2.1 Significantly different PARP1 expression levels are detected among lymphoblastic cell lines
Comparative expression of PARP1 in leukemic cell lines relative to the non-neoplastic cell model reveals significant overexpression in both SUP-B15 (p < 0.0001) and Raji (p < 0.05), with incremental 15- and 7-fold change in expression, respectively, while PARP1 expression level in the cell line Namalwa was very similar and not statistically different from the control (Fig. 1). SUP-B15 significantly overexpress PARP1 when compared to either of the neoplastic cell lines (p < 0.01), with Raji also presenting significantly higher expression when compared to Namalwa (p < 0.05).
PARP1 expression was normalized through the endogenous control ACTB and the non-neoplastic cell line MRC5 was used as the calibrator. Statistical differences were analyzed through ANOVA followed by Bonferroni’s multiple comparisons. NS: Not significant; *p < 0.05; **p < 0.01, ***p < 0.0001.
2.2 PARP inhibitor is highly cytotoxic to BCR-ABL p190 + cell line
Table 1 describes the IC50 of AZD2461, imatinib and doxorubicin after 72 hours of incubation with the cell lines SUP-B15, Raji and Namalwa. AZD2461 is the structural analogue of Olaparib with pan-PARP inhibitory activity and with modifications in functional groups that lessen its affinity of interaction with transmembrane efflux bombs of the ATP binding cassette subfamily B (ABCB) family12. Imatinib is the TKI with inhibitory activity over BCR-ABL1 chimeric protein and considered the gold-standard for the treatment of Ph + tumors 15. Finally, doxorubicin is a well-established and highly cytotoxic chemotherapeutic drug with action over rapidly dividing cells due to the ability to inhibit topoisomerase activity16, and was used as a comparative control to the inhibition values of the targeted therapies.
Table 1
Minimum inhibitory concentrations for 50% of the cells (IC50) in 72 hours incubation periods. Values were measured in nanomolar (nM) units.
Drug\Cell Line | SUP-B15 | Raji | Namalwa |
AZD2461 | 344.3 nM (240.1–493.7) (R2: 0.9358) | NR | 20938 nM (16152–27143) (R2: 0.9806) |
Imatinib | 329.2 nM (215.7–502.2) (R2: 0.9385) | 2509 nM (1622–3882) (R2: 0.9431) | 1296 nM (803.8–2089) (R2: 0.9284) |
Doxorubicin | 20.32 nM (11.41–36.19) (R2: 0.9233) | 85 nM (65.13–110.9) (R2: 0.9694) | 75.93 nM (49.63–116.2) (R2: 0.9622) |
NR: Not Reached; R2: Coefficient of determination.
The use of doxorubicin, as expected, was met with significant lower values of IC50 than the molecular targeted therapies due to the non-specificity of its action over the viability of rapidly dividing cells17, presenting cytotoxic activity in the order of 101 nanomolar (nM).
In the SUP-B15 cell line, the IC50 values of AZD2461 and imatinib were low and extremely similar to one another, demonstrating almost equivalent inhibition potential. Due to harboring BCR-ABL translocation, the inhibitory activity of imatinib over SUP-B15 is not a surprise and has already been described in the literature18, however, the inhibitory concentration achieved with AZD2461 of 344.3 nM in 72 hours, close to the 329.2 nM of inhibition of imatinib, suggests that the use of PARPis may be as effective as the treatment with TKI in models of ALL p190+.
In the cell lines Raji and Namalwa, used as comparative controls of other B-cell neoplasms with different PARP1 expression levels, the 72 hour IC50 of AZD2461 and imatinib showed great disparity. In both cell lines, imatinib presented some cytotoxic activity, probably due to its non-specific activity over other tyrosine kinase pathways that are essential in the maintenance of proliferative capabilities and in the survival of neoplastic clones19. However, the use of AZD2461 did not present relevant inhibitory capabilities, with an IC50 of over 20 µM for Namalwa while not even been able to reach an IC50 for Raji in the concentrations used in this assay.
