HAP1 loss confers l-asparaginase resistance in ALL by downregulating the calpain-1-Bid-caspase-3/12 pathway

Acute lymphoblastic leukemia (ALL) is a malignancy that mostly occurs in children, decreasing in frequency with age, before increasing again at ∼50 years of age.¹  High-intensity chemotherapy has improved the 5-year survival in pediatric (>2 years) ALL to >85%,²  and pediatric-based chemotherapy regimens have increased the survival of ALL patients up to 45 years of age.³  Infants (<12 months) and elderly ALL patients, however, have lower overall survival from factors such as mixed lineage leukemia/11q23 rearrangement, particularly in infants⁴ ; the presence of the Philadelphia chromosome and persistence of minimal residual disease, especially in adults⁵ ; and low tolerance for chemotherapy. As ALL treatment has intensified, the long-term survival particularly in children has remarkably improved but at the expense of increased toxicity that could lead to mortality. Even with more intensive therapy, a subset of patients still develops resistance, and refractory/relapsed children and adults have a 5-year survival estimate of ∼20%⁶  and <10%,⁷  respectively. To address these issues, interest has been directed at providing more selective and precise treatments. One approach is to stratify patients into risk groups based on clinical and laboratory profiles, but, thus far, these profiles have not been consistent in predicting the risk of relapse among groups.⁸  A promising approach is to identify molecular biomarkers of resistance to specific drugs, which would allow prediction of a patient’s response to these drugs. Thus, an appropriate ALL treatment regimen could be offered such that drugs that are expected to work in a subset of patients can be given at intensified levels but eliminated from the regimen of patients that are unlikely to respond to these drugs.

l-Asparaginase has been effectively used in ALL chemotherapy. Treatment regimens that include l-asparaginase result in higher remission induction in both pediatric and adult ALL compared with l-asparaginase–free regimens. l-Asparaginase works by catalyzing asparagine hydrolysis into aspartate and ammonia.⁹  Normally, asparagine is synthesized from other amino acids such as aspartic acid by asparagine synthetase.¹⁰  However, ALL cells have critically low levels of asparagine synthetase^11,12  and, therefore, are not able to synthesize de novo asparagine¹³  and rely heavily on plasma asparagine for protein synthesis and survival. l-Asparaginase treatment rapidly depletes plasma asparagine, causing asparagine deficit, which inhibits protein synthesis, leading to leukemic cell death.

Different l-asparaginase preparations have been tried in ALL patients,^14,15  including the Escherichia coli–derived native l-asparaginase (eg, Kidrolase) and pegylated (PEG) native l-asparaginase, and Erwinia chrysanthemi–derived l-asparaginase (Erwinase). The E colil-asparaginase and PEG-l-asparaginase are first-line treatments for ALL, but hypersensitivity and antibody-induced inactivation in up to 60% of cases have been reported.^16,17  In patients with E colil-asparaginase hypersensitivity, PEG-l-asparaginase or Erwinial-asparaginase is a prescribed second -line treatment, with Erwinial-asparaginase becoming a third-line drug when PEG-l-asparaginase is offered as a second line treatment.

Although the use of l-asparaginase clearly benefits a considerable subset of ALL patients, some relapse because of the development of resistance, leading to a refractory disease. Potential mechanisms of resistance to l-asparaginase have been suggested. As indicated, antibodies could inactivate the enzyme.¹⁷  Development of a mechanism to upregulate asparaginase synthetase activity has also been proposed because a population of resistant cells were found to have increased levels of asparagine synthetase.^18,19  Although this is consistent with the observed positive correlation between asparagine synthetase level and l-asparaginase resistance,²⁰  lack of correlation between these factors has also been described,^21,22  indicating the existence of other mechanisms that promote l-asparaginase resistance.²³  Indeed, a previous study has shown opioid receptor μ1-mediated mechanism for Erwinase resistance in ALL.²³  Here, we demonstrate a distinct mechanism whereby loss of huntingtin associated protein 1 (HAP1) confers l-asparaginase resistance in ALL cells. Our findings indicate that HAP1 loss prevents l-asparaginase–induced apoptosis by preventing the formation of the HAP1-huntingtin (Htt)-inositol 1,4,5-triphosphate receptor (InsP₃R) ternary complex, resulting in reduced [Ca²⁺]_i and downregulation of the Ca²⁺-mediated calpain-1-Bid-caspase-3/12 apoptotic pathway.

