Targeting of heme oxygenase-1 attenuates the negative impact of Ikaros isoform 6 in adult BCR-ABL1-positive B-ALL

The correlation between Heme oxygenase-1 (HO-1) and dominant-negative Ikaros isoform 6 (IK6) is unclear. Firstly, we detected that IK6 existed in 20 of 42 (47.6%) adult BCR-ABL1-positive B-lineage acute lymphoblastic leukemia (BCR-ABL1-positive B-ALL) by using reverse transcribed polymerase chain reaction (PCR) and nucleotide sequencing. IK6-positive patients had an unfavorable outcome compared with IK6-negative ones. Further study showed that the level of HO-1 expression was higher in IK6-positive patients’ samples than that in IK6-negative ones. And there was a strong correlation between the expression of IK6 and HO-1. The growth of primary CD34+ leukemic cells derived from our IK6-positive patients’ pool was prohibited by silencing HO-1, further promoting their apoptosis. Furthermore, primary CD34+ leukemic cells derived from IK6-positive patients shown poor responses to imatinib in comparison with wild-type (IK1) patients, suggesting that the expression of IK6 resisted to imatinib in adult BCR-ABL1-positive B-ALL. Importantly, inhibition of HO-1 also increased their sensitivity to tyrosine kinase inhibitors (TKIs). Finally, we found that IK6 activated downstream STAT5, and HO-1 was one of the downstream target genes of STAT5. In conclusion, HO-1 is an essential survival factor in BCR-ABL1-positive B-ALL with IK6, and targeting HO-1 can attenuate the negative impact of IK6.

Besides, IKZF1 deletions have been also considered as indicators of poor prognosis in B-ALL [5][6][7][8][9]. The IKZF1 gene encoding Ikaros, a DNA-binding zinc finger protein, involves in B-cell proliferation and differentiation [10,11]. IK1, known as IKZF1 wild type, contains 7 exons (exons 2 -8). It totally encodes 6 Cys-2 to His-2 zinc fingers, 4 at its N-terminus (F1, F2, F3, and F4) and 2 at its C-terminus (F5 and F6) [12,13]. Ikaros isoforms can also take a function of DNA-binding when there are at least three DNA-binding zinc fingers. However, the lack of two or more zinc-finger domains will impair the function of Ikaros proteins in a dominant-negative manner [14]. Mounting studies have revealed that the frequency of IKZF1 deletions is high (70%) in BCR-ABL1-positive B-ALL [9,15,16], and the most common one is IK6 which affects the DNA-binding domain in exons 4-7. IK6 exerts a dominant-negative effect over the unaffected allele, which results into a loss of the tumor suppressor [17,18]. Moreover, the expression of IK6 usually predicts an adverse clinical outcome [8] and the insensitivity to TKIs Research Paper www.impactjournals.com/oncotarget [9,13]. Lots of experimental data are available for pediatric B-ALL but for adults. Despite limited data, it appears to be true for adults, too [19,20]. The BCR-ABL1 fusion transcript itself is a strong adverse risk factor, and the existence of extra IK6 seems to exacerbate poor prognosis and risks of relapse. Thus, it is even wiser to figure out whether there are alternative therapies for patients with the expression of IK6.
HO-1 is a stress-related cytoprotective molecule. In our previous works, we found that it inhibited the apoptosis and promoted the proliferation of leukemic cells, being associated with tumorigenesis and drug resistance [21][22][23]. Recent studies show that HO-1 is considered as a survival molecule to play an important role in BCR-ABL1-positive leukemic cells, and HO-1-targeting drugs are able to synergize with TKIs to inhibit growth and promote apoptosis of leukemic cells [24][25][26]. Therefore, HO-1 is an essential survival factor and potential target in BCR-ABL1-positive leukemias.
Although IKZF1 deletions are under the spotlight in pediatric BCR-ABL1-positive B-ALL, there are few studies on adults. Therefore we chose adult B-ALL as the research object. As noted, IK6 is highly associated with the BCR-ABL1 fusion oncogene [13], meanwhile HO-1 is identified as a novel BCR-ABL1-driven survival molecule [24,25]. It is worthwhile to make a connection between HO-1 and IK6. There is assuming a correlation, whether targeting HO-1 can attenuate the negative impact of IK6 in adult BCR-ABL1-positive B-ALL, and how. To address these issues, we analyzed the expression of HO-1 in 42 adult BCR-ABL1-positive B-ALL. We further did studies on cell proliferation, apoptosis and cell cycle by increasing or decreasing the expression of HO-1 in primary CD34 + leukemic cells from our patients' pool. After the inhibition of HO-1, the experiment on the sensitivity of primary leukemic cells to TKIs was carried out. As previously demonstrated in murine B lymphocytes [27], IKZF1 deletions may synergize with STAT5 activation in the presence of BCR-ABL1 [12,27]. Therefore, it will be important to figure out the relationship among IK6, STAT5 and our target gene HO-1. Based on this, we studied the molecular pathways involved in BCR-ABL1-positive B-ALL with IK6.

