Glucocorticoid receptor gene mutations confer glucocorticoid resistance in B-cell precursor acute lymphoblastic leukemia

Glucocorticoid (GC) is a key drug in the treatment of B-cell precursor acute lymphoblastic leukemia (BCP-ALL), and the initial GC response is an important prognostic factor. GC receptors play an essential role in GC sensitivity, and somatic mutations of the GC receptor gene, NR3C1, are reportedly identified in some BCP-ALL cases, particularly at relapse. Moreover, associations of somatic mutations of the CREB-binding protein (CREBBP) and Wolf-Hirschhorn syndrome candidate 1 (WHSC1) genes with the GC-resistance of ALL have been suggested. However, the significance of these mutations in the GC sensitivity of BCP-ALL remains to be clarified in the intrinsic genes. In the present study, we sequenced NR3C1, WHSC1, and CREBBP genes in 99 BCP-ALL and 22 T-ALL cell lines (32 and 67 cell lines were known to be established at diagnosis and at relapse, respectively), and detected their mutations in 19 (2 cell lines at diagnosis and 15 cell lines at relapse), 26 (6 and 15), and 38 (11 and 15) cell lines, respectively. Of note, 14 BCP-ALL cell lines with the NR3C1 mutations were significantly more resistant to GC than those without mutations. In contrast, WHSC1 and CREBBP mutations were not associated with GC resistance. However, among the NR3C1 unmutated BCP-ALL cell lines, WHSC1 mutations tended to be associated with GC resistance and lower NR3C1 gene expression. Finally, we successfully established GC-resistant sublines of the GC-sensitive BCP-ALL cell line (697) by disrupting ligand binding and DNA binding domains of the NR3C1 gene using the CRISPR/Cas9 system. These observations demonstrated that somatic mutations of the NR3C1 gene, and possibly the WHSC1 gene, confer GC resistance in BCP-ALL.


Introduction
B-cell precursor acute lymphoblastic leukemia (BCP-ALL) is the most prevalent malignancy in children. Recently, the prognosis of BCP-ALL patients has dramatically improved: the 5-year overall survival rate exceeds 90 % in the majority of contemporary regimens [1][2][3]. However, the prognosis of the relapsed cases remains poor due to acquired resistance to standard chemotherapeutic agents [4][5][6]. To characterize the genetic mechanisms for chemoresistance in the relapsed pediatric ALL cases, comprehensive genomic analyses have recently been undertaken. As a result, the acquired mutations in the genes involving drug sensitivities have been clarified in the leukemia cells at relapse [7][8][9][10].
Glucocorticoid (GC) is an essential therapeutic agent in the treatment of BCP-ALL patients [11,12], and resistance to GC-monotherapy is one of the most important poor prognostic factors [13,14]. In the anti-leukemic activities of GC, glucocorticoid receptor (GR), encoded by the NR3C1 gene, plays a central role. GR translocates from cytosol to the nucleus through GC binding, and functions as a transcriptional regulator [15].
Once GR binds to its responsive DNA elements, GR transcribes pro-apoptotic genes such as BIM. Conversely, GR represses anti-apoptotic genes such as BCL2 and BCL-XL [16,17]. The acquisition of the NR3C1 gene mutation in GC-resistant mechanisms in the relapsed ALL cases has recently been reported [7,[18][19][20]. The significance of the NR3C1 mutations in the GC-sensitivity of ALL cells has been evaluated using ALL cell lines transfected with the expression vectors containing the wild-type or mutated-types of the NR3C1 cDNA [7,18,19]. The human NR3C1 gene has four splicing variants besides the functional isoform GRα, and their significance in GC sensitivity of ALL has been widely discussed [21,22]. Thus, to properly evaluate the significance of the NR3C1 mutations in the GC-sensitivity of ALL cells, the GC-resistant phenotype needs to be verified in the ALL cells with the intrinsic NR3C1 mutations. However, this issue still remains to be clarified.
