An Integrative Network Modeling Approach to T CD4 Cell Activation

Abstract

The adaptive immune response is initiated by the interaction of the T cell antigen receptor/CD3 complex (TCR) with a cognate peptide bound to a MHC molecule. This interaction, along with the activity of co-stimulatory molecules and cytokines in the microenvironment, enables cells to proliferate and produce soluble factors that stimulate other branches of the immune response for inactivation of infectious agents. The intracellular activation signals are reinforced, amplified and diversified by a complex network of biochemical interactions, and includes the activity of molecules that modulate the activation process and stimulate the metabolic changes necessary for fulfilling the cell energy demands. We present an approach to the analysis of the main early signaling events of T cell activation by proposing a concise 46-node hybrid Boolean model of the main steps of TCR and CD28 downstream signaling, encompassing the activity of the anergy factor Ndrg1, modulation of activation by CTLA-4, and the activity of the nutrient sensor AMPK as intrinsic players of the activation process. The model generates stable states that reflect the overcoming of activation signals and induction of anergy by the expression of Ndrg1 in the absence of co-stimulation. The model also includes the induction of CTLA-4 upon activation and its competition with CD28 for binding to the co-stimulatory CD80/86 molecules, leading to stable states that reflect the activation arrest. Furthermore, the model integrates the activity of AMPK to the general pathways driving differentiation to functional cell subsets (Th1, Th2, Th17, and Treg). Thus, the network topology incorporates basic mechanism associated to activation, regulation and induction of effector cell phenotypes. The model puts forth a conceptual framework for the integration of functionally relevant processes in the analysis of the T CD4 cell function.

Keywords: AMPK; Boolean model; CD28; CTLA-4; NDRG1; T CD4 cells; TCR; complex network.

Figure 1

A general 46-node network of the early biochemical interactions induced by TCR activation, co-stimulation, and phenotype-inducing cytokines. The continuous and dotted lines represent activator and inhibitory pathways, respectively. Inputs of the network are the activated TCR, the co-stimulatory molecules CD80/86, the nutrient micro-environment-sensing protein AMPK (triangles), and external cytokines (polygons). The set of functions defining the dynamical system for boolean modeling is shown in Supplementary Table 1. Green and red rectangles represent stimulatory and inhibitory nodes of the activation core, respectively. Purple and yellow rectangles represent phenotype-inducing transcription factors and cytokines produced endogenously, respectively. LAT, LAT-Gads-SLP-76 complex; IL2G, IL2 gene; IL4e, IL2e, IFNγe, IL10e, TGFβe, and IL21e, exogenous cytokines.

Figure 2

(A) The activation core of the network represents downstream signaling induced by TCR and CD28, encompassing the activity of CTLA-4, the anergy factor NDRG1, and the nutrient micro-environment-sensing protein AMPK. The effect of exogenous cytokines is not included. Green and red rectangles represents stimulatory and inhibitory nodes of the activation core, respectively. (B) Subnetwork of the core showing the convergence of the activity of NFκB, AP-1 and NFAT for the expression of the IL-2 gene (IL2G). IL2G induces the expression of its own high affinity receptor (CD25). CD25 mediates the activity of PDK1, mTOR, SOS, and STAT5, which positively feedback the expression of IL-2. The anergy-inducing factor Nrdg1 is induced by NFAT and inactivated by Akt. Colors of rectangles correspond to the type of components, as indicated in Figure 1.

Figure 3

Subnetworks representing the effect of exogenous cytokines on cell differentiation. Exogenous IL-4, IL-2, IFNγ, IL-10, TGF-β, and IL-21 were herein named IL-4e, IL-2e, IFNGe, IL-10e, TGFBe, and IL-21e (polygons). Specific cytokines induce the activity of particular transcription factors (T-bet, GATA3, RORγt, and Foxp3), which guide the differentiation of effector cell subtypes (Th1, Th2, and Treg cells produce IFNγ, IL-4, and TGF-β, respectively, whereas Th17 cells produce IL-21 and IL-17). Cross-inhibitory signals involved in the predominant induction of a cell subtype by particular cytokines are shown. Colors of rectangles correspond to the type of components, as indicated in Figure 1.

Figure 4

(A) Set of attractors generated by the 30-node network corresponding to the activation core and by considering the set of all possible initial conditions (2³⁰). Active and inactive nodes are shown in green and red, respectively. Outputs indicating activation were defined as the expression of the IL2G, AP-1, NFAT, and NF-κb transcription factors. The network dynamics generated attractors corresponding to states of no activation, TCR and CD28-induced activation, anergy (when TCR or CD28 were inactive), and immune checkpoint (when CTLA-4 and CTLA-4 dimers are expressed). (B) Set of attractors arising from the complete 46-node network (activation core and modules associated to induction of effector phenotypes). In this case, attractors were generated by setting the initial conditions of the activation state (TCR = 1, CD28 = 1, CD80/86L = 0). In addition, the indicated cytokines were set as positive inputs. AMPK was set as active for differentiation to Th2 and Treg cells. Differentiation to effector phenotypes was associated with the production of particular cytokines (INF-γ for Th1, IL-4 for Th2, IL-21 for Th17, and IL-10 and TGF-β for Treg subsets, respectively).

Figure 5

(A) Intermediate states leading to T lymphocyte activation and anergy. Active and inactive nodes are shown in green and red, respectively. The transition states from time steps 0 to 10 represent progressive activation. CTLA-4 and CTLA-4 dimers are expressed in a late state of activation (step 10). (B) Persistent activation leads to the sustained expression of CTLA-4 dimers; interaction of this molecule with CD80/86 sequentially inactivates multiple nodes (steps 11–18) until the immune checkpoint state is reached. (C) Intermediate states leading to the anergy attractor shown in Figure 4 in the absence of co-stimulation. (D) Turning on the Ndrg-1 node at the onset of activation through the TCR and CD28 (simulating activation of Ndrg-1-overexpressing cells) leads to partial activation but inhibition of IL-2 synthesis.

Figure 6

Time-dependent activity of CD80/86 due to the shift of its relative affinity for CD28 and CTLA-4 dimers. The magenta curve describes the linkage strength of the CD80/86-CD28 complex which shows a steep decrease at t_d ≃ 10. The green curve describes the anticorrelated behavior of the linkage strength of CD80/86-CTLA-4 dimers complex.

Figure 7

Graphic representations of basins of attraction for the activation core obtained with a simplified 12-node network. The rules involved in the simplified network were rigorously obtained by straightforward use of Boolean algebra as described in the Appendix. In the graphs, initial conditions are represented by the outside layers. From these states, the system transits through intermediate state layers before reaching the final steady state indicated by the loops. The largest basins of attraction are those related with no activation and immune checkpoint states.