Mouse melanoma B16-F10 cells (obtained from ATCC) were grown in Roswell Park Memorial Institute (RPMI) 1640 (Gibco, Grand Island, NY, USA, 11875093), mixed with 10% fetal bovine serum (FBS, Gibco, 26140079) and 1% penicillin/streptomycin (Sigma, San Francisco, CA, USA, P4333), in 37°C, 5% CO₂.

The National Research Council’s Guide follows all mouse experiments for the Care and Use of Laboratory Animals. 5×10 5 B16-F10 cells were injected subcutaneously into the flank of legs of the C57BL/6 mice (10 weeks age, male). Anesthesia regimens were used as reported (18).

Blood draws followed the reported protocol. The blood was mixed with an equal volume of pH 6.5 Tyrode’s buffer and centrifuged at 70g, 20 min, room temperature (RT), without brake. The 1-ml plasma-rich platelets (PRP) were separated into two parts: one was washed with 1.5 ml pH 7.4 Tyrode’s buffer and centrifuged at 1,200g for 10 min with brake, RT. The supernatant was discarded, and the pellet was resuspended in 400 μl pH 6.5 Tyrode’s buffer to obtain the TEPs. Another half of the PRP would be centrifuged directly at 1,200g for 10 min with brake, RT. The supernatant was then mouse plasma.

Glass coverslip ibidi sticky slides were prepared as reported (19). 104/μl mouse platelet together with 4 μM ADP (Aladdin) 30 μg/ml casein (Sigma), 40 μg/ml human fibrinogen (Sigma), 200 μM Cacl₂, and 2 μM U46619 (Enzo Life Sciences, Farmingdale, NY, USA) was mounted together in a total volume of 240 μl pH 7.4 Tyrode’s buffer. We used Fiji and chemotaxis migration to analyze cell tracks (9).

Platelet RNA isolation followed the kit protocol of Qiagen RNeasy Micro Kit (Cat. no. 74004). Using StringTie software to predict the transcripts of all samples, and then using RSEM software to calculate the results of the comparison of bowtie2, we got the number of reads of each transcript for each sample and performed FPKM (Fragments Per Kilobase per Million bases) conversion. The analysis software is the R language package DEseq2 (20). The screening threshold is FDR (false discovery rate) <0.05, log FC (fold change (condition 2/condition 1) for a gene) >1, or log FC<-1. Hypergeometric distribution was used for gene ontology (GO) enrichment analysis, and the GO term with Q-value ≤0.05 was selected as the significantly enriched GO entry. Gene Set Enrichment Analysis was performed using the GSEA software (https://www.broadinstitute.org/gsea/) with permutation = gene set, metric = Diff_of_classes, metric = weighted, #permutation = 2500.

All data are analyzed by the SPSS software and are presented as means ± SEM. Student’s unpaired t-test and repeated ANOVA are used in this study as the statistical methods, and p values<0.05 are considered statistically significant.

The retrospective study was conducted in Zhongshan Hospital, Shanghai. Thromboelastogram (TEG) ( Supplementary Figure 1A ) and platelet aggregation assay data were collected from 100 patients each, who were diagnosed with different solid cancers during January 2021 to September 2021 ( Table 1 ). Ethical approval (B2020-330R) was obtained from the Zhongshan Hospital Research Ethics Committee.

TEG data showed that cancer patients presented coagulation with a shorter reaction time (R) and higher coagulation index (CI) relative to the control group. The α angle is also more prominent in the cancer patient group than in the control group ( Supplementary Figures 1B – D ), whereas K time and maximum amplitude (MA) have no noticeable difference ( Supplementary Figures 1E, F ). The platelet aggregation assay showed that cancer patients in the AA group had a higher average, maximum aggregation rate, and more considerable mean platelet volume ( Supplementary Figures 1G, H, K ). However, in the ADP group, only mean platelet volume in the cancer patient group is significantly different ( Supplementary Figures 1I, J, L ). Collectively, these data represent that cancer patients are in a hypercoagulable state.

We established the platelet locomotion system based on Florian to map thrombosis formation (9, 19). We observed the whole locomotion process of mouse-derived platelets, including spreading, polarization, and migration ( Supplementary Videos 1 and 2 ). Unlike wild-type platelets (WTPs), many TEPs formed clots before the locomotion started. TEPs did migrate longer and faster than WTP ( Figures 1A, B ). Moreover, TEPs were more eager to migrate and more efficiently to form clots than WTP ( Figures 1C, D ). Here we name the clot when two or more platelets contact each other, which process that in vivo may form a thrombus. Migrating WTPs could contact each other without fusion ( Figure 1Ea green and white arrows and video 1). Instead, TEP could contact firmly (Figure 1Eb red arrows) and further fuse ( Figure 1Eb pink arrows); this may further attract other platelets or clots to form more extensive clots ( Figure 1Eb orange arrows, video 2). More interesting is that the big clot could still move at high speed ( Figure 1Eb yellow and pink arrows, video 2). The chemotaxis migration tool from ibidi depicts the cell migration track ( Figure 1F ). Conclusively, the TEPs may be more easily to form and consolidate thrombosis.