As with other possible genetic mechanisms involved in GC-resistance in ALL, mutations of the CREB-binding protein (CREBBP) and Wolf-Hirschhorn syndrome candidate 1 (WHSC1) genes have also been reported [7,[23][24][25]. The CREBBP gene mutations have reportedly induced impairment of histone acetylation ability, and subsequently disrupted transcriptional regulation of the target genes including the GC response genes [23]. WHSC1 is a histone methyltransferase specific for H3K27me2. The WHSC1 gene mutations have reportedly altered the GC transcriptional response, resulting in deactivation of pro-apoptotic BIM and BFM [26]. However, the significance of the CREBBP and WHSC1 gene mutations in GC-sensitivities of ALL cells still remains to be elucidated in their intrinsic genes.
In the present study, to verify the possible association of the NR3C1, CREBBP, and WHSC1 mutations with GC-resistance in their intrinsic genes, we analyzed a large series of BCP-ALL and T-ALL cell lines. We also developed GC-resistant sublines of the ALL cell lines by knocking out the NR3C1 gene using the CRISPR-Cas9 system.

AlamarBlue assay
Fifty percent inhibitory concentration (IC50) values of dexamethasone (Dex) and prednisolone (Pred) were determined using the ala-marBlue cell viability assay (Bio-Rad Laboratories, Hercules, CA) as previously reported [50]. Cells (0.1-4 × 10 5 ) were placed onto 96-well flat bottom plates in the presence or absence of seven separate concentrations of each drug in triplicate. The cells were cultured for 66 h to determine Dex and Pred sensitivities, and 20 μL of alamarBlue was then added. After incubation for an additional 6 h in the presence of ala-marBlue, the optimal density was read on a spectrophotometer at 570 nm using 600 nm as a reference wavelength. Cell viability was calculated by the ratio of the optical density of the treated wells to that of the untreated wells as a percentage. The concentration of each agent required to reduce the viability of the treated cells to 50 % of the untreated cells (IC50 value) was calculated and the median IC50 value of three independent assays was determined.

Real-time reverse transcription polymerase chain reaction (RT-PCR) assay
Total RNA was extracted from each cell lines with TRIzol reagent (Invitrogen, Waltham, MA), and reverse transcription was performed by using SuperScript IV VILO Master Mix (Thermo Fisher Scientific). Real-time reverse transcription polymerase chain reaction (RT-PCR) analyses of NR3C1 gene were performed using a TaqMan Probe Kit (Hs00353740_m1). As an internal control, gene expression levels of betaactin (ACTB) were also examined using a TaqMan Probe Kit (Hs99999903_m1).

NR3C1, CREBBP, and WHSC1 mutations in BCP-ALL cell lines
In 99 BCP-ALL cell lines, we performed target-exon sequences of the NR3C1, CREBBP, and WHSC1 genes, in which the covering exon was 100 %, 99.4 %, and 100 %, respectively. The median sequence coverage in the analyses was 311 (range: 65.5-1117). Identified mutations are plotted in Fig. 1a as a heatmap with establishment information of each cell line including the types of fusion genes. Among 99 BCP-ALL cell lines, mutations of the NR3C1, CREBBP, and WHSC1 genes were detectable in 14, 24, and 20 cell lines, respectively. Two cell lines had mutations in the all of the three genes, three cell lines had mutations in both the CREBBP and WHSC1 genes, and another two cell lines had mutations in both the NR3C1 and CREBBP genes. Among 15 NR3C1 mutations identified in 14 cell lines, 6 were missense mutations, 4 were frameshift insertions, 3 were nonsense mutations, 1 was a frameshift deletion, and 1 was an in-frame insertion (Fig. 1b). Six mutations were located in the glucocorticoid receptor domain, four were in the ligandbinding domain, and two were in the DNA binding domain (Fig. 1b, Table 1). Among 32 CREBBP mutations identified in 24 cell lines, 19 were missense mutations, 7 were nonsense mutations, 3 were frameshift deletions, 2 were frameshift insertions, and 1 was an in-frame deletion (Fig. 1c, Table 2). Five mutations were located in the KAT11 catalytic domain and three were in the Creb binding domain. Among 20 WHSC1 mutations identified in 20 cell lines, 19 were missense mutations and 1 was a splice-site mutation (Fig. 1d, Table 3). Eleven mutations were located in the SET domain.
