Agent-based model predicts that layered structure and 3D movement work synergistically to reduce bacterial load in 3D in vitro models of tuberculosis granuloma

Tuberculosis (TB) remains a global public health threat. Understanding the dynamics of host-pathogen interactions within TB granulomas will assist in identifying what leads to the successful elimination of infection. <i>In vitro</i> TB models provide a controllable environment to study these granuloma dynamics. Previously we developed a biomimetic 3D spheroid granuloma model that controls bacteria better than a traditional monolayer culture counterpart. We used agent-based simulations to predict the mechanistic reason for this difference. Our calibrated simulations were able to predict heterogeneous bacterial dynamics that are consistent with experimental data. In one group of simulations, spheroids are found to have higher macrophage activation than their traditional counterparts, leading to better bacterial control. This higher macrophage activation in the spheroids was not due to higher counts of activated T cells, instead fewer activated T cells were able to activate more macrophages due to the proximity of these cells to each other within the spheroid. In a second group of simulations, spheroids again have more macrophage activation but also more T cell activation, specifically CD8+ T cells. This higher level of CD8+ T cell activation is predicted to be due to the proximity of these cells to the cells that activate them. Multiple mechanisms of control were predicted. Simulations removing individual mechanisms show that one group of simulations has a CD4+ T cell dominant response, while the other has a mixed/CD8+ T cell dominant response. Lastly, we demonstrated that in spheroids the initial structure and movement rules work synergistically to reduce bacterial load. These findings provide valuable insights into how the structural complexity of <i>in vitro</i> models impacts immune responses. Moreover, our study has implications for engineering more physiologically relevant <i>in vitro</i> models and advancing our understanding of TB pathogenesis and potential therapeutic interventions.

结核病（Tuberculosis, TB）仍是全球公共卫生的重大威胁。解析结核肉芽肿内宿主-病原体相互作用的动态变化，有助于揭示感染被成功清除的关键机制。体外（<i>In vitro</i>）结核模型为研究肉芽肿动态变化提供了可控的实验环境。此前，我们构建了一种仿生3D球状肉芽肿模型，其细菌控制效果优于传统的单层培养模型。我们通过基于智能体的模拟（agent-based simulations）来预测这一差异的机制性原因。经校准的模拟模型能够预测与实验数据一致的异质性细菌动态变化。在一组模拟中，球状模型的巨噬细胞活化水平高于传统模型，从而实现了更优的细菌控制效果。球状模型中巨噬细胞活化水平的提升并非源于活化T细胞数量的增加，而是由于细胞在球状结构内的紧密邻近性，使得少量活化T细胞即可激活更多巨噬细胞。在另一组模拟中，球状模型不仅表现出更高的巨噬细胞活化水平，还伴有更强的T细胞活化，尤其是CD8+ T细胞。这种CD8+ T细胞活化水平的升高，被预测是由于这些细胞与活化它们的细胞之间的邻近性所致。模拟预测了多种细菌控制机制。去除单一机制的模拟结果显示，一组模拟呈现CD4+ T细胞主导的免疫应答，而另一组则表现为混合性或CD8+ T细胞主导的应答。最后，我们证实，球状模型中的初始结构与细胞运动规则协同作用，可降低细菌载量。这些发现为理解体外（<i>In vitro</i>）模型的结构复杂性如何影响免疫应答提供了重要启示。此外，本研究为构建更具生理相关性的体外（<i>In vitro</i>）模型、深化对结核发病机制及潜在治疗干预手段的理解提供了理论依据。