This study aims to characterize tumor-infiltrating macrophages (TAMs), myeloid-derived suppressor cells (MDSC), and the related molecular milieu regulating anti-tumor immunity in limited-stage neuroendocrine (NE)-high and NE-low small cell lung cancer. Primary tumors and matched lymph node (LN) metastases of 32 resected, early-stage SCLC patients were analyzed by immunohistochemistry (IHC) with antibodies against pan-macrophage marker CD68, M2-macrophage marker CD163, and MDSC marker CD33. Area-adjusted cell counting on TMAs showed that TAMs are the most abundant cell type in the TME, and their number in tumor nests exceeds the number of CD3 + T-cells (64% vs. 38% in NE-low and 71% vs. 18% in NE-high). Furthermore, the ratio of CD163-expressing M2-polarized TAMs in tumor nests was significantly higher in NE-low vs. NE-high tumors (70% vs. 31%). TAM density shows a strong positive correlation with CD45 and CD3 in tumor nests, but not in the stroma. fGSEA analysis on a targeted RNAseq oncological panel of 2560 genes showed that NE-high tumors exhibited increased enrichment in pathways related to cell proliferation, whereas in NE-low tumors, immune response pathways were significantly upregulated. Interestingly, we identified a subset of NE-high tumors representing an immune-oasis phenotype, but with a different gene expression profile compared to NE-low tumors. In contrast, we found that a limited subgroup of NE-low tumors is immune-deserted and express distinct cellular pathways from NE-high tumors. Furthermore, we identified potential molecular targets based on our expression data in NE-low and immune-oasis tumor subsets, including CD70, ANXA1, ITGB6, TP63, IFI27, YBX3 and CXCR2.

Short linear motifs (SLiMs) are small interaction modules within intrinsically disordered regions of proteins that interact with specific domains, and thereby regulate numerous biological processes. Their limited sequence information leads to frequent false positive hits in computational and experimental SLiM identification methods. We present SLiMMine, a deep learning-based method to identify SLiMs in the human proteome. By refining the annotations of known motif classes, we created a high-quality training dataset. Using protein embeddings and neural networks, SLiMMine reliably predicts novel SLiM candidates in known classes, eliminates ~80% of the pattern matching-based hits as false-positives, furthermore, it also functions as a discovery tool to find uncharacterized SLiMs based on optimal sequence environment. Finally, narrowing the broad interactor-domain definitions of known SLiM classes to specific human proteins enables more precise linking of predicted SLiMs to known protein-protein interactions. SLiMMine is available as a user-friendly, multipurpose web server at https://slimmine.pbrg.hu/ .