Life cycle process dependencies of positive-sense RNA viruses suggest strategies for inhibiting productive cellular infection

Positive strand (+)RNA viruses are the most common and clinically important human pathogens. Their life cycle processes are broadly conserved across many virus families but they employ different life cycle strategies for their growth in the cell. Upon RNA genome release into the cytoplasm post cellular entry, viral translation generates structural and non-structural proteins that induce intracellular remodelling, forming membrane compartments that foster viral replication leading to virus particle formation. We present a generalized dynamical model for intracellular (+)ssRNA virus growth that accounts for these critical steps. Our model can capture experimental growth dynamics for several RNA viruses as well as parse the effect of viral mutations and host cell permissivity. We show that Poliovirus (PV) employs rapid replication and virus assembly whereas Japanese Encephalitis virus leverages its higher rate of translation and efficient host membrane reorganization for enhanced viral dynamics compared to Hepatitis C virus. Since the slow membrane reorganization represents a crucial bottleneck for replication, stochastic simulations demonstrate that an infection event, even with multiple viral genomes, can go to extinction if all seeding viral RNA degrade before establishing robust viral replication. We estimate this probability of productive cellular infection, termed ‘Cellular Infectivity (Φ)’ using stochastic simulations. Φ varies for a virus-host pair with initial virus seeding and life cycle perturbations like increase in cytoplasmic RNA degradation and delay in compartment formation can reduce infectivity. Extent of synergy among these parameters while seemingly diverse for viruses is defined by Φ. Therefore, our model suggests new avenues for inhibition of viral infections by targeting early life cycle bottlenecks.

infection. Yet, they also display striking similarities in cellular life cycle dynamics, largely imitating 31 mechanisms of replication, translation, virus assembly as well as comparable interactions with the host 32 cell machinery. This has motivated the search for universal virus infection features that can be exploited 33 as broad spectrum approaches to inhibit them. 34 One common characteristic of most (+)RNA viruses is the induction of significant alterations of the 35 intracellular host membrane [1,2,3,4]. The vesicular membranous structures (also known as replica-36 rate, 1 h −1 [16,17]). Our estimates for (+)RNA export out of replication compartments is also similar 119 to previous estimates [16]. However, our estimates of virus production is 50-fold faster since the unac-120 counted delay in CM formation is subsumed in the assembly rate of the existing model [22]. Overall, we 121 recapitulate HCV experimental observations not built a priori into the model as well as match previous 122 estimates for comparable parameters. 123 To further validate our model, we evaluate the life cycle dynamics of subgenomic HCV (sgHCV) 124 transfected into Huh7 derived (Huh7-Lp and Huh7-Lunet) cells using our model [17] ( Figure S2). Our 125 estimates for k t , k r , N C (carrying capacity of RC CM ) and k c (the rate of formation of RC CM ) for 126 the subgenomic transfection are similar to corresponding estimates for full-genomic infection (Table S2, 127 Table S3) suggesting robustness of our model across different experimental systems for HCV. However, 128 the sgHCV system exhibits delayed RC formation and faster (+)RNA export out of CM likely due to lack 129 of structural proteins, transfection induced cellular artifacts or additional pre-processing of transfected compared to Huh7-Lp [46], suggesting efficient replication compartmentalization leads to higher cellular 133 permissivity (Table S3). 134 Comparative analysis of monopartite (+)RNA viruses 135 To understand the differences in life cycle traits among (+)RNA viruses, we further evaluate our model 136 with two distinct class of viruses for which comprehensive viral dynamics data exists, namely Enterovirus 137 (PV [19]) and Flavivirus (JEV [31]). Comparison of life cycle process parameters (Fig 2, Table S2) shows 138 that the replication rate (k r ) and export rate of (+)RNA from compartment (k e ) exhibit virus family 139 specific trends (Fig 2c). For example, Poliovirus RNA replicates rapidly (60-fold higher k r ) and re-enters 140 the cytoplasmic pool faster (16-fold higher k e ) than Flaviviridae family (HCV and JEV production and an associated early induction of membrane re-organization could be responsible for the 152 faster functional maturation of CM for JEV.

