Questions about comparative genomics of SARS coronavirus isolates

Sir

YiJun Ruan and colleagues' analysis (May 24, p 1779)¹ of the comparative genomics of coronavirus isolates from 14 patients with severe acute respiratory syndrome (SARS) is to be welcomed. Two questions, however, are begged by their survey.

The first question concerns genomic evolution of the SARS virus. The single-stranded RNA genome of the SARS virus assures genetic lability under moderate selective pressures and high rates of genetic drift. Droplets of respiratory-tract fluids in nasopharyngeal aerosols have volumes of 10⁻⁶–10⁻⁷ mL, so that the expected SARS virion population in a single droplet is between 0·1 and 10, even for patients with maximum degrees of viraemia. Thus, most infective doses are probably in the range of 10–10³ virions, whereas a patient's SARS virion-load at peak viraemia is about 10¹². Considering the 10³ second effective serum lifetime of a virion, a patient's 10⁵ second viraemic-term may see generation of about 10¹⁴ virions, or about 10¹² infective doses. Even with 10² successfully infective virions sourced per infected cell—a conservative upper-estimate—there are at least half a dozen viral generations per case history, or about 20 viral generations across the three case-history generations studied by Ruan and colleagues. Since the observed per-base replication error-rate of RNA polymerases is about 310⁻⁵ and the SARS viral genome has about 30 000 bases, the expected genome copying error-rate is about one base per viral generation, or about 20 base errors of aggregate genetic drift after 20 generations, roughly congruent with the 16 “observed twice” single nucleotide polymorphisms reported by Ruan and colleagues.

Crucially, however, these 14 case-isolates represent infections during March and early April, 2003, whereas Ruan and colleagues relate that the SARS epidemic began in Guangdong province in November, 2002, so that it has been propagating and mutating at least four—and perhaps as much as five—times longer than is represented by the time-span of all the analysed cases. Where is the four–fold larger genetic drift? Specifically, why is there such close genomic similarity between the Singapore cases and all of the overseas cases? Unless these all trace to the same index case in early March, which seems unlikely, their close genomic similarity is quantitatively inexplicable.

The second question concerns the ease with which the SARS virus propagates in vitro, a quite unusual, if not unique, characteristic for known human coronaviruses. This issue is at best thoroughly puzzling and at worst deeply troubling. How do Ruan and colleagues think that this set of viral propagation peculiarities arose?