Mathematical modeling of plus-strand RNA virus replication to identify broad-spectrum antiviral treatment strategies

Plus-strand RNA viruses are the largest group of viruses. Many are human pathogens that inflict a socio-economic burden. Interestingly, plus-strand RNA viruses share remarkable similarities in their replication. A hallmark of plus-strand RNA viruses is the remodeling of intracellular membranes to establish replication organelles (so-called “replication factories”), which provide a protected environment for the replicase complex, consisting of the viral genome and proteins necessary for viral RNA synthesis. In the current study, we investigate pan-viral similarities and virus-specific differences in the life cycle of this highly relevant group of viruses. We first measured the kinetics of viral RNA, viral protein, and infectious virus particle production of hepatitis C virus (HCV), dengue virus (DENV), and coxsackievirus B3 (CVB3) in the immuno-compromised Huh7 cell line and thus without perturbations by an intrinsic immune response. Based on these measurements, we developed a detailed mathematical model of the replication of HCV, DENV, and CVB3 and show that only small virus-specific changes in the model were necessary to describe the in vitro dynamics of the different viruses. Our model correctly predicted virus-specific mechanisms such as host cell translation shut off and different kinetics of replication organelles. Further, our model suggests that the ability to suppress or shut down host cell mRNA translation may be a key factor for in vitro replication efficiency which may determine acute self-limited or chronic infection. We further analyzed potential broad-spectrum antiviral treatment options in silico and found that targeting viral RNA translation, especially polyprotein cleavage, and viral RNA synthesis may be the most promising drug targets for all plus-strand RNA viruses. Moreover, we found that targeting only the formation of replicase complexes did not stop the viral replication in vitro early in infection, while inhibiting intracellular trafficking processes may even lead to amplified viral growth.

mathematical model. Using the model, we identified pan-viral similarities and virus-specific differences 157 in the life cycle of plus-strand RNA viruses that are represented by a unique set of model parameters.

158
The inter-viral differences among the plus-strand RNA viruses under investigation have been further 159 analyzed to study how these differences might be related to clinical disease manifestation, particularly 160 with regard to chronic versus acute infections. Our model suggests that the number of ribosomes 161 available for viral RNA translation may be a crucial factor for either acute or chronic infection outcome.
162 Furthermore, we studied broad-spectrum antiviral treatment options and found inhibiting viral 163 proteases involved in polyprotein cleavage, and RNA synthesis are promising drug targets.
164 165  [15,38] and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July https://doi.org/10.1101https://doi.org/10. /2022 Basic reproductive number (R 0 ) 1-3 (strain dependent) [ Measured in human and chimpanzee blood: 10 6 to 10 7 RNA per ml [43,45,46] In mouse blood: 10 6 RNA per ml [38]

Peak viral load
Measured in human liver: 10 8 RNA per g [43] Measured in human blood: 10 9 to 10 10 RNA per ml [44] In mouse heart: 10 11 to 10 12 RNA per g [38] Individuals with spontaneous clearance: and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. and is also made available for use under a CC0 license. was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July https://doi.org/10.1101https://doi.org/10. /2022 12 268 = concentrations throughout the course of infection. We studied hypothetical drug interventions by and is also made available for use under a CC0 license. was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. ; https://doi.org/10.1101/2022.07.25.501353 doi: bioRxiv preprint including the effects of direct acting antivirals (DAA) into the model. For this purpose, we simulated 383 putative drugs targeting (1) viral entry and internalization , (2) release of the viral RNA genome , (3)   384 formation of the translation initiation complex 1 , (4) viral RNA translation 2 , (5) polyprotein cleavage 385 , (6) replicase complex formation , (7) minus-and plus-RNA synthesis 4 and 4 , as well as (8)   386 virus particle production and release ( ). To introduce drug effects into the model, we assumed a drug 387 efficacy parameter 0 ≤ ≤ 1, and multiplied the parameters above by (1 - and is also made available for use under a CC0 license. was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. and is also made available for use under a CC0 license. was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022.

449
and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. The allocation of plus-strand RNA in the cytoplasm and within the RO, as predicted by our model, shows 452 interesting virus-specific differences (Fig 2 right panel) increasing the number of ribosomes in the HCV life cycle to those of CVB3 (from = 0.005 to 484 = 6.7 molecules per ml) increases the infectious virus load by three orders of magnitude (Fig 4A).

