scAEGAN: Unification of single-cell genomics data by adversarial learning of latent space correspondences.

Recent progress in Single-Cell Genomics has produced different library protocols and techniques for molecular profiling. We formulate a unifying, data-driven, integrative, and predictive methodology for different libraries, samples, and paired-unpaired data modalities. Our design of scAEGAN includes an autoencoder (AE) network integrated with adversarial learning by a cycleGAN (cGAN) network. The AE learns a low-dimensional embedding of each condition, whereas the cGAN learns a non-linear mapping between the AE representations. We evaluate scAEGAN using simulated data and real scRNA-seq datasets, different library preparations (Fluidigm C1, CelSeq, CelSeq2, SmartSeq), and several data modalities as paired scRNA-seq and scATAC-seq. The scAEGAN outperforms Seurat3 in library integration, is more robust against data sparsity, and beats Seurat 4 in integrating paired data from the same cell. Furthermore, in predicting one data modality from another, scAEGAN outperforms Babel. We conclude that scAEGAN surpasses current state-of-the-art methods and unifies integration and prediction challenges.

Deconvolution of the hematopoietic stem cell microenvironment reveals a high degree of specialization and conservation.

Understanding the regulation of normal and malignant human hematopoiesis requires comprehensive cell atlas of the hematopoietic stem cell (HSC) regulatory microenvironment. Here, we develop a tailored bioinformatic pipeline to integrate public and proprietary single-cell RNA sequencing (scRNA-seq) datasets. As a result, we robustly identify for the first time 14 intermediate cell states and 11 stages of differentiation in the endothelial and mesenchymal BM compartments, respectively. Our data provide the most comprehensive description to date of the murine HSC-regulatory microenvironment and suggest a higher level of specialization of the cellular circuits than previously anticipated. Furthermore, this deep characterization allows inferring conserved features in human, suggesting that the layers of microenvironmental regulation of hematopoiesis may also be shared between species. Our resource and methodology is a stepping-stone toward a comprehensive cell atlas of the BM microenvironment.

A Proteomic Atlas of Lineage and Cancer-Polarized Expression Modules in Myeloid Cells Modeling Immunosuppressive Tumor-Infiltrating Subsets.

Monocytic and granulocytic myeloid-derived suppressor cells together with tumor-infiltrating macrophages constitute the main tumor-infiltrating immunosuppressive myeloid populations. Due to the phenotypic resemblance to conventional myeloid cells, their identification and purification from within the tumors is technically difficult and makes their study a challenge. We differentiated myeloid cells modeling the three main tumor-infiltrating types together with uncommitted macrophages, using ex vivo differentiation methods resembling the tumor microenvironment. The phenotype and proteome of these cells was compared to identify linage-dependent relationships and cancer-specific interactome expression modules. The relationships between monocytic MDSCs and TAMs, monocytic MDSCs and granulocytic MDSCs, and hierarchical relationships of expression networks and transcription factors due to lineage and cancer polarization were mapped. Highly purified immunosuppressive myeloid cell populations that model tumor-infiltrating counterparts were systematically analyzed by quantitative proteomics. Full functional interactome maps have been generated to characterize at high resolution the relationships between the three main myeloid tumor-infiltrating cell types. Our data highlights the biological processes related to each cell type, and uncover novel shared and differential molecular targets. Moreover, the high numbers and fidelity of ex vivo-generated subsets to their natural tumor-shaped counterparts enable their use for validation of new treatments in high-throughput experiments.

Characterization of Macrophage Endogenous S-Nitrosoproteome Using a Cysteine-Specific Phosphonate Adaptable Tag in Combination with TiO2 Chromatography.

Protein S-nitrosylation is a cysteine post-translational modification mediated by nitric oxide. An increasing number of studies highlight S-nitrosylation as an important regulator of signaling involved in numerous cellular processes. Despite the significant progress in the development of redox proteomic methods, identification and quantification of endogeneous S-nitrosylation using high-throughput mass-spectrometry-based methods is a technical challenge because this modification is highly labile. To overcome this drawback, most methods induce S-nitrosylation chemically in proteins using nitrosylating compounds before analysis, with the risk of introducing nonphysiological S-nitrosylation. Here we present a novel method to efficiently identify endogenous S-nitrosopeptides in the macrophage total proteome. Our approach is based on the labeling of S-nitrosopeptides reduced by ascorbate with a cysteine specific phosphonate adaptable tag (CysPAT), followed by titanium dioxide (TiO2) chromatography enrichment prior to nLC-MS/MS analysis. To test our procedure, we performed a large-scale analysis of this low-abundant modification in a murine macrophage cell line. We identified 569 endogeneous S-nitrosylated proteins compared with 795 following exogenous chemically induced S-nitrosylation. Importantly, we discovered 579 novel S-nitrosylation sites. The large number of identified endogenous S-nitrosylated peptides allowed the definition of two S-nitrosylation consensus sites, highlighting protein translation and redox processes as key S-nitrosylation targets in macrophages.

A core of kinase-regulated interactomes defines the neoplastic MDSC lineage.

Myeloid-derived suppressor cells (MDSCs) differentiate from bone marrow precursors, expand in cancer-bearing hosts and accelerate tumor progression. MDSCs have become attractive therapeutic targets, as their elimination strongly enhances anti-neoplastic treatments. Here, immature myeloid dendritic cells (DCs), MDSCs modeling tumor-infiltrating subsets or modeling non-cancerous (NC)-MDSCs were compared by in-depth quantitative proteomics. We found that neoplastic MDSCs differentially expressed a core of kinases which controlled lineage-specific (PI3K-AKT and SRC kinases) and cancer-induced (ERK and PKC kinases) protein interaction networks (interactomes). These kinases contributed to some extent to myeloid differentiation. However, only AKT and ERK specifically drove MDSC differentiation from myeloid precursors. Interfering with AKT and ERK with selective small molecule inhibitors or shRNAs selectively hampered MDSC differentiation and viability. Thus, we provide compelling evidence that MDSCs constitute a distinct myeloid lineage distinguished by a "kinase signature" and well-defined interactomes. Our results define new opportunities for the development of anti-cancer treatments targeting these tumor-promoting immune cells.