Targeted deletions in ALKBH2, ALKBH3, and MLH1 DNA repair genes

The association of MSH2 and ALKBH2 was intriguing considering the interaction between MGMT and MSH2 (45). Although MSH2 functions at the recognition phase of MMR, elimination of Mlh1 at a later stage of MMR resulted in resistance and increased mutations in MEFs (46). The resistance of cells deficient in MGMT and MLH1 showed that MSH2 alone was not responsible for conveying resistance, but rather that a functional MMR system is necessary for sensitivity to methylation damage (33). Therefore, we targeted MLH1 to investigate the interaction between the ALKBH DRR proteins and the MMR pathway. We introduced targeted deletions that eliminated the active site of the protein in ALKBH2 and/or ALKBH3, both of which protect against methylation damage (Figs. 1A and S2) (9, 17). We also targeted MLH1 to explore whether an intact MMR system is critical for the response to methylation damage. To study the link between DRR and MMR, we identified KOs in clones with ALKBH2 or ALKBH3, or both (Fig. 1B). We also generated MLH1 KO cells with mutations in ALKBH2, ALKBH3 or both (Fig. 1C). To evaluate the presence of DRR protein MGMT, we used Western blots to show that MGMT remained present in all the targeted cell lines (Fig. S3).

Open in a separate window

Figure 1

Targeting and KO validation of MLH1, ALKBH2, and ALKBH3. A, schematic illustrating the target regions ALKBH2, ALKBH3, and MLH1. Indicated regions were cleaved using a dual single-guide RNA CRISPR/Cas9 targeting scheme. B and C, confirmation of MLH1, ALKBH2, and ALKBH3 KOs through immunoblotting. Fifty micrograms of whole cell extracts was subjected to immunoblotting to detect MLH1, ALKBH2, and ALKBH3 expression. β-actin is the loading control. Molecular weights are indicated on the left of the panels. The same β-actin bands are found in Fig. S3. ALKBH, alkylation B homolog; MLH1, Mlh homolog 1.

Loss of ALKBH2 and/or ALKBH3 DNA repair enzymes in cells results in enhanced sensitivity to MMS treatment

We sought to identify the role of the ALKBH DNA repair proteins in mediating cell viability in response to methylation damage. We hypothesized that ALKBH2 and/or ALKBH3 are/is required for inducing cellular sensitivity toward S_N2-methylating agents. Because S_N2-methylating agents yield a high percentage of 1meA and 3meC lesions relative to S_N1-agents, we began by investigating the response to this class of damaging agent (47). To address this hypothesis, we determined cell survival by treating KO cells with increasing concentrations of MMS (an S_N2 methylating agent) and staining with crystal violet at 3 days post treatment. We found that a loss of ALKBH2 and/or ALKBH3 results in an overall enhanced sensitivity to MMS compared to WT controls (Fig. 2). Specifically, both ALKBH2^ko and ALKBH3^ko cells were more sensitive to MMS than WT controls, but only above 500 μM MMS (Fig. 2, A, B, D, and E). At 1.0 mM MMS, both ALKBH2^ko and ALKBH3^ko cells are twice as sensitive to MMS compared to parental cells (Fig. 2E). These findings are consistent with previous findings, which found that loss of either ALKBH2 or ALKBH3 renders cells sensitive to methylation damage with MMS (17, 18, 48). Simultaneous loss of both ALKBH2 and ALKBH3 (ALKBH2^ko3^ko) results in enhanced sensitivity to MMS (Fig. 2C). At 0.4 mM MMS, there was a statistically significant decrease in survival in ALKBH2^ko3^ko cells, almost 3-fold, relative to parental cells (Fig. 2D). Calculation of the IC₅₀ confirmed enhanced sensitivity to MMS in ALKBH2^ko, ALKBH3^ko, and ALKBH2^ko3^ko cells (Table S1). These findings indicate that ALKBH2 and ALKBH3 have compensatory and/or overlapping activities for substrates when present; therefore, loss of both repair proteins exacerbates methylation damage.