The IK6 is highly expressed in BCR-ABL1-positive B-ALL
We analyzed the expression of different IKZF1 isoforms in 42 adult BCR-ABL1-positive B-ALL by using RT-PCR and nucleotide sequencing. 4 subtypes of IKZF1, IK1, IK2, IK4 and IK6 ( Figure 1A), were detected in our study. Results shown that patients containing either IK6 or IK4 accounted for 64.2% (27 of 42). IK4 was only expressed in 7 patients. There were 9 patients exclusively expressed IK6 in the 20 patients (47.6%) who expressed IK6, while the remaining 11 patients were accompanied with the expression of IK1 or IK2. Additionally, another 11 patients exclusively expressed IK1(26.2%), and the left 4 patients only present IK2 ( Figure 1B, Table 1). Data from gene sequencing also validated the structure of IK6 as described above ( Figure 1C). Usually, BCR-ABL1 fusion gene predominantly transcribes into either p210 or p190 oncoprotein, but our data further demonstrated the existence of IK6 did not limit to a specific BCR-ABL1 subtype (Table 2). Moreover, experimental results shown that IK6 expression was significantly correlated with higher white blood cell (WBC) count, minimal residual disease (MRD) and poor early clinical response, but neither gender nor age ( Table 2). We also found that IK6positive patients had a lower disease free survival (DFS) compared with IK6-negative ones, even when treated with imatinib ( Figure 1D), and with a higher incidence of relapse ( Figure 1E). These findings confirm that IK6 is highly expressed in adult BCR-ABL1-positive B-ALL, and it is closely associated with a poor outcome.

The expression of HO-1 is strongly correlated with IK6
We further analyzed the expression of HO-1 in patients' samples by using qRT-PCR and western blot. Compared to normal CD34 + cells, the level of HO-1 expression was higher in IK6-positive patients' samples, but there was no obvious increase in IK6negative samples. Furthermore, the expression of HO-1 in IK6-positive patients was 4 folds more than that in IK6-negative ones (Figure 2A). Then the expressions of Ikaros and HO-1 were detected by western blot. Primary leukemic cells expressed a 57kDa immunoreactive protein corresponding in size to IK1, a 55kDa immunoreactive protein corresponding in size to IK2 and a 47kDa immunoreactive protein corresponding in size to IK4. In contrast, we confirmed the presence of a smaller immunoreactive protein band of approximately 35 kDa, which corresponded in size to IK6 ( Figure 2B). HO-1 were also higher at the protein level in IK6-positive samples. Significantly, 15 of 20 (75%) IK6-positive patients were detected the overexpression of protein by western blot study ( Figure 2C, 2D). Because of the status of IKZF1 isoforms, there was a strongly statistically significant difference in HO-1 expression of the 20 IK6-positive patients. Specifically, 9 patients who expressed only IK6 expressed more HO-1 than those who co-expressed with another IKZF1 isoform (IK1 or IK2) at both mRNA and protein level (Figure 2A, 2D). In addition, HO-1 had no correlation with Ikaros(57KDa)(IK1), Ikaros(55KDa) (IK2) and Ikaros(47KDa)(IK4), but was positively correlated with Ikaros (35KDa) (IK6) ( Figure 2E). Further, co-immunoprecipitation assay in SupB15 cells showed that Flag-tagged IK6 interacted with Myc-tagged www.impactjournals.com/oncotarget HO-1. We further showed that endogenous Ikaros was coimmunoprecipitated with HO-1 in IK6-positive patients, but not in IK6-negative samples ( Figure 2F). Hence, these results show that the expression of HO-1 is strongly correlated with IK6.

