Although the co-development of companion diagnostics with molecular targeted drugs is desirable, truly efficient diagnostics are limited to diseases in which chromosomal translocations or overt mutations are clearly correlated with drug efficacy. Moreover, even for such diseases, few methods are available to predict whether drug administration is effective for each individual patient whose disease is expected to respond to the drug(s). We have previously developed a biosensor based on the principle of Förster resonance energy transfer to measure the activity of the tyrosine kinase BCR-ABL and its response to drug treatment in patient-derived chronic myeloid leukemia cells. The biosensor harbors CrkL, one of the major substrates of BCR-ABL, and is therefore named Pickles after phosphorylation indicator of CrkL en substrate. The efficacy of this technique as a clinical test has been demonstrated, but the number of cells available for analysis is limited in a case-dependent manner, owing to the cleavage of the biosensor in patient-derived leukemia cells. Here, we describe an improved biosensor with an amino acid substitution and a nuclear export signal being introduced. Of the two predicted cleavage positions in CrkL, the mutations inhibited one cleavage completely and the other cleavage partially, thus collectively increasing the number of cells available for drug evaluation. This improved version of the biosensor holds promise in the future development of companion diagnostics to predict responses to tyrosine kinase inhibitors in patients with chronic myeloid leukemia.