AUTHOR=Weinstock Laura D. , Forsmo James E. , Wilkinson Alexis , Ueda Jun , Wood Levi B. TITLE=Experimental Control of Macrophage Pro-Inflammatory Dynamics Using Predictive Models JOURNAL=Frontiers in Bioengineering and Biotechnology VOLUME=8 YEAR=2020 URL=https://www.frontiersin.org/journals/bioengineering-and-biotechnology/articles/10.3389/fbioe.2020.00666 DOI=10.3389/fbioe.2020.00666 ISSN=2296-4185 ABSTRACT=

Macrophage activity is a major component of the healthy response to infection and injury that consists of tightly regulated early pro-inflammatory activation followed by anti-inflammatory and regenerative activity. In numerous diseases, however, macrophage polarization becomes dysregulated and can not only impair recovery, but can promote further injury and pathogenesis, e.g., after trauma or in diabetic ulcers. Dysregulated macrophages may either fail to polarize or become chronically polarized, resulting in increased production of cytotoxic factors, diminished capacity to clear pathogens, or failure to promote tissue regeneration. In these cases, a method of predicting and dynamically controlling macrophage polarization will enable a new strategy for treating diverse inflammatory diseases. In this work, we developed a model-predictive control framework to temporally regulate macrophage polarization. Using RAW 264.7 macrophages as a model system, we enabled temporal control by identifying transfer function models relating the polarization marker iNOS to exogenous pro- and anti-inflammatory stimuli. These stimuli-to-iNOS response models were identified using linear autoregressive with exogenous input terms (ARX) equations and were coupled with non-linear elements to account for experimentally identified supra-additive and hysteretic effects. Using this model architecture, we were able to reproduce experimentally observed temporal iNOS dynamics induced by lipopolysaccharides (LPS) and interferon gamma (IFN-γ). Moreover, the identified model enabled the design of time-varying input trajectories to experimentally sustain the duration and magnitude of iNOS expression. By designing transfer function models with the intent to predict cell behavior, we were able to predict and experimentally obtain temporal regulation of iNOS expression using LPS and IFN-γ from both naïve and non-naïve initial states. Moreover, our data driven models revealed decaying magnitude of iNOS response to LPS stimulation over time that could be recovered using combined treatment with both LPS and IFN-γ. Given the importance of dynamic tissue macrophage polarization and overall inflammatory regulation to a broad number of diseases, the temporal control methodology presented here will have numerous applications for regulating immune activity dynamics in chronic inflammatory diseases.