Network Controllability Reveals Key Mitigation Points for Tumor-Promoting Signaling in Tumor-Educated Platelets

Int J Mol Sci 2025 AI 10 Explanations View Original
Original Paper (PDF)

Unable to display PDF. Download it here or view on PMC.

Plain-English Explanations
Pages 1-2
Platelets Educated by Tumors

Cancer reprograms platelets. Tumor cells alter platelet gene expression, protein content, and signaling activity through a process called platelet education, transforming normal platelets into tumor-educated platelets (TEPs) that actively support cancer spread.

TEPs promote metastasis by releasing growth factors, shielding circulating tumor cells from immune attack, inducing epithelial-to-mesenchymal transition, and helping cancer cells adhere to and cross blood vessel walls at distant sites.

This study analyzed a large-scale TEP gene expression dataset (GSE89843) originally collected from 402 NSCLC patients and 377 non-cancer donors to identify which signaling proteins drive pro-metastatic platelet behavior and can be targeted by existing drugs.

TL;DR: Tumor-educated platelets are reprogrammed by cancer cells and actively promote metastasis through altered signaling, making them a compelling therapeutic target.
Pages 3, 20, 21
Building a TEP Signaling Network

Multi-step transcriptome pipeline. Raw RNA sequencing data were processed with FastQC, fastp, and kallisto for quality control and quantification, followed by normalization and differential expression analysis using DESeq2, edgeR, and limma-voom.

To remove unwanted technical variation, the team applied RUVSeq using 1,491 empirically stable genes, then classified genes with log2 fold change beyond 0.58 and adjusted p-value below 0.05 as differentially expressed.

Differential gene expression data were integrated with high-quality, directed, and signed protein-protein interaction data from the OmniPath database, filtering from 134,282 raw interactions down to 12,963 high-confidence ones to build a platelet-specific signaling network of 962 interactions among 401 proteins.

A second, broader interactome was constructed using all available protein interactions, yielding 1,638 interactions among 600 platelet proteins, allowing proximity-based scoring to identify high-influence nodes beyond the curated directed network.

TL;DR: The team built a high-confidence TEP-specific signaling network by integrating RNA-seq transcriptomics with curated protein interaction data and applied multiple computational strategies to find therapeutic targets.
Pages 3-4
Transcriptome Shifts in NSCLC Platelets

219 genes changed in tumor-educated platelets. Compared to platelets from non-cancer donors, TEPs from NSCLC patients showed 111 upregulated and 108 downregulated genes, revealing a distinct cancer-associated transcriptional program.

Upregulated genes were enriched in pathways for extracellular matrix (ECM) interactions, focal adhesion, cytoskeleton organization, calcium signaling, and platelet activation, consistent with increased adhesive and pro-metastatic platelet behavior.

Downregulated genes were concentrated in immune-related processes including T-cell and B-cell signaling, NF-kB signaling, antigen processing, apoptosis, and ribosomal functions, suggesting TEPs actively suppress immune surveillance while boosting structural remodeling.

The TEP expression profile most closely resembled platelets from chronic pancreatitis patients, reflecting shared inflammatory signatures, while profiles from multiple sclerosis patients most closely resembled healthy controls.

TL;DR: TEPs in NSCLC upregulate ECM and adhesion genes while suppressing immune and apoptotic pathways, revealing a transcriptional signature tuned for metastasis support.
Pages 5-7
Gene Modules and Drug Targets

K-means clustering identified nine gene modules. Four modules were consistently upregulated in NSCLC TEPs: Module 1 linked to protein homeostasis, Module 3 tied to platelet activation and hemostasis, Module 7 associated with metabolic adaptation, and Module 8 connected to cytoskeletal remodeling.

Module 3, the platelet activation module, was uniquely enriched in Reactome pathways including hemostasis, platelet activation and aggregation, platelet degranulation, and calcium response, making it the most disease-relevant cluster for therapeutic targeting.

