Resistance of t(17;19)‐acute lymphoblastic leukemia cell lines to multiagents in induction therapy

Abstract t(17;19)(q21‐q22;p13), responsible for TCF3‐HLF fusion, is a rare translocation in childhood B‐cell precursor acute lymphoblastic leukemia(BCP‐ALL). t(1;19)(q23;p13), producing TCF3‐PBX1 fusion, is a common translocation in childhood BCP‐ALL. Prognosis of t(17;19)‐ALL is extremely poor, while that of t(1;19)‐ALL has recently improved dramatically in intensified chemotherapy. In this study, TCF3‐HLF mRNA was detectable at a high level during induction therapy in a newly diagnosed t(17;19)‐ALL case, while TCF3‐PBX1 mRNA was undetectable at the end of induction therapy in most newly diagnosed t(1;19)‐ALL cases. Using 4 t(17;19)‐ALL and 16 t(1;19)‐ALL cell lines, drug response profiling was analyzed. t(17;19)‐ALL cell lines were found to be significantly more resistant to vincristine (VCR), daunorubicin (DNR), and prednisolone (Pred) than t(1;19)‐ALL cell lines. Sensitivities to three (Pred, VCR, and l‐asparaginase [l‐Asp]), four (Pred, VCR, l‐Asp, and DNR) and five (Pred, VCR, l‐Asp, DNR, and cyclophosphamide) agents, widely used in induction therapy, were significantly poorer for t(17;19)‐ALL cell lines than for t(1;19)‐ALL cell lines. Consistent with poor responses to VCR and DNR, gene and protein expression levels of P‐glycoprotein (P‐gp) were higher in t(17;19)‐ALL cell lines than in t(1;19)‐ALL cell lines. Inhibitors for P‐gp sensitized P‐gp‐positive t(17;19)‐ALL cell lines to VCR and DNR. Knockout of P‐gp by CRISPRCas9 overcame resistance to VCR and DNR in the P‐gp‐positive t(17;19)‐ALL cell line. A combination of cyclosporine A with DNR prolonged survival of NSG mice inoculated with P‐gp‐positive t(17;19)‐ALL cell line. These findings indicate involvement of P‐gp in resistance to VCR and DNR in Pgp positive t(17;19)‐ALL cell lines. In all four t(17;19)‐ALL cell lines, RAS pathway mutation was detected. Furthermore, among 16 t(1;19)‐ALL cell lines, multiagent resistance was usually observed in the cell lines with RAS pathway mutation in comparison to those without it, suggesting at least a partial involvement of RAS pathway mutation in multiagent resistance of t(17;19)‐ALL.


