leprae reference genome (supplementary materials note 6) (2). De novo assembly avoids ascertainment biases in gene order that may result from mapping assemblies, although it is often unsuccessful in aDNA investigations owing to inadequate preservation. Few, if any, large insertions or deletions (InDels) have occurred during the ~1000 years that separate the ancient and modern M. leprae strains.

Fig. 1 Sources and origins of the five medieval M. leprae strains from Denmark, Sweden, and the United Kingdom for which whole-genome sequences were determined in this study.

Pictures of the bones or teeth and their radiocarbon dates are shown.

Fig. 2 De novo assembly of the ancient strain of M. leprae from skeleton Jorgen_625 and distribution of SNPs across all M. leprae genomes sequenced in this study.

(A) All the gaps between contigs are in repetitive regions that represent ~2% of the genome of the reference strain TN. There are no structural variations and no changes in synteny, and all the coding sequences in the forward (red) and reverse (blue) strands were as in the reference genome (metagenomic analysis is available in fig. S9). (B) The 755 SNPs observed among the 16 genomes of M. leprae are represented on the 3.27-Mb circular chromosome of the TN genome. The five ancient strains sequenced in this study are shown in the outermost circles followed by 10 modern strains, colored as indicated in the key. Figures were produced with DNAPlotter (29).

To assess the high representation of M. leprae sequences in Jorgen_625, the ratios were determined of M. leprae versus human sequences in libraries of 230 to 400 base pairs (bp) and 309 to 622 bp fragments (supplementary materials note 5) and found to be 3.3 and 9, respectively (supplementary materials note 6 and fig. S6), thus indicating less fragmentation of the M. leprae DNA. The fundamental structural difference between mycobacterial and eukaryotic cells likely accounts for better DNA preservation. Mycobacteria are surrounded by a robust, hydrophobic layer of mycolic acids that constitute more than 40% of the cell biomass (17). Exceptionally high amounts of mycolic acids that typify M. leprae were found in the tooth pulp from Jorgen_625 and, to a lesser extent, in those from SK2, SK8, SK14 (supplementary materials note 10), and Refshale_16 (fig. S5) (11). This finding implies that the lipid-rich cell wall protects mycobacterial DNA (18, 19) from hydrolytic damage (20), hence the longer fragment lengths and reduced nucleotide misincorporation patterns detected.

We extended the array-based enrichment and HTS to ancient leprosy specimens (supplementary materials note 5) from Winchester in the United Kingdom (SK2 and SK14) and, as a negative control, a matched skeleton (SK12) from the same cemetery with no osteological evidence of disease. All samples from skeletons with leprosy-associated lesions (SK2, SK8, and SK14) yielded high-quality M. leprae reads from distinct strains. In contrast, <1% of sequences from SK12 mapped to the M. leprae genome, mostly to conserved regions present in all mycobacteria (table S10). This analysis further confirmed the authenticity of our samples and eliminated the possibility of cross-contamination among skeletal remains.

A total of five medieval strains—namely 3077, Jorgen_625, Refshale_16, SK2, and SK8—satisfied our criteria for genome-wide comparison (>80% genome coverage, at least fivefold depth), enabling us to compare the genotypes of M. leprae strains from 11th- to 14th-century Europe with those of modern strains from different leprosy-endemic regions. For the comparison, the four available reference genome sequences (TN from India, Thai53 from Thailand, NHDP63 from the United States, and Br4923 from Brazil) (5) were supplemented with those of seven modern strains, each belonging to a specific SNP type or of a different geographic origin (Table 1 and table S11), obtained by using multiplexed array capture of DNA prepared directly from skin biopsies of leprosy patients or, in one case, following passage in an armadillo. Here, genome coverage ranged from 83 to 98% at 10- to 160-fold depth (Table 1 and table S9).

Table 1 Ancient and modern strains of M. leprae whose genomes were sequenced and compared in this study, together with the reference genomes.

BP, before present (in years); Ref, Reference genomes (5).

View this table:

Whole-genome reconstructions disclosed a remarkable level of conservation (supplementary materials notes 6 and 9). In total, only 755 SNP and 57 InDels (<7 bp) were found among the 16 M. leprae genomes (Fig. 2B and tables S12 to S14). The distribution of the specific SNPs across all 16 genomes revealed 122, 50, and 43 distinct SNPs in modern strains S15, S10, and S9, respectively, accounting for 40% of all the genetic diversity in M. leprae known to date. Strains S9 and S15 harbor the Thr53Ile mutation in their folP1 gene that confers dapsone-resistance (21). No new pseudogenes were found in ancient M. leprae, but five were discovered in the modern genomes. These all affect genes for conserved proteins (ML1270, ML1340, ML1761, ML0141, and ML0659) of unknown function. By far the most polymorphic gene, with a total of 11 SNPs (10 of which are nonsynonymous), is ML0411, which encodes an immunodominant serine-rich cell surface antigen (2), and this may reflect pressure from the host immune system.

The phylogenetic relatedness of all 16 genomes was assessed with outgroup analysis by using maximum parsimony, maximum likelihood, neighbor-joining, and Bayesian phylogeny inference (Fig. 3 and fig. S7). Most strains clustered into four major branches—which is consistent with the SNP typing scheme (5)—except for S9 and S10, which form the deepest lineages, and a new branch, named branch 0 (Table 1). Strain S15 also displayed deep divergence and could not be unambiguously placed among the five branches (Fig. 3). These three strains (S9, S10, and S15) were found to branch off closest to the common ancestor of M. leprae, M. tuberculosis, M. ulcerans, and M. avium (supplementary materials note 7 and figs. S7 and S8).