HO-1 inhibition increases the sensitivity of leukemic cells to TKIs in BCR-ABL1 + -IK6 patients
We also did experiments on patients' leukemic cells to figure out the effects of inhibiting HO-1 on the sensitivity to TKIs, such as imatinib or nilotinib. Primary CD34 + leukemic cells directly from pIV, pV and pVI shown poor responses to imatinib in comparison with their counterparts (pI, pII and pIII) (IC50, mean±SD, 395.1±54.3nM VS 242.5±27.9nM) ( Figure 4A), suggesting that the expression of IK6 probably resisted to imatinib. Importantly, silencing HO-1 in primary leukemic cells of pIV, pV and pVI sensitized their responses to imatinib with a significant   decrease in the IC50 from 393±53.9nM to 185.3±52.8 nM ( Figure 4B). Similar trends of IC50 (18±1.8nM to 7.6±2.7nM) also shown in nilotinib group, the most potent second-generation TKIs ( Figure 4C). Inhibition of HO-1 by ZnppIX, a pharmacologic inhibitor of HO-1, in place of silencing HO-1 by siRNA lowered the IC50 to imatinib as well ( Figure 4D). These results suggest that HO-1 inhibition in combination with TKIs may be a rational strategy for treating BCR-ABL1 + -IK6 patients.

IK6 activates STAT5 and HO-1 in BCR-ABL1 + -IK6 leukemic cells
We compared the subcellular localization of Ikaros protein in normal CD34 + cells and primary leukemic cells from patients by confocal microscopy. Results showed that BCR-ABL1 + -IK6 leukemic cells caused Ikaros protein to be retained in the cytoplasm, in contrast to the normal localization of Ikaros in the nucleus observed in both normal CD34 + cells and BCR-ABL1 + -IK1 leukemic cells. We also found that forced expression of IK6 in BCR-ABL1 + -IK1 leukemic cells caused partial Ikaros protein to be retained in the cytoplasm ( Figure 5A). These findings indicate an ability of IK6 to disrupt the nuclear transcription factor activity of Ikaros, which is consistent with previous reports [12,28].
Then we used flow cytometry and qRT-PCR to look for molecular pathways that are differentially activated in IK6-positive leukemic cells. We found that IK6-transduced BCR-ABL1 + -IK1 or BCR-ABL1 + -IK2 or BCR-ABL1 + -IK4 leukemic cells contained consistently higher levels of pSTAT5 by comparison with those derived from either matched control-transduced cells or IK6transduced normal CD34 + cells. Effects of IK6 on other signaling intermediaries in BCR-ABL1-positive B-ALL cells were more heterogeneous ( Figure 5B, 5C). In light of these findings, we hypothesized that the expression of IK6 in BCR-ABL1-positive B-ALL leukemic cells may disrupt mechanisms regulating JAK2-STAT5 activity in them. Consistent with this postulate was the finding that IK6-expressing BCR-ABL-positive B-ALL leukemic cells contained decreased levels of SOCS2 and SH2B3 (LNK) transcripts, both of which encode important known negative regulators of JAK-STAT signaling [28]. Notably, none of these molecular effects were apparent in the IK6transduced normal CD34 + cells ( Figure 5D). These suggest that IK6 enhances STAT5 signaling in BCR-ABL1 + -IK6 leukemic cell.
Next, our results of western blot were consistent with confocal examination. Our data also demonstrated that HO-1 was mainly expressed in the nucleus, too ( Figure 5E). Secondly, the expression of IK6 was not only associated with the increase of pSTAT5 in both cytoplasm and nucleus, but also the increase of HO-1 in nucleus ( Figure 5E). However, the expressions of Ikaros and pSTAT5 were not significantly changed after silencing HO-1 in BCR-ABL1 + -IK6 leukemic cells ( Figure 5F). Although AZ960 can block the phosphorylation of STAT5 [29,30], the expression of Ikaros protein was not significantly changed after treating BCR-ABL1 + -IK6 leukemic cells with 0.2uM AZ960 for 48 hours. Contrarily, the expression of HO-1 was significantly decreased. Similar results were found after treated by SH-4-54, inhibitor of STAT5 [31] (Figure 5G). These indicate that IK6 activates downstream STAT5, and HO-1 may be one of the downstream target genes of STAT5.
In brief, all of these data demonstrate that Ikaros protein is mainly synthesized and localized in the cytoplasm due to IK6, and HO-1 is activated by the IK6/STAT5 signaling pathway in BCR-ABL1 + -IK6 leukemic cells.

