Statistical modelling of virus growth curves indicates that CH058 CC has a lower growth rate than its TF progenitor but is not otherwise more sensitive to IFN.

(A) TMZR5 cells were pre-stimulated with different doses of IFNα14 for 24 hours prior to infection with CH058 TF or CC virus and sampled daily to monitor virus spread. GFP-positive cells were enumerated using flow cytometry. Blue and orange points show means (+/- standard error) across 4 experimental replicates, while translucent points show individual observations. Viral spreading replication experiments took place on two occasions, a typical result is shown. (B) Model fit for the differential sensitivity model, in which viruses are allowed to vary in both baseline growth rate (i.e., growth rate in the absence of interferon) and in their sensitivity to interferon. (C) Model fit for the constant sensitivity model, in which viruses differ only in their baseline growth rate. In both B and C, opaque points show predictions from the fitted model, along with 95% confidence intervals. Translucent points and lines underneath show the observed (experimental) replicate growth curves also present in A. For each treatment, only the initial timepoints until at least one replicate curve decreased by >30% relative to its preceding observation were used in model fitting (i.e. points after HIV-1 had overwhelmed the culture were not considered). (D) Effect sizes for both differential and constant sensitivity models (maximum likelihood estimates and 95% confidence intervals). Model fit was compared by likelihood ratio test and AIC, with results shown in the legend. (E) Illustration of effects making up the fitted growth rates in the differential sensitivity model. Effects were modelled as additive, allowing us to separate discrete contributions to the growth rates needed to recapitulate the in vitro data as illustrated in B. Points show maximum likelihood estimates, while error bars show 95% confidence intervals. An inset in the final panel (right) illustrates expected patterns under different hypothesized scenarios; in particular, if the CC virus was more sensitive to IFN, its achieved growth rate would have diverged from that of the TF virus at increasing IFN doses.