Ancient Mycobacterium leprae genomes reveal a potential origin and dispersal of the ancestral Branch 0

Abstract

Although previous studies have examined numerous ancient Mycobacterium leprae genomes, the available strains remain concentrated in western Eurasia. Ancient genomes from eastern Eurasia have remained virtually unexplored, limiting our ability to reconstruct the evolutionary history of M. leprae. In this study, we report the first two ancient M. leprae genomes recovered from this region. Phylogenetic analyses place both genomes within the ancestral Branch 0, and the observed phylogeographical patterns are consistent with the hypothesis that this lineage originated in southern East Asia. Together, these findings expand the known ancient genomic diversity of M. leprae and suggest that population movements across the Tianshan region may have contributed to its wider dissemination.

The first ancient Mycobacterium leprae genomes from eastern Eurasia are reported, both within ancestral Branch 0, suggesting a potential southern East Asian origin and possible contribution of human movements across the Tianshan region to its spread.

Introduction

Leprosy is a deforming and chronically stigmatizing infectious disease caused primarily by M. leprae, transmitted mainly via nasal droplets and has profound consequences for patients, both in terms of physiological function and social identity¹. Although skeletal manifestations and historical records provide valuable insights into the disease^2,3, genomic data are indispensable for reconstructing the pathogen’s evolutionary history in detail.

Phylogenetic analyses of M. leprae genomes have divided global strains into five major Branches (0-4)⁴. Subsequent studies further resolved Branch 2 into sub-lineages 2E, 2H, and 2F, and identified a previously unrecognized Branch 5. Modern M. leprae strains exhibit varying diversity across geographic regions: high diversity characterizes endemic areas, such as India (Branches 1 and 2) and Brazil (Branches 1, 3, and 4), whereas lower diversity is found in Madagascar (Branch 1) and West African countries (Branch 4)^5,6.

To date, dozens of ancient M. leprae genomes have been recovered, nearly all from medieval Europe, with Abusir1630 from Egypt, belonging to Branch 4, as the sole exception^4,7–11. The genetic diversity observed among medieval European strains (Branches 0, 2F, 3, and 4) has led to two competing origin models. Model 1 suggests a western Eurasian origin, as most medieval European strains are ancestral to modern strains. In contrast, Model 2 argues for an extra-European origin, citing the broad distribution of ancestral Branch 0 strains across modern East Asia and Pacific islands and suggesting that European strains were introduced repeatedly from multiple regions during or before the medieval period¹⁰. However, the strong geographical sampling bias, particularly the absence of ancient genomes from eastern Eurasia, precludes a definitive assessment of these models.

Despite numerous historically documented records of leprosy in China¹², there is no documentation of large-scale leprosy epidemics comparable to those observed in medieval Europe¹³. Moreover, soft tissues are rarely preserved in archeological remains, so researchers typically rely on dental pulp as an alternative source for ancient pathogen screening. However, M. leprae primarily affects the skin and peripheral nerves, exhibits a highly focal pattern of infection, and only rarely causes bacteremia, which substantially limits the sensitivity of dental pulp for detecting this pathogen. In addition, the stigmatization of leprosy, leading to condemnation, isolation, or even expulsion of individuals since ancient times, makes it challenging to obtain ancient M. leprae samples, except from special sites such as leprosaria. As a result, no ancient M. leprae genome has yet been obtained from eastern Eurasia, and this significant sampling bias fundamentally limits our understanding of the evolutionary history of this pathogen.

To address this gap, we systematically screened 1800 human skeletal remains from archeological sites across China for traces of M. leprae DNA. We identified two leprosy-positive individuals and successfully reconstructed high-quality genomes from both, which belong to the ancestral Branch 0 and represent the oldest known strains of this lineage. These genomes provide the first direct evidence of ancient M. leprae in eastern Eurasia and enable a more comprehensive investigation of the evolutionary history, geographic origin, and dispersal routes of this pathogen.

