Levels of sENG are higher in ALL and potentially interfere with TRC105 efficacy. (A-C) Levels of sENG in leukemic cells. Measurements were performed using standard quantitative ELISA. (A) Concentration of sENG in the plasma of AML (n = 11) and ALL (n = 18) patients. Each data point represents a single patient sample. **P < .01 by Student t test. (B) Levels of sENG in the serum of mice that had been injected with ALL or AML blasts, and subjected to TRC105 treatment, or not, for a period of 12 weeks. Error bars indicate SEM for each cohort (n = 4). ***P < .001 by Student t test. (C) Levels of sENG in leukemic cell lines Nalm-6, Semk-2, and HL-60. Supernatant of these cell cultures was collected on day 4. Error bars indicate SEM from 2 independent experiments. (D) Western blot analyses for MMP-14 in lysates from indicated cancer cell lines. Active form of MMP-14 is observed in Semk-2 as well as a positive control MDA-MB-231, but not in HL-60. GAPDH was used as loading control. (E) Schematic representation outlining the effect of sENG on TRC105 treatment. In AML, which is characterized by low levels of sENG, the monoclonal antibody TRC105 is free to bind to the membrane form of endoglin (left panel). However, in ALL (right panel), which exhibits high levels of sENG, TRC105 is decoyed, resulting in less TRC105 available to bind to mENG, and thereby less therapeutic effect.