A comprehensive map of the dendritic cell transcriptional network engaged upon innate sensing of HIV

Abstract

Transcriptional programming of the innate immune response is pivotal for host protection. However, the transcriptional mechanisms that link pathogen sensing with innate activation remain poorly understood. During infection with HIV-1, human dendritic cells (DCs) can detect the virus through an innate sensing pathway leading to antiviral interferon and DC maturation. Here, we developed an iterative experimental and computational approach to map the innate response circuitry during HIV-1 infection. By integrating genome-wide chromatin accessibility with expression kinetics, we inferred a gene regulatory network that links 542 transcription factors with 21,862 target genes. We observed that an interferon response is required, yet insufficient to drive DC maturation, and identified PRDM1 and RARA as essential regulators of the interferon response and DC maturation, respectively. Our work provides a resource for interrogation of regulators of HIV replication and innate immunity, highlighting complexity and cooperativity in the regulatory circuit controlling the DC response to infection. Key Wordsnetwork inference, DNA sensing, innate signaling, cGAS, STING, IRF3, NF-κB, chromatin modification, ATAC-seq, RNA-seq of a TF's predicted gene targets ( Figure 3C). This step approximates the effect of unmeasured parameters, such as post-translational regulation and protein-protein interactions, on TF activity in a condition-dependent manner. We were thus able to model changes in TF activity that are semi-independent of changes in TF expression, as observed for IRF3 ( Figure S3A, S3B). IRF3 activity was predicted to increase in response to LPS, pIC and HIV sensing, which are all known to drive IFN production through IRF3 phosphorylation. Our inference model also predicted that activity and expression for STAT2, IRF7, and RELA correlated with innate stimulation, as expected, given their well-defined roles in the innate response (Cao, 2016).Once TF activity was estimated, the network structure prior was then used to bias model selection of TF-to-gene target regulatory interactions towards edges with prior information during network inference ( Figure 3D). To improve the overall performance and stability of network inference, several aspects of the method (and key inputs) were tested, such as: different sources of priors (publicly available ENCODE data vs our ATAC-seq data), different TF binding motif databases (HOCOMOCO (Kulakovskiy et al., 2013) vs CisBp 2.0 (Weirauch et al., 2014)), and different model selection methods (Bayesian Best Subset Regression (BBSR) vsElastic Net (EN)). Additionally, since every computational run is subject to stochasticity in the inference procedure, we evaluated whether network performance could be improved by combining hundreds of individual computation runs (selecting model components such as TF activity estimates that were stable across random subsamples of the structure prior; see STAR Methods). To evaluate the performance of networks inferred using the parameters described above, we used area...