Topology preserving stratification of tissue neoplasticity using Deep Neural Maps and microRNA signatures

Background Accurate cancer classification is essential for correct treatment selection and better prognostication. microRNAs (miRNAs) are small RNA molecules that negatively regulate gene expression, and their dyresgulation is a common disease mechanism in many cancers. Through a clearer understanding of miRNA dysregulation in cancer, improved mechanistic knowledge and better treatments can be sought. Results We present a topology-preserving deep learning framework to study miRNA dysregulation in cancer. Our study comprises miRNA expression profiles from 3685 cancer and non-cancer tissue samples and hierarchical annotations on organ and neoplasticity status. Using unsupervised learning, a two-dimensional topological map is trained to cluster similar tissue samples. Labelled samples are used after training to identify clustering accuracy in terms of tissue-of-origin and neoplasticity status. In addition, an approach using activation gradients is developed to determine the attention of the networks to miRNAs that drive the clustering. Using this deep learning framework, we classify the neoplasticity status of held-out test samples with an accuracy of 91.07%, the tissue-of-origin with 86.36%, and combined neoplasticity status and tissue-of-origin with an accuracy of 84.28%. The topological maps display the ability of miRNAs to recognize tissue types and neoplasticity status. Importantly, when our approach identifies samples that do not cluster well with their respective classes, activation gradients provide further insight in cancer subtypes or grades. Conclusions An unsupervised deep learning approach is developed for cancer classification and interpretation. This work provides an intuitive approach for understanding molecular properties of cancer and has significant potential for cancer classification and treatment selection.

Single-cell RNA-seq of out-of-thaw mesenchymal stromal cells shows tissue-of-origin differences and inter-donor cell-cycle variations

Background Human Mesenchymal stromal cells (hMSCs) from various tissue sources are widely investigated in clinical trials. These MSCs are often administered to patients immediately after thawing the cryopreserved product (out-of-thaw), yet little is known about the single-cell transcriptomic landscape and tissue-specific differences of out-of-thaw human MSCs. Methods 13 hMSC samples derived from 10 “healthy” donors were used to assess donor variability and tissue-of-origin differences in single-cell gene expression profiles. hMSCs derived and expanded from the bone marrow (BM) or cord tissue (CT) underwent controlled-rate freezing for 24 h. Cells were then transferred to the vapor phase of liquid nitrogen for cryopreservation. hMSCs cryopreserved for at least one week, were characterized immediately after thawing using a droplet-based single-cell RNA sequencing method. Data analysis was performed with SC3 and SEURAT pipelines followed by gene ontology analysis. Results scRNA-seq analysis of the hMSCs revealed two major clusters of donor profiles, which differ in immune-signaling, cell surface properties, abundance of cell-cycle related transcripts, and metabolic pathways of interest. Within-sample transcriptomic heterogeneity is low. We identified numerous differentially expressed genes (DEGs) that are associated with various cellular functions, such as cytokine signaling, cell proliferation, cell adhesion, cholesterol/steroid biosynthesis, and regulation of apoptosis. Gene-set enrichment analyses indicated different functional pathways in BM vs. CT hMSCs. In addition, MSC-batches showed significant variations in cell cycle status, suggesting different proliferative vs. immunomodulatory potential. Several potential transcript-markers for tissue source differences were identified for further investigation in future studies. In functional assays, both BM and CT MSCs suppressed macrophage TNFα secretion upon interferon stimulation. However, differences between donors, tissue-of-origin, and cell cycle are evident in both TNF suppression and cytokine secretion. Conclusions This study shows that donor differences in hMSC transcriptome are minor relative to the intrinsic differences in tissue-of-origin. hMSCs with different transcriptomic profiles showed potential differences in functional characteristics. These findings contribute to our understanding of tissue origin-based differences in out-of-thaw therapeutic hMSC products and assist in the identification of cells with immune-regulatory or survival potential from a heterogeneous MSC population. Our results form the basis of future studies in correlating single-cell transcriptomic markers with immunomodulatory functions.