Due to the non-responsiveness of Raji and Namalwa to the proposed treatment, cytotoxic assays of 48 and 24 hours of treatment were only carried out with the SUP-B15 cell line, as well as all the subsequent phenotypic and molecular characterization assays. The obtained results are described in Table 2 and the inhibition patterns mimic those observed in the 72-hour treatments, with doxorubicin presenting considerably higher cytotoxic activity in comparison to the targeted therapies and an equivalence in the inhibitory potential of AZD2461 and imatinib.
Table 2
Minimum inhibitory concentrations for 50% of the cells (IC50) of SUP-B15 cell line in 24 and 48 hours of incubation. Values were measured in nanomolar (nM) units.
Drug\Period | 24 hours | 48 hours |
AZD2461 | 3925 nM (2502–6157) (R2: 0.8763) | 1421 nM (766.9–2632) (R2: 0.8087) |
Imatinib | 3966 nM (2398–6558) (R2: 0.8612) | 1230 nM (747.5–2024) (R2: 0.8905) |
Doxorubicin | 141.6 nM (93.92–213.6) (R2: 0.9305) | 57.33 nM (36.53–89.97) (R2: 0.9342) |
R2: Coefficient of determination.
Taking into consideration the three treatment intervals, the inhibitory values of AZD2461 and imatinib in incubation with SUP-B15 were not significantly different when analyzed through Student’s t-test, demonstrating the equivalence of both drugs in treating this ALL p190 + model in vitro.
2.3 AZD2461 promotes similar cell cycle arrest profile to that of imatinib in SUP-B15
Analysis of cell cycle arrest induced by treatment strategies allows for the identification of a drug’s potential cytostatic properties, inhibiting neoplastic clones proliferation20–22. Figure 2 describes cell cycle arrest profiles of either AZD2461 or imatinib in 24-hour sub-inhibitory treatment of SUP-B15 cell line in comparison to the non-treated control, and very similar patterns of inhibition may be seen, with considerable arrest at G0/G1 phase (p < 0.0001).
Cells were treated with sub-inhibitory concentrations of 1,5 µM of either AZD2461 or imatinib in a 24-hour incubation period and the graphs represent the means from three distinct experiments. Cell cycle arrest profile was compared among the treated experiments and the non-treated control through ANOVA followed by Bonferroni’s multiple comparisons. Fluorescence emission was detected after sample processing with DNA intercalating agent, propidium iodide (PI). A) Control experiment treated with equivalent volume of DMSO. B) Imatinib treated experiment. C) AZD2461 treated experiment. D) Percentage of cells in each phase of the cell cycle after the treatment and statistical comparison with the non-treated control. Dip: Pseudodiploid population; An1: Hypodiploid population; NS: Not significant; ***p < 0.0001.
In our non-treated control experiment, most events of SUP-B15 cell line were detected on G0/G1 phase – 47,37% –, with a great amount also detected on S phase – 43.98% – and only a minority of cells undergoing mitosis on phase G2/M – 8,66%. This low presence of SUP-B15 cells in mitotic phase goes along with the already described on the literature and is representative of the extensive duplication time of SUP-B15, of approximately 46 hours23. Furthermore, a tumor sub-population of apparently hypodiploid cells, represented by the yellow peaks in Fig. 2, was detected in the analysis of the control cell cycle, constituting 4.21% of events, although the percentage of this population in each phase of the cell cycle was not able to be determined due to fluorescence overlap with the pseudodiploid population of greater frequency.
