Parthenolide and DMAPT induce cell death in primitive CML cells through reactive oxygen species

Abstract Tyrosine kinase inhibitors (TKI) have become a first‐line treatment for chronic myeloid leuakemia (CML). TKIs efficiently target bulk CML cells; however, they are unable to eliminate the leukaemic stem cell (LSC) population that causes resistance and relapse in CML patients. In this study, we assessed the effects of parthenolide (PTL) and dimethyl amino parthenolide (DMAPT), two potent inhibitors of LSCs in acute myeloid leukaemia (AML), on CML bulk and CML primitive (CD34+lin−) cells. We found that both agents induced cell death in CML, while having little effect on the equivalent normal hematopoietic cells. PTL and DMAPT caused an increase in reactive oxygen species (ROS) levels and inhibited NF‐κB activation. PTL and DMAPT inhibited cell proliferation and induced cell cycle arrest in G0 and G2 phases. Furthermore, we found cell cycle inhibition to correlate with down‐regulation of cyclin D1 and cyclin A. In summary, our study shows that PTL and DMAPT have a strong inhibitory effect on CML cells. Given that cell cycle arrest was not dependent on ROS induction, we speculate that this effect could be a direct consequence of NF‐κB inhibition and if this mechanism was to be evaded, PTL and DMAPT induced cell death would be potentiated.

proliferation, differentiation and alter cell adhesion. Together, these features characterize CML cells and drive the progression of CML. 5 Diagnosis in 90% of cases occurs at chronic phase (CP), where patients have abnormal cell counts but myeloid cells retain differentiation ability. 6 If not treated, patients progress to an accelerated phase and finally to blast crisis (BC), 6 which is invariably fatal.
Knowledge of the molecular biology of CML has provided the development of tyrosine kinase inhibitors (TKIs), able to directly target Bcr-Abl, and inhibit its downstream activity, offering patients in CP the possibility of achieving remission and prolong survival. TKI's, however, are not able to target LSCs, a rare, quiescent, self-renewing cell population that initiates and sustains CML. 6 Recently, Nieborowska-Skorska and colleagues have reported that LSC from CML patients present high levels of ROS compared to normal primitive cells. 7 This has been related to defects in mitochondria of CML cells, which cause the accumulation of ROS, driving genomic instability and acquired resistance to TKIs. 7 This raises the interest in seeking targets other than Bcr-Abl that may be able to eliminate LSC, which are not eliminated with current TKI therapies.
Medicinal plants with anti-inflammatory properties are rich reservoirs of bioactive compounds. 8 Parthenolide (PTL) is a sesquiterpene lactone extracted from the plant Tanacetum parthenium, used by folk medicine for anti-inflammatory purposes, arthritis, migraine, as an anticoagulant, and a digestive aid, for centuries. 9 Guzman and collaborators have reported that PTL is able to target activity against LSC in AML. 10 To improve the pharmacological properties of PTL, an orally bioavailable analogue was generated, dimethyl amino parthenolide (DMAPT) that showed similar activity to PTL in AML LSC 11 ; nevertheless, the effect on CML cells, including both primitive primary cells, as well as CML cell lines, is unknown. The mechanism of action of PTL has been reported to rely on a number of pathways, two of which are the inhibition of NF-κB via the alkylation of critical cysteine residues 12 and the induction of ROS by affecting the glutathione system. 13 We present evidence that PTL and DMAPT are able to abrogate bulk CML cells and a primitive CML population (CD34 + lin − ) by a mechanism dependent on ROS induction and NF-κB pathway inhibition. The effect on ROS levels is transitory, reaching a peak at 1 hour of treatment.
In addition, we observed the up-regulation of HMOX-1 which is activated by Nrf2 transcription factor, described as an integrator of cellular stress signals 14 and promotes survival of a fraction of CML cells. These cells demonstrate a cell cycle arrest in G2/M and G0 cell cycle phase, which was accompanied by changes in Cyclin D and A, and down-regulation of CDK2.

| Chemicals and reagents
PTL and DMAPT were kindly provided by Drs. Cesar Compadre and Peter Crooks, respectively (University of Arkansas for Medical Sciences). Both agents were kept at -20 C at 40 mmol/L in DMSO. CML

| Stem and progenitor cell enrichment
Mononuclear cells (MNC) from primary samples were isolated using Ficoll-Paque Plus (Pharmacia Biotech, Uppsala, Sweden). Primary CD34 + lin − were enriched and cultured from MNC after negative selection as previously described. 15

| Cell cultures for cell viability and colony forming cells (CFC) assays
Cell viability was assessed by trypan blue exclusion. Cells were cultured in the presence or absence of PTL or DMAPT, after 24 or 48 hours, cells were harvested and counted with trypan blue at 0.4% in a Neubauer chamber. For CFC assays, 5000 viable CD34 + lin − cells were cultured for 14 days in methylcellulose-based semisolid medium (Methocult Classic with cytokine, STI) after 24 hours of PTL or DMAPT treatment. Colonies were evaluated using an inverted microscope.

