Regulation of drug metabolizing enzymes in the leukaemic bone marrow microenvironment

Abstract The bone marrow (BM) microenvironment contributes to drug resistance in acute myeloid leukaemia (AML) and multiple myeloma (MM). We have shown that the critical drug metabolizing enzymes cytochrome P450 (CYP) 3A4 and cytidine deaminase (CDA) are highly expressed by BM stroma, and play an important role in this resistance to chemotherapy. However, what factors influence the chemoprotective capacity of the BM microenvironment, specifically related to CYP3A4 and CDA expression, are unknown. In this study, we found that the presence of AML cells decreases BM stromal expression of CYP3A4 and CDA, and this effect appears to be at least partially the result of cytokines secreted by AML cells. We also observed that stromal CYP3A4 expression is up‐regulated by drugs commonly used in AML induction therapy, cytarabine, etoposide and daunorubicin, resulting in cross‐resistance. Cytarabine also up‐regulated CDA expression. The up‐regulation of CYP3A4 associated with disease control was reversed by clarithromycin, a potent inhibitor of CYP3A4. Our data suggest that minimal residual disease states are characterized by high levels of stromal drug metabolizing enzymes and thus, strong microenvironment‐mediated drug resistance. These results further suggest a potential role for clinically targeting drug metabolizing enzymes in the microenvironment.


| INTRODUC TI ON
Most patients with acute myeloid leukaemia (AML) and other haematologic malignancies achieve complete remissions (CRs) with initial chemotherapy, but eventually relapse and die of their disease. 1 The mechanisms responsible for the resistance of minimal residual disease (MRD), the disease cells present in CR that leads to relapse, remain under study. In addition to intrinsic mechanisms exhibited by MRD including stem cell characteristics, 2,3 it is now clear that specialized microenvironments or niches play important roles in extrinsic drug resistance. 1,4,5 Our group previously showed that bone marrow stromal cells (BMSCs) protect normal human haematopoietic stem cells and AML cells from the pro-differentiating effects of retinoic acid by expressing the retinoid-inactivating enzyme, cytochrome P450 (CYP)26. 7, 8 We also found that stromal CYP3A4 similarly protected AML and multiple myeloma (MM) cells from various chemotherapeutic agents. 9,10 However, any mechanisms responsible for regulating stromal CYPs are unknown.
The leukaemic BM is a pro-inflammatory, cytokine-rich environment, 11 and many of these factors, such as IL6, play important roles in AML biology. 12,13 Cytokines and inflammation, especially related to cancer, have been shown to suppress hepatic and intestinal CYP levels. 16,17 For the initial treatment of newly diagnosed AML patients, the '7 + 3' regimen, which combines a 7-day continuous intravenous infusion of cytarabine with a short infusion of an anthracycline given on days 1-3, remains the most commonly used regimen. Etoposide is another agent used in many induction regimens. 21,22 All three of these agents are substrates for CYP3A4, 23,24 and cytarabine is also inactivated by cytidine deaminase (CDA). 25,26 Chemotherapeutics can induce CYP3A4 activity in human liver cells. 28 Thus, the clinical status of the AML and its treatment could theoretically influence the expression of CYP3A4 and CDA in the BM microenvironment, and hence impact associated drug resistance. Accordingly, we suggested that effects of tumour burden and chemotherapy on the tumour microenvironment could play a role in treatment resistance, particularly in the setting of MRD. Here, we find that BMSC CYP3A4 and CDA are not only influenced by the status of the AML and its treatment, but clinically targeting drug metabolizing enzymes in the microenvironment also holds promise in modulating the adverse effects of MRD on extrinsic drug resistance.

| Isolation of primary BMSCs
Primary BMSCs were derived from normal allogeneic BM donors granting informed consent as approved by the Johns Hopkins Medical Institutes Institutional Review Board, as we have previously described. 7 Briefly, mononuclear cells isolated from BM of normal volunteers were
Samples were plated in duplicate onto 35-mm 2 tissue culture dishes and incubated in a humidified atmosphere at 37°C and 5% CO 2 .
Colonies consisting of more than 40 cells were scored at 10-15 days using an inverted microscope.