Peripheral blood samples were obtained from pediatric ALL patients at the Second Affiliated Hospital (Harbin, China), following the protocol approved by the Ethics Committee at Harbin Medical University (HMUIRB20150023). Patient peripheral blood samples diluted to double its original volume with phosphate-buffered saline (PBS) were layered on Ficoll-Paque (2:1 ratio) and then centrifuged at 1800 rpm (Beckman Allegra X-15R, SX4750) for 20 minutes at room temperature. Buffy coats were diluted with PBS, centrifuged at 1200 rpm (Beckman Allegra X-15R, SX4750), and white blood cell pellets were washed with PBS. Cells were resuspended in RPMI and viability was assessed by trypan blue exclusion assay.

ALL cells were cultured in RPMI supplemented with 10% fetal bovine serum (both Gibco, ON, Canada) and 100 units/mL each of penicillin and streptomycin. Kidrolase, a first-line l-asparaginase formulation, was from the Oncology Clinic at the Foothills Hospital in Calgary and Aru Narendran at the University of Calgary. Fluorescent dyes were from Thermo Fisher Scientific (Burlington, ON, Canada). Saponin and 2,5-di-tert-butylhydroquinone (TBHQ) were from Sigma (Oakville, ON, Canada), 1,2-Bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis(acetoxymethyl ester) (BAPTA-AM) was from Abcam (Toronto, ON, Canada), and InsP3 was from Millipore (Etobicoke, ON, Canada). Antibodies to HAP1 (A-11), InsP₃R (E-8), Htt (3E10), calnexin (AF18), calpain-1 (6C-12), Bid (B-3), caspase-3 (E-8), caspase-12 (1611), PARP1 (F-2), glyceraldehyde-3-phosphate dehydrogenase (GAPDH; 0411), and actin (I-19 or C-2) were from Santa Cruz Biotech (Dallas, TX). Antibody to cleaved caspase-3 (D1-75) was from Cell Signaling (Whitby, ON, Canada). Trypan blue was from Invitrogen (Life Technologies Co., ON, Canada).

The SEM leukemia cell line originating from the peripheral blood of a relapsed 5-year-old female ALL patient²⁴  are (1) pediatric B-cell ALL cells, (2) grows well in vitro, and (3) have high retrovirus infection efficiency; thus, we used these cells as a model system for our study. SEM cells were infected with retrovirus carrying pRS-short hairpin RNA (shRNA) library²³  or pRS vector alone. Infected cells were selected using puromycin (2 μg/mL) for 3 weeks (first screening); selected cells were cultured in the presence or absence of l-asparaginase (10 mIU/mL) for 2 weeks in the second round of screening. Single colonies were then isolated by colony formation assay in soft agar. Cells were mixed with 0.8% low melting point agarose in RPMI, containing 20% fetal bovine serum, 2 μg/mL puromycin, and 10 mIU/mL l-asparaginase and then plated on top of 0.8% normal melting point agarose for an additional week. Single colonies were isolated and amplified individually and as a pool.

Presence of shRNA inserts in cells infected with retrovirus carrying the shRNA library were determined by polymerase chain reaction (PCR) using genomic DNA (gDNA) as a template and pRS forward (1333-1355; CCCTTGAACCTCCTCGTTCGACC) and reverse (1953-1975; GAGACGTGCTACTTCCATTTGTC) primers. Reactions were carried out at 94°C for 5 minutes, 35 cycles of 94°C for 30 seconds, 60°C for 30 seconds, 72°C for 1 minute, and 72°C for 10 minutes. PCR products were resolved in 1% agarose gel and the 642-bp bands were cut, extracted, and sequenced using pRS-sequence primer (1697-1716; GCTGACGTCATCAACCCGCT) at the University of Calgary (AB, Canada) Core DNA services. Candidate genes were identified by comparing the barcodes with sequencing results. As proof of concept to validate our shRNA library screening results, we used 2 types of HAP1-depleted SEM cells for our mechanistic experiments: (1) cells infected with retrovirus carrying shHAP1 target sequence 1 (GAAGTATGTCCTCCAGCAA) and (2) a clone of cells (from the screening analysis) infected with retrovirus carrying shHAP1 target sequence 2 (GCTGATTTGCAGGAGGCA).