| Gene expression assay
Cells from cell lines or CD34 + lin − from primary samples were cultured for 6 hours with or without PTL or DMAPT. Cells were harvested and RNA was extracted using RNA-easy Kit (Qiagen, Hilden, Germany), according to manufacturer's instructions. Quantitative RT-PCR was performed using Taqman RNA to Ct 1 step assay (Applied Biosystems, Foster City, CA), and Taqman assays to evaluate the expression of the following genes: NFKB1 (Hs00765730_m1), HMOX-1 (Hs01110250_m1), PLAU (Hs01547054_m1) and GAPDH (Hs02758991_g1).

| Cytometry Analysis and Statistical Analiysis
All Flow cytometry assays were done in a BD FACS Verse, BD FACS CANTO or BD FACS LSR II. Flow cytometer data was analysed with Flow-jo Software.
All data is expressed as Mean ± SEM.
Statistical analysis for each experiment is specified in each figure.
Graph and statistical analysis was done using Graphpad Software.

| PTL and DMAPT decrease viability of CML bulk and progenitor cells
To assess the effect of PTL and DMAPT on viability, three CML cell lines (K562, Meg-01, and KCL-22) were exposed to increasing concentrations of the compounds for 24 and 48 hours. As a normal control, MNC from NBM were used, as well as AML cell line HL60, previously reported as PTL sensitive. 16 We found a significant decrease in CML cell line viability by trypan blue (Figure 1A), the LC50 we observed (7.5 μmol/L at 24 hours, and 5 μmol/L at 48 hours) were similar to the LC50 reported for other leukaemias 10 ( Figure 1A), importantly this was not observed in MNC from NBM cells, we did however observe a significant reduction in NBM cell viability at the highest concentration of DMAPT (20 μmol/L). Interestingly, an increase in the number of viable cells was observed in MNC from NBM exposed to low concentrations of PTL (2 μmol/L). This is in concordance with a previous report by Li-Weber and colleagues 17 in which PTL at low concentrations (2 μmol/L) induced apoptotic inhibition on T cells. This may also be because of an increase in ROS levels, as an increase in proliferation has been reported by Day and Susuki for fibroblasts exposed to low levels of ROS. 18 After determining the LC50 for CML cell lines, we chose to treat primary MNC from CML and NBM with 5 and 7.5 μmol/L PTL and DMAPT, for 24 and 48 hours. The TKI imatinib (IMT, at 2.5 μmol/L) was used to compare effect on primitive and bulk CML cells to parthenolide and DMAPT treatment. As shown in Figure   . The IC50 is shown in the right side of the graph. B, represents the viability of MNC from normal (NBM) and CML primary samples after at 5 and 7.5 μmol/L of PTL and DMAPT for 24 and 48 h of culture, symbol represents significance between NBM and CML groups, determined by Student's t-test. C, indicates the per cent of CFC from normal (NBM) and CML CD34 + lin − cells exposed to PTL and DMAPT in liquid culture by 24 h and then sub-cultured in methylcellulose for 14 days, stars represent significance with UT group, determined using Dunnet's multiple comparison test. *P < .05 ** P < .01, *** P < .001

| PTL and DMAPT induce cell death in CML cells
To determine if viability loss was related to apoptosis, CML cell lines K562, KCL-22 and Meg-01 were treated with 7.5 and 10 μmol/L PTL or DMAPT, and showed an increase in cell death, compared with untreated ( Figure 2A,B). The AML cell line HL60 was used as control and showed the highest death level.
To observe this effect in primary samples, MNC from NBM and CML patients (CP and BC), were exposed to PTL and DMAPT.  Figure 2E), suggesting that cell death associated with PTL and DMAPT treatment is mostly caspase independent, as has been reported for other anticancer agents in leukaemia cells, 19 as well as has been reported for osteosarcoma cells where PTL induces autophagy and mitophagy associated with the increase in reactive oxygen species. 20 Interestingly, in the case of HL60, AML cell line the cell death index appear to be associated with caspase 3 activation ( Figure 2D).

