AZD6738

DNA damage response inhibitor combinations exert synergistic antitumor activity in aggressive B cell lymphomas

ABSTRACT
The DNA damage response (DDR) kinases ATR, Chk1 and Wee1 play vital roles in the response to replication stress and in maintaining cancer genomic stability. Inhibitors of these kinases are currently under clinical investigation. Mantle cell lymphoma (MCL) and diffuse large B cell lymphoma (DLBCL) are aggressive lymphomas whose clinical outcome is still largely unsatisfactory. These cell lymphoma subtypes are highly dependent on both Chk1 and Wee1 for survival. We investigated the activity of the ATR inhibitor AZD-6738 as single agent and in combination with either Chk1 (AZD- 7762) or Wee1 (AZD-1775) inhibitors in several preclinical models of MCL and DLBCL. This study included preclinical in vitro activity screening on a large panel of cell lines, both as single agent and in combination, and validation experiments on in vivo models. Cellular and molecular mechanisms of the observed synergistic effect as well as pharmacoynamic analysis of in vivo samples were studied. AZD- 6738 exerted a strong synergistic cytotoxic effect in combination with both AZD-7762 and AZD-1775 in the two lymphoma subtypes regardless of their p53, MYC and ATM mutational status. These DNA damage response inhibitor combinations, similarly to the Chk1/Wee1 inhibitor combination, caused a marked S phase delay, with an increase in cyclin dependent kinases (CDKs) activity, increased DNA damage and decreases in Wee1, MYC and RRM2 protein levels. The synergistic in vitro activity translated to striking in vivo antitumor activity. DDR-DDR inhibitor combinations could potentially offer promising novel therapeutic strategies for B-cell lymphoma patients.

INTRODUCTION
The DNA damage response (DDR) kinases ATR, Chk1 and Wee1 are required during normal S phase to avoid deleterious DNA breakage, and to maintain cancer cell survival under replication stress (RS) (1). RS may be caused by improper control of replication initiation, associated with molecular features very common in cancer (e.g. constitutive activation of oncogenes such as RAS, MYC, cyclin E, cyclin D; inactivation of key oncosuppressors such as p53, RB1, CDKN2A) (2). RS activates the ATR/CHK1 pathway. The ATR kinase is activated by the ssDNA generated during RS, phosphorylates CHK1 and activates the S phase checkpoint, preventing the collapse of the replication fork, the generation of DSB and the formation of new origin firing, allowing the maintenance of DNA damage at a tolerable level (3,4). The inhibition of the ATR/CHK1 pathway leads to an inappropriate initiation of DNA replication, consumption of replication factors, fork stalling and fork collapse (5). The proper timing of DNA replication is also controlled by Wee1, that inhibits cyclin dependent kinase 2 (CDK2) responsible for the regulation of replication origin firing. As a result, ATR, CHK1 and Wee1 directly control the proper DNA replication rate during S phase progression and appropriately respond to RS(6). Inhibitors of Chk1, Wee1 and ATR are currently under clinical investigation, mainly in combination with chemotherapy or radiotherapy (5,7). The combined targeting of these DDR proteins with other targeted drugs (without chemotherapy) is approaching the clinical setting (e.g. the PARP inhibitor olaparib, the p38/MAPK kinase inhibitor ralimetinib) (5). Recent data from the literature and our own support the hypothesis that combining inhibitors against these DDR components with non redundant roles could be of therapeutic value in preclinical models (8-11).