Cells were treated with different concentrations of l-asparaginase with or without BAPTA-AM (0.5 μM) for the indicated times. Surviving cell fractions were quantified using Alamar blue assay (Invitrogen). Fifty percent inhibitory (IC₅₀) values were calculated after plotting l-asparaginase dose-dependent survival of leukemic cells.

Immunoprecipitation (IP) of clarified cell lysates in RIPA buffer was performed using the indicated antibodies. IP samples or cell lysates were resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotted by also using the indicated antibodies. Western blot images were obtained using the ChemiDoc Imager (Bio-Rad) and using the optimal exposure setup. No enhancements were performed. Ratios of protein bands of interest vs actin were determined after densitometry using National Institutes of Health Image J 1.61.

See supplemental Materials and methods, available on the Blood Web site, for preparation of ER fractions, measurement of InsP₃-induced endoplasmic reticulum (ER) Ca²⁺ release, single-cell Ca²⁺ imaging, measurement of resting [Ca²⁺]_i, analyses for apoptosis and necrosis, and statistical analysis.

To identify resistance biomarkers for l-asparaginase, unbiased genome-wide RNAi screening²⁵  for 24 000 distinct shRNAs was performed on the SEM cell line. SEM cells infected with retrovirus carrying the pRS-shRNA library were initially treated with puromycin for 3 weeks to select for infected cells, and subsequently with l-asparaginase for 2 weeks. As shown in Figure1A, SEM parental cells (*) infected with retroviral vector alone (*+pRS) were sensitive to l-asparaginase (IC₅₀ = 177 ± 10.96 mIU/mL, n = 3) with only ∼20% of cells surviving after addition of 400 mIU/mL l-asparaginase. Conversely, cells infected with the shRNA library-containing vector (*+pRS-shRNA Lib) showed resistance to l-asparaginase (IC₅₀ = 372 ± 15.99 mIU/mL, n = 3) with ∼50% of cells surviving in 400 mIU/mL l-asparaginase, and an IC₅₀ that is significantly greater (twofold; P < .01) than that of control *+pRS cells. By PCR analysis of gDNA from pooled l-asparaginase–resistant cells (ie, infected with pRS-shRNA library), and using primers flanking the shRNA inserts, we detected a 642-bp PCR product following the first (puromycin) and second (l-asparaginase) screenings of these cells (Figure 1B). This observation indicates that the pRS-shRNA library-infected cells carry the retroviral shRNA insert, which accounts for cell survival after puromycin and l-asparaginase treatment. Two successive screening rounds and subsequent barcode sequencing of the shRNA insert PCR products from l-asparaginase–resistant clones led to the identification of HAP1²⁶  (Figure 1C) as a candidate resistance biomarker for l-asparaginase. As shown in Figure 1D, HAP1 is expressed in SEM cells (*) as well as in C1,²⁷  MOLT3,²⁸  TIB202,²⁹  and POETIC2²³  ALL cell model systems.

To determine whether HAP1 loss does indeed confer l-asparaginase resistance in SEM cells, these cells were infected with retrovirus carrying shHAP1 #1. As shown in Figure 2A, shHAP1 #1 effectively diminished the expression of HAP1 messenger RNA (mRNA; left) and protein (right), and HAP1-depleted cells exhibited resistance to l-asparaginase (Figure 2B). l-asparaginase IC₅₀ in HAP1-depleted cells (IC₅₀ = 611 ± 43.40 mIU/mL, n = 3) was significantly higher (P < .01) compared with that in control cells (IC₅₀ = 187 ± 11.80 mIU/mL, n = 3). More than 60% of HAP1-depleted cells survived the maximum l-asparaginase treatment (400 mIU/mL), whereas <20% of control cells survived this treatment. To support these findings, we also used 1 of the 3 l-asparaginase–resistant SEM clones that carry HAP1 shRNA #2 (target sequence: GCTGATTTGCAGGAGGCA; Figure 1C), which were isolated from our shRNA library screening. As shown in supplemental Figure 1, HAP1 depletion by infection with retrovirus carrying shHAP1 #2 also caused l-asparaginase resistance in SEM cells, further validating our shRNA library screening results. To examine the possibility that decreased cell death in l-asparaginase–treated HAP1-depleted cells could be due to a general reduction in apoptotic response of HAP1-deficient cells, ALL cells were treated were doxorubicin, another pro-apoptotic ALL drug. As shown in supplemental Figure 2, control and HAP1-depleted cells show similar sensitivity to doxorubicin, suggesting that HAP1 loss may specifically confer resistance to l-asparaginase.