Species (ROS) in CML Cells
Previous reports demonstrated that ROS levels are increased in CML primitive cells. 7 Considering that PTL and DMAPT have been reported as ROS inducers in different cancer models, 21  to both agents ( Figure 3D). This effect could be attributed to the lack of ROS induction, which would not activate the Nrf2 pathway responsible for HMOX-1 transcription.

| PTL and DMAPT induced cell death is ROS dependent
We next analysed whether cell death in CML cells is ROS dependent. K562 and KCL-22 cell lines were pre-treated with the ROS scavenger NAC at 15 mmol/L, for 1 hour; the cells were then cultured in fresh media with or without PTL and DMAPT at 7.5 or suggest that the major effect on ROS is mediated by superoxide anion ( Figure 4A). At 24 hours of treatment, cell death was determined. NAC pre-treatment was able to rescue cells from death induced by PTL or DMAPT ( Figure 4C), indicating that ROS induction is a key factor in cell death induced by these agents ( Figure 4B).

| PTL and DMAPT inhibit NF-κB pathway in CML cells
PTL and DMAPT have been reported as NF-κB pathway inhibitors, although the molecular mechanisms have not been completely elucidated, a report by García Piñeres and colleagues suggest that PTL blocks NF-κB pathway by alkylation of cysteine residues in p65. 12 Other authors have reported that the mechanism relies on an upstream inhibition of the NF-κB pathway, by inhibiting the activity of IKK. 10 Furthermore, PTL activity has been reported to rely on the alkylation capacity of critical serine residues. 12 To verify if PTL and DMAPT were able to inhibit NF-κB, K562 cells were treated for 6 hours with PTL or DMAPT at 7.5 and 10 μmol/L and p65 total F I G U R E 3 PTL and DMAPT induce Reactive Oxygen Species and activate Haemoxigenase-1 transcription. Leukaemic cell lines and progenitor cells from CML patients and NBM were cultured in the presence or absence of PTL and DMAPT. ROS levels were quantified with DCFDA stain by flow cytometry. A representative histogram from normal and CML CD34 + lin − cells treated for 1 h is shown in A. B, represents the ROS levels in CML and NBM cells, DCFDA MFI levels were normalized to untreated cells to evaluate ROS level changes with treatment. C, represents changes in ROS levels in CML cell lines at 1, 3 and 6 h of exposure to PTL. Haemoxigenase-1 transcription was determined by qPCR in primary CML and normal bone marrow CD34 + lin − cells, as well as CML (K562, Meg-01, KCL-22) and AML (HL60) cell lines, exposed for 6 h to PTL or DMAPT. Results represent the fold change relative to basal transcription without treatment, D. Significance between DCFDA fold change in NBM CD34+ and leukaemia cell lines was determined using an Unpaired Student's T test. **P < .01 and ****P < .0001 F I G U R E 2 PTL and DMAPT induce cell death in CML cells. Leukaemic cell lines and primary samples from CML patients and normal bone marrow were cultured for 24 h at different concentrations of PTL and DMAPT and the cell death index was analysed. A, Shows flow cytometry plots from K562 and Meg-01 CML and HL60 AML cell line (used as a positive control). B, Shows representative plots from 3 primary samples (Normal Bone Marrow and CML from a chronic phase and a blast crisis patient). ex in Untreated conditions or exposed to 7.5 or 10 μmol/L of DMAPT. Lower and left arrows indicate fluorochrome stain, Upper arrow indicates DMAPT concentration. C, Represents the cell death index detected in K562, Meg-01, KCL-22 and HL60 cell lines and D Indicates the average of 3 replicates with primary samples from normal bone marrow (NBM), CML Chronic Phase (CML-CP) and CML Blast Crisis (CML-BC) mononuclear cells. E, Represents the cell death index increase (YoPro-1 and Propidium Iodide) in cells from cell lines and a CD34+ enriched CML-CP primary sample, pre-treated with the pan-caspase inhibitor Z-vad and PTL or DMAPT. F, Shows the activation of Caspase-3 after pre-treatment with Z-vad and treatment with PTL or DMAPT for 24 h. Cell death index was determined by dividing the frequency of dead cells in PTL or DMAPT treated cells by the frequency of dead cells detected in untreated group, (considered as 1). Significance between cell death index in UT and treated in cell line experiments was determined by Dunnet's multiple comparison test and between NBM treated and CML treated cells was determined by Tukey's multiple comparison test. *P < .05 **P < .01, ***P < .001, ****P < .0001 FLORES-LOPEZ ET AL.