DISCUSSION
Few studies have investigated adult B-ALL, but available data suggest that IK6 might occur at an even higher frequency in adults [13,19,20]. In this study, we found there were several IKZF1 deletion patterns in adult BCR-ABL1 + B-ALL patients by using RT-PCR, in which IK6 was the most common type accounting for 47.6%. Our study confirms the statistics reported by Iacobucci et al on the frequency of IK6 (~50%) in adult B-ALL [13,32]. Although lack of the investigation on IK6 expression in BCR-ABL1-negative B-ALL, according to references, the frequency of all abnormal IKZF1 isoforms is around 15% in BCR-ABL1-negative B-ALL [9,15,16], we believe that IK6 is highly expressed in BCR-ABL1-positive B-ALL based on all the facts. In addition, we found that early clinical response was poor in BCR-ABL1-positive B-ALL patients with IK6, and they usually had a higher relapse rate. Thus, IK6 is indeed related to poor prognosis.
The correlation between IK6 and HO-1 has not been reported yet. We studied the expression and role of HO-1 in BCR-ABL1 + -IK6 leukemic cells to explore their relationship. Results showed that HO-1 was highly expressed in BCR-ABL1 + -IK6 leukemic cells. And there was a strongly correlation between the expression of IK6 and HO-1. In BCR-ABL1 + -IK6 leukemic cells, silencing HO-1 resulted in a decreased proliferation and a substantial pro-apoptotic effect, validated by the finding of a significantly enriched subG1 population, an increased proportion of Annexin V-positive cells and an increased expression of Cleaved Caspase3. But it remained almost unchanged in IK6-negative leukemic cells by overexpression of HO-1. This suggests that HO-1 is dependent on IK6 to exert its antiapoptosis and proliferation on cells. It is also need to confirm that HO-1 can promote the proliferation and inhibit apoptosis of BCR-ABL1 + -IK6 leukemic cells in vivo. www.impactjournals.com/oncotarget We also found that BCR-ABL1 + -IK6 leukemic cells had a poor response to TKIs. This is consistent with a previous study [13]. These suggest that IK6 expression in BCR-ABL1 + B-ALL may contribute to TKI resistance. A major clinical challenge in the treatment of BCR-ABL1positive B-ALL is resistance to TKI [33][34][35][36]. Therefore, a number of novel agents are needed to be investigated to overcome drug-resisitance. In our study, silencing HO-1 increased BCR-ABL1 + -IK6 leukemic cells' sensitivity to TKIs and led to apoptosis. This phenomenon was also supported by the observation that ZnPPIX substantially augmented the growth-inhibitory effects of imatinib on primary leukemic cells. Moreover, HO-1 inhibitors showed no major effects on viability of normal cells