| INTRODUCTION
For childhood B-cell precursor acute lymphoblastic leukemia (BCP-ALL), chromosomal translocation is strongly associated with therapeutic outcome. 1,2 t(17;19)(q21-q22;p13) is a rare translocation and presents in less than 1% of childhood BCP-ALL cases. 3 Clinically, prognosis of t(17;19)-ALL is extremely poor even in recently intensified chemotherapy. 4 In t(17;19)-ALL, the TCF3 (E2A) gene on 19p13 fuses to the HLF gene on 17q21-22 in-frame. 5,6 TCF3-HLF fusion acts as a transcription factor through the transactivation domains of TCF3 and a DNA-binding and dimerization basic leucine zipper (bZIP) domain of HLF. 7,8 t(1;19)(q23;p13), which is another translocation involving the TCF3 gene, is quite common translocation and presents in approximately 5% of childhood ALL cases. 9 Prognosis of t(1;19)-ALL has been dramatically improved in recently intensified chemotherapy. [10][11][12] In t(1;19)-ALL, the TCF3 gene fuses in-frame to the PBX1 gene on 1q23. 13,14 TCF3-PBX1 acts as the transcription factor through the transactivation domains of TCF3 and a homeobox DNA-binding domain of PBX1. 13,15 Although both fusion transcription factors share the transactivation domains of TCF3, TCF3-HLF and TCF3-PBX1 regulate different downstream target genes by binding to different consensus nucleotide sequences through the bZIP domain of HLF and the homeobox domain of PBX1, respectively. 16 Thus, distinctive prognosis between t(17;19)-ALL and t(1;19)-ALL may be attributed at least partially to differences in the transcriptional activities of TCF3-HLF and TCF3-PBX1.
Recent comprehensive genetic analyses of t(17;19)-ALL and t(1;19)-ALL using patient-derived xenografts revealed significant differences between the molecular landscape of the two groups; deletions of PAX5 and VPREB1 and mutations of TCF3 and RAS pathway genes such as NRAS, KRAS, and PTPN11 were more frequently observed in t(17;19)-ALL samples. 17 These observations suggest an unconfirmed possibility that these additional genetic abnormalities may be involved in the poor therapeutic response of t(17;19)-ALL in association with TCF3-HLF. Consistent with the dismal outcome with chemotherapy, drug response profiling of patientderived t(17;19)-ALL xenografts on human mesenchymal stroma cells using a coculture system revealed resistance to several standard chemotherapeutic agents such as vincristine (VCR) and cytarabine. 17 However, t(17;19)-ALL xenografts are significantly more sensitive to glucocorticoids than other high-risk pre-B and T-ALL xenografts including t(1;19)-ALL. 17 Thus, further analyses are required to verify the drug response profiling of t (17;19)-ALL in comparison with that of t(1;19)-ALL.
In this study, using a panel of t(17;19)-ALL and t(1;19)-ALL cell lines, we analyzed drug response profiling in a simple liquid culture system. Prior to analyses of the cell lines, we prospectively examined the levels of minimal residual disease (MRD) during induction therapy in a newly diagnosed cell lines were found to be significantly more resistant to vincristine (VCR), daunorubicin (DNR), and prednisolone (Pred) than t(1;19)-ALL cell lines. Sensitivities to three (Pred, VCR, and l-asparaginase [l-Asp]), four (Pred, VCR, l-Asp, and DNR) and five (Pred, VCR, l-Asp, DNR, and cyclophosphamide) agents, widely used in induction therapy, were significantly poorer for t (17;19)

| AlamarBlue cell viability assay
To determine IC 50 s of DNR, VCR, prednisolone (Pred), dexamethasone (Dex), l-asparaginase (l-Asp), cyclophosphamide (CPM), and selumetinib, an alamarBlue assay was performed. 20 The sources of the drugs are shown in Table S2. For CPM sensitivity, mafosfamide (MAF), an active analog of CPM, was used. Cells (1-4 × 10 5 ) were plated onto a 96-well flat-bottom plate in triplicate in the absence or presence of seven concentrations of each drug. The cells were cultured for 44 hours to determine the DNR, VCR, and CPM sensitivities and for 68 hours to determine Pred, Dex, l-Asp, and selumetinib sensitivities, and, then, 20 µL of alamarBlue was added. After a 6-hours additional incubation with alamarBlue, absorbance at 570 nm was monitored by a microplate spectrophotometer using 600 nm as a reference wavelength. Cell survival was calculated by expressing the ratio of the optical density of the treated wells to that of the untreated wells as a percentage. The concentration of agent required to reduce the viability of the treated cells to 50% of the untreated cells was calculated, and the median of three independent assays was determined as IC 50 . The median of the IC 50 s measured by three independent assays was determined.

| Flow cytometric analysis
To detect apoptotic events, cells were cultured in the absence or presence of DNR or VCR in combination with or without verapamil, cyclosporine A (CyA), or nilotinib for 24 hours, and stained with a fluorescein isothiocyanateconjugated Annexin-V (BioLegend, San Diego, CA) and actinomycin-D (Sigma-Aldrich, St Louis, MO). Cell surface expression of P-gp was analyzed using a phycoerythrinconjugated anti-P-gp antibody. For the functional assay of P-gp-mediated efflux of calcein-AM (CAM), HALO1 cells were incubated with 0.25 mmol/L of CAM for 10 minutes at 37°C in the absence or presence of velapamil, CyA, or nilotinib. The stained cells were analyzed by flow cytometry (FACSCalibur, BD Biosciences, San Jose, CA).