Mantle cell lymphoma (MCL) is an incurable form of lymphoma with a median overall survival of only 4-5 years. Its clinical course usually involves early relapses, with no responsiveness to current standard therapies (12,13). Diffuse large B cell lymphoma (DLBCL) is the most common aggressive lymphoma in Western countries, with approximately 40% of patients having refractory disease or disease that will relapse after an initial response to first -line therapy. Unfortunately the majority of patients with relapsed DLBCL succumb to the disease (14). Deregulation of the cell cycle is the pathogenic hallmark of MCL due to both the chromosomal translocation t(11;14), that leads to overexpression of cyclin D1, and to enhanced expression of the oncogene MYC and inactivation/genetic alteration of cell cycle inhibitors (12,15,16). MYC aberrations are also present in more than 40% of DLBCL and define patients with a more aggressive phenotype and a poor outcome (14,17). Thus both MCL and DLBCL are characterized by having a high endogenous level of replication stress, due to constitutive activation of such oncogenes (e.g. Myc, cyclin D1) and oncosupressors (e.g. p53) that cause improper DNA replication initiation. As our recent data showed a dependence of MCL and DLBCL on Chk1 and Wee1 for survival (9,11) and ATR inhibitors are strongly active in experimental systems with high oncogenic replicative stress (18,19), we investigated the cytotoxic effect of the ATR inhibitor AZD6738 as single agent and in combination with the Chk1 inhibitor AZD7762 and the Wee1 inhibitor AZD1775. AZD6738 was active at low concentrations in MCL and DLBCL cell lines. We showed that AZD6738 can act synergistically with AZD7762 and AZD1775, in vitro and in vivo. These findings indicate the therapeutic value of DDR-DDR inhibitor combinations, warranting their testing in MCL and DLBCL patients for whom new therapeutic strategies are urgently needed.

Cell lines derived from mature B cell lymphomas were used, as previously specified (9,20). Cell lines had been authenticated by the authors in the last six months. The STR profiles were compared with the American Type Culture Collection database or the German Collection of Microorganisms and Cell Cultures database.AZD6738, AZD1775 and AZD7762 were kindly provided by Astrazeneca, and PF-00477736 and VE- 822 were commercially available (Axon Medchem). More details on cell lines and drugs are given in the Supplementary Methods. Quantification of the effect of the treatments For the initial screening 36 B cell lymphoma cell lines (20) were seeded in 384-well plates at a density of 2000 cells, using a VIAFLO 96/384 channel pipette (Integra Biosciences AG, Zizers, Switzerland); compounds were distributed using a D300e Digital Dispenser (Tecan, Männedorf, Switzerland), and anti-proliferative activity was calculated after 72h of treatment, as previously described (20). For the validation step 96 well plates were used. For the assessment of the combined treatment, each cell line was treated simultaneously with serial concentrations of the drugs. Results were examined by isobologram analysis with Calcusyn Software (Biosoft, Cambridge, UK) and Combination Index (CI) values at the IC50 were calculated to assess the efficacy of the combination (8,9). All the experiments were done at least twice and with six replicates in each experimental group, as already described (9). Additional specifications on the treatment procedures are included in the Supplementary Methods.Proteins were extracted as previously described (21). Briefly, total cellular protein extracts were obtained using a lysis buffer containing 10 mM Tris-HCl pH 7.4, 150 mM Na Cl, 0.1% Nonidet NP-40, 5 mM EDTA, 50 mM NaF with the addition of proteinase inhibitors. 50 µg of proteins were loaded and separated on SDS-polyacrylamide gel and transferred to a nitrocellulose membrane (PROTRAN, Schleicher and Shull) Immunoblotting was done using antibodies, as detailed in Supplementary Methods and visualized using Odissey FC Imaging System (Li-COR).

