Blocking STAT3 S727 phosphorylation ameliorates K-RasG12D–driven MPNs. Control animals (M:F/+) lacking K-Ras^G12D, Mx:STAT3F/SA (M:F/SA), Mx:K-Ras^G12DLSL:STAT3F/+ (M:K:F/+), or MxK-Ras^G12DLSL:STAT3F/SA (M:K:F/SA) mice were injected with poly(I:C) and peripheral white blood cells enumerated every 2 weeks (A). Each point represents the mean of cohorts of 5 animals, with error bars representing 1 standard deviation from the mean. *P < .05 and ****P < .0001 for comparisons between the M:K:F/+ and M:K:F/SA cohorts (calculated by 2-way ANOVA with the Tukey multiple comparison test). On the day animals were sacrificed due to morbidity (week 14 for Mx:K-Ras:F/+) or at the end of the experiment in the case of all other cohorts (week 20), peripheral white blood cells (B) or granulocytes (C) were counted and plotted. Each point represents cell numbers from an independent animal. **P < .01 and ***P < .001 for comparisons between the M:K:F/+ and M:K:F/SA cohorts (1-way ANOVA with the Tukey multiple comparison test). (D) Animals were sacrificed and tissues collected for histological analysis (20 weeks post poly(I:C) injection for Mx:F/+, Mx:F/SA, Mx:K-Ras:F/SA; 14 weeks for Mx:K-Ras:F/+). The peripheral blood (subpanels a-d) and bone marrow smears (subpanels e-h) were prepared with Wright-Giemsa stain and imaged with a 60× objective (inset: 2× digital enlargement). Spleen (subpanels h-o) and liver sections (subpanels p-w) were stained with hematoxylin and eosin and imaged with a 10× or 40× objective, as indicated. ANOVA, analysis of variance.