Stephen K. Anderson

Estrogen receptor-negative (ER-) breast tumors are aggressive subtypes with limited therapeutic options. Cyclooxygenase-2 promotes chronic inflammation, immune suppression, and therapeutic resistance. Whether COX2 signaling organizes tumor microenvironments across oxygen gradients remains unclear. Restricted Exchange Environment Chambers (REECs) were used to explore the COX2 impact on spatial tumor architecture across oxygen gradients. Baseline COX2+, B7H4+, and CD44v6+ cells exhibited distinct distributions, with CD44v6 predominance. Inflammatory stimulation increased and spatially reorganized COX2+, B7H4+, and COX2+CD44v6+ populations. While COX2+ cells occupied oxygenated-to-intermediate zones, B7H4+ populations were enriched in intermediate-to-hypoxic regions, and CD44v6 remained broadly distributed. Diclofenac disrupted this organization and reduced clustering of COX2+, B7H4+, and COX2+CD44v6+ phenotypes. Poor-outcome patients showed high COX2 at the tumor-stroma interface as well as altered COX2+, B7H4+, and CD44v6+ clustering phenotypes. These findings identify COX2-dependent microenvironments as clinically relevant features of therapeutic resistance. Diclofenac preferentially depleted hypoxic compartments preserved following cisplatin treatment alone, supporting the targeting of spatially organized inflammatory niches to improve therapeutic response.

Plasmacytoid dendritic cells (pDCs) produce robust type I interferons (IFN-I) within hours of viral sensing, while epithelial cells at mucosal surfaces mount a delayed response dominated by type III interferons (IFN-III). Both cell types express pattern recognition receptors that activate similar downstream transcription factors, yet they produce distinct subsets of IFNs. The mechanisms underlying these differences have remained unclear. Here, using Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq) in primary human pDCs and intestinal epithelial cells, we show that IFN-I and IFN-III gene loci carry opposing, constitutively established chromatin accessibility landscapes that determine cell-type-specific interferon induction. The IFN-I locus is broadly open in pDCs and constitutively closed in epithelial cells, while the IFN-III locus displays the reciprocal pattern. Motif enrichment analysis of accessible regions at the IFN-I locus in pDCs revealed unexpected and significant enrichment of ETS family binding motifs alongside IRF motifs, which determines the cell-type-specific locus accessibility. The ETS factor PU.1 and IRF8 bind IFN-I promoters, at composite ETS-IRF elements positioned at the Positive Regulatory Domain IV (PRDIV) site of IFNB1 promoter and adjacent to the TATA-proximal IRF motif of IFNA promoters. IFNL gene promoters lack ETS recognition sequences. PU.1-IRF8 composite factor binding extends across intergenic regions of the IFN-I locus, where candidate enhancer elements marked by H3K4me1, H3K27ac, and RNA Pol II occupancy were identified. We propose that this network of ETS-IRF composite-element-anchored enhancers maintains the IFN-I locus in a constitutively poised state in pDCs, licensing the rapid and robust IFN-I response that defines pDCs. In epithelial cells, the absence of PU.1 and IRF8 renders the IFN-I locus epigenetically silent, while the IFN-III locus is constitutively open. Despite this, the delayed IFN-III gene expression in epithelial cells is due to intrinsically weaker promoter activity relative to IFN-I. These findings reveal that the divergent IFN induction of pDCs and epithelial cells is determined by the chromatin architecture prior to infection. Overall, these observations show that lineage-specific ETS-IRF regulatory factors and promoter strength determine the cell-type-specific IFN activation.

Tumor analysis of NOS2 and COX2.

Spatial dot plot of CD8+/-NOS2+/-COX2+/- phenotypes in tumor from deceased patient.

Immune and tumor quantification

Modulation of EpCAM and CD44v6+.

Supplementary Fig. 4. Supplementary Fig. 4. Spatial UMAP analysis of CD8+/-NOS2+/-COX2+/- phenotypes in Deceased vs Alive patient tumors. Single cell neighborhood profile summary.

Supplementary Table I summarizes pathological features of tumor immune microenvironment where NOS2+ inflamed regions are significantly higher in tumors from Deceased patients.