Molecular determinants of plaque size as an indicator of dengue virus attenuation

The in vitro replication rate of pathogens is a common measure of fitness. A rapidly replicating virus in culture is interpreted as having greater fitness than slower growing strains1,2,3. Correspondingly, the size of plaques, which are caused by necrosis or apoptosis upon infection of a cell monolayer, is a common proxy measure of viral fitness. This empirically derived measurement of fitness is used in the development of live attenuated vaccines (LAVs), where strains that produce small plaques are repeatedly selected and purified for clinical development1,4,5. While this empirical approach in vaccine development has yielded successful vaccines including measles6, mumps7 and yellow fever8,9, it has also led to failed clinical trials, where candidate vaccines caused acute disease. These undesired outcomes suggest that in vitro replication rate alone is not a reliable indicator of a virus’ clinical fitness10,11. We recently showed that an epidemiologically fitter strain of dengue virus (DENV) that emerged during the 1994 dengue outbreak in Puerto Rico paradoxically replicated more slowly than the virus it displaced12. The fitter virus “sacrificed” its genomic RNA to form more subgenomic RNA, which downregulated type-I interferon expression by binding a RIG-I signalling pathway intermediate. The reduced type-I interferon signalling likely then allowed DENV to spread to more susceptible cells, eventually reaching higher viraemia levels for more efficient vector-borne transmission12. Our findings thus indicate that viral replication rate is not invariably correlated with viral fitness in an epidemiological or clinical setting. To gain clinically informed insights into the molecular determinants of attenuation, we examined a LAV candidate that caused acute illness in phase I clinical trials. The strains in the tetravalent dengue LAV developed at Mahidol University were selected for passaging primarily based on the small plaque phenotype; DENV in cells surrounding small plaques were purified and expanded after every five passages4. Other empirically defined markers used to determine attenuation were temperature sensitivity and reduced neurovirulence in suckling mice4,5. Despite this uniform approach, the vaccine candidates had contrasting clinical outcomes. DENV serotype 3 (DENV-3) vaccine strain DENV-3 PGMK30FRhL3 (hereafter referred to as PGMK30) was generated by passaging wild-type DENV-3 16562 thirty times in primary green monkey kidney cells (PGMK) and three times in primary foetal rhesus lung (FRhL) cells. In a human safety trial, however, all volunteers who received tetravalent LAV formulation containing PGMK30 developed symptoms and signs consistent with acute dengue, with detectable DENV-3 viraemia10. Following the disappointing clinical trial results, this DENV-3 vaccine strain was further passaged in Vero cells using the same Mahidol University protocol to yield the Vero-derived Vaccine (VDV3) strain, which again met the same phenotypic criteria for attenuation previously used. When VDV3 was administered to healthy volunteers, it again caused disease in recipients13,14. In contrast, the DENV-2 vaccine strain, PDK53, which was generated from wild-type virus (16681) after 53 passages in primary dog kidney (PDK) cells was clinically safe and has since been used as a backbone to construct chimeric vaccines for the other DENV serotypes15,16. Although follow up investigations on PGMK30 found differences in host cell responses compared to wild-type DENV-313, the underlying reasons for why PGMK30 caused acute illness in vaccine recipients despite meeting the empirical criteria for attenuation were never determined. Understanding the mechanisms that operate to yield divergent clinical outcomes despite the similar small plaque phenotype in these two LAV candidates could enable a more objective and accurate approach to selecting attenuated strains for development into LAVs.