Iterative community-driven development of a SARS-CoV-2 tissue simulator

Key results

1. SuggestedWith default v5 parameters, simulated viral replication peaks around day 4 and virus is cleared by about day 6. The authors describe v5 as the first version able to eliminate infection without destroying the entire epithelium.

   Fig 5.2, N = 12 replicates over 15 days: interferon rises quickly and persists. CD4+ T cells enter briefly, while CD8+ T cells stay for all 15 days. Ig production only ramps up after day 12, once most virus is gone, so antibody gives no major protection in the first infection. · Overview of v5 model results with base…; Discussion; Figure legends

2. SuggestedAdding transit delays for dendritic cells (tissue to lymph node) and T cells (lymph node to tissue) postpones T cell arrival and lowers epithelial survival at day 15, with little change in macrophage, neutrophil or fibroblast recruitment.

   Four settings were compared: no delay; dendritic cell delay only (0.5 days); dendritic cell 0.5 days plus T cell 0.25 days (base); doubled delays (1 day and 0.5 days). Epithelial survival falls as delays increase (Fig 5.3B). The authors attribute this to CD8+ T cell-driven viral clearance. Replicate count not stated. · Impact of a time delay; Discussion of current model version; Figure legends

3. SuggestedTissue was re-challenged from starting immune states interpolated between naive and post-infection ('educated'), with f the fraction of prior immunity retained. Protection holds after substantial waning: of the three levels tested, only f = 0.002 shows more damage and a viral load approaching the naive case.

   At f = 1: very little damage; CD8+ T cells still present at re-inoculation. At f = 0.03125: little added damage, and CD8+ T cells are recruited about 2 days earlier than at f = 0.002. Antibodies were present at the start in all three cases, and their overall level varied by at least two orders of magnitude between them. Authors: clearance with minimal damage despite more than 30-fold waning. · Impact of prior immune responses on subsequent…; Discussion of current model version; Figure legends

4. SuggestedIn a four-parameter sensitivity analysis of 15-day v5 runs, the ACE2 binding rate constant had the largest influence (toward more surviving cells), followed by the viral diffusion coefficient and the paracrine interferon secretion rate. The T cell recruitment rate had little effect on day-15 live cell count.

   Latin hypercube samples within one order of magnitude of default values. A 4-dimensional linear regression of log10 final cell count on log10 parameters used only runs with more than 0 and fewer than the maximum surviving cells (Fig 5.5). Coefficient values and sample size are not given in the prose. · Global sensitivity analysis; Figure legends

5. ClaimedSummaries of earlier versions (data in supplementary material, not shown in this text) report that lower ACE2 expression could paradoxically speed tissue-wide spread, and that rapid or pre-existing CD8+ T cell responses are protective.

   v1: for viruses that rapidly produce and exocytose new virions, extra virion release at cell death did not significantly speed spread, and lower cell tolerance to viral load sped tissue loss but did not drastically change spread. v2: lower ACE2 slowed per-cell replication but sped spread. v3: rapid recruitment and pre-existing CD8+ T cells were protective. v4: interferon strongly affected plaque spread. · Discussion; Summary of Versions 1-4