HAP1 interacts with Htt, which together with InsP₃R, forms the HAP1-Htt-InsP₃R complex that was implicated in mediating ER Ca²⁺ release.³⁰  This suggests the importance of HAP1-Htt-InsP₃R complex in the maintenance of intracellular Ca²⁺ homeostasis, which is critical for cell survival and perturbation of which can trigger apoptotic cell death.³¹  Because HAP1 loss induces l-asparaginase resistance, we sought to examine whether such loss and expected deficiency in the HAP1-Htt-InsP₃R complex formation affect HAP1-Htt-InsP₃R-mediated Ca²⁺ release from the ER. First, we noted that ER fractions from SEM cells infected with retrovirus carrying shHAP1 were depleted of HAP1 (Figure 3A). Levels of InsP₃R and HTT were not much different in ER fractions from control and HAP1-depleted cells, and therefore, were unaffected by HAP1 depletion. Purity of ER fractions was determined by analysis for enrichment of the ER marker, calnexin, and lack of contamination by the cytoplasmic protein, GAPDH. The presence of Htt in both the S43 and ER fractions was expected because Htt is also a cytosolic protein.³²  IP of HAP1, Htt, or InsP₃R and subsequent immunoblotting showed the presence of all proteins in all immunoprecipitates from the ER fraction of control cells (Figure 3B, left), but markedly reduced amounts of HAP1 in immunoprecipitates from the ER of shHAP1-infected cells (Figure 3B, right), demonstrating that HAP1 loss leads to loss of HAP1/Htt/InsP₃R complex formation in the ER. Given that the complex was implicated to mediate ER Ca²⁺ release,³⁰  we examined whether HAP1 loss affects ER Ca²⁺ release. To do so, cells (5 × 10⁴) were loaded with an ER Ca²⁺ probe, Mag-Fluo-4-AM, and then treated with IP3 after permeabilization with saponin to facilitate IP3 access to IP3R. Ca²⁺ release was then measured every 2 seconds. We found that InsP₃-induced InsP₃R-mediated Ca²⁺ release from the ER was significantly (P < .05) reduced in HAP1-depleted cells compared with control cells, particularly at 100 nM InsP3 treatment (Figure 3C), indicating that HAP1 loss, leading to loss of HAP1/Htt/InsP₃R complex formation, inhibits InsP₃-activated InsP₃R-regulated ER Ca²⁺ release.