Finally, we evaluated the association of NR3C1, CREBBP, and WHSC1mutations with karyotypes in 99 BCP-ALL cell lines and compared with that in the clinical samples using PeCan database (https://pecan.stjude.cloud). Among 14 cell lines with the NR3C1 mutations, the incidences of MEF2D fusion (18 % vs 29 %) and TCF3-HLF (4 % vs 21 %) tended to be higher, while number of patients with NR3C1 mutations in PeCan database was too small (n = 5) to verify the association (Supplement Fig. 1a-d). Among 24 cell lines with the CREBBP mutations, the incidence of MEF2D fusion (18 % vs 29 %) tended to be higher, while that of MLL fusion (14 % vs 4 %) tended to be lower (Supplement Fig. 1a, e). Consistently, in the PeCan database, no case with MLL fusion was included (Supplement Fig. 1b, f). Among 20 cell lines with the WHSC1 mutations, the incidence of TCF3-PBX1 (16 % vs 35 %) and ETV6-RUNX1 (6.1 % vs 10 %) tended to be higher (Supplement Fig. 1a, g). Consistently, in the PeCan database, the incidence of TCF3-PBX1 and ETV6-RUNX1 were relatively higher as previously reported (Supplement Fig. 1b, h) [51,52]. Of note, although statistically insignificant, the NR3C1 gene mutations tended to be more frequently observed in the cell lines established at relapse (10/55 cell lines; 18 %) than those established at diagnosis (2/31 cell lines; 6 %) ( Fig. 2a and b). In contrast, the CREBBP and WHSC1 gene mutations were almost equally observed in the cell lines established at relapse (16 % and 20 %, respectively) and in those established at diagnosis (31 % and 19 %, respectively) ( Fig. 2a and b). These results indicated that the mutations of the NR3C1, CREBBP, and WHSC1 genes were relatively common in BCP-ALL cell lines and that the NR3C1 mutation tended to be more frequently observed in the cell lines established at relapse.

Association of NR3C1, CREBBP, and WHSC1 mutations with GC sensitivities in BCP-ALL cell lines
Next, we analyzed the possible association of the NR3C1, CREBBP, and WHSC1 mutations with GC sensitivities in 98 BCP-ALL cell lines, except for HBL3 due to its poor proliferation in regular culture condition. As we previously reported in 72 BCP-ALL cell lines [53], the IC50 values of dexamethasone (Dex) and prednisolone (Pred) were significantly correlated with each other (r 2 = 0.67, p < 0.0001) (Supplement Fig. 2).
In  (Fig. 3a). Similarly, although the association of the Pred-resistant phenotype with the NR3C1 mutation was statistically insignificant (10/14 mutated cell lines vs 39/84 unmutated cell lines; p = 0.15 in Fisher's exact test), the IC50 values of Pred in 14 NR3C1 mutated cell lines (median IC50; 52.4 μM) were significantly higher than those in 84 NR3C1 unmutated cell lines (median IC50; 0.31 μM) (p = 0.012) (Fig. 3b). These results indicated that the NR3C1 gene mutation is clearly associated with the GC-resistant phenotype of BCP-ALL cell lines in their intrinsic gene.