In the experiments of AZD2461 or imatinib treatment, a significant arrest in phase G0/G1 was observed in comparison to the non-treated control (p < 0.0001), with 72,6% and 75,12% of events in G0/G1, respectively to the treatments, and no statistically significant difference being detected in the arrest induced between both treatments. In both treatments, the arrest was induced with detriment of S phase cells, being significantly decreased in comparison to the control (p < 0.0001), with 19.3% of events in S phase for the AZD2461 treated experiment and 14,9% of events in S phase for the imatinib treated experiment, once again not being detected statistically significant difference between S phase events of both treatments. Finally, in the analysis of G2/M phase, no difference was detected comparing either treatment with the control, with 8,36% and 9,98% of cells in G2/M phase for AZD2461 and imatinib, respectively.
Moreover, the same tumor sub-population of hypodiploid clones detected in the control experiment was also observed with greatly defined fluorescence peak in the imatinib treated experiment, representing 8,92% of total events, while almost not detected after AZD2461 treatment, representing only 1,7% of events. When submitted to analysis of variance, however, the differences in sub-population frequency did not show statistical difference.
The presented data indicates that PARP1 inhibition through the use of AZD2461 has cytostatic capabilities comparable to those of imatinib treatment in ALL p190 + cell models, inhibiting neoplastic clone proliferation due to G0/G1 phase arrest. A tendency for AZD2461 to be cytotoxic to yet-to-be characterized tumor sub-populations, even in sub-inhibitory concentrations, was also observed, although further analyzes are needed to elaborate on this hypothesis.
2.4 AZD2461 induces early apoptotic markings in SUP-B15 cell line after sub-inhibitory treatment
Fluorescent marking with annexin-V reagent conjugated with fluorescein isothiocyanate (FITC) fluorochrome was used to identify exposed phosphatidylserine, a marker of early apoptosis, on the outer membrane of cells after treatment incubation period24,25. 7-AAD, a DNA intercalating agent that is unable to permeate the cytoplasm of cells with intact cytoplasmic membrane, was used as a counterstain, and events marked with both fluorochromes represent late apoptotic/necrotic cells26,27. Events not marked with any fluorochromes represent viable non-apoptotic cells (Fig. 3).
Cells were treated with sub-inhibitory concentrations of 1,5 µM of either AZD2461 or imatinib in a 24-hour incubation period and the graphs represent the means from three distinct experiments. Cell cycle arrest profile was compared among the treated experiments and the non-treated control through ANOVA followed by Bonferroni’s multiple comparisons. A) Control experiment treated with equivalent volume of DMSO. B) Imatinib treated experiment. C) AZD2461 treated experiment. D) Population frequencies and significance of variations. NS: Not significant; ***p < 0.0001.
The use of imatinib in sub-inhibitory concentrations of 1,5 µM did not have any effect of apoptosis induction after 24 hours of incubation, not being statistically different from the non-treated control experiment. Meanwhile, the use of AZD2461 in the same concentration was able to induce considerable increase in the frequency of both early apoptosis and late apoptosis/necrosis populations in comparison to control and to imatinib treated experiments (p < 0.0001).
The results indicate that AZD2461 acts over apoptosis induction much more prematurely than imatinib in this model of ALL p190+, however, data interpretation must take into account the aforementioned presence of a tumor hypodiploid sub-population and the potential cytotoxic effect of AZD2461 over this population, raising the hypothesis if AZD2461 actually induces early apoptosis over the cell line as a whole or over specific and differentiated neoplastic clones.
Nevertheless, AZD2461 potential to induce higher rates of apoptosis than the treatment nowadays considered to be the gold-standard, imatinib, represents an advance into alternative and complementary therapies for a subset of hematological malignancies which struggle with recurrent cases of tumor resistance and refractoriness.