Results

Screening and genome reconstruction

We screened 1800 human skeletal remains collected from various archeological sites distributed across present-day China (Supplementary Data 1 and Supplementary Fig. 1). Two individuals tested positive for M. leprae: XSSIM2B from the Shihuyao site in Xinjiang (Historical Era) and DLZCM04 from the Dangliuzhuangcun site in Shanxi (Iron Age), based on the presence of reads mapping to the M. leprae TN reference genome (NC_002677.1).

High-depth shotgun sequencing enabled the reconstruction of two ancient M. leprae genomes, A533 (from XSSIM2B) and D318 (from DLZCM04), with a mean coverage of 11.14× and 31.44×, respectively, uniformly distributed across the genome (Supplementary Data 2 and Supplementary Fig. 2). Read length distributions were consistent with ancient DNA fragmentation (Supplementary Data 2), and both genomes displayed characteristic deamination damage profiles with elevated C-to-T misincorporation frequencies at read termini (Supplementary Figs. 3, 4). Reads mapped to the reference genome showed a progressive decrease in the number of aligned reads as the number of allowed mismatches increased (Supplementary Fig. 4), consistent with authentic ancient DNA, in which genuine endogenous reads cluster tightly around the reference genome.

Direct ¹⁴C measurements on tooth samples confirmed their antiquity. After calibration, XSSIM2B yielded a date range of 1711-1550 calibrated years before present (cal BP), and DLZCM04 yielded 2340-2157 cal BP (Supplementary Fig. 5). Overall, these results indicate that A533 and D318 were the first ancient M. leprae genomes sequenced from eastern Eurasia.

Comparative genomic analysis

To evaluate evolutionary changes, we conducted comparative genomic analyses of the newly reconstructed genomes against all published ancient and modern M. leprae genomes (Supplementary Data 3). No notable differences were detected in virulence-associated genes between our ancient genomes and published comparators (Supplementary Data 4 and Supplementary Fig. 6). To explore potential functional shifts, we conducted a detailed single-nucleotide polymorphism (SNP) analysis on the newly reconstructed genomes. A total of 138 to 141 SNPs with potential effects were identified. Notably, A533 harbored seven unique SNPs and D318 harbored one (Supplementary Data 5). Four of the unique SNPs in A533 are located within coding genes associated with a transcription-repair coupling factor, a putative sugar phosphorylase, a NAD+ synthase, and a conserved hypothetical protein, suggesting minor changes in DNA repair and core metabolic pathways (Supplementary Data 5).

To further explore genomic variability, we quantified mutation frequencies across all protein-coding genes and intergenic regions in the full M. leprae dataset and applied Fisher’s exact test with Bonferroni correction to identify mutation-enriched regions. This analysis detected 36 mutation-enriched regions in our M. leprae dataset, comprising 25 coding genes and 11 intergenic regions (Fig. 1 and Supplementary Fig. 7). Notably, ML0411 in Branch 0 and ML1334 in Branch 4 were prominently identified as highly mutated genes. ML0411 encodes a serine-rich PPE family protein, a mycobacteria-specific class that may contribute to antigenic diversity and immune evasion; ML1334 encodes a conserved membrane protein featuring characteristic calcium-binding repeat domains and predicted transmembrane regions.

Fig. 1. Heatmap of mutation-enriched regions identified by Fisher’s exact test with Bonferroni correction in ancient and modern M. leprae genomes (n = 192).

This heatmap illustrates the presence (red) or absence of mutation-enriched regions across multiple genomes. Columns correspond to specific genomic regions identified as mutation-enriched regions, while rows represent M. leprae genomes. Along the y-axis, a phylogenetic tree depicts the evolutionary relationships among the genomes, providing context for the observed mutation-enriched regions patterns. Ancient samples are highlighted by red dots on the phylogenetic tree. This figure displays only mutation-enriched genes; a complete version, including intergenic regions, is provided in Supplementary Fig. 7.