485
In the same way, decreasing the number of ribosomes in the CVB3 life cycle to those of HCV (from 486 3 = 6.7 to 3 = 0.005 molecules per ml) decreases the CVB3 virus load by three orders of 487 magnitude (Fig 4B). In contrast, when increasing the viral RNA synthesis rates of HCV to those of CVB3 488 (from 4 = 4 = 1.1 to 4 = 4 = 50 ℎ -1 ), the viral load did not increase. However, decreasing 489 the viral RNA synthesis rates of CVB3 to those of HCV (from

499
and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. ; https://doi.org/10.1101/2022.07.25.501353 doi: bioRxiv preprint 24 500 The subsequent processes of vRNA replication depend on successful viral protein production. Viral non-501 structural proteins are crucial for the formation of the replicase complex and its formation rate , 502 which has been selected as virus specific. Here, HCV seems to be more efficient and better adapted to 503 the Huh7 cell line, showing a 10-and 4-times faster formation rate compared to DENV and CVB3, 504 respectively. Furthermore, our estimated replicase complex formation rates suggest that the formation 505 of double membrane vesicles may be more efficient (HCV and CVB3) compared to the formation of 506 invaginations (DENV). However, the maximum number of replicase complexes as well as the 507 degradation of species within the RO ( ) were not selected as virus-specific, especially since the viral 508 RNA synthesis rates were initially set as virus-specific (

Sensitivity analysis and drug intervention
532 Having a detailed model of the intracellular replication of plus-strand RNA viruses, we next addressed 533 the question of which processes shared across all viruses showed the highest sensitivity index to 534 potential drug interventions (Fig 6). Our sensitivity analysis suggests that model parameters associated 535 with vRNA translation ( 2 ) and synthesis within the RO ( 4 and 4 ) are highly sensitive for all viruses.

541
and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. ; https://doi.org/10.1101/2022.07.25.501353 doi: bioRxiv preprint 27 542 Interestingly, over the course of infection, DENV and CVB3 showed a time-dependent sensitivity pattern 543 beginning with viral entry ( ) being sensitive, followed by the release of the viral genome ( ).
544 However, both model parameters were not sensitive for HCV, possibly due to practical non-545 identifiability (see above). Moreover, vRNA translation and replication seem to start around 5 or 20 h pi 546 in CVB3 and DENV, respectively, suggesting viral entry as a rate limiting process.

548
There are also some interesting differences between the three viruses. While the formation of the 549 translation initiation complex ( 1 ) showed a higher sensitivity in HCV, vRNA translation ( 2 )  As a next step, we aimed to analyze if any processes can be targeted leading to a 99% reduction in 560 extracellular virus upon inhibition. We therefore studied the effects of inhibiting core processes of the 561 viral life cycle (Fig 7). We then simulated in silico the administration of a hypothetical drug at two cleared. Note that we define a virus infection as being cleared if extracellular virus is reduced by more 566 than 99%. By testing both drug administration time points, we found that at the beginning of infection 567 (0 h pi) inhibiting any process led to an eradication of the virus (Fig 7). Since the viral replication 568 machinery is not established, viral entry and vRNA release may be possible drug targets, however, an 569 almost 100% inhibition (~1) was necessary to block the infection process (S1 Table). Obviously, in-silico 570 drugs targeting virus entry and vRNA release at a time point after an established viral infection, is not 571 able to reduce the viral load. However, for both drug administration time points, targeting vRNA 572 translation as well as vRNA synthesis showed the strongest effect, and thus are the most promising drug 573 targets (S1 Table). Interestingly, targeting the formation of the replicase complexes could not clear (or and is also made available for use under a CC0 license. was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July https://doi.org/10.1101https://doi.org/10. /2022 28 574 even reduce) CVB3 infection with a drug administration given at steady state (S1 Table) was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July https://doi.org/10.1101https://doi.org/10. /2022 29 586 polyprotein cleavage used in combination with drugs that inhibit vRNA synthesis or formation of the 587 replicase complex at steady state (Figs 8 and 9 and S1 and S2, Figs, S1 Table). We identified the "sweet 588 spot" for efficient viral eradication (by more than 99%). Our model predicted that HCV and DENV 589 showed a comparable pattern of viral clearance to a combination of two drugs, while for the clearance 590 of CVB3 higher drug efficacies were necessary to clear the infection. Inhibiting vRNA synthesis in 591 combination with vRNA translation or polyprotein cleavage by more than 90% was an efficient 592 combination for HCV and DENV (Fig 8B and 8C, S1  Table).