Caspase-3 activity was measured by enzymatic assay using a fluorogenic substrate for caspase-3, Ac- DEVD-AMC (acetyl Asp-Glu-Val-Asp 7-amido-4-methylcoumarin), as already described (8). Briefly, lysed extracts were mixed with the apoptosis activity buffer (Hepes pH 7.5 20 mM, glycerol 10%, DTT 10 mM) in a black 96-well plate and incubated at 37°C for 5 minutes. The substrate was then added at a final concentration of 12.5 μM. Fluorescent AMC production was measured at excitation 370 nm, emission 460 nm, using a plate reader (Infinite M200, TECAN). Activity was expressed as the linear change in fluorescence units per hour and normalized for the protein concentration. The increase in the apoptotic signal of treated samples was measured as a percentage of untreated control samples at each experimental time point.To analyze DNA content distribution by flow cytometry, OCILY7 and JEKO-1 cells were fixed in 70% ethanol 8, 24 and 48 h after treatments either singly or combined. Fixed cells were washed with PBS and stained with 25 µg/ml Propidium Iodide (PI) (Calbiochem) in PBS plus 25 µl of 1 mg/ml RNAse (Sigma) in water. After 2h of incubation at room temperature, at least 10,000 cells for each sample were acquired by FACSCalibur (Becton Dickinson) flow cytometer. Cell cycle percentages in the different phases were obtained by flow cytometric histograms as previously described (22). For two- parameter flow cytometry analysis of DNA content and p-S10 histone H3 about 2 x 106 cells fixed in Ethanol 70% were washed with PBS and permeabilized in Triton X-100 0.25% in PBS for 10 min on ice. Then, cells were washed and incubated with 100 μL of anti p-S10 histone H3 (Cell SignalingTechnology #9706) diluted 1:100 in PBS containing 0.5% BSA for 2 h at room temperature. After washing with PBS, cells were incubated with Alexa-fluor488 (goat anti-mouse, Molecular Probe #A-11017) diluted 1:500 in PBS + 0.5% BSA for 1 h at room temperature. After the incubation with antibody, cells were centrifuged, resuspended in 2.5 μg/mL PI in PBS plus 25 μl of 1 mg/ml RNase in water, incubated overnight and analyzed.

Gene expression profiling was done using the HumanHT-12 v4 Expression BeadChip (Illumina, San Diego, CA, USA) as previously reported (23). GenomeStudio software and then imported in the Partek Genomics Suite 6.4 and quantile normalized.Total RNA was extracted from cell lines and tissues using a Maxwell® RSC simplyRNA Cells Kit and Maxwell® RSC simplyRNA Tissue Kit with the Maxwell® RSC Instrument (Promega, Milan, Italy). RNA retro-transcription and mRNA quantification was done as previously described (24). Primer specifications are given in Supplementary Table 1.Five-week-old female NCr-nu/nu mice were obtained from Envigo s.r.l. Italy and maintained under specific pathogen-free conditions. Procedures involving animals and their care were conducted in conformity with institutional guidelines, in compliance with national and international laws and policies and with guidelines for the welfare and use of animals in cancer research (25). Details of in vivo procedures and pharmacodynamic studies are given in the Supplementary Methods.Statistical significance was determined with GraphPad Prism 7.02 (GraphPad Software, San Diego, CA, USA). The legends to the figures specify which tests were done. The Pearson test was used to establish the correlations between different variables. A p-value below 0.05 was taken as significant. Groups of functionally related genes expressed differently between classes were identified as previously described (20), using the gene expression profiles of the untreated cell lines from the GSE94669 dataset and the Gene Set Enrichment Analysis (GSEA) tool and MSigDB 5.2 gene-sets, (26) with a threshold based on FDR < 0.1. Raw data are available at the National Center for Biotechnology Information Gene Expression Omnibus https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE64820 database. RESULTS We investigated the cytotoxic effect of AZD6738 in a large panel of lymphoma cell lines: 36 mature B- cell lymphoma cell lines comprising ten MCL and 26 DLBCL cell lines (seven activated B cell (ABC)- DLBCL and 19 germinal center B (GCB)-DLBCL), treating them with increasing doses of the ATR inhibitor AZD6738, for 72 h. Figure 1A reports the AZD6738 IC50s in all the cell lines tested. AZD6738 anti-proliferative activity weakly correlated with that of another ATR inhibitor, VE-822 (R= 0.541, P= 0.00065) and of the Chk1 inhibitor AZD7762 (R= 0.49, P=0.002378), but not with the Wee1 inhibitor AZD1775 (Supplementary Figure 1A-C). We then validated the results of the initial screening in 21 of these cell lines in 96 well plates (Figure 1B) with similar results. Specifically we re-screened all the MCL cell lines and we compared them with an equal number of DLBCL cell lines, chosen mainly on the basis of the IC50 values obtained by the first screening and selecting 5 DLBCL-ABC and 6 DLBCL-GCB with IC50 spanning from the lowest to the highest values, with the idea to catch representative cell lines. The JEKO-1R cell line with acquired resistance to the Chk1 inhibitor PF- 00477736 (27) was cross resistant to the ATR inhibitor (Supplementary Figure 1D).Considering the wide range (100 fold) of sensitivity to AZD6738 (from 143 nM to 10400 nM) in this panel of cell lines, and to gain some idea of the molecular determinants associated with the response, we compared the baseline gene expression profiles of the 11 most sensitive (IC50< 600nM) with the 11 most resistant (IC50 > 2500 nM) B-cell lymphoma cell lines (Supplementary Figure 2A). Sensitive cell lines showed enrichment of the genes related to chromosome segregation, cell cycle checkpoint, damaged DNA binding and cyclin genes; less sensitivity to the drug was related to enrichment in oxidative phosphorylation and reactive species genes, hydrogen transport, fatty acid metabolism and PI3K/AKT/mTOR signaling (Figure 1C and Supplementary Tables 2-3).