Phylogenetic analysis

We reconstructed a maximum likelihood (ML) tree to assess the phylogenetic relationship between the newly reconstructed genomes and 36 published ancient genomes, along with 154 modern genomes^{4,6–11,14–21}. Only genomes with an average coverage of at least 5× were included (Supplementary Data 3). Phylogenetic reconstruction places both newly reconstructed genomes, A533 and D318, within the most basal lineage, Branch 0 (Fig. 2b and Supplementary Fig. 8). Branch 0 is clearly divided into two sister lineages: Branch 0A, comprising ancient and modern Eurasian strains, and Branch 0B, consisting of modern Pacific Island strains. The genome D318 from Shanxi occupies a basal position to the modern East Asian lineage, designated here as Branch 0A1. A533 from Xinjiang clusters with five medieval European genomes reported in previous studies^8,10. Within this ancient lineage, designated here as Branch 0A2, A533 occupies the basal position.

Fig. 2. Geographic distribution and phylogenetic relationships of Branch 0 M. leprae strains.

A The map shows the sampling locations and inferred dispersal routes of M. leprae Branch 0 lineages across Eurasia and the Pacific Islands. Arrows indicate hypothesized migration routes and do not represent directly observed movements. The basemap used in this figure is in the public domain and accessible through the Natural Earth website (https://www.naturalearthdata.com/downloads/10m-raster-data/). B Maximum Clade Credibility time-calibrated phylogenetic tree integrates both ancient and modern M. leprae genomes, with the newly reconstructed genomes from this study highlighted in red. Blue horizontal bars represent the 95% HPD interval of the most recent common ancestor, and the node labels indicate the posterior values. Color coding is consistent across (A) and (B).

Bayesian time-aware phylogenetic reconstruction was performed in BEAST to estimate divergence times and assess temporal signal. Temporal signal was evaluated using TempEst and further validated by a Date Randomization Test (DRT), in which the 95% highest posterior density intervals (HPDIs) of twenty randomized datasets did not overlap with the real estimates, confirming a detectable clock signal in the data (Supplementary Figs. 9, 10). Model selection using Bayes Factors strongly favored the Coalescent Bayesian Skyline model with a relaxed lognormal clock over alternative clock and demographic models (Supplementary Data 6). Time to the most recent common ancestor (tMRCA) of all M. leprae sequences was estimated at 4199 y BP (3340-5195 y 95% Highest Posterior Density (HPD)), approximately 300 years older than previous estimates⁸. The MRCA of branch 0 was estimated as 3589 y BP (2915-4492 y 95% HPD). Phylogenetic positions of A533 and D318 were consistent between the ML and Maximum Clade Credibility (MCC) time-calibrated phylogenetic trees, confirming their robustness.

Taken together, these findings indicate that the ancient East Asian M. leprae genomes represent deeply rooted evolutionary lineages and reveal previously unrecognized connections between ancient European and Xinjiang M. leprae strains.

Human genomic analysis

To characterize the host genetic backgrounds of the M. leprae-infected individuals, we analyzed the genomic structure of individual XSSIM2B from Xinjiang and individual DLZCM04 from Shanxi. The principal component analysis (PCA) shows that these two individuals occupy distinct positions in the Eurasian background (Supplementary Fig. 11a). XSSIM2B lies centrally along PC1 on the admixture cline bridging western Eurasian (Sintashta_MLBA/Gonur1_BA) and eastern Eurasian (Ulaanzuukh_SlabGrave) populations. This individual clusters with local Iron Age and Historic Era Xinjiang samples, indicating a mixed ancestry typical of Xinjiang. In contrast, DLZCM04 falls within the cluster formed by ancient Yellow River farmers spanning the Middle Neolithic to the Iron Age (‘YR_MN’, ‘YR_LN’, and ‘YR_LBIA’), reflecting a more homogeneous regional ancestry. ADMIXTURE analysis is consistent with the PCA result and further refines the binary genetic structure (Supplementary Fig. 11b). XSSIM2B carries a heterogeneous ancestry composed of western Steppe (blue), Central Asian Oasis (red), and Northeast Asian (green) components, reflecting the intense genetic admixture characteristic of the Tianshan corridor. DLZCM04, in turn, shows a homogeneous genetic profile dominated by the northeastern (green) and southern (yellow) East Asian-related component, with negligible Western Eurasian influence.

