A104-23 Differential Transcription Factor Activities in Hyper and Hypo-Inflammatory ARDS Endotypes - Potential Therapeutic Insights From Regulatory Genomics

Rationale

ARDS can be categorized as hyperinflammatory or hypoinflammatory endotypes using blood biomarkers with the hyperinflammatory associated with higher mortality (1, 2); however, the underlying mechanism of these two endotypes are less-well known. We recently profiled the functional activity of transcription factors (TF) by measuring the genomic cis-regulatory elements (cRE) from the blood of COVID-ARDS patients (3). We were able to identify TF programs (TFP) associated with severe disease, highlighting potential therapeutic molecular targets that regulate downstream inflammatory programs during ARDS pathogenesis (3). We developed a novel machine learning pipeline that can identify TF activity in other cohorts as long as they have a transcriptomic dataset. To elucidate this, we applied our method to a recent ROSE-PETAL transcriptomic cohort by performing an observational analysis applying our TFPs to profile the transcription factor (TF) activity of the hyper/hypo-inflammatory endotypes. We hypothesized that the two endotypes would have different TF activation profiles.

Methods

We profiled cRE activity (capped-short RNA-seq) and transcriptomics (total RNA-seq) from matching specimens from a UCSD ARDS cohort (N = 97), enabling us to leverage supervised machine-learning (Support Vector Machine) to identify protein-coding genes whose expression reliably predicted TF-specific regulatory activity with paired empiric data. We used the 432 genes that predicted the 14 TFPs with high confidence (0.682 < R2< 0.957) to assess TFPs activities in the blood transcriptomics of ARDS patients in the ROSE-PETAL trial (n = 134, Sinha 2024). The activity of each TFP was calculated by averaging the expression of genes predictive for that TFP. We then computed the differential TFP activity between patients annotated as hyper- or hypo-inflammatory and expressed in log2 fold change (LogFC).

Results

Two TF programs showed higher activity in the hyperinflammatory ARDS phenotype. These are the inflammatory STAT/BCL6 and proliferation-related E2F TF programs. Two TFPs had higher activity in the hypoinflammatory group, including the anti-viral type 1 interferon (T1ISRE/STAT) and TF for steroid-sensing glucocorticoid receptors (glucocorticoid responsive elements, GRE).

Conclusions

Our novel approach identified a transcription factor activation profile for the hyperinflammatory group that correlates with disease severity. STAT/BCL6 TFP had increased relative activity, which suggests a potential role for FDA-approved STAT inhibitors in modulating immune response in that group. These findings have exciting implications in developing effective targeted therapies in patients with ARDS.

This abstract is funded by: NIH5T32