2.5 BCR-ABL p190 expression is modulated by treatment with AZD2461 or imatinib
Aiming to determine if the proposed treatments influence biomarker expression levels, a qPCR assay of the treated samples quantifying mRNA for BCR-ABL p190+, a hallmark of SUP-B15 cell line, and PARP1 was performed, concomitant with PARP1 protein fluorescent marking through flow cytometry.
qPCR analysis reveals both treatments with AZD2461 (p < 0.01) or imatinib (p < 0.0001) to significantly increase BCR-ABL p190 expression levels in comparison to the non-treated SUP-B15 control group, 3- and 4.5-fold, respectively (Fig. 4). Expression upregulation between both treatments was also significantly different (p < 0.05), although the same tendency for BCR-ABL p190 mRNA increase was detected, revealing similar patterns of neoplastic attempts to evade cytotoxic cell death. Over PARP1 expression, however, neither treatment had any statistically significant modulatory effects, remaining close to control mRNA transcript levels.
Cells were treated with sub-inhibitory concentrations of 1,5 µM of either AZD2461 or imatinib in a 24-hour incubation period and the graphs represent the means from three distinct experiments. BCR-ABL p190 and PARP1 expression was normalized through the endogenous control ACTB. Expression levels were measured in SUP-B15 cell line comparing the non-treated control experiment and the experiments after the proposed treatments. Statistical differences were analyzed through ANOVA followed by Bonferroni’s multiple comparisons. NS: Not significant; **p < 0.01, ***p < 0.0001.
To prove if AZD2461 or imatinib treatments induced post-transcriptional modifications on PARP1 levels, cells were stained with anti-PARP1 antibodies and analyzed by flow cytometry for mean fluorescence intensity (MFI) quantification (Fig. 5). Comparison of relative MFIs reveals increased PARP1 in cell populations treated with AZD2461 (p < 0.05), while imatinib-treated populations had no statistical difference to the control group. A tendency for lower levels of PARP1 may be seen in imatinib-treated group, however, as evidenced by an increase in statistical significance when comparing both treated groups to each other (p < 0.01).
Cells were treated with sub-inhibitory concentrations of 1,5 µM of either AZD2461 or imatinib in a 24-hour incubation period and the graphs represent the means from three distinct experiments. PARP1 levels were measured through the mean fluorescence intensity (MFI) of the fluorochrome Alexa Fluor® 647 and compared between the treated experiments and the non-treated control through ANOVA followed by Bonferroni’s multiple comparisons. MFIs ratios are represented as percentages relative to 100% of fluorescence of the non-treated control. NS: Not significant; *p < 0.05; **p < 0.01.
A clear trail may be seen in the AZD2461-treated graph, revealing a small percentage of events with depletion of PARP1 levels and, taken together, results from the expression and PARP1 intracellular levels seem contrasting. However, our flow cytometry analysis indicates that AZD2461 acts directly over PARP1 protein, inducing degradation and consequent loss of catalytic function, much sooner than it may act over gene expression.
2.6 Array comparative genomic hybridization (aCGH) revealed that PARP1 is amplified in the SUP-B15 cell line
In order to further characterize the cell model prioritized in this study, we conducted a comprehensive genome-wide analysis using aCGH to screen copy number alterations (CNAs) in the SUP-B15 cell line. Interestingly, this cell line harbored 28 CNAs (4 gains and 24 losses) with sizes ranging from 0.11 Mb to 86 Mb (Fig. 6; Supplementary Table S1). Large CNAs (> 20 kb) were preferentially observed on chromosomes 1, 4, 8, and 14. In addition to these broad CNAs, we also identified numerous localized gains and losses. Some of these affected chromosomal regions where genes were previously shown to be altered in B-ALL, such as PAX5 at 9p13.2 (loss)28–30, CDKN2A/B at 9p21.3 (loss)29,30, and MYC at 8q24.2131 (Supplementary Table S2). Interestingly, the SUP-B15 cell line harbors a gain in the PARP1 gene, localized in the chromosome 1q42.12, supporting its high mRNA expression observed in the real-time PCR data (Fig. 1).
