A mechanistic integrative computational model of macrophage polarization: Implications in human pathophysiology

Abstract

Macrophages respond to signals in the microenvironment by changing their functional phenotypes, a process known as polarization. Depending on the context, they acquire different patterns of transcriptional activation, cytokine expression and cellular metabolism which collectively constitute a continuous spectrum of phenotypes, of which the two extremes are denoted as classical (M1) and alternative (M2) activation. To quantitatively decode the underlying principles governing macrophage phenotypic polarization and thereby harness its therapeutic potential in human diseases, a systems-level approach is needed given the multitude of signaling pathways and intracellular regulation involved. Here we develop the first mechanism-based, multi-pathway computational model that describes the integrated signal transduction and macrophage programming under M1 (IFN-γ), M2 (IL-4) and cell stress (hypoxia) stimulation. Our model was calibrated extensively against experimental data, and we mechanistically elucidated several signature feedbacks behind the M1-M2 antagonism and investigated the dynamical shaping of macrophage phenotypes within the M1-M2 spectrum. Model sensitivity analysis also revealed key molecular nodes and interactions as targets with potential therapeutic values for the pathophysiology of peripheral arterial disease and cancer. Through simulations that dynamically capture the signal integration and phenotypic marker expression in the differential macrophage polarization responses, our model provides an important computational basis toward a more quantitative and network-centric understanding of the complex physiology and versatile functions of macrophages in human diseases.

Fig 1. Multiple signaling axes regulate macrophage phenotype and polarization.

Macrophage polarization is dynamically controlled by different receptor-mediated signaling pathways, cell stress, transcriptional and post-transcriptional regulators (e.g. miRs), which collectively lead to differential expression of a panel of macrophage phenotype markers (including both intracellular and secreted proteins). Arrow indicates activation,–| symbol indicates inhibition. Green shapes indicate intracellular proteins, orange shapes indicate secreted products. This figure is an overview of model formulation; full mechanistic details of the computational model are presented in S1 Fig and S1 and S2 Tables. It should also be noted that this figure only describes a subset of the M1- and M2-related pathways and markers.

Fig 2. IFN-γ-mediated signaling controls macrophage phenotype.

In response to different doses of IFN-γ treatment, the model simulations are compared with corresponding literature time-course data including (A) degradation of receptor-bound IFN-γ [59], (B) phosphorylation of receptor-associated JAK [15], (C) phosphorylation of STAT1 [60], (D) expression of IRF-1 [61], (E) expression of iNOS [62], (F) levels of secreted TNFα [17], (G) levels of secreted IL-12 [16], (H) levels of secreted CXCL-9 [16], (I) intracellular mRNA expression of CXCL-10 [63], (K) expression of miR-3473b [42], plus single timepoint measurements including intracellular levels of (J) itaconic acid at 18 h [64], (L) PTEN at 36 h (in response to miR-3473b mimic transfection, see also S2H Fig) [42], and (M) HIF-1α (in response to IFN-γ with or without hypoxia) [48]. (A-M) All literature data are measured in macrophage cell lines and values are for protein levels unless noted otherwise. Y-axes show normalized expression respectively (A-E, G-I, K: simulations and data are normalized to the maximum expression; F, L: normalized to the no-treatment/time 0 expression; J: normalized to the expression at 18 h; M: normalized to the expression under IFN-γ treatment with hypoxia). S–simulation, D–literature data, Utr–untreated, Trd–IFN-γ treated, Hyp–hypoxia.

Fig 3. IL-4-mediated signaling controls macrophage phenotype.

Comparison between model simulations and literature experimental data on IL-4 induced (A) STAT6 phosphorylation [65], (B-C) IRF-4 upregulation (time-course and at 24 h) [66, 67], (D) AKT activation [68], (E) PPARγ expression at 18 h [69, 70], (F-G) Arg-1 expression (time-course and at 24 h) [67, 71], (H) IL-10 secretion at 24 h [72], (I) VEGF secretion at 24 h [73], (J) downregulation of TNFα secretion at 24 h [72], and (K) HIF-2α stabilization (in response to IL-4 with or without hypoxia) [48]. (A-K) All literature data are measured in macrophage cell lines and values are for protein levels unless noted otherwise. Y-axes show normalized expression respectively (A, B, D, F: simulations and data are normalized to the maximum expression; C, G: normalized to the expression at 24 h post-treatment; E, H, I, J: normalized to the no-treatment/time 0 expression; K: normalized to the expression under IL-4 treatment with hypoxia). S–simulation, D–literature data, Utr–untreated, Trd–IL-4 treated, Hyp–hypoxia.

Fig 4. Hypoxia promotes M1 and M2 marker expression.

Model simulation and literature experimental data from macrophages on hypoxia-induced (A) time-course stabilization of HIF-1α and (B) HIF-2α under 3% O₂ [75, 76], (C) sustained stabilization of HIF-1/2α at 24 h under 0.5% O₂ [77], (D) upregulation of iNOS and (E) Arg-1 proteins at 8 h under 1% O₂ [78], (F) increase in TNFα secretion at 24 h under 0.3% O₂ [79], (G) increase in IFN-γ secretion over time under 1% O₂ [47], (H) increase in VEGF secretion at 24 h under 1% O₂ [80], and (I) inhibition of miR-93 abundance at 12 h under 2% O₂ [53]. (J) Enforced overexpression of miR-93 (see also S4G Fig) leads to decreased IFN-γ secretion at 12 h under 2% O₂ [53]. (A-J) All literature data are measured in macrophage cell lines and results are for protein levels unless noted otherwise. Y-axes show normalized expression respectively (A, B: simulations and data are normalized to the maximum expression; C: normalized to the expression at 24 h under hypoxia; D-I: normalized to the normoxic/time 0 expression; J: normalized to the hypoxia-induced expression at 12 h without miR-93 mimic treatment). S–simulation, D–literature data, Utr–normoxia/untreated, Trd–treated with miR-93 mimic, Hyp–hypoxia, O₂ –oxygen.