We then examined the effect of HAP1 loss on ER Ca²⁺ release upon l-asparaginase treatment by single-cell Ca²⁺ imaging of ratiometric Fura-2-loaded cells in Ca²⁺-free/EGTA-containing KRH buffer. We found that l-asparaginase treatment of control cells elicited a prompt increase in [Ca²⁺]_i that declined gradually over 20 minutes (Figure 4A). Conversely, no significant increase in [Ca²⁺]_i was observed following l-asparaginase treatment of HAP1-depleted cells, indicating that HAP1 loss inhibits l-asparaginase-induced ER Ca²⁺ release from the ER. Treatment with TBHQ, an ER Ca²⁺ pump inhibitor, caused a similar increase in [Ca²⁺]_i in control and HAP1-depleted cells, indicating that these cells were viable during analysis. We then explored the possibility that HAP1 loss may also affect external Ca²⁺ entry and Ca²⁺ extrusion upon treatment with l-asparaginase. To do so, internal Ca²⁺ stores in Fura-2-AM–loaded cells were initially emptied by treatment with TBHQ in Ca²⁺-free/EGTA-containing buffer. Cells were then treated with l-asparaginase in the presence of TBHQ and subsequently with 2 mM Ca²⁺, which was added to the external buffer, to initiate Ca²⁺ entry (Figure 4B). Ca²⁺ extrusion was also measured in the presence of TBHQ following a switch to Ca²⁺-free/EGTA-containing buffer with TBHQ. Initial treatment with TBHQ, which causes the release of internal Ca²⁺ stores, allowing the measurement of internal Ca²⁺ store capacity, indicated that such capacity was significantly reduced in HAP1-depleted cells compared with control cells (Figure 4B-C). In the presence of l-asparaginase, Ca²⁺ entry following external Ca²⁺ addition was also significantly lower in HAP1-depleted cells compared with control cells (Figure 4B,D). The rate of Ca²⁺ influx was slower in HAP1-depleted cells, as indicated by increased half-life of influx (Figure 4B,E). In addition, the integrated Ca²⁺ signals, which correspond to the calculated area under the curve (ie, from the beginning to the end [back to baseline] of Ca²⁺ signal) was reduced in HAP1-deficient cells (Figure 4B,F). However, the rate of Ca²⁺ extrusion upon switching to Ca²⁺-free/EGTA-containing buffer with TBHQ, was similar in control and HAP1-depleted cells as shown by their similar slope of [Ca²⁺]_i decline. Similar results were observed in the l-asparaginase–resistant SEM clone that carry HAP1 shRNA 2 (supplemental Figure 4). Resting [Ca²⁺]_i levels was significantly lower in HAP1-depleted cells compared with control cells (Figure 4G). Together, our findings suggest that loss of HAP1 inhibits both l-asparaginase–induced ER Ca²⁺ release and external Ca²⁺ influx, avoiding a lethal rise in [Ca²⁺]_I and causing ALL cell resistance to the drug.

We then examined the expression of some of the components of the Ca²⁺-mediated apoptotic pathway in HAP1-depleted cells. As shown in Figure 5A (left and right), levels of activated calpain-1, t-Bid, and caspase-3 and caspase-12 were considerably reduced in HAP1-depleted cells compared with control cells. The caspase-3–mediated PARP1 cleavage product was also reduced in HAP1-depleted cells.

To further examine the link between l-asparaginase–induced apoptosis and [Ca²⁺]_i increase, control cells were treated with l-asparaginase (10 mIU/mL) or the Ca²⁺ chelator BAPTA-AM, alone or both. l-asparaginase caused 18% apoptosis (Figure 5B; P < .05), which was inhibited by BAPTA-AM (P < .05) to a level equivalent to that of apoptosis induced by BAPTA-AM itself, indicating that l-asparaginase–induced apoptosis is primarily mediated by Ca²⁺. Although l-asparaginase also induced necrosis, this was not affected by BAPTA-AM treatment, suggesting that this is a Ca²⁺-independent process. HAP1-depleted cells showed a much reduced (P < .05) l-asparaginase–induced apoptosis (Figure 5C; 3%) compared with control cells (Figure 5B; 18%), and l-asparaginase–induced necrosis was also reduced (P < .05) in HAP1-depleted cells (2.5%) compared with control cells (6%), albeit the latter difference is relatively modest. Thus, HAP1 loss inhibits both l-asparaginase–induced apoptosis and necrosis.

To assess the clinical potential of HAP1 loss as a functional indicator of l-asparaginase resistance, we examined HAP1 levels and corresponding l-asparaginase IC₅₀ values in leukemic cells isolated from 6 ALL patients. As shown in Figure 6A, HAP1 level is inversely proportional to l-asparaginase IC₅₀: the lower the HAP1 level, the higher the l-asparaginase IC₅₀. We also noted that nonrelapsed patients have relatively higher HAP1 levels and lower IC₅₀ values compared with relapsed patients (Figure 6B). Together, these findings support a clinically functional link between loss of HAP1 and l-asparaginase resistance in ALL.

In this study, we demonstrate that loss of HAP1 results in the development of ALL cell resistance to l-asparaginase. Consistent with our RNAi screening, which identified HAP1 as a biomarker for l-asparaginase resistance in ALL, we determined that knocking down HAP1, which we found to be expressed in all ALL cells tested, causes l-asparaginase resistance in our ALL model system. This proof of concept approach validates our data from the RNAi screening and showed that following treatment with l-asparaginase, HAP1-deficient cells have significantly greater survival than control cells.