Fig. 3 Phylogeny of medieval and modern M. leprae.

(A) Phylogenetic relationship of M. leprae genomes using a maximum parsimony tree, including M. avium as an outgroup. Geographic origin and SNP type are given at each branch tip. Bootstrap node support is shown in gray, and nucleotide substitutions on each branch is in bold. (B) Bayesian phylogenetic tree calculated with BEAST 1.7.1 (23), including all ancient strains with radiocarbon dates, inferred from a total of 516 genome-wide variable positions. Divergence time intervals are shown on each node in years B.C.E. and C.E. Posterior probabilities for each node are shown in gray. Labeling colors vary from red to blue based on the tip age for each branch.

Two new phylogeographic conclusions can be drawn from our reconstructed M. leprae phylogeny. First, three ancient strains (3077, Refshale_16, and SK8) belonging to branch 2 form a tight cluster with very short branch lengths (Fig. 3A). Modern branch 2 strains have previously been reported only in Iran and Turkey (5), pointing to a possible link between Middle-Eastern and medieval European strains. Second, branch 3 strains have not been found in the Middle East (5), whereas this genotype has been detected in European skeletons (5, 7, 8), including SK2. The striking closeness of Jorgen_625 and SK2 (Fig. 3A) with the NHDP63 strain, and 52 other branch 3 strains from the United States (22), is consistent with the European origin of leprosy in the Americas (4).

Branch shortening (supplementary materials note 8) is commonly observed when ancient sequences are included in tree reconstructions owing to their lower number of derived positions. The five ancient M. leprae strains do indeed have shorter branch lengths than those of modern strains, which have accumulated more substitutions. The average distance of the ancient strains to the most recent common ancestor (MRCA) was 19.8 nucleotides, whereas it was 27.5 nucleotides for the modern strains, a statistically significant difference (Mann-Whitney U test: U = 7.50, P < 0.05).

Divergence times for the M. leprae strains were estimated by using the Bayesian inference software BEAST (23), with models of both a strict and a relaxed clock in order to compensate for possible rate variation among lineages observed in other bacterial pathogens (24). For both models, radiocarbon dates for ancient skeletal samples and isolation dates for the modern bacteria were used as tip calibration (supplementary materials note 8). Using the relaxed clock and 16 M. leprae genomes, we estimated a rate of 8.6 × 10−9 substitutions per site per year [1.32 × 10−8 to 3.61 × 10−9 95% highest posterior density (HPD)]. Exclusion of S15 permitted the use of a strict clock, yielding an estimated rate of 6.13 × 10−9 substitutions per site per year (8.56 × 10−9 to 3.38 × 10−9 95% HPD). The calculated mutation rate of M. leprae by use of direct fossil calibration is thus close to that of the related human pathogen M. tuberculosis by use of tip calibrations from modern strains [5.4 × 10−9 substitutions per site per year (24)]. It remains to be seen whether long-time fossil calibration as applied here to M. leprae will produce similar results for M. tuberculosis. The resulting divergence times for the MRCA for all the M. leprae strains are 2871 years ago (1350 to 5078 years ago) and 3126 years ago (1975 to 4562 years ago) for the relaxed and strict clock models, respectively (Fig. 3B). This is consistent with the oldest known and accepted osteological evidence for leprosy from 2000 BCE. India (25).

The sudden decline of leprosy in 16th-century Europe was almost certainly not due to the medieval European strain of M. leprae losing virulence, as evidenced by the close similarity between the modern NHDP63 strain and ancient Jorgen_625 and SK2 strains. Extraneous factors—for example, other infectious diseases such as plague or tuberculosis, changes in host immunity (26, 27), or improved social conditions—may have accounted for its decline. Our finding of well-preserved mycolic acids in a 14th-century tooth together with sufficient DNA to generate a whole de novo genome sequence may enable investigation of ancient leprosy bacilli and other pathogens well beyond the inferred maximum age for mammalian DNA of around 1 million years (28). There is thus a real prospect that the prehistoric origins of M. leprae can be retraced.

Supplementary Materials

Materials and Methods

Figs. S1 to S9

Tables S1 to S14

References (3073)

References and Notes

  1. Acknowledgments: We are grateful to the following people for providing samples, support, and advice: C. Bauer, H. Burbano, J. Hinds, S. Junker, M. Kodio, A. Fomba, C. Lanz, P. McLaren, J. Rougemont, and S. Schreiber. All raw read files have been deposited in the trace archive of the National Center for Biotechnology Information Sequence Read Archive under accession no. SRP022139. This work was supported by the European Research Council (ERC-APGREID), the Carl Zeiss Foundation, the Fondation Raoul Follereau, the Swiss National Science Foundation (Brazilian Swiss Joint Research Program), the Deutsche Forschungsgemeinschaft Priority Program 1335 Scalable Visual Analytics, the Central Innovation Program (grant KF2701103BZ1), the Graduate School Human Development in Landscapes, the Excellence Cluster Inflammation at Interfaces, the Medical Faculty of the Christian-Albrechts-University Kiel, a British Academy Small Research Grant, the Leverhulme Trust (Grant F/00094/BL), and the Social Sciences and Humanities Research Council of Canada (postdoctoral fellowship grant 756-2011-0501). Human remains were obtained under license from the UK Ministry of Justice.
View Abstract

Stay Connected to Science

Navigate This Article