Since both p53 and MYC have already been found to be associated with a strong response to ATR inhibitors (19,28,29), we looked for a putative association of these molecular features with the degree of sensitivity to AZD6738 in 26 of the 36 B cell lymphoma cell lines tested (with known functional status for p53 and MYC), as specified in Supplementary Figure 2B. No association could be found with inactivation of p53 and/or genetic alterations of MYC (translocation, amplification) (30-34), as these were observed in both sensitive and resistant cell lines (Supplementary Figure 2B). Additionally, no association was detected between sensitivity to AZD6738 and ATM deficiency (deletions/lack of expression, where data where available) in these cell lines in contrast with prevous data (28,29,35). Indeed two cell lines with no protein expression of ATM (among those shown in Supplementary Figure 3) (Z-138 and SP-53) displayed at least a four fold difference in sensitivity to AZD6738 (IC50 respectively 143.6 nM and 756.2 nM). Moreover, two MCL cell lines with known deletion of ATM (Granta-519 and MAVER1) (36,37) were not the most sensitive.We then investigated whether drug sensitivity was associated with endogenous DNA damage (γH2AX and p-S317Chk1levels), the protein levels of the oncogenic protein MYC, of the DNA damage response protein ATM (both phosphorylated form and total protein), of Wee1 and of the cell cycle inhibitors p16, p18 and p27 in the 21 cell lines used in the validation (Supplementary Figure 3). The majority of the MCL and DLBCL cell lines had high levels of constitutive DNA damage (high levels of γH2AX and pChk1), as already reported (11,38). There was a significant correlation between ɣH2AX levels and sensitivity to AZD6738 (P=0.044) (Supplementary Figure 4A). The correlation was even stronger (R=0.70; P= 0.022) in the MCL subgroup and was observed not only with the sensitivity to AZD6738, but also with that of the Chk1 inhibitor AZD7762 (R= 0.85; P=0.001) (Supplementary Figure 4B), and with another Chk1 inhibitor, PF-00477736 (R=0.758; P=0.0109) (data not shown). No correlation was also found between AZD-6738 sensitivity and ATM and ATR expression at mRNA levels (Supplementary Figure 4C and 4D). Figure 1B (lower part) reports the p53 and MYC alterations and expression of the cell cycle inhibitors in these 21 cell lines; overall, most of these cell lines present multiple aberrations/deletions in the genes/proteins involved in the control of replication initiation and G1/S transition, but none of the combinations of these factors seem to be associated with the response to AZD6738.