Next, we analyzed the association between the WHSC1 mutation and GC sensitivities. In contrast to the NR3C1 mutation, no statistically significant differences were observed in the IC50 values of Dex or Pred between 20 WHSC1 mutated cell lines and 78 WHSC1 unmutated cell lines (Fig. 3c, d). Since mutation status of the NR3C1 gene was highly associated with the GC sensitivities in BCP-ALL cell lines, we focused next on 84 NR3C1 unmutated cell lines. Although statistically insignificant, 18 WHSC1 mutated cell lines were more resistant to Dex (p = 0.060 in Mann-Whitney U test) and Pred (p = 0.077) than 66 WHSC1 unmutated cell lines (Fig. 3e, f). We finally analyzed the association of the CREBBP mutation with GC sensitivities. However, no statistically significant differences were observed in the IC50 values of Dex or Pred between 24 CREBBP mutated cell lines and 74 CREBBP unmutated cell lines (Fig. 3g, h). Moreover, even when focusing on the 84 NR3C1 unmutated cell lines, no statistically significant differences were observed in the IC50 values of Dex or Pred between 20 CREBBP mutated cell lines and 64 CREBBP unmutated cell lines (Fig. 3i, j). These results suggested that the mutation of the WHSC1 gene, but not the CREBBP gene, is associated with the GC-resistant phenotype in the NR3C1 unmutated BCP-ALL cell lines.

Association of NR3C1, CREBBP, and WHSC1 mutations with Dexamethasone sensitivities in T-ALL cell lines
We also analyzed the associations of the NR3C1, CREBBP, and

Association of the NR3C1 gene expression levels with GC sensitivities in BCP-ALL cell lines
We quantified the NR3C1 gene expression levels in 98 BCP-ALL cell lines except for HBL3 by RT-PCR with the TaqMan probe targeting at exons 4-5 that are commonly transcribed in all of the 4 splice variant forms of GR [53]. The NR3C1 gene expression levels in 98 BCP-ALL cell lines showed weak negative correlations with the IC50 values of both Dex (r 2 = 0.14, p = 0.0001) and Pred (r 2 = 0.15, p < 0.0001) (Fig. 4a, b).  The NR3C1 gene expression levels in 50 Dex-sensitive cell lines (median relative expression level; 1.0) were significantly higher (p < 0.0001 in Mann-Whitney U test) than those in 48 Dex-resistant cell lines (0.59) (Fig. 4c). Similarly, the NR3C1 gene expression levels in 49 Pred-sensitive cell lines (median relative expression level; 1.01) were significantly higher (p < 0.0001 in Mann-Whitney U test) than those in 49 Pred-resistant cell lines (0.62) (Fig. 4d). The NR3C1 gene expression levels in 14 NR3C1 mutated cell lines (median relative expression level; 0.55) tended to be lower (p = 0.078 in Mann-Whitney U test) than those in 84 NR3C1 unmutated cell lines (0.86) (Fig. 4e). When focused on the 84 NR3C1 unmutated cell lines, the NR3C1 gene expression levels in 48 Dex-sensitive and 45 Pred-sensitive cell lines were significantly higher than those in the 36 Dex-resistant (p < 0.0001) and 39 Pred-resistant (p = 0.0001) cell lines, respectively (Fig. 4f, g). These results indicated that mutations and lower gene expression levels of the NR3C1 gene were independently associated with GC resistance in BCP-ALL cell lines. Next, we evaluated the association of the WHSC1 and CREBBP mutations with the NR3C1 gene expression level. Interestingly, consistent with their relatively lower GC sensitivities, the NR3C1 gene expression levels in 20 WHSC1 mutated cell lines (median relative expression level; 0.62) were significantly lower (p = 0.018 in Mann-Whitney U test) than those in 78 WHSC1 unmutated cell lines (0.87) (Fig. 4h). When focused on the 84 NR3C1 unmutated cell lines, the NR3C1 gene expression levels in 18 WHSC1 mutated cell lines (median relative expression level: 0.62) were significantly lower (p = 0.0085 in Mann-Whitney U test) than those in 66 WHSC1 unmutated cell lines (0.98) (Fig. 4i).
In contrast, the NR3C1 gene expression levels in the 24 CREBBP mutated cell lines (median relative expression level; 0.92) were almost similar to those in the 74 CREBBP unmutated cell lines (0.79) (Supplement Fig. 3a). Even when focused on the 84 NR3C1 unmutated cell lines, the NR3C1 gene expression levels in the 20 CREBBP mutated cell lines (0.97) were almost similar to those in the 64 CREBBP unmutated cell lines (0.83) (Supplement Fig. 3b). These results indicated that lower NR3C1 gene expression level is a potential mechanism for GC resistance in BCP-ALL cell lines with the WHSC1 gene mutation.

