On the origin of smallpox: correlating variola phylogenics with historical smallpox records

Abstract

Human disease likely attributable to variola virus (VARV), the etiologic agent of smallpox, has been reported in human populations for >2,000 years. VARV is unique among orthopoxviruses in that it is an exclusively human pathogen. Because VARV has a large, slowly evolving DNA genome, we were able to construct a robust phylogeny of VARV by analyzing concatenated single nucleotide polymorphisms (SNPs) from genome sequences of 47 VARV isolates with broad geographic distributions. Our results show two primary VARV clades, which likely diverged from an ancestral African rodent-borne variola-like virus either approximately 16,000 or approximately 68,000 years before present (YBP), depending on which historical records (East Asian or African) are used to calibrate the molecular clock. One primary clade was represented by the Asian VARV major strains, the more clinically severe form of smallpox, which spread from Asia either 400 or 1,600 YBP. Another primary clade included both alastrim minor, a phenotypically mild smallpox described from the American continents, and isolates from West Africa. This clade diverged from an ancestral VARV either 1,400 or 6,300 YBP, and then further diverged into two subclades at least 800 YBP. All of these analyses indicate that the divergence of alastrim and variola major occurred earlier than previously believed.

Fig. 1.

Geographic location of VARV isolates. The VARV genome sequences used in this study are from ref. and GenBank (accession nos. DQ 441438, DQ 441439, and DQ 441441); the isolates are derived from geographically distinct smallpox outbreaks between the mid-1940s and 1977. Outbreak locations are grouped and color keyed by regions; indigenous country cases are labeled as circles, and imported cases are labeled as squares. All the European isolates are imported cases from endemic regions, of which two had known origins (YUG72_164 originated in the Middle East, and UNK52_but originated in South America). The high similarity between IND64_vel4 and IND64_vel5 (4) to Central Asia/Middle East isolates suggests that the two India 1964 VARV isolates probably were imported cases that were isolated from a Christian Mission Hospital in Vellore, India (41).

Fig. 2.

Tree topologies generated by using maximum likelihood analysis of VARV plus CMLV and TATV SNP matrices. (A) The VARV portion of rooted VARV phylogram was expanded horizontally to display the details of VARV subclades and Bootstrap values. (B) The VARV phylogram generated from concatenated VARV plus CMLV and TATV SNP matrices, using TATV as an outgroup. The major clades and subclades were designated by name. The consistency index (CI) is 0.97, and the retention index (RI) is 0.98.

Fig. 3.

Hypotheses of the origin of VARV. Event 0, The high similarity of VARV to CMLV and TATV genome sequences suggests that TATV and CMLV share a more recent common ancestor with VARV than do other known Orthopoxvirus species. Event 1, hypothesis A: An ancestral VARV diverged into two primary clades (P-I and P-II) and evolved independently in the Old World and the New World by following its human hosts. Event 1, hypothesis B: The divergence of P-I/P-II in Africa. Event 2, hypotheses A and B: The long single branch and subsequent radiation of P-I suggest that the ancestral VARV of P-I probably originated in Northeast Asia and that the smallpox endemics of P-I started to spread to East Asia, the Middle East, and India as local populations became sufficiently large. Event 2a, The diversification and migration of P-I throughout Asia. Event 2b, Western exploration brought VARV major to South Africa from the Indian subcontinent, which then spread northward to Central Africa and the Horn of Africa (Somali Peninsula). Events 3 and 4, hypothesis A: The diversification of P-II in the New World followed by its reintroduction into Africa; hypothesis B: The diversification of P-II in Africa, followed by the later introduction of one subclade into the New World.