Asparaginase activates the G-protein coupled protease-activated receptor 2, resulting in activation of phospholipase C, which cleaves phosphatidylinositol bisphosphate into diacylglycerol and InsP₃.³³  InsP₃ binds to the InsP₃R ligand-gated Ca²⁺ channel on the ER surface; such binding triggers the opening of the Ca²⁺ channel, releasing ER Ca²⁺ into the cytoplasm.³⁴  Opening of the InsP₃R Ca²⁺ channel by InsP₃ requires HAP1-Htt interaction with the InsP₃R carboxy terminus, implicating the HAP1-Htt-InsP₃R complex as a functional mediator of InsP₃-induced Ca²⁺ release from the ER.³⁰ 

In ALL cells, l-asparaginase treatment resulted in a prompt increase in [Ca²⁺]_i, which is consistent with the increased [Ca²⁺]_i response of pancreatic acinar cells to asparaginase,³³  although the latter showed repetitive Ca²⁺ spikes in the absence of external Ca²⁺. The discrepancy in response may be due to differences in cellular contexts. Nonetheless, we found that HAP1 loss, which causes loss of HAP1-Htt-InsP₃R complex formation, inhibited not only the InsP₃-induced Ca²⁺ release from the ER, but also prevented the l-asparaginase–induced ER Ca²⁺ release. Although the extent of inhibition of ER Ca²⁺ release from HAP1 loss appears to be greater in l-asparaginase–induced than InsP₃-induced release, this may be due to the measurement of Ca²⁺ in the ER vs cytoplasm using 2 different Ca²⁺ probes, Mag-Fluo-4-AM (distribution coefficient = 22 μM) and Fura-2-AM (distribution coefficient = 224 nM), with different sensitivity or the measurement of Ca²⁺ using spectrofluorometer vs ratiometric single-cell microscopic imaging, or the involvement of additional factors in l-asparaginase–evoked InsP₃R sensitization. Nevertheless, our findings suggest that HAP1 loss is sufficient to prevent l-asparaginase–mediated Ca²⁺ release from internal stores, which could promote ALL cell apoptosis.

Interestingly, loss of HAP1 reduces l-asparaginase–induced extracellular Ca²⁺ influx into ALL cells by 18%, but has no effect on the rate of Ca²⁺ extrusion, suggesting that HAP1 loss affects Ca²⁺ influx through Ca²⁺ channels in the plasma membrane.³⁵  Thus, although l-asparaginase treatment in ALL cells causes both an immediate ER Ca²⁺ release and excessive extracellular Ca²⁺ influx, resulting in a toxic rise in [Ca²⁺]_i that triggers apoptotic cell death, loss of HAP1 inhibits both l-asparaginase–mediated effects, preventing lethal elevation in [Ca²⁺]_i, and causing ALL cell resistance to the drug.

Apoptotic cell death is mediated by a series of proteolytic events that involve cysteine proteases. Calpain-1 is activated by Ca²⁺, triggering activation of caspase-12.³⁶  The pro-apoptotic Bid and the apoptosis executioner, pro-caspase-3, are then targeted and cleaved, generating their activated forms, t-Bid and caspase-3,³⁷  respectively. PARP1 is a recognized caspase-3 substrate that is cleaved during apoptosis. Our studies showed that loss of HAP1 in ALL cells caused reduced levels of l-asparaginase–induced cleavage of calpain-1, Bid, procaspase-3 and procaspase-12, and PARP1, and significantly reduced l-asparaginase–induced apoptosis. This finding indicates that l-asparaginase treatment results in [Ca²⁺]_i-mediated apoptosis. The ability of BAPTA-AM to almost completely reverse this process indicates that l-asparaginase–induced apoptosis is essentially mediated by Ca²⁺. Although l-asparaginase also causes cell death in pancreatic acinar cells by promoting [Ca²⁺]_i increase,³³  we further demonstrate a link between Ca²⁺ and l-asparaginase–induced apoptosis through our finding in ALL cells that HAP1 loss prevents l-asparaginase–induced apoptosis by downregulating the Ca²⁺-mediated calpain-1-Bid-caspase-3/12 apoptotic pathway. The lack of BAPTA-AM effect on l-asparaginase–induced necrosis indicates that this process is independent of Ca²⁺. Nonetheless, reduced l-asparaginase–induced apoptosis and necrosis in HAP1-deficient cells indicates that HAP1 loss inhibits both the apoptotic and necrotic pathways induced by l-asparaginase treatment. It appears that HAP1 loss specifically confers ALL cell resistance to l-asparaginase because cells depleted of HAP1 showed no resistance to doxorubicin.