Establishment of GC-resistant sublines of BCP-ALL cell lines by disrupting the NR3C1 gene
Since most of the NR3C1 mutations (12/15; 80 %) observed in this study affected either the ligand binding or DNA binding domains (Fig. 1b), we tried to establish GC-resistant sublines of the GC-sensitive BCP-ALL cell line by disrupting ligand binding and DNA binding domains of the NR3C1 gene using the CRISPR/Cas9 genome editing system (Fig. 5a). We electroporated the plasmid containing both Cas9 cDNA and sgRNA targeting at the PAM site adjacent to codons 473-474 (hotspot mutation sites located in the middle of exon 4 that encodes DNA-binding domain) of the NR3C1 gene. We used the GC-sensitive 697 cell line (IC50 values of Dex and Pred; 3 nM and 3.6 nM, respectively), which has no mutation in the NR3C1, CREBBP, and WHSC1 genes. After 10-day expansion of the transfected cells in the absence of GCs, the cells were cultured in the presence of 90 nM of Pred (25 times higher concentration than the IC50). Following 7-day selection with Pred, the Predresistant subline was expanded. Sanger sequencing of genomic PCR products of the subline, which were subcloned into the TA-cloning vector, revealed various insertion or deletion at the target site resulting in one to four amino acids insertion or frameshift mutations (Fig. 5b). The obtained subline showed a highly GC-resistant phenotype compared with its parental cell line; the IC50 values of Dex and Pred were > 250 nM and > 54.1μM, respectively. When treated with 200 nM of Dex or 3.5 μM of Pred, more than 90 % of the parental cells underwent cell death. In contrast, no significant apoptotic cell death was induced in the obtained subline. These results demonstrated that 1-4 amino acid insertions and frameshifts in exon 4 adjacent to codons 473-4 induced GC resistance in the BCP-ALL cell lines.

Discussion
In the present study, we performed target exome sequencing of the NR3C1, CREBBP, and WHSC1 genes in 99 BCP-ALL cell lines, and found their mutations in 14, 24, and 20 cell lines, respectively. These mutations were not mutually exclusive: two cell lines had mutations in the all of the three genes, and five cell lines had mutations in two of the three genes. Among 87 BCP-ALL cell lines with establishment information, mutations of the NR3C1, CREBBP, and WHSC1 genes were observed in 6 %, 31 %, and 19 % of the 32 cell lines established at diagnosis, whereas 18 %, 16 %, and 20 % of the 55 cell lines were established at relapse, respectively.
Thus, although statistically insignificant, the NR3C1 gene mutation was relatively more common in the cell lines established at relapse. In the previous study of paired clinical samples obtained at diagnosis and at relapse in 67 BCP-ALL cases, mutations of the NR3C1, CREBBP, and WHSC1 genes were observed in 4.5 %, 17.9 %, and 1.5 % of the samples at diagnosis compared to 9 %, 23.9 %, and 4.5 % of the samples at relapse, respectively [8]. Thus, the incidence of the CREBBP mutations in the BCP-ALL cell lines were almost similar to that in the clinical samples. In contrast, incidence of the NR3C1 and WHSC1 mutations in the BCP-ALL cell lines seemed to be higher than that in the clinical samples. In particular, the WHSC1 mutation was observed approximately 10 times more commonly in the BCP-ALL cell lines than in the clinical samples, suggesting that the WHSC1 mutation may be somehow associated with growth advantage in vitro.