| Real-time RT-PCR analysis
Total RNA was extracted using the Trizol reagent (Invitrogen, Carlsbad, CA), reverse transcription was performed using a random hexamer (Amersham Bioscience, Buckinghamshire, United Kingdom) by Superscript II reverse transcriptase (Invitrogen), and then incubation with RNase (Invitrogen). For quantitative real-time PCR, triplicated samples containing cDNA with TaqMan Universal PCR Master Mix (Applied Biosystems) and Gene Expression Product listed in Table  1 were amplified following manufacturer's protocol using UOCB1 as a control. As an internal control for relative gene expression, quantitative real-time PCR for ACTB was performed.

| In vivo analysis of drug sensitivity
Six-week-old female NSG (NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ) mice were purchased from Jackson Laboratory (Bar Harbor, ME, USA). The experiment was performed in a specific pathogen-free unit after approval of protocols for animal care and experiment by the Tokyo Medical and Dental University animal care and use committee (approved No. A2017113). HALO1 cells (1 x 10 4 ) were injected into the tail vein to establish xenografts. One day after injection of HALO1 cells, each of the five mice were treated with 0.5 mg/ kg of DNR alone or 0.5 mg/kg of DNR in combination with 50 mg/kg of CyA for five consecutive days. DNR and CyA were further diluted with phosphate-buffered saline (PBS) and intraperitoneally injected into the mice. CyA was injected one hour before administration of DNR. The control group of five mice was administered PBS only.

| Target deep sequencing of RAS pathway genes
Target deep sequencing of RAS pathway genes including the PTPN11, NRAS, KRAS, and NF1 were analyzed using SureDesign software (Agilent Technologies, Santa Clara, CA). Libraries were prepared using the HaloPlex Target Enrichment System (Agilent Technologies), followed by paired-end sequencing on a MiSeq instrument (Illumina, San Diego, CA). Bioinformatic analysis was performed using the SureCall software (Agilent Technologies). Common germline polymorphisms reported in public databases were excluded and nonsense, frameshift, splice site, nonsynonymous variants were considered as mutations. Minimal allele frequency for mutation calling was set at 0.3.

| MRD analysis of a newly diagnosed t(17;19)-ALL case
We prospectively evaluated MRD levels in a t(17;19)-ALL case, treated with a high risk (HR) regimen in the TCCSG L04-16 study, 18 using real-time RT-PCR targeting of TCF3-HLF ( Figure 1). Higher levels of TCF3-HLF chimeric mRNA were continuously detected during intensified induction therapy consisting of Pred, VCR, DNR, l-Asp, and CPM. Bone marrow relapse was confirmed in the patient at

Genes
TaqMan probes the end of early intensification therapy. The patient received haploidentical transplantation from his mother, but regrowth of leukemic blasts was confirmed in the bone marrow on day 36. He was treated with donor lymphocyte infusions and obtained immediate remission. His bone marrow remained in complete remission for over 3 years. 23 We also performed prospective evaluation of MRD levels in 16 consecutive cases of t(1;19)-ALL, treated with an identical regimen, by real-time RT-PCR targeting of TCF3-PBX1. TCF3-PBX1 chimeric mRNA constantly decreased during induction therapy and became undetectable on day 43 except in two cases (12.5%).

| Multiagent resistance in t(17;19)-ALL cell lines
To comprehensively evaluate sensitivity of t(17;19)-ALL cell lines to multiple agents used in induction therapy, we analyzed the combined sensitivities to three (Pred, VCR, and l-Asp), four (Pred, VCR, l-Asp, and DNR) and five (Pred, VCR, l-Asp, DNR, and CPM) agents (Table 3) according to previous reports. 21,22 Total scores of sensitivities in t(17;19)-ALL cell lines were significantly higher than those in t(1;19)-ALL cell lines (P = 0.019 for three agents, P = 0.011 for four agents, and P = 0.039 for five agents; Figure 3), indicating that t(17;19)-ALL cell lines were far more resistant to the multiple agents commonly used in induction therapy than t(1;19)-ALL cell lines.

| Expression of ABC transporters in t(17;19)-ALL cell lines
The IC 50 s of DNR in t(17;19)-ALL and t(1;19)-ALL cell lines were closely correlated with that of VCR (R 2 = 0.58, P = 000,091) ( Figure 4A). Since both DNR and VCR are sensitive to ABC transporters, 24