AZD6738 synergizes with both the Chk1 inhibitor AZD7762 and the Wee1 inhibitor AZD1775 Based on previous data showing the high susceptibility of B cell lymphomas to the DDR inhibitors against Chk1 and Wee1, we investigated the effects of combined AZD6738 treatments with either the Chk1 inhibitor AZD7762 or the Wee1 inhibitor AZD1775 in 11 cell lines out of the 21 used in the re- screening panel (5 MCL cell lines and 6 DLBCL cell lines), based on their sensitivities to the three inhbitors. Combination indexes at the IC50 of each drug combination were extrapolated (Figure 2A). Both drug combinations were strikingly synergistic in all the cell lines investigated, regardless of the lymphoma subgroups and of MYC, p53 and ATM status. Interestingly the synergistic effect was also observed in the AZD6738 insensitive cell lines (SUDHL2, MINO). There was a negative correlation between CI at the IC50 of the Wee1+ATR inhibitor combination and Wee1 protein levels (R= -0.66; P= 0.026), similarly to what has been reported with Chk1/Wee1 dual inhibition in DLBCL cell lines
(11). The negative correlation was even stronger when only DLBCL cell lines were considered both at protein (R= -0.929; P= 0.0073) and mRNA levels (R= -0.887; P= 0.0184) (Figure 2B).We further investigated the two drug combinations (ATR/Chk1 and ATR/Wee1 inhibition) to clarify the cellular and molecular mechanisms of the observed synergistic effect, and compared them with the previously characterized Chk1/Wee1 dual inhibition (9,31). We selected the OCILY-7 cell line (DLBCL) and the JEKO-1 cell line (MCL). Figure 3A shows the cell growth of the two cell lines every 24h up to 96h when treated with drug concentrations exerting a strong synergistic effect when combined (concentrations specified in the Figure legend). There was no effect on cell growth after a single treatment, but starting from 24h, the three combinations strongly reduced cell growth up to 96h in both cell lines (more than 90% of growth inhibition). The combined treatments induced clear cell cycle perturbation in both cell lines (Figure 3B and Supplementary Figure 5). OCILY-7 cells treated with the combinations were able to exit G1, but accumulated in early S phase.

This effect was already detectable at 8h of treatment in the single drug samples, but was much stronger with the drugs combined. At 24h cells treated with the combinations were still delayed in their progression through S phase, but some were able to reach G2-M phase, remaining blocked there also at 48h. The biparametric staining of p-S10 Histone H3 and DNA (Supplementary Figure 6A) demonstrated that cells that were in mitosis during the first hours of treatment (still detectable at 8h) were able to divide or undergo cell death and only a small amount of p-S10 Histone H3-positive cells (less than 1%) had a DNA content between 2N and 4N, excluding the hypothesis of a massive premature entrance into mitosis.The presence of sub-G1 events, detected at 48h (Figure 3B and Supplementary Figure 5) , together with the increased caspase-3 activity at 24h and 48h with all the combinations (Supplementary Figure 6B) confirmed that cell death was occurring. Single drug treated JEKO-1 cells were almost unaffected by the inhibitors, while cells treated with the combinations stopped growing and DNA distributions highlighted the S phase delay. The effect and its kinetics were similar for all the combinations and the accumulation of cells became evident at 24h, clearly increasing at 48 h in cells treated with the Chk1/Wee1 inhibitor combination. Between 24h and 48h cell death occurred, as demonstrated by the slight increase in sub-G1 events and the high caspase-3 activity detected from 8 h and further increasing at the subsequent time points (Supplementary Figure 5). The three drug combinations induced a strong decrease in MYC, Wee1 and RRM2 protein levels and an increased DNA damage, measured by upregulation of ɣH2AX protein levels in both cell lines (Figure 3C). MYC and Wee1 mRNA levels did not decrease after the drug combinations, suggesting that this expression is modulated at the protein levels. There was a slight decrease of RRM2 expression at mRNA levels in OCILY-7 cells at 48h after treatment with AZD6738 combined with AZD1775 and AZD7762 (Supplementary Figure 6C).