In the anti-leukemic activities of GC, GR plays a central role [11,12]. Indeed, in the present analysis of 98 BCP-ALL cell lines, lower NR3C1 gene expression level was significantly associated with GC resistance, as we previously reported in 72 BCP-ALL cell lines [53]. In addition to the lower NR3C1 gene expression level, the NR3C1 gene mutations were significantly associated with GC resistance, since the IC50 values of Dex and Pred in 14 NR3C1 mutated cell lines were significantly higher than those in 84 NR3C1 unmutated cell lines. Consistent with the GC-resistance of the NR3C1 mutated cell lines in the present study, it has been reported that the introduction of mutated-types of the NR3C1 cDNA with expression vector rendered GC-resistance in the GC-sensitive ALL cell lines [18,19]. In this context, it should be noted that the human NR3C1 gene has four splicing variants besides the functional isoform GRα [53]. Under these circumstances, previous analyses using expression vectors evaluated the mutated cDNAs of the main GRα isoform only. Importantly, in the present study, we confirmed the GC-resistance in the majority of NR3C1 mutated cell lines, in which various splicing isoforms of the NR3C1 gene are expressed [53]. Thus, this is the first direct confirmation regarding the impact of the NR3C1 gene mutation on the GC sensitivities of BCP-ALL in the intrinsic NR3C1 gene.
Among 14 NR3C1 mutated BCP-ALL cell lines, P30/OHK with C644R (IC50 of Dex and Pred; 5.2 nM and 7.9 nM, respectively) and THP5 with S377Lfs*5 (11.4 nM and 1.8 nM, respectively) were relatively sensitive to GCs. Of note, the majority (8 cell lines) of the 12 GC-resistant mutated cell lines had homozygous mutations while both P30/OHK (mutated sequences; 48.6 %) and THP5 (47.7 %) had the heterozygous mutation. Thus, wild-type GR is produced by an intact allele and may be involved in partial GC sensitivity of P30/OHK and THP5. In a protein function prediction tool (PROVEAN, http://provean.jcvi.org/) [55], C644R is predicted to be functionally deleterious (PROVEAN score; − 10.3 < − 2.5 cutoff value). Moreover, S377Lfs*5 is also deleterious due to a loss of the C-terminal domain of GR as a result of the frameshift mutation. Thus, heterodimers of the mutated and intact GR as well as homodimers of the mutated GR are likely to be non-functional in these two cell lines. In this context, other acceleration factors involving GC sensitivity may also contribute to partial GC sensitivity of these two cell lines in spite of the heterozygous mutation of the NR3C1 gene.
In contrast to the NR3C1 mutations, no statistically significant associations with GC-resistance were observed in BCP-ALL cell lines with the WHSC1 and CREBBP mutations. Since the NR3C1 mutation status was strongly associated with the GC-resistant phenotype, we further focused on the 84 NR3C1 unmutated cell lines. Of note, among 84 NR3C1 unmutated cell lines, the WHSC1 mutated cell lines tended to be more resistant to GC in comparison with the unmutated cell lines. Moreover, the NR3C1 gene expression levels in the WHSC1 mutated cell lines were significantly lower than those in the unmutated cell lines. These our observations in BCP-ALL cell lines were supported by a recent paper by Li J et al., in which introduction of WHSC1 mutation by genome editing into RCH-ACV cell line with the wild type WHSC1 sequence induced Dex-resistant phenotype, while revision of the WHSC1 mutation to wild type sequence by genome editing in the Dex-resistant ALL cell lines with the E1099 K mutation conferred Dex sensitivity [56]. Although there was no direct evidence for the association of WHSC1 and NR3C1 in GC resistance in clinical samples, recent report by Fig. 3. Association of the NR3C1 (a-b), WHSC1 (c-f), and CREBBP (g-j) gene mutations with the GC sensitivities in BCP-ALL cell lines. In panels (a), (b), (c), (d), (g), and (h), the IC50 values of Dex (a, c, and g) and Pred (b, d, and h) were compared between the cell lines with mutation (+) and those without mutation (-) in 98 BCP-ALL cell lines. In panels (e), (f), (i) and (j), the IC50 values of Dex (e and i) and Pred (f and j) were compared between the cell lines with mutation (+) and those without mutation (− ) in the 84 NR3C1 unmutated cell lines. In each panel, vertical axis indicates log-scaled IC50 value of Dex or Pred and p-value in Mann-Whitney U test is indicated on the top.
