CyTOF analysis reveals HSC defect is due to reduced canonical and noncanonical TGF-β signaling. (A-C) CyTOF analysis. (A) Representative SPADE tree of c-Kit⁺ cells (gated on nonapoptotic live Lin⁻c-Kit⁺ cells) from BM cells from 1 representative control mouse. Each node in the tree represents 1 cluster of phenotypically similar cells, whereas node size corresponds to the number of cells. Nodes are colored according to the mean expression of the indicated marker in that particular node. A core of 9 markers was used to generate the SPADE trees, as well as to assign clusters to specific cell fractions. c-Kit expression in all nodes is shown. (B) Expression of the 9 markers used for cluster classification. (C) Mean expression of Eng in cKO Eng mice (right tree) and control mice (left tree) shows that Eng is highly expressed in HSC and MEP fractions from control mice (arrows) but not in the respective cell fractions from mice transplanted with cells lacking Eng (arrows) (n = 4 per group). (D-E) Expression levels of pSMAD2/3 and pMAPKAPK2 (or pMK2), the canonical and noncanonical downstream effectors of TGF-β, respectively, in cKO Eng mice and control mice reveal reduced activation of the TGF-β signaling pathway in the absence of Eng. (D) Heat map shows Eng, pSMAD2/3, and pMK2 arcsinh ratios for HSCs and MEPs from control mice and cKO Eng mice. Each row represents an individual animal (n = 4 per group). Raw expression was arcsinh transformed and fold change was calculated to the column’s maximum value (arcsinh ratio). (E) Quantification of raw expression for pSMAD2/3 and pMK2 in HSCs and MEPs from control mice (red) and cKO Eng mice (blue). Box plot whiskers represent minimum and maximum values for each cohort. *P < .05, **P < .005.