Fig 2: Noncanonical STAT5B phosphorylation confers creatine-instigated gene expression in MKs.a Volcano plot of 5812 differentially phosphorylated peptides in freshly isolated MKs from vehicle- or creatine-treated mice (n = 3 mice per group). b Top 3 phosphorylated TFs in freshly isolated MKs from vehicle- or creatine-treated mice (n = 3 mice per group). Arrow points to STAT5B phosphorylated peptides. c Venn diagram of top phosphorylated TFs in creatine-treated MKs and predicted TFs that may regulate platelet activation-associated genes. d Diagram of human and mouse STAT5B amino acid sequence. Creatine-induced phosphorylation site S127 and canonical phosphorylation sites Y699 and S731 were marked. MKs (e) and 293T cells (f) were transfected with empty vector and STAT5B-FLAG vector with wt, S127A mutant, triple mutants, or S127D mutant. These groups were then treated with vehicle- or creatine in vitro. Western blot of phosphorylation of STAT5B-FLAG in various groups. FLAG marks the loading amount in each lane. Quantifications of signal in various groups (n = 3 blots per group). g–i SiRNA-GATM (human)-transfected 293T cells were further transfected with wt or S127A mutant STAT5B-FLAG vector, and empty vector or CKB overexpression vector (g). Si-scrambled- or si-CKB (human)-transfected 293T cells were further transfected with empty vector, wt STAT5B-FLAG vector, or S127A mutant STAT5B-FLAG vector, and were treated with or without creatine (h). 293T cells were transfected with empty vector and STAT5B-FLAG vector with wt or S127A mutant, and further transfected with empty vector or CKB overexpression vector (i). Western blot of phosphorylation of STAT5B-FLAG in various groups. FLAG marks the loading amount in each lane. j ChIP assay of STAT5B binds to different promoter regions of P2ry12 and Gp6. Nonimmune IgG was used for input. Coding region (exon) of each gene served as controls (n = 3 samples per group). Schematic diagram of the DNA binding model (created in BioRender. Ruibo, C. (2026) https://BioRender.com/f37rihj). Statistical significance in (a) was determined by unpaired two-tailed Student’s t test. Statistical significance in (e, f) was determined by one-way ANOVA. Statistical significance in (j) was determined by two-way ANOVA. Source data are provided as a Source data file. Data presented as mean ± s.d.

Fig 3: Exogenous creatine supplementation increases hyperactive platelets in humans.a Schematic model of 11 healthy male and female volunteers taking 20 g creatine powder per day for 14 days. Peripheral blood samples were collected at various timepoints for further detections. b Circulating creatine concentrations at day 0, 3, 7, and 14 after receiving creatine supplement in volunteers (n = 11 volunteers per group). c Platelet aggregometry analysis of platelet-rich plasma in response to collagen, epinephrine (EPI), arachidonic acid (AA), or ADP. Quantifications of maximum aggregation rate, average aggregation rate, and platelet adhesion rate in various groups (n = 11 volunteers per group). d Four volunteers were randomly selected, and circulating CD41+ cells were isolated. Heatmap of mRNA expression of platelet receptor- and granule-associated genes in isolated CD41+ cells at various timepoints (n = 4 volunteers per group). e Quantification of protein levels of GP6 and P2RY12 in isolated platelets at various timepoints (n = 4 volunteers per group). f, g Schematic diagram of the adoptive platelet transfer model. Platelets were isolated from volunteers before exogenous creatine supplementation and after treatment for 7 days. Donor platelets were co-injected with melanoma cells in immunocompromised recipient mice. Representative lung pictures, H&E histological analysis and quantifications of lung metastasis (n = 5 mice per group). Dashed lines mark the borders between tumor (T) and lung (L) tissues. Scale bar in upper panels, 1 cm. Scale bar in middle panels, 2 mm. Scale bar in lower panels, 50 μm. h Circulating creatine is taken up by megakaryocytes via the creatine transporter SLC6A8. Elevated intracellular creatine upregulates CKB expression, which in turn phosphorylates STAT5B at the noncanonical site S127. Nuclear-translocated phosphorylated STAT5B drives gene transcription to generate hyperactive platelets. These hyperactive platelets facilitate tumor metastasis by protecting CTCs from blood shear stress, evading immune surveillance, inhibiting anoikis, and promoting CTC extravasation into distant organs. Schematics in (a, f, h) were created in BioRender. Ruibo, C. (2026) https://BioRender.com/35axa7t. Statistical significance in (b, c, e) was determined by one-way ANOVA. Statistical significance in (g) was determined by unpaired two-tailed Student’s t test. Source data are provided as a Source data file. Data presented as mean ± s.d.

Fig 4: Creatine promotes CKB expression for platelet hyperactivity in MKs.a Intracellular creatine concentrations in freshly isolated MKs from vehicle- or creatine-treated mice (n = 3 mice per group). Brain tissues from the same group served as controls. b Western blot of CKB in freshly isolated MKs from vehicle- or creatine-treated mice. Brain tissues from the same group served as controls. β-ACTIN marks the loading amount in each lane. c Intracellular phosphocreatine concentrations in freshly isolated MKs from vehicle- or creatine-treated mice (n = 3 mice per group). d Intracellular creatine concentrations in MKs treated with or without creatine in vitro (n = 3 samples per group). e QPCR quantification of Ckb mRNA expression in MKs treated with or without creatine in vitro (n = 3 samples per group). Western blot of CKB in MKs treated with or without creatine in vitro. β-ACTIN marks the loading amount in each lane. f Intracellular phosphocreatine concentrations in MKs treated with or without creatine in vitro (n = 3 samples per group). g Western blot of CKB, GP6, and P2RY12 in MKs treated with or without creatine or phosphocreatine in vitro. β-ACTIN marks the loading amount in each lane. h Heatmap of platelet receptor- and granule-associated genes in MKs treated with or without creatine or phosphocreatine in vitro (n = 3 samples per group). i Western blot of GP6 and P2RY12 in si-scrambled- or si-Ckb-transfected MKs treated with or without creatine in vitro. β-ACTIN marks the loading amount in each lane. j Heatmap of platelet receptor- and granule-associated genes in si-scrambled- or si-Ckb-transfected MKs treated with or without creatine in vitro (n = 3 samples per group). k Western blot of GP6 and P2RY12 in empty vector- or Ckb overexpression vector-transfected MKs in vitro. β-ACTIN marks the loading amount in each lane. l Heatmap of platelet receptor- and granule-associated genes in empty vector- or Ckb overexpression vector-transfected MKs in vitro (n = 3 samples per group). Statistical significance in (a, c–f) was determined by unpaired two-tailed Student’s t test. Source data are provided as a Source data file. Data presented as mean ± s.d.