Li J et al. focused on this issue using the PDX model [56]. Consistent with the findings in ALL cell lines, Dex-treatment failed to promote survival of the mice inoculated with WHSC1 E1099 K mutated ALL samples, while it significantly promoted survival of the mice inoculated with ALL sample with wild type WHSC1. They also demonstrated that introduction of WHSC1 E1099 K mutation by genome editing into ALL cell line downregulated basal NR3C1 gene expression level, which was associated with the accumulation of H3K27me3-binding at the NR3C1 gene promoter. Consistently, in the present study, the NR3C1 gene expression levels in the 10 cell lines with the WHSC1 E1099 K mutation were relatively lower than those without the mutation, regardless of the NR3C1 mutations (p = 0.081, Supplement Fig. 4). These results strongly suggested that WHSC1 is an upstream regulatory factor of the NR3C1 gene and that the WHSC1 mutation subsequently induces GC-resistance by downregulating NR3C1 gene expression level.
In contrast to the WHSC1 mutations, even when focused on the NR3C1 unmutated cell lines, we could not confirm the association of the mutation status of the CREBBP gene with the GC-sensitivities and the NR3C1 gene expression levels. Association of CREBBP mutations with GC-resistance was originally suggested by the findings that all of the five T-ALL cell lines with CREBBP mutations were resistant to Dex [23]. However, in this original report, three of four T-ALL cell lines without CREBBP mutation were also resistant to Dex. Consistently, in the present study, 13 out of 14 CREBBP mutated T-ALL cell lines were resistant to Dex, whereas 6 out of 8 CREBBPunmutated T-ALL cell lines were resistant to Dex. Thus, most of T-ALL cell lines were resistant to Dex regardless of their mutational status of the CREBBP gene. Accordingly, one background aspect for difference in the association of CREBBP mutations with GC sensitivity between previous observation [23] and present study might be difference between T-ALL cell lines and BCP-ALL cell lines.
Finally, we tried to evaluate the significance of the NR3C1 mutation in the GC-sensitivity by disrupting ligand binding and DNA binding domains of the NR3C1 gene using the CRISPR/Cas9 genome editing system in GC-sensitive cell line. We targeted the PAM site adjacent to codons 473-474 of the NR3C1 gene, and selected the GC-resistant sublines in the presence of Pred at a 25 times higher concentration than IC50. The obtained subline was confirmed to be various acquired insertions or deletions at the target site of the intrinsic NR3C1 gene resulting in 1-4 amino acids insertion or frameshift mutations, respectively. In the inserted amino acids as a result of non-homologous endjoining, no common functional features were observed in terms of electrical charge or water solubility. These mutations were located at the C-terminal end of the zinc finger domain for dimerization [57] and the adjacent R477 residue plays a critical role in the GC transactivation activity [58]. Thus, these mutations may abrogate transactivation activity of the GC-GR complex by disrupting dimerization in the obtained GC-resistant subline.
In conclusion, our analyses provided direct evidence showing that the loss-of-functional mutations of the intrinsic NR3C1 gene induce GC resistance in BCP-ALL cell lines. Of clinical importance, the NR3C1 mutations were more commonly observed in the cell lines established at relapse than at diagnosis. These observations in BCP-ALL cell lines suggest that the acquisition of the loss-of-functional mutation of the NR3C1 gene may render the GC-resistant phenotype and, subsequently, induce expansion of the mutated clone during chemotherapy (including GC therapy), as observed in our CRISPR-Cas9 mediated NR3C1 gene disruption model.
Although further evaluations are required, our observations also suggest that the WHSC1 mutation may provide some growth advantages as well as the GC-resistant phenotype of BCP-ALL cells. In addition, these large series of BCP-ALL cell lines are useful tools to evaluate the involvement of somatic gene mutations in the drug-resistant phenotype of poor prognostic leukemia.

Author contribution
Minori Tamai