DNR and VCR of t(17;19)-ALL cell lines
To test the involvement of P-gp in resistance to DNR and VCR in t(17;19)-ALL cell lines, we evaluated the functional drug-efflux activity using calcein-AM (CAM), an ABC transporter-dependent dye. 26 We performed flow cytometric analyses of CAM staining in the presence or absence of ABC transporter inhibitors such as verapamil, 27 CyA, 28,29 and nilotinib. 30,31 In HALO1 cells, CAM staining level was remarkably intensified in the presence of verapamil, CyA, or nilotinib in a dose-dependent manner ( Figure 5A). We next evaluated the effects of nilotinib on DNR and VCR sensitivity in HALO1, as well as in 697 cells [a P-gp-negative t(1;19)-ALL cell line], using flow cytometry. Nilotinib alone did not induced apoptosis in either HALO1 cells or in the 697 cells ( Figure 5B,C). DNR and VCR induced apoptosis in HALO1 cells more effectively in the presence of nilotinib, while sensitivities to DNR and VCR in the presence of nilotinib were unchanged in 697 cells. We further analyzed the effect of nilotinib on DNR and VCR sensitivity in two Pgp-positive t(17;19)-ALL cell lines (HALO1 and UOCB1) using an alamarBlue cell viability assay. Sensitivities to DNR and VCR were significantly enhanced by nilotinib in both cell lines ( Figure 5D). Sensitivity to VCR was also significantly enhanced by verapamil in both cell lines ( Figure  S3). To directly verify involvement of P-gp in DNR resistance of HALO1 cells, we next established P-gp knocked out HALO1 cells using the CRISPR/Cas9 system with a CD4 reporter. 20 P-gp expression was knocked out in nearly half of the CD4-positive population ( Figure S4). 20 Then, we treated the cells with DNR (12.5 ng/mL), VCR (25 ng/mL), or Dex (250 nmol/L), for 72 hours. Two-color analysis of P-gp expression and Annexin V-binding revealed that P-gp-negative population was sensitive to DNR and VCR (cell viabilities: 19.2% and 22.0%, respectively) whereas P-gp-positive population was resistant (63.2% and 66.7%, respectively). In contrast, both P-gp-negative and P-gp-positive populations were equally resistant to Dex. These in vitro observations demonstrated an involvement of P-gp in DNR and VCR resistance in P-gp-positive t(17;19)-ALL cell lines.

| Involvement of P-gp in daunorubicin resistance of t(17;19)-ALL cell line in vivo
We finally tried to confirm the involvement of P-gp in DNR resistance of t(17;19)-ALL in vivo using NSG mice. After inoculation of HALO1 cells into NSG mice, we treated mice with DNR alone or DNR in combination with CyA for 5 days ( Figure 5F). Although treatment with DNR alone did not improve survival (median survival: 34 days) in comparison with untreated control (33 days), the combination of DNR and CyA significantly improved survival (36 days, P = 0.018 in Kaplan-Meier analysis).

| Frequent RAS pathway mutations in t(17;19)-ALL cell lines
Association of gene mutation in the RAS pathway with poor therapeutic outcome in childhood ALL is controversial. 32 Figure 6A). Mutations in PTPN11, NRAS, KRAS, and NF1 genes were detectable in one, two, two, and none of the four t(17;19)-ALL cell lines, respectively, and gene mutation in RAS pathway was detectable in all t(17;19)-ALL cell lines. In contrast, mutations in PTPN11, NRAS, KRAS, and NF1 genes were detectable in none, three, four, and one of 16 t(1;19)-ALL cell lines, respectively, and gene mutation in RAS pathway was detectable in seven out of 16 t(1;19)-ALL cell lines. Incidence of RAS pathway mutation tended to be higher in t(17;19)-ALL cell lines than in t(1;19)-ALL cell lines (P = 0.094 in chi-square test).