The three drug combinations were tested in vivo in nude mice bearing OCILY-7-xenografts. Doses and schedules were based on previous data (8-10,39,40). No significant effect on tumor growth was observed in mice treated with AZD6738, AZD7762 and AZD1775 in monotherapy, except for one mouse in the AZD7762 treated group, whose tumor regressed after treatment and the mouse was sacrified at day 150 without tumor. On the contrary, all the three combinations were quite active, significantly delaying tumor growth, including tumor regressions, and increasing overall survival (Figure 4A). Single treatments were well tolerated, as suggested by the fact that there were almost no body weight changes, although two toxic deaths were recorded in the AZD7762 group. The combinations caused body weight loss which was rapidly regained upon drug withdrawal (Supplementary Figure 7A). and all the combined treatments resulted in very low T/Cs, indicative of high antitumor activity (Supplementary Figure 7B). Interestingly, tumor responses were clearly observed also at the third treatment cycle for the AZD7762+AZD1775 and AZD6738+AZD1775 combinations. In addition, in these latter groups long term survivals (no tumor regrowth five months after transplantation) were recorded (three and two mice out of seven respectively in the AZD7762+AZD1775 and AZD6738+AZD1775 combinations) (Figure 4A). The AZD6738+AZD1775 combination was also active when tested in a late stage setting, in which treatment started at tumor weights of 580 mg (as opposed to 150 mg in the previous experiment), causing tumor regression or stabilization (Supplementary Figure 7C). The drugs were able in vivo to inhibit their respective targets: ATR (decreased pS317 of Chk1), Chk1 (increased p-S317 of Chk1 and decreased pY15 CDK1) and Wee1 (decreased pY15 CDK1) (Supplementary Figure 8A). MYC, Wee1, RRM2 and pCDK2 protein levels clearly decreased after all three comnbinations, though it was more marked after AZD6738+AZD1775 and AZD1775+AZD7762, which exerted higher antitumor activity.

No clear-cut increase in ɣH2AX was detected after the combinations as compared to single drugs in vivo (Figure 4C).AZD1775+AZD6738 was also tested in JEKO-1 MCL xenografts (Figure 5A). Again, using the same drug schedules as in the previous experiment there was no significant body weight loss (Supplementary Figure 8B), maximun loss being 5.7%. In this model tumor growth inhibition was significant in animals treated with the drugs as monotherapy, but the antitumor effect was greater in mice treated with the combination, as suggested by the lower T/C% (Supplementary Figure 8D). In addition, tumor regressions were observed in all the animals treated with the combination (Supplementary Figure 8D). Again, tumors were still responsive to the third drug combination cycle. The AZD6738 and AZD1775 targets were inhibited in vivo (Figure 5C). The increased Chk1 activation (increased pChk1) exerted by AZD1775 was partly neutralized by AZD6738. MYC, Wee1 and RRM2 protein levels decreased clearly after the drug combination, while the decreases in pCDK2 and the increases in ɣH2AX were less clear (Figure 5C).

DISCUSSION
Here, we investigated the role of the DDR kinase ATR as a potential therapeutic target in B cell lymphomas, both MCL and DLBCL. AZD6738 is a novel oral selective inhibitor of ATR under early clinical trial development both in monotherapy in hematologic malignancies with reported ATM deficiencies (NCT01955668) and in combination with chemotherapy (e.g. carboplatin, paclitaxel), DNA damage repair agents (olaparib) and new anticancer agents (MEDI4736) in advanced solid tumors (for recent reviews see (5,7,41)). We studied the cytotoxic activity of AZD6738 in a wide panel of MCL and DLBCL cell lines, both as single agent and in combination with other DDR (Chk1 and Wee1) inhibitors. MCL and DLBCL displayed a wide range of sensitivity to AZD6738, with IC50 ranging from 143 nM to 10400 nM. AZD6738 sensitivity was unrelated to the genetic and/or mutational status of the known oncogenes and/or oncosuppressors previously found in synthetic lethality with ATR (such as p53, ATM, MYC) (19,28,29,35), and not even associated with the combination of alteration/lack of expression of more than one molecular feature (MYC, p53, p16, p18, p27) related with the increased G1/S transition.The AZD6738 IC50 in the most sensitive cell lines were lower than those recently reported in the chronic lymphocytic leukemia cell lines (29) and gastric cancer cells with ATM or p53 dysfunction (39), corroborating the evidence that at least in this experimental setting the highest sensitivity to AZD- 6738 is not strictly and exclusively associated with ATM mutational status/expression. Hocke S. et al(42) identified POLD1 and other DNA repair genes, other than ATM, as synthetically lethal with ATR. In a recent paper Menezes DL et al (43) demonstrated that the ATM deficient cancer cell line GRANTA-519 is more sensitive to the ATR inhibitor WO2010/073034 than the ATM proficient JVM-2 cell line, but these are not identical experimental systems and thus the different sensitivity may be due to other molecular features. Our data indeed demonstrated that very sensitive lymphoma cell lines (e.g UPN1, OCILY-10) display ATM expression and that two cell lines lacking ATM expression (Z-138 and SP-53) differ in sensitivity to AZD6738 (of at least 4 fold) (Figure 1B). Others, not yet identified molecular features, are responsible for the sensitivity to ATR inhibitor in these lymphoma cell lines. Unexpectedly, we found that the most sensitive cell lines to both ATR and Chk1 inhibitors were those with lower ɣH2AX levels. One might hypothesize that in cell lines with low constitutive ɣH2AX levels, the proteins responsible for this phosphorylation (e.g. DNAPK and/or ATM activity (44)) have lower activity, and this might explain the greater susceptibility to the ATR inhibitor, but this remains to be demonstrated. The baseline gene expression profiles indicated that cell cycle-related gene-sets and genes related to DNA damage response were associated with higher sensitivity to the ATR inhibitor, while the gene expression profiles of the most resistant cell lines were enriched in genes related to the pro-survival pathway PI3K/AKT/mTOR signalling, and genes associated with metabolism. This pattern of expression is similar to that related to Chk1 inhibitors sensitivity (9). There was a weak correlation between the sensitivity to Chk1 and ATR inhibitors, also supporting the cross resistance to AZD6738 of a MCL cell line, JEKO-1R (with acquired resistance to the Chk1 inhibitor (27)).