Within these upregulated modules, several druggable proteins were identified, including ITGA2B (targeted by abciximab and tirofiban), FCGR2A (targeted by cetuximab and bevacizumab), CACNA1D, and ATOX1, all with existing FDA-approved drug options.

Gene Set Enrichment Analysis further identified ECM proteoglycan and cell junction organization pathways as highly enriched, spotlighting additional targets such as filamin A (FLNA), amyloid-beta precursor protein (APP), and fibronectin (FN1).

TL;DR: Clustering analysis resolved four upregulated gene modules in NSCLC TEPs, each containing FDA-targetable proteins with direct roles in platelet activation and cancer-promoting adhesion.
Pages 8-10
Network Controllability Analysis

Controllability identifies which nodes can steer the network. The research team applied graph-theory-based maximum matching to determine which proteins are essential for transitioning the platelet signaling network from any initial state to a desired therapeutic state.

The analysis classified 86 proteins as critical nodes, meaning they are required in every minimal driver set and carry high control capacity. An additional 196 nodes were intermittent, and 119 were redundant with alternative signaling paths covering their role.

Separately, 62 proteins were classified as indispensable nodes, meaning their removal disrupts signal transduction through the network. These nodes had significantly higher degree, closeness centrality, and betweenness centrality than critical nodes, reflecting their structural importance.

Importantly, differential expression levels did not differ significantly between critical and indispensable nodes, confirming that network position rather than expression magnitude determines which proteins are most essential for therapeutic control.

TL;DR: Network controllability analysis classified 86 critical and 62 indispensable proteins in the TEP signaling network, identifying the most strategically important intervention points regardless of expression fold change.
Pages 10-11
Central TEP Subnetwork and Top Drugs

Shortest-path mapping created a focused therapeutic subnetwork. By calculating edge weights from fold-change data and tracing shortest paths from critical to indispensable nodes using Dijkstra's algorithm, the team built a central TEP subnetwork of 188 nodes and 501 interactions.

Within this subnetwork, 22 indispensable nodes were identified, 16 of which are known drug targets addressable by 75 existing compounds. The top drugs by number of targets covered were fostamatinib (5 targets), minocycline (4 targets), and acetylsalicylic acid or aspirin (3 targets).

Fostamatinib, an FDA-approved SYK kinase inhibitor, emerged as the leading single agent because it simultaneously inhibits JAK2, PTK2, CAMK1, MAPK14, and PRKCD, all key nodes in platelet activation and ITAM signaling pathways.

After filtering for drug interaction safety, the team proposed a three-drug combination of fostamatinib, aducanumab (targeting APP), and aspirin (targeting CASP3) as a clinically actionable multi-target regimen to suppress TEP-driven metastasis.

TL;DR: Fostamatinib targeting five key network nodes emerged as the top single agent, and a combination with aducanumab and aspirin was proposed as a safe, multi-target TEP-suppressing regimen.
Pages 12-13
Expanded Interactome Reveals New Targets

Broader network analysis uncovered additional high-value nodes. Removing the directed-network restriction and including all protein-protein interactions expanded the platelet network to 1,638 interactions among 600 proteins, roughly 1.5 times larger than the curated network.

Node weights were assigned by combining differential expression with network connectivity (degree). The top-weighted proteins included SRC, PRKCA, LYN, STAT1, and PIK3R1, all well-known kinases in platelet and immune signaling.

A proximity-based scoring method then ranked proteins by closeness to high-weight nodes. The highest-scoring proteins were SHIP1 (INPP5D), PLCG2, and CBL, and this approach revealed additional targetable nodes including P2RY12, FCGR2A, BTK, and JAK1 and JAK3.

JAK1 was the most pharmacologically accessible, with ten FDA-approved inhibitors including ruxolitinib, tofacitinib, and baricitinib. BTK, P2RY12, and FCGR2A were also targetable by multiple approved drugs, broadening the therapeutic options beyond the curated network results.

