Fig 1.
Outline of in vitro experimental procedure.
a) Patients with active TB have bronchoalveolar fluid (BALF) and peripheral blood collected. Alveolar macrophages (AM) are isolated from BALF and PBMCs from peripheral blood. AM and PBMCs are recombined in vitro to create traditional and spheroid in vitro models, along with corresponding in silico traditional and spheroid simulations. b) For the 3D spheroid models, AM are infected and combined with magnetic beads before being magnetically levitated. Autologous T cells (i.e., the CD3+ fraction of the PBMCs) are added at day 2, forming a cuff around the macrophage core. This produces a spheroid of cells that is floating in the center of the well. c) Spheroid culture has lower CFU counts at day 6 than traditional culture. (Adapted from Kotze et al. 2021 [11]).
Fig 2.
Example spatial layout of 4 simulation cases: a) spheroid structure with 3D movement, b) spheroid structure with gravity-limited movement, c) no structure with gravity-limited movement seen from the side, and d) no structure with 3D movement seen from the side. All simulations are done with 3 dimensions, but the ones with gravity-limited movement represent experimental “2D” models.
Table 1.
Parameters that are varied during calibration.
Initial ranges are either determined by literature, estimated through preliminary simulations (e), or broadened to the full mathematically possible range (f). The collection of calibrated parameter sets can be found in the supplementary material: parameter descriptions are given in S1 Table. This table is reproduced from Petrucciani et al. [15].
Fig 3.
Spheroid (blue) and traditional (green) simulations have been calibrated in a paired manner to in vitro data. a) Bacterial count dynamics for spheroid and traditional simulations over the 6-day time course. The blue and green bars on the right-hand side represent the ranges of experimental calibration data (calculated from fold changes). The black x is the point used to differentiate between similarly and differentially controlled simulations. We divide the simulations into two groups with traditional CFU greater than 6000 (differentially controlled simulations) or less than or equal to 6000 (similarly controlled simulations) at time step 550 (about day 2.29). b) Due to the overlap in the experimental data for CFU, some spheroid and traditional simulations perform similarly. c) The similarly controlled runs have a higher baseKillingProbability than the differentially controlled runs. Two patterns emerge in the differentially controlled simulations: d) simulations with two major decreases and e) simulations with one major decrease.
Fig 4.
a) Bacterial dynamics for DC-1 simulations.
The time range of interest (day 3–4) is outlined. These are the same time courses as shown in Fig 3C. Proportions of bacterial killing due to b) activated macrophages, c) base macrophages, and d) CD8+ T cell killing from day 3 to 4. Comparing total counts of e) activated macrophages, f) STAT1 activated macrophages, g) NF-κB activated macrophages, h) activated T cells, i) activated CD4+ T cells, and j) activated CD8+ T cells between the spheroid and traditional simulations at day 4. ** p< = 1e-2, *** p< = 1e-3.
Fig 5.
The mean distance from macrophages to nearest a) T cells, b) activated T cells, c) activated CD4+ T cells, and activated CD8+ T cells. Each dot represents an average of all cells in a single simulation. The distribution of the mean distances across the set of simulations is compared between spheroid and traditional at day 4. e) A slice through the center of a representative spheroid simulation from the DC-1 simulations at day 4 colored by cell type. f) The same simulation with macrophages in blue, activated T cells enlarged and in yellow, and all other cells in grey. ** p< = 1e-2, **** p< = 1e-4.
Fig 6.
a) Bacterial count over time for DC-2 simulations. The initial deviation of spheroid and traditional simulations occurs during day 3 to 4 (black rectangle). These are the same time courses as shown in Fig 3D. Proportions of bacterial killing due to b) activated macrophages, c) base macrophages, and d) CD8+ T cell killing from day 3 to 4. Comparing total counts of e) activated macrophages, f) STAT1 activated macrophages, g) NF-κB activated macrophages, h) activated T cells, i) activated CD4+ T cells, and j) activated CD8+ T cells between the spheroid and traditional simulations at day 4. *** p< = 1e-3.
Fig 7.
Mean distance between TBS CD8+ T cells and (a) macrophages or (b) STAT1 activated macrophages. Each dot represents the mean distance of all cells for a single simulation. c) A slice through the center of a representative spheroid simulation from the DC-2 simulations at day 4 colored by cell type. d) The same simulation with CD8+ T cells in blue, STAT1 activated macrophages in orange, and all other cells in grey. * p< = 0.05, **** p< = 1e-4.
Fig 8.
Simulations were run with IFN-γ secretion virtual knockouts from CD4+ T cells, CD8+ T cells, or both.
A simulation knocking out CD8+ T cell killing of infected macrophages was also performed. The original simulation and the 4 knockout scenarios were compared by looking at a-c) bacterial counts at day 6, d-f) activated macrophage counts at day 4, and g-i) activated CD8+ T cell counts at day 4. These outputs are shown for the DC-1 spheroid simulations (n = 118), DC-2 spheroid simulations(n = 71), and DC-2 traditional simulations(n = 71).
Table 2.
Bacterial count relative to control for DC-1 and -2 spheroid and traditional virtual knockouts and the predicted dominant T cell fraction in these bacterial responses as based on the virtual knockout results.
Fig 9.
The movement rules were swapped for the spheroid and traditional simulations, creating spheroid gravity and traditional floating.
a) Bacterial counts at day 6 in spheroid, spheroid gravity, traditional, and traditional floating scenarios. Proportions of bacterial killing due to b) activated macrophages, c) base macrophages, and d) CD8+ T cells from day 3 to 4. Total counts of e) activated macrophages, f) activated CD4+ T cells, and g) CD8+ T cells on day 4. h) Mean distance from macrophages to activated T cells on day 4. i) Mean distance from CD8+ T cells to STAT1 macrophages on day 4. This analysis is done with DC-1 simulations, the same analysis for DC-2 simulations is in the S8 Fig. * p< = 0.05, ** p< = 1e-2, *** p< = 1e-3.