Taken together, these results demonstrate that while both individuals were victims of leprosy, they derived from different genetic and geographic backgrounds: XSSIM2B represents a locally admixed individual from the Central Asian corridor, and DLZCM04 represents an individual with relatively stable Yellow River-related ancestry.

Discussion

The geographical distribution and genetic diversity of ancient leprosy have long been of great interest, as they offer crucial insights into the pathogen's evolutionary trajectory and the historical movement of populations. However, the lack of ancient M. leprae genomes from eastern Eurasia has left this region as a major blind spot in our understanding of the pathogen’s global history. In this study, we reconstructed the first two ancient M. leprae genomes from eastern Eurasia, providing an important foundation for investigations into the transmission and evolutionary history of M. leprae in this region.

Comparative genomic analysis of the newly reconstructed ancient M. leprae genomes revealed only a small number of SNPs with potential functional effects, primarily involving DNA repair and basic metabolic pathways. Virulence-associated genes were overall highly conserved across the entire M. leprae dataset. Mutation-enriched regions, including ML0411 in Branch 0 and ML1334 in Branch 4, indicate localized, lineage-specific diversification rather than widespread adaptive change. These findings suggest that while M. leprae has maintained remarkable genomic stability over time, specific genes have experienced localized, lineage-specific diversification.

Overall, M. leprae has maintained remarkable genetic conservation since its MRCA (Fig. 2a), with ancient strains from both eastern and western Eurasia showing high similarity, suggesting that its geographic dissemination was possibly driven by human migration and contact rather than by adaptive evolution. This functional stability holds important implications for paleopathological research. While medieval Europe has left abundant textual and osteoarchaeological records of leprosy, corresponding references in ancient East Asia (such as ‘lai’, ‘li’ or ‘dafeng’) are sparse and ambiguous. Based on the genomic homogeneity revealed in this study, we speculate that the clinical manifestations of leprosy in ancient East Asia were likely comparable to those documented in medieval Europe, providing a potential theoretical framework for cross-regional comparisons of leprosy.

Phylogenetic reconstruction places both newly reconstructed genomes, A533 and D318, within the most basal lineage, Branch 0 (Fig. 2b, Supplementary Figs. 8, 12). The genome D318 from Shanxi occupies a basal position to the modern East Asian lineage (Branch 0A1), indicating a long-term, continuous endemic of this lineage in East Asia. A533, recovered from Xinjiang, clusters at the base of Branch 0A2, which otherwise comprises five medieval European genomes. This topology establishes a direct phylogenetic link between ancient M. leprae strains in the Tianshan region and those in Europe, suggesting that the evolutionary history of Branch 0 may not align with a traditionally Eurocentric perspective.

To contextualize this trans-Eurasian dispersal, we integrated the pathogen phylogeny with host genomic analysis. PCA and ADMIXTURE analyses of the genome of individual XSSIM2B (the individual infected with M. leprae strain A533) revealed a higher proportion of East Asian ancestry compared with earlier populations in the region (Supplementary Fig. 11). This genetic pattern is consistent with evidence from contemporaneous Xinjiang populations, reflecting extensive human movements and gene flow from eastern to western Eurasia during the historical period²².

This regional background of population mobility and admixture may have facilitated the dispersal of M. leprae. Against the backdrop of extensive interaction between eastern and western Eurasian populations, the Branch 0A2 lineage may have been introduced into the Tianshan region at a time that may coincide with the historical period or may even predate it. Subsequently, this lineage may have been gradually disseminated westward through continued human movement, eventually giving rise to the medieval European Branch 0A2 strains (Fig. 2a).

