Cohorts of untreated leukemic mice (Figure 4B) or from mice undergoing relapse during continuous treatment with dasatinib (Dasa), ruxolitinib (Rux), and dexamethasone (Dex) (Figure 4D) were euthanized when moribund. BM cells from 5 to 7 individual leukemic mice from each group were harvested from their long bones, genomic DNA was extracted, and BCR-ABL KD sequences were amplified from 200 ng of DNA using nested PCR. Individual PCR products were subcloned into plasmids transfected into Escherichia coli, and bacteria were streaked onto plates containing isopropyl β-D-1-thiogalactopyranoside to form individual recombinant colonies, 10 to 20 of which were picked at random from each such sample. Recloned BCR-ABL KD DNAs were sequenced in both directions to avoid sequencing errors.

ND = not done.

Differences between the Dasa + Rux and Dasa + Rux + Dex cohorts are significant (P = 0.015, Fisher’s 2-tailed exact test).

Mutations identified in 20% or more of the sequenced clones are designated in single-letter amino acid code (with mutational frequencies in parenthesis). For example, “Y253F (36)” means that a mutation encoding a phenylalanine for tyrosine substitution at codon 253 was identified in 5 of 14 clones (36%) from a single leukemic BM sample. In no case was more than a single KD mutation detected in a relapse sample. T315I and F317L are gatekeeper mutations that prevent dasatinib binding; Y253F, E255K, and Q252H are P-loop mutations. F401L has not been characterized.