Our finding that ALL patient leukemic cells with lower HAP1 levels are more resistant to l-asparaginase treatment than those with higher HAP1 levels supports the clinical significance of HAP1 loss as a novel biomarker for l-asparaginase resistance. However, considering the different genetic backgrounds and presence of asparaginase antibodies, further analyses in ALL patients is necessary. Interestingly, analysis of publicly available Oncomine multiple microarray datasets³⁸  for HAP1 levels in ALL vs corresponding normal samples revealed that in 9 independent analyses, consisting of 1056 ALL patient samples, 49.9% (527/1056) had significantly (P < .05) downregulated levels of HAP1. However, there are no data on patient treatment history or genetic background and, therefore, what truly accounts for HAP1 downregulation is unknown. Nonetheless, the high percentage of ALL patients with reduced leukemic cell HAP1 levels may reflect the significance of HAP1 in ALL progression. Further analysis of a large-scale cancer genomics dataset in cBioPortal for Cancer Genomics³⁹  revealed the occurrence of HAP1 deletions or mutations in hematopoietic and other cancers: HAP1 is deleted in 0.5% of acute myeloid leukemia, and mutated in 3.4% of diffuse large B-cell lymphoma, 2.1% of lymphoid neoplasm diffuse B-cell lymphoma, and 1.0% of multiple myeloma. It is also deleted in 2.9% of metastatic breast cancer, 4.5% of adrenocortical carcinoma, and 2.1% of prostate adenocarcinoma, and mutated in 8.1% of melanoma treated with pembrolizumab and nivolumab. These findings may imply HAP1 involvement in the development of other types of cancer and potentially, in their development of drug resistance. Our further effort to analyze the link between HAP1 level (or deletion or mutation) and treatment outcome in clinical samples using published gene expression databases,^40-42  including the NCBI Gene Expression Omnibus database,⁴³  was unsuccessful; the microarray data showed that HAP1 expression level in samples was too low to be considered as reliable or having insufficient data.

In conclusion, our investigations identify HAP1 as a useful biomarker for l-asparaginase resistance in ALL and suggest that HAP1 can be exploited to define high-risk ALL patients. In addition to using HAP1 expression for stratifying patients to predict l-asparaginase treatment outcome, Hap1 may serve as a target for new therapies. It will be interesting to know whether HAP1 loss-mediated dysregulation of [Ca²⁺]_i is also used to develop resistance to other ALL drugs.

The authors thank the Oncology Clinic at the Foothills Hospital in Calgary and Aru Narendran for providing l-asparaginase. The authors also thank Wayne Chen for the helpful discussions on our Ca²⁺ studies and allowing us to use his Ca²⁺ imaging equipment and Deepa Bhojwani at the University of Southern California for assisting us in analyzing HAP1 levels, treatment response, and patient outcome using the NCBI Gene Expression Omnibus database.

This work was supported by a grant from the Canadian Institutes of Health Research (MOP-123400) (K.-Y.L.).

Contribution: S.K. and J.K.L. performed most of the experiments and drafted the manuscript; J.W. provided blood samples obtained from children diagnosed with leukemia; X.W. isolated leukemic cells from patient’s blood samples and performed the experiment described in Figure 6; X.G., Y.J., S.F., and H.-G.B. contributed significantly to the analysis and interpretation of data and/or provided constructive comments on experimental design and/or the preparation and writing of the manuscript; K.-Y.L. conceived the study; and J.L.R. and K.-Y.L. critically revised the manuscript for important intellectual content and wrote the final version of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Ki-Young Lee, Department of Cell Biology and Anatomy, University of Calgary, 3330 Hospital Dr NW, Room 359, HMRB, Calgary, AB T2N 4N1, Canada. e-mail: kylee@ucalgary.ca.