Our phylogeographic analysis provides a more detailed picture of the evolutionary history of M. leprae. Using a Bayesian time-aware phylogenetic framework, we estimated the tMRCA of Branch 0A, which includes ancient and modern Eurasian strains, and Branch 0B, which represents modern Pacific Islands strains, to be approximately 3589 y BP (2915-4492 y 95% HPD). This timeframe aligns with two major waves of human movement that originated from southern East Asia: the southward expansion of Austronesian-speaking populations and the northward movement of rice-farming groups^23,24. The distribution of Branch 0B strongly overlaps with the Austronesian expansion route into the Pacific Islands. Meanwhile, the ancestral Branch 0A strains (A533 and D318) found in northern East Asia correspond well with the period of “north–south population integration” around 4000-3000 years ago, during which northward-moving rice farmers likely carried M. leprae into northern regions.

Taken together, we hypothesize that Branch 0 initially circulated in southern East Asia. From there, it appears to have diverged: Branch 0B moved southward into the Pacific Islands, whereas Branch 0A expanded northward. Within Branch 0A, the sub-lineage represented by D318 (Branch 0A1) appears to have persisted in northern East Asia; while the sub-lineage represented by A533 (Branch 0A2) later spread westward into Europe (Fig. 2a), a dispersal pattern that is consistent with Model 2, at least regarding Branch 0: it may have originated in southern East Asia, with sublineage 0A2 subsequently dispersing to Europe. While our ancient genomes derive from northern East Asia, this hypothesis relies on the extant distribution of Branch 0B in the Pacific Islands and its correlation with Austronesian expansions. Therefore, we acknowledge the possibility of an alternative scenario: Branch 0 may have initially emerged in northern East Asia, later spread southward through interactions between northern and southern populations, and eventually reached the Pacific Islands. Future studies that recover ancient M. leprae genomes from other key regions, such as southern East Asia, will be essential for evaluating alternative origin scenarios and shedding light on the pathogen’s early transmission routes.

In conclusion, by confirming the ancient presence of M. leprae in eastern Eurasia, our study expands the pathogen’s known geographic scope beyond the traditional focus on European pandemics, offering a more comprehensive global perspective. This deeper temporal and geographical context also reshapes our understanding of the cultural responses to leprosy. Whether it is the “lai” or “li” in ancient Chinese records or the leprosaria in medieval Europe, the fear, stigma, and isolation associated with leprosy seem to be a cross-cultural theme throughout human history. Notably, our findings confirm the ancient presence of M. leprae in East Asia, which makes the absence of records describing large-scale leprosy epidemics in ancient Chinese sources, unlike the well-documented outbreaks in medieval Europe, particularly striking. We hypothesize that this contrast may reflect differences in public health practices, societal responses, or genetic susceptibility among populations, but further interdisciplinary evidence will be required to evaluate these hypotheses. By providing molecular evidence, our study highlights the value of integrating archeology, history, and genetics to better understand the socio-cultural history of leprosy.

Materials and methods

Archeological context

All ancient samples screened in this study were obtained from human skeletal remains excavated from archeological sites across East Asia. The samples consisted primarily of teeth, with a small number of bone samples also included. We screened 1800 human skeletal remains from sites across East Asia (Supplementary Fig. 1), and detected authentic M. leprae DNA in two individuals: XSSIM2B and DLZCM04.

The individual XSSIM2B was excavated from the Shihuyao site, located in Shihezi, Xinjiang, China. The site was excavated in 2017 by the Xinjiang Institute of Cultural Relics and Archeology, the Shihezi Bureau of Culture and Sports, and the Xinjiang Corps Museum, marking the first large-scale archeological excavation conducted in the Manas River Basin. Based on tomb structure, unearthed artifacts, and radiocarbon dating results, these burials can be divided into three periods: the Late Bronze Age, the Early Iron Age, and the Tang to Mongol-Yuan period²⁵.

The individual DLZCM04 was excavated at the Dangliuzhuangcun site in Datong, Shanxi, China. The site was excavated in 2025 by the Datong Institute of Archeology. Based on the characteristics of the archeological remains, the cemetery is estimated to date from the Warring States to the Qin dynasty.

The radiocarbon dating was carried out in the Graphitization Lab, Guangzhou Carbonage Technology Co., Ltd. (Supplementary Fig. 5). The ¹⁴C dates were calibrated using OxCal v4.4.4²⁶ using the IntCal20 atmospheric curve²⁷.