In contrast, we found no correlation between the sensitivity to ATR and Wee1 inhibitors in these cell lines. Taken together these data reinforce the fact that, although the three proteins belong to the DDR pathway and share similar functions, they also act independently in the DNA damage response pathway. Strong synergistic activity was observed with AZD6738/AZD1775 and AZD6738/AZD7762 combinations in all the cell lines (DLBCL and MCL); in addition even in tumors less sensitive to the inhbitors as single agents, the combinations were synergistic. Our findings not only corroborated recent data on the synergistic effect with Chk1 and ATR inhibitors (10), but also clearly indicate a therapeutic synergism of the combinations with Wee1 and ATR inhibitors, in support of very recent data that showed this synergism in triple negative breast cancer (45). We investigated the molecular mechanism at the basis of the synergy of the investigated drug combinations and compared it with the Chk1+Wee1 inhibitor combination previously characterized (9). All the three drug combinations blocked cell growth, caused general mis-coordiantion of the cell cycle mainly due to substantial S phase delay, induced DNA damage, and activated apoptosis. In the search for biomarkers of response, we found that all the DDR- DDR inhibitor combinations downregulated MYC, Wee1 and RRM2 protein levels both in vitro and in vivo. The decrease in Wee1 protein levels is not transcriptionally dependent and could be due to an ubiquitin ligase activity associated with the increased CDK activity induced by the drug combination, as recently suggested (46,47). We have already reported that MYC protein levels were destabilized after Chk1/Wee1 dual inhibition (11).

This effect was confirmed after the other DDR-DDR inhibitor combined treatments as well. Still to be clarified are the molecular mechanisms. The effect on MYC protein further stresses the therapeutic potential of these treatments in MYC dependent lymphoma subtypes. We also observed a significant decrease of RRM2 protein, likely to be due to increased CDKs activity caused by the drug combination, as recently reported (48,49) and observed in pancreatic cancer cells after Chk1 inhibitor in combination with gemcitabine (50). The decreased RRM2 levels leading to dNTP exhaustion can partially explain the appreciable S phase delay after the drug combinations, which consequently induced DNA damage and cell death. The in vitro synergism translated into striking in vivo antitumor activities in two different xenograft models. Indeed in both models the combinations were very active, regardless the activity as single agent. Our data provide the first evidence that the combination of ATR and Wee1 inhibitors is effective and well tolerated. This new drug combination exerts significant antitumor activity even in a late stage of disease.

In conclusion, these findings suggest that DDR-DDR inhibitor combinations are feasible and strongly effective in MCL and DLBCL. Despite the limitations due to the lack of data on primary lymphoma samples and/or patient derived xenografts, these data do suggest these combinations warrant clinic investigation as new and effective therapeutic strategies AZD6738 in aggressive B cell lymphoma patients.