| Statistical analysis
Statistical analysis was performed by using two-tailed unpaired student t test to compare the averages of two groups and calculate the P value.

| Active AML up-regulates BMSC CYP3A4 and CDA
We previously showed that most CYPs, including CYP3A4, as well as CDA were all highly expressed in BMSCs, but not AML and MM cells. 10 To model the BM niche in AML patients, HL-60, Kasumi-1 or OCI-AML3 cells were co-cultured with human BMSCs for 72 hours and the expression of CYP3A4 and CDA in stroma cells was assessed by RT-qPCR. All three AML lines significantly suppressed the expression of both CYP3A4 and CDA in both the human BMSC line F/STRO ( Figure 1A and B respectively) and the primary human BMSCs ( Figure 1C and D respectively).

| Cytokines associated with AML down-regulate BMSC CYP3A4 and CDA
Active AML BM is a pro-inflammatory environment, associated with aberrant cytokine signaling. 14 (Figure 2A and B respectively) and the primary human BMSCs ( Figure 2C and D respectively).

| AML induction chemotherapy up-regulates BMSC CYP3A4 and cytarabine also up-regulates CDA
Three of the most commonly used chemotherapy drugs for remission induction in AML were assessed for their ability to induce the expression of drug metabolizing enzymes in BMSCs, as they have  Figure 3A). Similar results were also seen in primary BM stroma ( Figure 3B). To further confirm the specificity of the drug-induced up-regulation, CYP26A1 and CYP26B1, enzymes involved in retinoid but not chemotherapy inactivation, were measured and found not to be significantly affected by cytarabine, etoposide or daunorubicin ( Figure S3).

| Cytarabine induces stroma-mediated crossresistance of AML cells to etoposide via CYP3A4
Drugs used for AML induction are often given sequentially. 39,40 To model sequential AML therapy, F/STRO BMSCs were pre-incubated with cytarabine for 72 hours, and then after removing the drug, co-cultured with the AML cell line HL-60 and etoposide. As we previously showed, 10 F/STRO BMSCs protected the HL-60 cells from etoposide, and CYP3A4 knockdown by shRNA partially reversed this protection ( Figure 4A). Pre-incubation of F/ STRO cells with cytarabine further augmented the BMSCs ability to protect the HL-60 cells against etoposide, while cytarabine preincubation had no effect on etoposide sensitivity when CYP3A4 was knocked down ( Figure 4B). Clarithromycin, a potent CYP3A4 inhibitor, 10,42,43 similarly reversed the protective effect of the BMSCs, including after pretreatment with cytarabine ( Figure 4A).
Clarithromycin also had no effect on drug resistance after CYP3A4 knockdown, providing evidence it was working through inhibiting CYP3A4 (Figure 4).

| D ISCUSS I ON
There is increasing evidence of the crosstalk between malignant cells and their surrounding microenvironment 9,44 but the full extent of the functional consequences of these interactions remains unknown. Previously, our group showed that expression of CYP enzymes appears to be at least partly responsible for the well-recognized ability of BMSCs to protect AML cells from chemotherapy. 8,10 The BM during leukaemia therapy is a dynamic environment with changes related to treatment and tumour burden; we suggested that these changes likely can modulate drug metabolizing enzymes in the BM microenvironment, and as a consequence, drug resistance. We found that both AML cells, as well as inflammatory cytokines that are elevated during active disease, decrease CYP3A4 and CDA gene expression level in BMSCs similar to the effects of inflammatory cytokines on liver CYP3A4 expression. 18 In addition, pretreatment of BMSCs with cytarabine induced AML drug resistance to etoposide. As CYP3A4 is responsible for metabolizing about half of the chemotherapy drugs currently in use, 45 it is perhaps not surprising that this mechanism of drug resistance led to cross-resistance. Clinical trials testing this concept are in progress.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest.

AUTH O R CO NTR I B UTI O N S
All authors critically read the manuscript. MS: designed research, performed research, analysed data, wrote the paper; Y-TC, DH: performed research; RJJ, GG: designed research, analysed data, wrote the paper.