DNA extraction and library preparation

DNA was extracted in the cleanroom facilities dedicated to ancient DNA work at Jilin University's ancient DNA laboratories. Briefly, after grinding, 50 mg of bone powder were incubated in 900 µL ethylenediaminetetraacetic acid (EDTA; 0.5 M) and 100 µL proteinase K (20 mg/mL) under gentle rotation at 50 °C overnight. After centrifugation, the supernatant was combined with binding buffer and subsequently purified using the MinElute PCR Purification Kit (QIAGEN) following the manufacturer’s instructions. DNA was eluted in a final volume of 70 μl per sample, of which 20 μl were used for downstream library preparation. Double-stranded, non-UDG-treated DNA libraries were constructed from the extracted DNA and negative extraction controls were included at each step to monitor potential contamination. Shotgun sequencing was performed on an Illumina HiSeq X10 platform in the 150-bp paired-end design.

Screening of M. leprae DNA

Raw sequencing read quality was evaluated using FastQC v0.11.8 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Adapter sequences were removed, and paired-end reads were merged using AdapterRemoval v2.3.1²⁸, retaining reads with a minimum length of 30 bp and a minimum base quality score of 20. The processed reads were first mapped to the human reference genome (hs37d5) using EAGER v1.92.50²⁹, and non-human reads were converted back into FASTQ format. Taxonomic classification of the resulting reads was then followed by metagenomics analysis using MALT v0.5.3³⁰. The output files (rma6) were examined in MEGAN v6.24.4³¹ and further processed with HOPS v0.34³², searching against a custom database comprising genomes from a wide range of commonly encountered bacterial and viral taxa.

Data processing of M. leprae

Following the identification of leprosy-positive samples, dental pulp was re-sampled for library construction, and deep shotgun sequencing was subsequently performed. The raw sequencing data were aligned to the M. leprae reference genome (NC_002677.1) using the BWA-aln algorithm v0.7.12 with parameters “-n 0.01 -l 32”, as implemented in the EAGER v1.92.50²⁹. The PCR duplicates were removed using DeDup v0.12.5²⁹, and ancient DNA damage patterns were characterized using mapDamage v2.0.6³³.

The Genome Analysis Toolkit (GATK) UnifiedGenotyper v3.5³⁴ was used to generate a mapping assembly and SNP calling, using the “EMIT_ALL_SITES” option. MultiVCFanalyzer v0.85.2³⁵ was used for collating homozygous SNPs (the position was covered by at least 5 reads, the genotype quality of the call was at least 30, and the minimum SNP allele frequency was 90%). Positions located in repetitive regions, rRNA and tRNA regions and the positions covered by sample SK12⁴ were excluded. Finally, the homozygous SNPs were annotated using SnpEff v4.3³⁶.

Virulence analysis and identification of mutation-enriched regions

We compiled a bed file containing the coordinates for known virulence factors of M. leprae TN. Using the output BAM files, we calculated the percentage of each gene covered using bedtools v2.25.0³⁷. To identify mutation-enriched regions, we determined mutation frequencies across all protein-coding genes and intergenic regions in our M. leprae dataset. For each gene and intergenic region, we performed a Fisher’s exact test to compare the observed mutation frequency with that of other regions, and statistical significance was evaluated using a Bonferroni correction for multiple testing. Regions showing significantly elevated mutation frequencies (adjusted p-value < 0.05) and at least a twofold increase in mutation rate were classified as mutation-enriched regions (Supplementary Data 7). Detailed SNPs within mutation-enriched regions are provided in Supplementary Data 8.

Processing of published genomes

All previously published M. leprae samples were processed using the same approach described above. For strains lacking sequencing reads (TN, Br4923), reads were simulated using Genome2Reads v1.0¹⁰. The same was done for Mycobacterium lepromatosis, which was used as the outgroup.

