Decosus: An R Framework for Universal Integration of Cell Proportion Estimation Methods.

The assessment of the cellular heterogeneity and abundance in bulk tissue samples is essential for characterising cellular and organismal states. Computational approaches to estimate cellular abundance from bulk RNA-Seq datasets have variable performances, often requiring benchmarking matrices to select the best performing methods for individual studies. However, such benchmarking investigations are difficult to perform and assess in typical applications because of the absence of gold standard/ground-truth cellular measurements. Here we describe Decosus, an R package that integrates seven methods and signatures for deconvoluting cell types from gene expression profiles (GEP). Benchmark analysis on a range of datasets with ground-truth measurements revealed that our integrated estimates consistently exhibited stable performances across datasets than individual methods and signatures. We further applied Decosus to characterise the immune compartment of skin samples in different settings, confirming the well-established Th1 and Th2 polarisation in psoriasis and atopic dermatitis, respectively. Secondly, we revealed immune system-related UV-induced changes in sun-exposed skin. Furthermore, a significant motivation in the design of Decosus is flexibility and the ability for the user to include new gene signatures, algorithms, and integration methods at run time.

An Antibody Targeting ICOS Increases Intratumoral Cytotoxic to Regulatory T-cell Ratio and Induces Tumor Regression.

The immunosuppressive tumor microenvironment constitutes a significant hurdle to immune checkpoint inhibitor responses. Both soluble factors and specialized immune cells, such as regulatory T cells (Treg), are key components of active intratumoral immunosuppression. Inducible costimulatory receptor (ICOS) can be highly expressed in the tumor microenvironment, especially on immunosuppressive Treg, suggesting that it represents a relevant target for preferential depletion of these cells. Here, we performed immune profiling of samples from tumor-bearing mice and patients with cancer to demonstrate differential expression of ICOS in immune T-cell subsets in different tissues. ICOS expression was higher on intratumoral Treg than on effector CD8 T cells. In addition, by immunizing an Icos knockout transgenic mouse line expressing antibodies with human variable domains, we selected a fully human IgG1 antibody called KY1044 that bound ICOS from different species. We showed that KY1044 induced sustained depletion of ICOShigh T cells but was also associated with increased secretion of proinflammatory cytokines from ICOSlow effector T cells (Teff). In syngeneic mouse tumor models, KY1044 depleted ICOShigh Treg and increased the intratumoral TEff:Treg ratio, resulting in increased secretion of IFNγ and TNFα by TEff cells. KY1044 demonstrated monotherapy antitumor efficacy and improved anti-PD-L1 efficacy. In summary, we demonstrated that using KY1044, one can exploit the differential expression of ICOS on T-cell subtypes to improve the intratumoral immune contexture and restore an antitumor immune response.