| PTL and DMAPT inhibit proliferation and arrest Cell Cycle of CML cells
Current objectives in CML therapy include the eradication of LSC remaining after TKI therapy and that cause relapse in patients. 26 Although TKI treatments are very effective targeting Bcr-Abl Significance between treated cells and control in MFI of ROS levels and cell death index was determined using an unpaired Student t-test. *P < .05 **P < .01, ***P < .001, ****P < .0001 activation, the quiescent leukaemic pool has been reported as not sensitive to this inhibition. 27 Recently, our group reported that the G0/G1 phase fraction increases in CML cells after being treated with TKIs, and that this is dependent on nuclear localization of CDK inhibitors p21 and p18. 15

| Cell cycle arrest by PTL and DMAPT is related to cyclin down-regulation
To further evaluate the mechanism behind the cell cycle arrest in Recently, an interest has grown on the ability of parthenolide and its analogue DMAPT, to target malignant populations. Several reports have described effect on neoplastic cells from colorectal, 30 lung and bladder cancer. 31 Of particular importance is their reported capacity to target cancer stem cells, such as acute leukaemia 10 and breast cancer stem cells. 32 The knowledge that different cancers depend on a stem cell-like population to maintain the disease which are resistant to therapy, thus causing relapse and or metastasis, has made it a priority in cancer research to find methodologies to target this population to find better treatments.
The overall data from this work indicate a robust effect of PTL and DMAPT on CML cells. CML cells exposed to PTL and DMAPT loss 50% viability by 24 hours, at a concentration around 7 μmol/L. We also found that the induction of cell death depended on ROS increase. Indeed, the ROS scavenger NAC was able to rescue cells from death and this correlated with absence of ROS induction.
Interestingly, this effect was observed in CML CD34 + lin − cells, but not on their normal counterpart. NBM primitive cells showed lower basal levels of ROS than their CML counterpart, this could indicate higher levels of reduced glutathione in normal populations, which would make normal primitive cells resilient to PTL and DMAPT effect as has been observed in CD34+ cells from AML samples, 13 and proves a key mechanism to be exploited and further studied. Significance between UT and treated cells in proliferation assay was determined using an unpaired Student t-test. Significance between Cell cycle phase was determined using Tukey's multiple comparison test. *P < .05 **P < .01, ***P < .001, ****P < .0001 DMAPT effect on CML cells (without affecting normal cells in a significant manner) is induction of ROS.
One of the hallmarks of CML, is the increase in cell proliferation. 36 TKIs, which directly target the activation of Bcr-Abl, inhibit proliferation of bulk CML cells; however, a fraction of LSC has been shown to enter quiescence and avoid apoptosis. 15 Here, we found that a fraction of CML cells is able to survive PTL and DMAPT  1). B, shows the effect of PTL and DMAPT in CDK2 and P21 expression in cells CD34 + lin − from a chronic phase CML primary sample. Significance between UT and PTL or DMAPT was determines using an unpaired Student t-test, *P < .05 **P < .01 the arrest on cell cycle to be reversed by NAC pre-treatment; however, to our surprise we found that while cells were consistently rescued from cell death with NAC pre-treatment, they readily suffered a cell cycle arrest that could not be explained by ROS induction.
NF-κB activity in CML has been previously identified as a critical pathway, particularly in imatinib resistant cell lines, 37 and suggested as a therapeutic target. Since we found cell cycle changes to be independent of ROS induction, we sought to further address the mechanism behind cell cycle arrest after sesquiterpene lactone treatment. Thus, we looked into protein levels of cell cycle regulators.
Surviving cells from K562 cell line and CML CD34 + lin − cells after 48 hours treatment showed changes in cyclin D1 and A, p21, and CDK2. K562 cells showed down-regulation of Cyclin D, which is critical in the G1 to S transition. This effect would be consistent with the inhibition of NF-κB as has been reported by Hinz and colleagues, 38 also it is the strongest known link between NF-κB pathway and cell cycle, as mentioned by Joyce and collaborators. 39 We also found a decrease tendency in CDK2 and Cyclin A, which is mostly linked to G2/S cell cycle progression, which would suggest an involvement with the K562 arrest observed in the G2 phase. On the other hand, CD34 + lin − cells from a CML patient showed a marked induction of p21 inhibitor and an increase in CDK2. Interestingly, we did not observe a G2 cell cycle arrest or an S phase burden in this cell population; instead, a cell cycle arrest was strongly observed in the G0/G1 compartment. The absence of a G2 arrest could be explained by an up-regulation of CDK2, however further assessment would be necessary to understand the mechanism behind the cell cycle arrest induced by sesquiterpene lactone.
Taken together, our data show a strong inhibitory effect of PTL and DMAPT on bulk CML cells and on a more primitive CD34 + lin − cell population. We also observed that PTL-induced cell death is accompanied by cell cycle arrest in surviving cells. Our findings indicate that cell cycle arrest we observed after PTL and DMAPT treatment is ROS independent, and changes in cell cycle regulators would be consistent with NF-κB inhibition. We speculate that evading cell cycle arrest could increase and potentiate PTL and DMAPT effect on the CML model. Finally, it is important to comment that it is necessary to consider an in vivo approach to consider PTL and DMAPT as selective molecules to eliminate CML primitive cells.