Phylogenetic analysis

The phylogenetic placement was performed based on the SNP alignment. After removing sites with less than 80% coverage across the dataset, the remaining variant positions were used to generate a ML phylogenetic tree using IQ-TREE v1.6.10³⁸ and 1000 bootstraps using the following command: iqtree -s snpAlignment.fasta -nt AUTO -bb 1000. The built-in ModelFinder module selected TVM + F + ASC + R3 as the best-fit substitution model for maximum-likelihood phylogenetic reconstruction. The ASC correction accounts for ascertainment bias arising from the exclusion of invariant sites in the SNP alignment. We estimated tMRCA and substitution rates through application of Bayesian methods using BEAST v2.6.6³⁹. To ensure the robustness of parameter estimation and to account for model uncertainty, we systematically compared a range of molecular clock and demographic models. Specifically, we evaluated both strict and relaxed lognormal clock models along with four demographic models: Coalescent Bayesian skyline, Coalescent Constant Population, Coalescent Exponential Population and Birth Death Model. Comparing the ratios of marginal likelihoods using Bayes factors strongly supported the Coalescent Bayesian skyline with a relaxed lognormal clock over the other models (Supplementary Data 6). We then tested whether there is a correlation between either genetic and geographical distance or genetic and temporal distance using TempEst v1.5.3⁴⁰. To further evaluate whether tip calibration alone could provide reliable temporal inference, we conducted a DRT. Using the best-supported model, twenty randomized datasets were generated with the R package TIPDATINGBEAST⁴¹. For the parameter of interest (tree height), the 95% HPDIs of the randomized datasets did not overlap with the real estimates (Supplementary Fig. 10). These results indicate that the data contain a detectable temporal signal.

The dated phylogeny was performed under the Coalescent Bayesian Skyline model with a relaxed lognormal clock and a GTR substitution model. After excluding the hypermutator strains⁶, a concatenated SNP alignment of 2718 sites was used. The total Markov chain length was set to 300,000,000, and computed sampling every 30,000 steps. The MCC time-calibrated phylogenetic tree was chosen using TreeAnnotator v2.6.6³⁹ with the first 10% burn-in. All the parameters have a higher ESS value than 200. Because the alignment contained only variable sites, substitution rates estimated from the SNP dataset were rescaled to genome-wide rates by accounting for the proportion of variable sites in the reference genome. Furthermore, dates of strains for discrete phylogeographic analysis in the BEAST v1.10.4⁴² to speculate the spread routes of M. leprae (Supplementary Fig. 12). The resulting trees were visualized with FigTree v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/).

Population genomic analysis

We contextualized the genetic background of the samples by performing PCA and ADMIXTURE analyses. The dataset was constructed by merging our samples with selected worldwide populations from the AADR v54.1.p1⁴³ database (Supplementary Data 9). PCA was computed using smartpca v16000⁴⁴. Ancient samples were projected onto the principal components inferred from modern Eurasian populations using the lsqproject: YES option to mitigate missing data bias. For ancestry estimation, we utilized the ML-based tool ADMIXTURE v1.3.0⁴⁵. The analysis was performed in unsupervised mode for K values ranging from K = 2 to K = 20. SNPs with a minor allele frequency below 1% were excluded, and linkage disequilibrium pruning was applied using the parameters --indep-pairwise 200 25 0.2 in PLINK v1.90⁴⁶. The cross-validation (CV) error procedure was utilized to assess the best-fitting model for the data (Supplementary Fig. 13).

Ethics statement

Due to the nature of the samples analyzed, ethical approval was not required. All samples were collected with permission from the representative bodies/host institutions.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Y.C., B.S., and Z.Z. conceived the study. Y.N., J.Z., and X.X provided archeological information and archeological materials. S.F. and X.Z. performed the laboratory work. S.F., X.L., X.Z., and D.Z. performed the analyses. S.F., B.S., and Y.C. wrote the manuscript.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editor: Rupali Sathe. A peer review file is available.

Funding

This work was supported by National Natural Science Foundation of China (Grant No. 42372017 and 82100076).

Data availability

Raw FASTQ files for M. leprae generated in this study have been deposited in the Genome Sequence Archive in National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (https://ngdc.cncb.ac.cn/gsa/browse/CRA034742).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s42003-026-10683-1.