Figure 4: (Continued ) HO-1 inhibition increases the sensitivity of leukemic cells to TKIs in BCR-ABL1 + -IK6 patients.
C. Equal numbers of IK6, IK6/siHO-1, IK6/vector' were incubated with different concentration of nilotinib (8nM, 12nM, 24nM, 48nM, 96nM). D. Cells were treated with 10uM ZnPPIX and differrant concentration of imtinib as indicated above. www.impactjournals.com/oncotarget [24][25][26]. And water-soluble pharmacologic inhibitors of HO-1 has been developed and tested in solid tumors [24,[37][38][39][40] in the past few years. Therefore, it provides a theoretical basis for HO-1 inhibitors as a novel agent. All of these indicate that HO-1 inhibitors can be considered as a specific for treating TKIs-resistance, especially BCR-ABL1 + -IK6 patients. But a bigger sample pool of patients is required for further investigations on the verification of HO-1 expression and role in BCR-ABL + -IK6 leukemic cells.   Next, we want to know the pathway of HO-1 involved in antiapoptosis of BCR-ABL1 + -IK6 leukemic cells. Firstly, experimental results showed that IK6 activated HO-1 and STAT5 signaling in BCR-ABL1 + -IK6 leukemic cell. But we need to figure out how STAT5 reacts with HO-1. Lentivirus-mediated silencing HO-1 didn't change the expressions of STAT5 and pSTAT5 in BCR-ABL1 + -IK6 leukemic cells. But the expression of HO-1 was significantly decreased after blocking STAT5 or STAT5 phosphorylation. This shows that IK6 leads to STAT5 activiation, and HO-1 may be one of the downstream target genes of STAT5, which inhibits cell apoptosis [ Figure 6]. Our data demonstrate that the signaling pathway (IK6/STAT5/HO-1) suppress the  apoptosis of leukemic cells, suggesting that the activation of STAT5 is also closely involved into BCR-ABL1 + -IK6 leukemic cells' apoptosis. This accords with STAT5 as a key player involved in apoptosis [41].
It is reported that if IK6 leads to the activation of STAT5, STAT5 inhibitors may offer an alternative strategy for the treatment of BCR-ABL1 + -IK6 patients [9]. Nevertheless, there have not yet been any convincing demonstrations of STAT5 inhibitors that are both safe and sufficiently specific for human use [42]. However, we found that inhibiting HO-1 reduced the proliferation and induced the apoptosis of BCR-ABL1 + -IK6 leukemic cells, which enhanced the sensitivity of cells to TKIs. Although HO-1 is in the downstream of STAT5, it can be regarded as one of the targets in treating BCR-ABL1positive B-ALL with IK6. In conclusion, IK6 results in the activation of STAT5 and HO-1 in the presence of BCR-ABL1. Therefore, we believe that there is a close relationship among IK6, STAT5, HO-1 and BCR-ABL1. However, we still need to figure out how these four genes react with each other.
In summary, inhibition of HO-1 prohibits the proliferation and induces the apoptosis of BCR-ABL1 + -IK6 leukemic cells, it also increased the sensitivity of cells to TKIs. Therefore, targeting HO-1 can attenuate the negative impact of IK6 in adult BCR-ABL-positive B-ALL.

Patients samples
Bone marrow (BM) samples were obtained from 42 adult BCR-ABL1-positive B-ALL patients at diagnosis whose blast cells were expressed CD34. Patients' characteristics are shown in Table 2. BM-derived CD34 + cells from healthy donors were served as controls. Mononuclear cells (MNC) were isolated by Ficoll-Hypaque (Sigma USA). CD34 + progenitor cells (>85% pure) from patients and healthy donors were enriched by using immunomagnetic cell sorting system (MACS) (Miltenyi Biotech, Auburn CA), and frozen in Iscove modified Dulbecco medium (IMDM; Gibco) with 50% fetal bovine serum (FBS; Gibco) and 10% dimethyl sulfoxide (DMSO), then stored in liquid nitrogen, as previously described [43]. This study was approved by the Ethics Committee of Guiyang Medical University in accordance with the Declaration of Helsinki, and all subjects signed consent forms.

Reverse-transcribed polymerase chain reaction analysis
Total RNA was extracted by using TRIzol Reagent ((Invitrogen, Carlsbad, CA). RNA was reverse-transcribed into cDNA by using a reverse transcription kit (Takara, Japan). The first round of polymerase chain reaction (PCR) was performed by using specific primer of IKZF1, 5'-CACATAACCTGAGGACCATG-3' (forward) and 5'-AGGGCTTTAGCTCATGTGGA-3'(reverse). Specifically, 2ul cDNA, 12.5uL 2×Taq Master Mix (TianGen Biotech, Beijing, China), 10pmol of each primer, and distilled water were mixed into a final volume of 25uL. The first round condition was 5min at 95°C, followed with 35 cycles at 95°C for 30s, and 57°C for 30s, then 72°C for 90s, finally stopped after 72°C for 7 min. The first round product was used as the template for the second round PCR amplification. The primers 5'-ATGGATGCTGATGAGGGTCAAGAC-3' (foward) and 5'-GATGGCTTGGTCCATCACGTGG-3' (reverse) were designed and synthesized according to the literature [13]. Condition was as follow: 5min at 95°C, followed with 35 cycles at 95°C for 30s, and 62°C for 30s, then 72°C for 30s, finally stopped after 72°C for 7 min. The RNA integrity was confirmed by cDNA amplification of the β-Actin at the level of mRNA.

