Population Genomics of Mycobacterium tuberculosis in Ethiopia Contradicts the Virgin Soil Hypothesis for Human Tuberculosis in Sub-Saharan Africa

Abstract

Colonial medical reports claimed that tuberculosis (TB) was largely unknown in Africa prior to European contact, providing a "virgin soil" for spread of TB in highly susceptible populations previously unexposed to the disease [1, 2]. This is in direct contrast to recent phylogenetic models which support an African origin for TB [3-6]. To address this apparent contradiction, we performed a broad genomic sampling of Mycobacterium tuberculosis in Ethiopia. All members of the M. tuberculosis complex (MTBC) arose from clonal expansion of a single common ancestor [7] with a proposed origin in East Africa [3, 4, 8]. Consistent with this proposal, MTBC lineage 7 is almost exclusively found in that region [9-11]. Although a detailed medical history of Ethiopia supports the view that TB was rare until the 20(th) century [12], over the last century Ethiopia has become a high-burden TB country [13]. Our results provide further support for an African origin for TB, with some genotypes already present on the continent well before European contact. Phylogenetic analyses reveal a pattern of serial introductions of multiple genotypes into Ethiopia in association with human migration and trade. In place of a "virgin soil" fostering the spread of TB in a previously naive population, we propose that increased TB mortality in Africa was driven by the introduction of European strains of M. tuberculosis alongside expansion of selected indigenous strains having biological characteristics that carry a fitness benefit in the urbanized settings of post-colonial Africa.

Figure 1

Topology Obtained by Bayesian Analyses as Described in Experimental Procedures Note that no branch length information is used. The topology is highly congruent with the topology from neighbor-joining analysis (see Figure S1). For both analyses, bootstrap values and Bayesian posterior probability values were higher than 95% in almost all nodes. The tree is rooted with Mycobacterium africanum strains that are the most basal clades within the Mycobacterium tuberculosis complex [3]. External branches are color coded according to the geographic region of the patient from which the isolate was collected. Black dots indicate that the strain was isolated in Ethiopia. The groups identified within each lineage correspond to the groups delineated using a set of diagnostic SNPs as explained in Experimental Procedures. Groups with no diagnostic SNP or that do not form a monophyletic group within the sub-lineage were labeled LX.ETHX. Geographic region: CAM, Central America; CAS, Central Asia; EAF, East Africa; EAS, East Asia; EEUR, Eastern Europe; EUR, Europe; NAM, North America; SAF, South Africa; SAM, South America; SAS, South Asia; SEA, Southeast Asia; WAF, West Africa; WAS, West Asia.

Figure 2

Principal-Component Analysis Using the SNP Matrices Derived from Whole-Genome Analysis (A) The PCA shown is for lineage 4; the corresponding PCA for lineages 3 and 1 are shown in Figures S2 and S3, respectively. The colors represent strains with known African origin (blue) or Eurasian and American origin (red) from a global reference collection and strains with Ethiopian origin (green). (B) The correlation between genetic diversity within a geographic area and the geographic distance from Addis Ababa. Error bars indicate the variance in diversity indices within a region. Geographic region: AF, Africa; SAS, South Asia; EUR, Europe; SEA, Southeast Asia; EEUR, Eastern Europe; EAS, East Asia; CAS, Central Asia; AM, America.

Figure 3

Contour Maps Derived from Point Estimation of the Frequency of Each Lineage in the Different Sampling Locations The results are based on a previously published M. tuberculosis collection [9]. Note that there are different scales for each lineage, reflecting their maximum and minimum frequency across the country.

Figure 4

Representation of Dating Events for MTBC Sub-lineages Included in This Study Dots (•) represents the age of the most recent common ancestor of the different Ethiopian sub-lineages circulating today. Diamonds (♦) represent the split of those Ethiopian groups from the closest non-Ethiopian strains in the global dataset described in [3]. Note that for lineage 7, the split from other MTBC was the third millennium BCE and is not represented. Light gray highlights the time of first sporadic European contacts in Ethiopia (16^th–19^th centuries CE). Dark gray highlights the time frame for a more continuous contact between Ethiopia and foreign nations (19^th century CE and onward). The data presented show only the results for the MTBC-6 model in which the whole complex is predicted to be around 6,000 years old [25]; the results for the MTBC-70 model where all lineages were already established by 4,000 years ago are not shown. (See Table S2.)