153
Virus assembly and generation defined by k a is significantly (> 10 5 fold) faster for PV compared  highlights three distinct dynamical regimes for the viruses (Fig 3 a). The initial establishment (E) phase 174 is sensitive to the delay in formation of CM (τ F ), and displays minimal replication due to shortage of 175 CM. The next growth (G) phase represents the rapid increase in viral RNA production and is influenced 176 by parameters governing the increase of (+)RNA in the cytoplasm. Growth phase is sensitive to changes 177 in viral replication rate (k r ), the kinetics of (+)RNA export from CM (k e ) and the rate of cytosolic 178 degradation of (+)RNA (µ R ) that determine the formation of dsRNA (replication intermediate

185
Differences in the TSA profiles across the viruses are more evident when V T (Fig 3b) and R cyt 186 are considered ( Figure S9). it is sub-linear in case of JEV (Fig. 3d). The increase in V T with viral (+)RNA levels is super-213 linear and sub-quadratic, for JEV and HCV, respectively whereas it is linear for PV (Fig. 3d). The (k a R cyt >> µ P ) leads to R cyt -independent level of P S (P S ≈ kt η S ka ), resulting in linear relation between The analysis also suggests that when comparing HCV and JEV, 217 higher k t .k a estimate contributes to faster assembly of R cyt . Thus R cyt (and consequently P N S = ktRcyt µ P ) 218 increases sub-linearly with increase in N C for JEV.

219
To evaluate the effect of the compartment formation kinetics on the viral dynamics, we compare 220 various NS4B mutants shown to be defective in inducing membrane re-organization [34,58]. Using τ F 221 as the fitting parameter (details in SI SM3), the model is able to accurately recapitulate the normalized 222 protein dynamics observed for these sgHCV mutants [5] (Fig 3e). The estimated τ F for the NS4B sgHCV

254
The effects of these inhibitors on model parameters are summarized in Table S5.

255
Synergy among strategies reducing cellular infectivity 256 Since the life cycle parameters that limit Φ collaborate in complex ways, we further characterized the 257 synergy (Ψ) among them. We use the Bliss independence criterion [66] to evaluate this synergy since 258 these life cycle events occur independently at the molecular level. Apart from reducing Φ independently, 259 τ F and µ R positively synergize when combined (Ψ τ F ,µ R > 1) for the viruses (Figs. 5a, Figure S12a individually or in combination. But, we observe high synergy when Φ is farther from these two extreme 269 outcomes. Figure 5c shows that Ψ τ F ,µ R decreases with {(1 − p 0 ) + p 12 }, a surrogate for how far the 270 system is from either of the two deterministic limits where p 0 and p 12 denote Φ in unperturbed and 271 doubly perturbed conditions, respectively (SI S4). Similar synergy and associated negative correlation Figure S12c-g). Therefore, interventions that target 273 membrane reorganization can be combined with other antiviral inhibitors to target early life cycle events 274 in order to achieve effective viral clearance. 275 We incorporated the dynamics of replication compartment (CM) formation accompanying cellular infec- replication kinetics for JEV, is more sensitive to perturbations compared to translation whereas the 295 reverse is true for PV. Virus production seem to be robust against perturbation to assembly rate (k a ) 296 for all the three viruses. Coupled with steady state analysis, this suggests that genomes are packaged 297 more efficiently than they are degraded.

298
In our formalism, we also find dynamics of CM formation (broadly captured by τ F and k c ) to be a 299 key kinetic barrier in the early stage of the (+)RNA virus life cycle and it has been aptly described as 300 the 'load and choke point' [17]. We demonstrate that ability of viruses to successfully establish infection . Similarly, we speculate that the infectivity of a virus in a host cell could 312 also determine the permissiveness of the host cell line [17,70,71].

313
Some of the early-infection parameters can also control cellular infectivity more effectively in combi-314 nation, displaying higher order effects due to their mutual action on common viral entities or processes. . In due to discretization of f CM to 6.5%.

355
To calculate (Bliss) synergy, Ψ, between two parameters, a 1 and a 2 , to reduce Φ we define p 0 , p 1 , p 2 356 and p 12 as Φ corresponding to no perturbation, perturbation in parameter a 1 , perturbation in parameter 357 a 2 , and simultaneous perturbations in parameter a 1 and a 2 , respectively. g X denotes p X p0 for X ∈ 1, 2, 12.
6×10 −3 * Values reported here indicate the median of the distribution estimated from data fitting (Fig. 2c)     initial condition R cyt = 1000 and considering constant relative error in observation. Data source: [17].

397
Here the parameter estimates correspond to that estimated in the third iteration of iABC.