Further, these CNAs were subjected to functional annotation by the g:Profiler232 tool to predict the biological processes they were involved. Gains were associated with immune system activation, encompassing immunoglobulin binding and complex formation, antigen binding, complement activation and mediation of immune response, and transmembrane signaling (Fig. 7A). In contrast, losses are primarily associated with hemoglobin metabolism through oxygen-carrying capacity, haptoglobin binding, and hemoglobin complex formation (Fig. 7B). Overall, our findings contribute to a deeper understanding of the functional consequences of CNAs in B-ALL, revealing their involvement in immune system regulation and hemoglobin metabolism. This knowledge can serve as a foundation for further investigations, ultimately advancing our understanding of the molecular mechanisms underlying these biological processes and potentially providing insights into developing targeted interventions and therapies.
Figure 6. Chromosome’s ideogram showing CNAs identified in SUP-B15 cell line.
Dark blue indicates gains, red represents losses, and light blue represents condensed heterochromatin.
Key significant terms enriched in gene A) gains and B) losses are annotated in the table below. The X-axis displays the functional terms, and the Y-axis shows − log10 of the FDR-adjusted p-value from the enrichment test. (GO: MF) molecular function, (GO: BP) biological process, (GO: CC) cellular component. GO size represents the number of total genes in each specific ontology.
2.7 Differential PARP1 expression profile among p190 + ALL and CML patients and dataset analysis
Attempting to gather evidence of PARP1 relevance in the landscape of BCR-ABL p190 + tumors in the clinical practice, a cohort of 60 patient samples, including both ALL and also CML patients, all of which were quantified and attested as positive for BCR-ABL p190 isoform, had their PARP1 expression measured through qPCR. Overall patient’s characteristics, including available age, white blood cell (WBC) count, immunophenotype, karyotype and qPCR cycle threshold (Ct), are reported in the supplementary material (Supplementary Table S3).
Analyzing ALL p190 + patient samples, clear differential expression of PARP1 may be seen, as evidenced by the significant 2-fold increase compared to healthy donors (p < 0.01; Fig. 8A), while statistical analysis of CML p190 + patient samples highlight repressed expression, evidenced by a 0.35-fold change when compared to healthy donors (p < 0.01; Fig. 8B). To corroborate with our findings from the patient sample analysis, data from the Microarray Innovations in Leukemia (MILE) study was gathered and stratified to discriminate between PARP1 levels in B-ALL with t(9;22), T-cell ALL and samples from bone marrow of healthy donors (Fig. 8C). Indeed, the findings support our claim that PARP1 is overexpressed in Ph + ALL and also show that this is not the case for T-ALL, tying it to the occurrence of B-cell phenotypes.
Furthermore, investigations into single-cell transcriptomics utilizing available online datasets33–35 highlight the overexpression of PARP1 as coming directly from malignant cells in cohorts of c-ALL cases (Fig. 9). Interestingly, the generated heatmap also brings into evidence the naturally occurring higher PARP1 expression in B-cells as compared to other non-neoplastic cell subtypes.
A) Box plot of PARP1 expression in ALL p190 + patient samples compared individually (left) followed by pooled analysis of the average fold change (right). B) Box plot of PARP1 expression in CML p190 + patient samples compared individually (left) followed by pooled analysis of the average fold change (right). Healthy blood donors were used as comparative controls. Statistical differences were analyzed through Student’s T test or Wilcoxon tests. ALL: Acute lymphoblastic leukemia; C-ALL: Childhood Acute Lymphoblastic Leukemia; CML: Chronic myeloid leukemia; NS: Not significant; **p < 0.01; ***p < 0.0001.
A) Heatmap of PARP1 expression among cell subtypes in each dataset. B) Single-cell view of cell populations detected and overlapping PARP1 expression in each dataset.
The results highlight that PARP1 increased expression may not be solely associated with BCR-ABL p190 presence, but rather with pathogenesis of B-ALL in general. These data also suggests that PARP1 relevance and overexpression in B-ALL might be a result of a genetic program inherited from the B-cell lineage. This possibility is further reinforced by the higher expression of PARP1 in B cells rather than in myeloid cells and by the downregulation of PARP1 in CML.