Genomic diversity of Yersinia pestis from Yunnan Province, China, implies a potential common ancestor as the source of two plague epidemics – Communications Biology

Sampling of Y. pestis in Yunnan Province spanning two epidemic-silent cycles

In this study, 356 strains were isolated from 11 prefectures or cities in Yunnan Province, China, and Myanmar, near the China-Myanmar border, between 1952 and 2016, spanning two epidemic-silent cycles (Fig. 1a). Among them, 70 strains were isolated from the Epidemic1 period, one from Silent1, 282 from Epidemic2, and three from Silent2 (Table 1). To facilitate subsequent analysis, these strains were classified into Cycle1 (Epidemic1 and Silent1) and Cycle2 (Epidemic2 and Silent2) groups, according to their sampling dates (Fig. 1b). Specifically, we observed the greatest number of human cases during Cycle1 in 1954, and following a 25-year silent period, the plague epidemic reemerged in 1982, and then spread throughout the province, peaking in 1983, 1990, and 1996^15,16.

Genomic diversities of Y. pestis strains during the two cycles

To infer the genomic diversities and population structure of Y. pestis in Yunnan during the two epidemic-silent cycles, we reconstructed the phylogeny of 356 isolates from the Yunnan R. tanezumi plague. Ninety-six percent of the strains (343 strains) were assigned to phylogroup 1.ORI2, which is associated with the third historic pandemic (Supplementary Fig. 1a). We excluded 13 strains scattered on other branches from further analysis and focused on the 343 strains in 1.ORI2 (Supplementary Data 1). The reconstructed phylogeny, based on 263 high-quality single nucleotide polymorphisms (SNPs), showed that these 343 1.ORI2 strains could be subdivided into two main clades, Sublineage1 and Sublineage2 (Figs. 2 and 3a). We discovered that Sublineage1 predominantly included strains from Cycle1 (59/60, 98.33%), characterized by three clade-specific SNPs, whereas most strains in the Sublineage2 clade came from Cycle2 (280/282, 99.29%), with one clade-specific SNP. Notably, four strains were isolated during two plague silent periods. Among them, one (Silent1, Cycle1) was located in the Sublineage1 clade and descended from Epidemic1 strains, and the other three (Silent2, Cycle2) were clustered together and located in the Sublineage2 clade and descended from Epidemic2 strains (Fig. 3a).

Large and red text, Sublineage1 and Sublineage2; smaller letters, sub-clades (for example, 1.ORI2.2.3.1). Circle colors indicate the sampling locations.

a A minimum spanning tree of 343 Yunnan strains (rooted with CO92 strain in 1.ORI1). Circle colors indicate the sampling date, and their size is proportional to the number of strains. Large, red text, Sublineage1 and 2; smaller letters, sub-clades (for example, 1.ORI2.2.3.1). Between nodes, numbers of single nucleotide polymorphisms (SNPs), apart from one SNP, are indicated by thick and thin black lines, respectively. b Pairwise SNP distance distributions within the two main clades. The center line of the boxplot is the median, bounds of the box are the first and third quartiles, circles are the outliers. The plot is based on result in Supplementary Data 3.

The phylogeny indicated that the two sublineages were sister clades because they split from a common ancestor, rather than one descending from the other. In addition, the two sublineages showed dissimilar genomic diversity. Although collected within a relatively short time interval, the pairwise SNP distance among strains in Sublineage1 (median = 6) was higher than those in Sublineage2 (median = 5; p < 0.001, Wilcoxon test) (Fig. 3b). Therefore, we inferred that the two plague epidemic periods in Yunnan were caused by two distinct lineages with a common ancestor and evolved in parallel.

Spread and origin inference of Y. pestis in Yunnan plague epidemic

To infer the spread pattern of Y. pestis during the two epidemics in Yunnan, we integrated the information of phylogenetic relationships, sampling dates, and geographic distribution of the 343 Y. pestis strains. Phylogeographic analysis revealed that strains of Cycle1, mostly corresponding to Sublineage1, were distributed in western Yunnan, including Dehong Dai-Jingpo Autonomous Prefecture (DH), the Dali Bai Autonomous Prefecture (DL), and Baoshan City (B) (Figs. 1 and 2). Most strains from Cycle1 were isolated from DL (52/61, 85.25%); only 14.75% were from DH and B, suggesting that cross-region transmission events occurred among these regions during the Epidemic1 period.

Sublineage2, including >99.29% of Cycle2 strains, can be further subdivided into four sub-clades (1.ORI2.2.3.1–1.ORI2.2.3.4) (Figs. 2 and 3a). Strains of the most basal lineage (1.ORI2.2.3.1) were mainly distributed in western Yunnan (Figs. 2 and 4a); 1.ORI2.2.3.2 strains were distributed in western and central Yunnan, 1.ORI2.2.3.3 strains were distributed in western, central, southern, and eastern Yunnan with a relatively shorter sampling time interval but a larger geographical span than the former two lineages; 1.ORI2.2.3.4, had a wide geographical range similar to that of 1.ORI2.2.3.3 (Figs. 2 and 4a). Bayesian-based phylogeographic analysis showed that the two widely distributed lineages, 1.ORI2.2.3.3 and 1.ORI2.2.3.4, exhibited a transmission trend from west to east (Fig. 4b, c). Additionally, most of the strains located on the deep branches were isolated from DH, a prefecture located in western Yunnan bordering northern Myanmar, where the plague epidemic lasted for the longest period. Thus, we inferred that the Epidemic2 plague, might have originated from DH Prefecture or its bordering countries, and then spread from the western to the central, southern, and eastern regions of Yunnan.

a Geographic source of Sublineage2 strains. Spread routes are indicated by red lines with an arrow. Solid lines indicate direct Yunnan spread event routes based on phylogeographic analysis, while dotted lines indicate the possible or indirect spread event routes. The blue pentagon indicates the possible origin region of plague epidemics in Yunnan, which encompasses the most diverse strains. Filled circles represent numbers of strains whose groupings are indicated by colors (see legend at the right). Maximum clade credibility phylogenetic trees of 1.ORI2.2.3.3 (b) and 1.ORI2.2.3.4 (c) strains. Branch lengths are scaled to years. Branch colors indicate the sampling locations. Branch widths represent posterior probability value.

To further investigate whether the plague epidemic in Yunnan was caused by strains introduced from other countries, we collected 470 publicly available genomes (Supplementary Data 2) and used them to build a phylogeny with the newly sequenced genomes from this study (Supplementary Data 1). The results indicated that 1.ORI2 included 350 Yunnan strains, one Myanmar isolate, five Vietnam isolates, 57 South America isolates, one Zimbabwe isolate, and one Guangxi isolate from China (Supplementary Figs. 1a and 2). A high-resolution phylogenetic tree, containing only 1.ORI2 strains showed that the five Vietnam strains^19,20, isolated in 1962 (Saigon-Nhatraung-62-3), 1967 (140-Dalat), 1986 (P-13273, P-13272), and 1988 (P-14709) were grouped together and formed a distinct clade (Supplementary Figs. 1b and 2, Supplementary Data 2). The Vietnam clade is embedded within the Yunnan clade in the overall phylogeny. The sampling dates of the five Vietnam strains correspond with the Silent1 and Epidemic2 periods in Yunnan, leading us to hypothesize that the introduction to Vietnam likely originated from Yunnan. However, due to the limited sampling in Southeast Asian countries, it is also conceivable that the common ancestor for both the Vietnam clade and Yunnan’s Sublineage1 and Sublineage2 might have originated elsewhere in Southeast Asia.