Cell transfection
Recombinant vector were designed and purchased from Biomics Biotechno logies Co., Ltd (Nan tong Jiangsu, China). Cells were transfected according to a previous literature [46] and routine protocols [47]. Briefly, cells were plated onto 12 well plates at 2.5 x 10 5 cells/ well and infected with lentiviral stocks at a multiplicity of infection (MOI) of 10 in the presence of polybrene (10ug/ ml), then analyzed by fluorescence microscopy (Olympus, Tokyo, Japan) and western blot 72h post infection. Prior to transfection, cells were cultured in the presence of 3-cytokine mix for 7 days.

Proliferation assays
After transfection for 72h, cells were washed and continued to culture in IMDM medium with 20% FBS plus 100 U/ml penicillin and 100 mg/ml streptomycin in the absence of 3-cytokine mix. Cell proliferation was analyzed by the Trypan-blue dye exclusion assay. Cells were cultured in 24-well plates at 2×10 4 /ml. The number of viable cells was counted after 24, 48 and 72h by Trypan blue exclusion.

Apoptosis assays
Cells which were cultured in the absence of 3-cytokine mix were harvested at 72h and washed with phosphate-buffered saline (PBS), then stained with the Annexin-V/7-AAD apoptosis kit (7sea biotech, Shanghai, China) according to manufacturer's instructions. Apoptotic cells were detected by using a FACScan flow cytometry (Becton-Dickinson, Franklin lakes, NJ, USA), and the data were analyzed by using cell fit software.

Cell cycle analysis
Flow cytometry was used to detect the cell cycle distribution. Cells which were cultured in the absence of 3-cytokine mix were harvested at 72h, and washed with PBS. Cells were then fixed by ice-cold 70% ethanol at 4°C overnight followed by washing with PBS at the second day and staining with 50ug/ml propidium iodide, finally dissolved in 100ug/ml RNase A. All reported values were means of three independent mearsurments with SD.

Sequencing analysis
PCR products were purified with QIAquick PCR purification kit (QIA-GEN) according to the manufacturer's protocol. The purified products reacted with Big Dye Terminator DNA sequencing kit (Applied Biosystems) and sequenced by an ABI 3500 Genetic Analyzer (Applied Biosystems).

Western blot
Total proteins were extracted by lysing cells in RIPA buffer containing 1 mM Phenylmethanesulfonyl fluoride (Solarbio Science & Technology, Beijing, China). According to manufacturer's instructions, cytoplasmic and nuclear proteins were extracted using the nuclear and cytoplasmic protein extraction kit (Beyotime, Shanghai, China). Western blot was performed as described [21,22]. We used Lamin B and Actin as control to quantify our nuclear and cytoplasmic fractions. Their optical densities were analyzed with Quantity One software.

Confocal microscopy
Cells were blocked and then stained with fluorochrome-labeled antibodies. For intracellular Ikaros detection, cells were first permeabilized in detergentbuffered paraformaldehyde and then stained as above using appropriate secondary antibodies. Stained cells were cytospun onto coated slides and mounted with Vectashield containing 4,6 diamidino-2-phenylindole (DAPI). Images were acquired with a FluoView confocal laser scanning microscope (Olympus) and processed with ImageJ.

Co-immunoprecipitation binding assay
For co-immunoprecipitation, SupB15 cells in 10 cm dishes were transiently transfected with different plasmids. It was performed as previously described [49,50]. For western blot analysis, anti-Flag and anti-Myc monoclonal antibody (Sigma) were used as primary antibody. To detect endogenous interaction of Ikaros and HO-1, anti-Ikaros and anti-HO-1 antibody were incubated with Dynbeads (Invitrogen), and similar procedures were performed.

Statistical analysis
All data were analyzed by SPSS ver.19.0 software package (SPSS, Chicago, USA). Statistical significance among groups was determined by analysis of variance (ANOVA) and Student's t-test. The Chi-square test was used to assess the association between IKZF1 status and clinical features. The correlation between the expression of HO-1 and Ikaros was built up by the application of Spearmant. Kaplan-Meier survival curves were plotted and analyzed with the log-rank test. P values less than 0.05 were considered statistically significant. Each experiment was performed in triplicate, data were presented as mean ± standard deviation (SD).