EPCO-12. STING IS SILENCED IN GBM BY PROMOTER METHYLATION THAT IS ESTABLISHED BY TISSUE OF ORIGIN

STING (Stimulator of interferon genes) is an initiator of endogenous antitumor immunity. Loss of STING expression has been observed in various cancer types. Here we describe a mechanism of STING suppression that is common to brain tumors, occurring through regional promoter hypermethylation (RPH) near its transcriptional start site. The RPH-mediated suppression of STING expression is selective to normal brain and neuroectoderm-derived tumors, but not present in most other tumor types and cells in the brain tumor microenvironment. A pan-cancer comparison revealed that STING RPH levels in tumors correlate with the respective tissue of origin, including tumors metastasizing to the brain. We demonstrate that STING RPH suppression in glioma cells may be reversed by treatment with decitabine, a DNA methyltransferase inhibitor that is used in the treatment of hematologic malignancies. Collectively, our results suggest that the epigenetic silencing of STING occurs early in brain development and may provide an immunosuppressive context for the genesis of brain tumors. Furthermore, our findings that STING suppression in brain tumor cells can be reversed by epigenetic modulation have significant therapeutic implications for the treatment of glioblastoma and potentially other STING-silenced, immunologically-cold cancers.

Differential expression of RNASE4 in cancer of the vulva.

In these brief notes we document work using published microarray data (1, 2) to pioneer integrative transcriptome analysis comparing vulvar carcinoma to its tissue of origin, the vulva. We report the differential expression of ribonuclease A family member 4, encoded by RNASE4, in cancer of the vulva. RNASE4 may be of pertinence to understanding transformation and disease progression in vulvar cancer (3).

Differential expression of LTBP4 in cancer of the vulva.

In these brief notes we document work using published microarray data (1, 2) to pioneer integrative transcriptome analysis comparing vulvar carcinoma to its tissue of origin, the vulva. We report the differential expression of latent transforming growth factor beta binding protein 4, encoded by LTBP4, in cancer of the vulva. LTBP4 may be of pertinence to understanding transformation and disease progression in vulvar cancer (3).

Differential expression of ACACB in cancer of the vulva.

In these brief notes we document work using published microarray data (1, 2) to pioneer integrative transcriptome analysis comparing vulvar carcinoma to its tissue of origin, the vulva. We report the differential expression of acetyl-CoA carboxylase beta, encoded by ACACB, in cancer of the vulva. ACACB may be of pertinence to understanding transformation and disease progression in vulvar cancer (3).

Differential expression of KLHL3 in cancer of the vulva.

In these brief notes we document work using published microarray data (1, 2) to pioneer integrative transcriptome analysis comparing vulvar carcinoma to its tissue of origin, the vulva. We report the differential expression of kelch-like family member 3, encoded by KLHL3, in cancer of the vulva. KLHL3 may be of pertinence to understanding transformation and disease progression in vulvar cancer (3).

Differential expression of CHPT1 in cancer of the vulva.

In these brief notes we document work using published microarray data (1, 2) to pioneer integrative transcriptome analysis comparing vulvar carcinoma to its tissue of origin, the vulva. We report the differential expression of choline phosphotransferase 1, encoded by CHPT1, in cancer of the vulva. CHPT1 may be of pertinence to understanding transformation and disease progression in vulvar cancer (3).

Differential expression of ZNF629 in cancer of the vulva.

In these brief notes we document work using published microarray data (1, 2) to pioneer integrative transcriptome analysis comparing vulvar carcinoma to its tissue of origin, the vulva. We report the differential expression of zinc finger protein 629, encoded by ZNF629, in cancer of the vulva. ZNF629 may be of pertinence to understanding transformation and disease progression in vulvar cancer (3).

Differential expression of LHX6 in cancer of the vulva.

In these brief notes we document work using published microarray data (1, 2) to pioneer integrative transcriptome analysis comparing vulvar carcinoma to its tissue of origin, the vulva. We report the differential expression of LIM homeobox 6, encoded by LHX6, in cancer of the vulva. LHX6 may be of pertinence to understanding transformation and disease progression in vulvar cancer (3).