The hypercoagulable state is a pathophysiological phenomenon that is affected by multiple factors. We next examined whether the change of plasma or both platelet and plasma finally leads to the hypercoagulable state ( Supplementary Video 3 ). Different from their migration in Tyrode’s buffer, platelets formed cilium-like pseudopodium and moved without complete attachment to the substrate. Moreover most platelets migrated within a small range in the plasma.

Given the previous reports that plasma contains many coagulation factors, we estimated that the TEPs would behave more robustly when they migrate in plasma (21). However, the TEP migrating track is shorter and with a relatively slower speed ( Figures 2A, B ). More obviously, very few TEPs can migrate to form clots ( Figure 2C ) in their physical plasma environment ( Supplementary Video 4 ).

To check whether the plasma interferes with the migration of TEPs, we conducted the WTP migration assay within the B16 mouse-derived plasma (TPL). The results showed that compared with the WTP+WTPL group, WTP did migrate slower and shorter in TPL ( Figures 2A, B ). Compared with the TEP+TPL group, the WTP can still migrate faster in TPL, but the migrating distance has no statistical difference. To further explain the finding that the TPL may inhibit the migration of platelets, we then performed the TEP migration within the WTPL. The results showed that only the WTP+WTPL group could move faster and longer than WTP. On both migrating distance and speed sides, the TEP+WTPL group showed no noticeable difference from the other two groups. The chemotaxis migration data presented a disorderly and unsystematically cell track ( Figures 2D – G and Supplementary Videos 3 – 6 ). These data also supported our hypothesis that plasma isolated from tumor-burdened mice may contain more inhibitors to prevent platelet migration.

We assumed that the significant difference between TEP and WTP may be due to the tumor’s education. The total mRNA from WTPs and TEPs was captured to unravel the ultimate reason. The volcano plots showed that many genes in WTPs were regulated when they were activated ( Figure 3A ), and TEPs differ significantly from WTPs in differential gene numbers ( Figure 3B ). Activated TEPs differ less from the original TEPs in the regulated gene numbers ( Figure 3C ). But the activated TEPs did undergo a noticeable change in gene set compared to the activated WTPs ( Figure 3D ). Gene ontology (GO) analysis showed that TEPs showed a significant difference in cell response to stimuli, actin filament, and extracellular vesicles ( Supplementary Figures 2A, B ). The RNA transcript of activated WTPs and TEPs indicated that genes associated with cell adhesion, oncogenes, and immune cell cytokines were upregulated in activated TEPs ( Supplementary Figures 2A, B ).

The gene heat map showed that TEPs highly expressed cell adhesion genes like Itgb2, Tgfbi, Gp5, and Ache ( Figure 3E ). The cell migration-associated skeleton genes, Dnaaf5, Rhog, Parvb, and Tubb5, were significantly enriched in the activated TEPs. The actin-filament genes were upregulated, but those cytoskeleton genes like tubulin and cilium were highly expressed. The highly expressed calcium channel, ATP synthesis, and mitochondrial genes in TEPs may facilitate their potent migration. The vesicle assembly, endocytosis genes, and pro-/oncogenes were upregulated in the activated TEPs. However, the TEP also contains some antitumor genes like Lgals3bp, Ndrg1, Camta3, and Pbrm1. Except these, we also found that TEP expressed Hjv, Psme4, Sbno2, and Zfpm. These genes participate mainly in the biological process like iron homeostasis, DNA damage, osteoblast differentiation, megakaryocyte, and red cell differentiation. The ubiquitination signal pathway genes also were upregulated in TEPs.

Our Gene Set Enrichment Analysis (GSEA) revealed that ATP synthesis, excitatory synapse assembly, postsynapse assembly, and respiratory transport chain were upregulated in activated platelets compared to inactivated wild-type platelets ( Supplementary Figure 3A ). TEPs differed strongly from wild-type platelets in upregulated gene expression associated with the glucose catabolic, microtubule polymerization regulation, vesicle lumen, and abnormal platelet volume ( Supplementary Figure 3B ). The intrinsic pathway of blood coagulation, the assembly of the mitochondrial respiratory chain complex, and the assembly of the motile cilium were strongly expressed in the activated TEPs compared with the TEPs ( Supplementary Figure 3C ). The activated TEPs also exhibited marked upregulation of coagulation, homotypic cell–cell adhesion, actin filament, and vesicle lumen compared with the activated wild-type platelets ( Supplementary Figure 3D ).