TL;DR: Expanding the network to include all protein interactions revealed additional high-value drug targets including JAK1, BTK, and P2RY12, each addressable by multiple FDA-approved agents.
Pages 17-18
Five High-Confidence Therapeutic Targets

Four complementary strategies converged on five key genes. Across transcriptomic clustering, GSEA, network controllability, and proximity scoring, the proteins ITGA2B, FLNA, GRB2, FCGR2A, and APP consistently emerged as central to TEP-mediated cancer promotion in NSCLC.

ITGA2B encodes the alpha-IIb subunit of the platelet integrin complex that mediates aggregation and bridges platelets to tumor cells via fibrinogen and fibronectin, and its high expression in NSCLC TEPs makes it both a diagnostic marker and therapeutic target.

FCGR2A and GRB2 both operate in the ITAM signaling axis triggered when tumor-derived IgG and immune complexes engage platelets. Selectively disrupting this pathway could suppress TEP-driven cancer progression while preserving normal platelet hemostatic function.

The five key genes were validated across three independent datasets including two NSCLC cohorts, a breast cancer cohort, and a pancreatic cancer cohort, with comparable log fold changes and statistical significance across all datasets despite some variation in ranking.

TL;DR: Five proteins, ITGA2B, FLNA, GRB2, FCGR2A, and APP, were validated across multiple independent cancer datasets as consistently dysregulated, FDA-targetable hubs of tumor-promoting platelet signaling.
Pages 18-19
Fostamatinib as Anti-Metastatic Agent

Fostamatinib targets the same pathways tumors use to hijack platelets. Tumor cell-induced platelet aggregation is mediated through ITAM-coupled receptors GPVI, CLEC-2, and FCGR2A. Fostamatinib inhibits SYK, the kinase directly downstream of all three receptors, selectively disrupting cancer-driven platelet activation.

Beyond platelet effects, fostamatinib demonstrated direct anti-cancer activity in vitro by inhibiting receptor tyrosine kinases, the PI3K-AKT pathway, and immune checkpoints PD-L1 and CD47, reducing tumor cell proliferation and inducing apoptosis in NSCLC and AML cell lines.

Clinical phase I and II trials have evaluated fostamatinib in NSCLC, colorectal, ovarian, thyroid, and renal cell carcinoma, and its safety and efficacy in immune thrombocytopenia are already established, reducing barriers to repurposing for cancer-associated platelet hyperactivation.

Aspirin complements fostamatinib by inhibiting COX-1 to reduce thromboxane A2, limiting inflammation-driven platelet aggregation, and releasing T-cells from TXA2-mediated immunosuppression, thereby adding anti-metastatic and immunomodulatory effects to the combination.

TL;DR: Fostamatinib is proposed as a dual-action anti-platelet and anti-cancer agent that selectively blocks the ITAM pathway tumors exploit, supported by existing clinical trial evidence and complemented by aspirin and aducanumab.
Pages 19, 23
Precision Targeting of Metastasis-Promoting Platelets

This study provides a novel in silico roadmap for TEP-targeted therapy. By combining transcriptomics, network controllability, and proximity scoring, the team identified FDA-approved drugs that can disrupt the specific signaling reprogramming that cancer induces in platelets without broadly impairing normal hemostasis.

The proposed combination of fostamatinib, aducanumab, and aspirin operates through non-overlapping mechanisms and has no known cross-drug interactions, enabling a safer low-dose combination regimen that could be tested directly in preclinical NSCLC models.

Future experimental work should include functional validation of the five key target genes, direct measurement of fostamatinib effects on platelets from NSCLC patients, and construction of larger-scale NSCLC signaling networks to further refine these findings.

The broader implications extend beyond lung cancer, as the top five TEP target genes showed consistent dysregulation in breast and pancreatic cancer datasets as well, suggesting that targeting platelet education may represent a cross-cancer strategy against metastasis.

TL;DR: A network-based in silico analysis identified fostamatinib plus aducanumab plus aspirin as a promising combination to selectively disrupt tumor-educated platelet signaling and limit NSCLC metastasis.
Citation: Open Access, 2025. Available at: PMC12609506.