| Association between RAS pathway mutation and sensitivity to MEK inhibitor
A recent report revealed that ALL samples with KRAS mutation are sensitive to inhibitors of MAP kinases in vitro. 33 Thus, we tested sensitivity of t(17;19)-ALL cell lines to selumetinib, a MEK inhibitor that has been reported to be active against ALL with the KRAS mutation. We determined IC 50 of selumetinib in t(17;19)-ALL and t(1;19)-ALL cell lines using an alamarBlue cell viability assay ( Figure 6B). The IC 50 of

| Relationship between RAS pathway mutation and multiagent resistance
We finally analyzed a possible association of RAS pathway mutation with drug resistance in t(17;19)-ALL and t(1;19)-ALL cell lines, since RAS pathway mutation was observed more frequently in t(17;19)-ALL cell lines. Association of RAS pathway mutation with sensitivity to each of the five drugs was not statistically significant in t(1;19)-ALL cell lines ( Figure S5), but t(1;19) cell lines with RAS pathway mutation tended to be more resistant to l-Asp than those without it (P = 0.050 in Mann-Whitney test). Then, we compared the total score of three, four, and five drug sensitivities of t(1;19)-ALL cell lines with RAS pathway mutation with those without it ( Figure 6C). Although statistically insignificant, total scores of three, four, and five drug sensitivities tended to be higher in t(1;19)-ALL cell lines with RAS pathway mutation than in those without it. Additionally, multidrug resistance to four (total score ≥ 10) and five drugs (total score ≥ 13) was significantly more common in t(1;19)-ALL cell lines with RAS pathway mutation (four out of seven cell lines: 57.1%) than in those without it (none of nine cell lines: 0%) (P = 0.019 in chi-square test).     for GRα and GRγ isoforms), shows significant correlation with IC 50 of Pred in 72 BCP-ALL cell lines. 37 Of note, GR gene expression level in t(17;19)-ALL cell lines was almost similar to that in t(1;19)-ALL cell lines (data not shown), suggesting that some mechanism(s) besides the GR gene expression level may be associated with resistance to Pred in t(17;19)-ALL cell lines. Recently, P-gp expression has been reported to be associated with resistance to glucocorticoids in inflammatory bowel disease. 38 However, an association of higher P-gp expression with resistance to glucocorticoids is unlikely at least in HALO1 cells, since knockout of P-gp expression by CRISPR-Cas9 did not overcome Dex resistance.
A previous report revealed that genes in RAS pathway are frequently mutated in clinical samples of t(17;19)-ALL cases but not in t(1;19)-ALL cases. 17 In the present study, RAS pathway mutation was detected relatively more frequently in t(17;19)-ALL cell lines than in t(1;19)-ALL cell lines; all four t(17;19)-ALL cell lines and seven of the 16 t(1;19)-ALL cell lines had mutations. Thus, RAS pathway mutation seems to be more frequent in the cell lines than in the clinical samples, suggesting that RAS pathway mutation may be advantageous for in vitro cell growth and/or cell survival of the cell lines. Of note, two t(17;19)-ALL cell lines and four t(1;19)-ALL cell lines with the KRAS mutation were relatively more sensitive to selumetinib, a MEK inhibitor, than two t(17;19)-ALL cell lines and 12 t(1;19)-ALL cell lines without the mutation. This higher sensitivity to selumetinib seems to be consistent with the above hypothesis that RAS pathway mutation may provide an advantage for in vitro cell growth and/or cell survival of the cell lines. Furthermore, among 16 t(1;19)-ALL cell lines, multidrug resistance was significantly more common in the cell lines with RAS pathway mutation than those without it. These observations suggest that frequent RAS pathway mutation may be involved at least partly in the aggressive clinical course of t(17;19)-ALL.
In summary, our observations of a large panel of cell lines revealed that t(17;19)-ALL cell lines were significantly more resistant to multiple agents in induction therapy in comparison with t(1;19)-ALL cell lines. Although there are some limitations in using the cell lines in drug sensitivity studies, our findings seem to be consistent with the clinical notion that t(17;19)-ALL is resistant to intensified induction therapy in comparison with t(1;19)-ALL. Thus, these cell lines may be optional tools to study the mechanism(s) for drug resistance and to verify the activities of newly developed compounds in t(17;19)-ALL.