Fig 5. Pathway feedbacks and cross-talks in M1-M2 regulatory network.

Overexpression of SOCS1 and SOCS3 in macrophages can downregulate activation of (A) STAT1 by IFN-γ and (B) STAT6 by IL-4. Silencing of SOCS3 promotes (C) IFN-γ-induced M1 marker expression while it minimally affects (D) IL-4-induced M2 marker expression (relative fold changes are labeled). (A-D) Overexpression is modeled as 50x initial level with normal (1x) production, and silencing is modeled as 0 initial level with 0 production. (E) Simulated dose response of iNOS and Arg-1; relative protein levels measured at 12 h are plotted and labeled (the baseline condition is represented by the 0.01 ng/ml case). (F) Upon IFN-γ stimulation followed by the addition of IL-4 (at 4 h), cellular IRF-1 level is downregulated compared to IFN-γ only; (G) Upon IL-4 stimulation followed by the addition of IFN-γ (at 1 hr), cellular activation of AKT is downregulated compared to IL-4 only. (H) The addition of a second stimulus IFN-γ (after 24 h of IL-4 stimulation) would antagonize the expression pattern of M1 and M2 markers induced by IL-4 (see also S5G Fig). (I) Similarly, IL-4 added after 24 h of IFN-γ stimulation would antagonize the marker expression pattern induced by IFN-γ. When macrophages are stimulated with IFN-γ and IL-4 simultaneously, the simulated expression of (J) M1 and M2 markers as well as (K) the activation of a number of M1 and M2 signature proteins (see also S5I Fig) are collectively induced with distinct temporal profiles. (L) Dynamic protein expression patterns (after 12, 24 and 48 h of stimulation) of M1 and M2 markers in macrophages under seven different stimulation conditions (A+B means simultaneous stimulation, expression levels are normalized to the untreated/time 0 levels and then log2 transformed). (A-L) All simulation results are protein levels (except CXCL10 is mRNA level). (C-E, H-K) Y-axes show relative expression respectively (C-D, H-K: normalized to untreated/control/time 0 levels; E: normalized to maximum levels at 50 ng/ml). Simulated treatment doses are 10 ng/ml IFN-γ and 10 ng/ml IL-4 for (A-D), 10 ng/ml IFN-γ and 20 ng/ml IL-4 for (F-G), 20 ng/ml IFN-γ and 20 ng/ml IL-4 for (H-I), 10 ng/ml IFN-γ and 5 ng/ml IL-4 for (J-L). Utr–untreated, hyp–hypoxia (2% oxygen for L).

Fig 6. Global sensitivity analysis and simulated therapeutic strategies to repolarize macrophages in hypoxia.

(A-B) Sensitivity indices (top 25 positive and negative PRCC values with p<0.05) of model parameters that control M1 and M2 marker expression in terms of the M1/M2 score (a ratio-based estimate of M1 phenotypes relative to M2 phenotypes, see Materials and Methods for more details) in hypoxia (2% O₂). In the parameter descriptions, ‘X_RC’ means receptor complex formed by ligand X, receptor and JAK, ‘X/Y’ means complex formed by X and Y. Simulated time-course expression of M1 and M2 markers when macrophages are subjected to (C-D) hypoxia, (E-F) hypoxia with IFN-γ inhibition, (G-H) hypoxia with HIF-1α inhibition, (I-K) hypoxia with STAT1 inhibition, and (L-M) hypoxia with IRF-1 inhibition. Inhibition of IFN-γ, HIF-1α and IRF-1 is simulated by setting the respective production rates to 10% of their original values (STAT1 inhibition is simulated as a 90% decrease in the binding rate between STAT1 and activated IFN-γ receptor complex). Species name denoted with * means expression in hypoxia plus treatment (species name without * means expression in hypoxia alone). (C-M) Marker expression levels are normalized to their respective t = 0 values (e.g. normoxia, unstimulated). (A-B) More details about the parameters listed can be found in S1 Table using the labels (positive–k127, kf63, kr70, kf64, kf17, k33, k61, k77, k45, kf44, kf42, k37; negative–k99, kr42, k78, kf8, kr44, kf95, k71, kf13, ka77, kf7, kf52, kf70, kr64; order is from top to bottom as displayed). (C-M) All simulation results are protein levels (except CXCL10 is mRNA level).

Fig 7. Targeting IL-4 signaling axis in macrophages in tumor.

Simulated time-course expression of M1 and M2 markers when macrophages are subjected to (A-B) high IL-4 production (10x of original value), (C-D) high IL-4 production with IL-4/receptor blockade (90% decrease in the binding rate between IL-4 and its receptor), (E-F) high IL-4 production with STAT6 inhibition (90% decrease in the binding rate between STAT6 and activated IL-4 receptor complex), and (G-H) high IL-4 production with PHD inhibition (90% decrease in the binding rate between PHD and O₂). Species name denoted with * means expression in high IL-4 production plus treatment (species name without * means expression in high IL-4 production alone). (A-H) Marker expression levels are normalized to their respective t = 0 values (e.g. normoxia, unstimulated). All simulation results are protein levels (except CXCL10 is mRNA level).