598
Overall, we found the lowest pan-viral critical drug efficacy was for the combined inhibition of vRNA 599 synthesis and polyprotein cleavage with a required 98% effectiveness for each drug (Figs 8C and 9C, S1 600 Table,). Note that we also tested in silico the combination therapy of inhibiting translation complex 601 formation, vRNA translation, and polyprotein cleavage together with replicase complex formation.

602
However, higher critical drug efficacy constants were needed to clear the infection (S1, S2 Figs and S1 603 Table).

604
and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022.

621
Mathematical modeling of viral dynamics has a long history and has been applied to a variety of viral 622 infectious diseases [25]. Population based models considering susceptible and infected cell populations, and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. conditions. We compare viral replication mechanisms as well as pan-viral similarities and virus-specific 637 differences, which may help to understand acute or chronic infection outcome that in turn may be an 638 initial step towards the development of broad-spectrum antiviral treatment strategies. was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. ; https://doi.org/10.1101/2022.07.25.501353 doi: bioRxiv preprint initiation complex 1 seems to be a non-identifiable process in the model structure, as it was also non-656 identifiable in our previous DENV model [55]. Further, the model processes of virus entry and vRNA 657 genome release, and , were practically non-identifiable for HCV. A possible explanation for both 658 processes being non-identifiable may be insufficient experimental measurements for HCV to uniquely 659 estimate both rate constants, e.g., the lack of intracellular protein concentration measurements for 660 HCV. However, since both parameters were identifiable for CVB3 and DENV and both processes were 661 selected as virus-specific, and , they remained virus-specific in the final model.

663
Virus specific differences and pan-viral similarities and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted https://doi.org/10.1101https://doi.org/10. /2022  was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. may be directly associated with a higher production of viral proteins. However, the more ribosomes 795 available for cellular mRNA translation and thus the production of proteins of the immune response, the 796 higher may be the intracellular degradation of viral components, resulting in a limitation in viral 797 resources. Ribosome availability and its control may thus be a crucial factor for viral replication 798 efficiency.

800
To analyze this aspect further, we asked whether we could make virus production in HCV more efficient 801 or CVB3 less efficient. Increasing the in-silico ribosome availability in HCV to that of CVB3 increased the 802 viral load by three orders of magnitude. In contrast, a 50-fold increase in the HCV RNA synthesis rate 803 had no effect on the viral load in steady state due to a limited availability of the viral RNA polymerase 804 in the replication organelle [19]. In contrast, using only 0.07% of ribosomes for CVB3 RNA translation, 805 thus setting the ribosome level to the number of ribosomes used in HCV, decreased the CVB3 viral load 806 by three orders of magnitude. Interestingly, the coronaviruses nonstructural proteins, including those of 807 SARS-CoV-2, target multiple processes in the cellular mRNA translation, causing a host cell translation 808 shut off similar to CVB3 and DENV [117,118]. Therefore, a repression or complete shut-off of the host 809 mRNA translation machinery may be a key-feature of acute viral infections.

811
Comparing in vivo viral dynamics with those of in vitro experiments is challenging. Nevertheless, we 812 found comparable pattern of viral dynamics: reported in vivo and our in vitro experiments. In vivo, HCV and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022. ; https://doi.org/10.1101/2022.07.25.501353 doi: bioRxiv preprint understand and optimize therapy. Clin Liver Dis. 2003;7: 163-178. doi:10.1016/S1089-1055 3261(02)00063-6 and is also made available for use under a CC0 license.
was not certified by peer review) is the author/funder. This article is a US Government work. It is not subject to copyright under 17 USC 105 The copyright holder for this preprint (which this version posted July 25, 2022.