Open in a separate window

Figure 2

Cell Survival in HAP1-knockout cells after treatment with MMS. Cell survival of ALKBH2^ko, ALKBH3^ko, MLH1^koALKBH2^ko, MLH1^koALKBH3^ko, and MLH1^koALKBH2^ko3^ko cells was determined after treatment with S_N2 methylating agent, MMS. Cells were treated with increasing concentrations of MMS for 1 h, then incubated in fresh-medium. Three days after treatment, cells were fixed with neutral buffered formalin, followed by staining with 0.1% crystal violet solution, and the dye was resolubilized with 10% acetic acid. Relative survival was determined by setting the untreated controls as the baseline. Survival of (A) ALKBH2^koversus MLH1^koALKBH2^ko, (B) ALKBH3^koversus MLH1^koALKBH3^ko, and (C) ALKBH2^ko3^koversus MLH1^koALKBH2^ko3^ko cells was determined. Survival of all cells after treatment with (D) 0.4 mM or (E) 1.0 mM MMS was analyzed using crystal violet staining. Statistical analysis was carried out using an ordinary one-way ANOVA test. All experiments were performed in triplicate (n = 3). p < 0.05, p < 0.01, p < 0.001, p < 0.0001. ALKBH, alkylation B homolog; MMS, methyl methanesulfonate; MLH1, Mlh homolog 1.

Simultaneous loss of MLH1, ALKBH2, and ALKBH3 leads to resistance to MMS treatments at low concentrations

Next, we sought to identify whether the interaction with the MMR system that was previously observed (Table 1) with ALKBH2 is important for establishing a response to methylation damage. To address this, we used the MLH1^ko ALKBH2^ko cell line to determine cell survival after MMS treatment. Also, because ALKBH2 and ALKBH3 may have compensatory roles, we investigated a potential role for MMR in cells deficient in ALKBH2, ALKBH3, or both. We found that loss of MLH1 in ALKBH2^ko cells (MLH1^koALKBH2^ko) resulted in enhanced sensitivity toward MMS, specifically at 1.0 mM MMS (Fig. 2, A and E). In contrast, MLH1^koALKBH3^ko and MLH1^koALKBH2^ko3^ko cells demonstrated resistance to MMS treatments compared to their MLH1-expressing controls (Fig. 2, B and C). At 0.4 mM, MLH1^koALKBH2^ko3^ko cells showed a statistically significant increase in survival as compared to ALKBH2^ko3^ko cells (Fig. 2D). At 1.0 mM MMS, both MLH1-expressing and MLH1-deficient ALKBH2^ko3^ko cells show the same sensitivity (Fig. 2E). The IC₅₀ for ALKBH2^ko3^ko and MLH1^koALKBH2^ko3^ko cells were 387 and 449 μM MMS, respectively (Table S1). In contrast, at 1.0 mM MMS, MLH1^koALKBH3^ko cells were resistant to treatment as compared to MLH1-expressing controls (Fig. 2E). The IC₅₀ for ALKBH3^ko and MLH1^koALKBH3^ko cells were 469 and 509 μM MMS, respectively (Table S1). These results are surprising because a direct interaction between an MMR protein and ALKBH3 was not observed. In conclusion, these findings indicate that functional MLH1 is required for sensitivity at low MMS concentration in the absence of ALKBH3, but not ALKBH2.

Loss of ALKBH2 and/or ALKBH3 renders cells sensitive to TMZ

Next, we sought to determine the survival of the previously established cell lines after treatment with an S_N1-methylating agent. In contrast to S_N2-agents, S_N1-agents have a different distribution of lesions upon reacting with DNA, and as they are widely used in a clinical setting, understanding the cellular response to these agents has relevance for treatment. We hypothesized that loss of either ALKBH2 or ALKBH3 or both results in enhanced sensitivity to S_N1-methylation agents such as TMZ. To address this hypothesis, ALKBH2^ko, ALKBH3^ko, and ALKBH2^ko3^ko cells were treated with TMZ, then allowed to recover for 3 days, and cell survival was analyzed using crystal violet staining. We found that cell survival of ALKBH2^ko and ALKBH3^ko cells remained relatively unchanged when cells were treated within a range of 0 to 500 μM of TMZ (Fig. 3, A and B). The IC₅₀ of ALKBH2^ko and ALKBH3^ko cells were 446.0 and 452.9 μM TMZ, respectively (Table S1). However, ALKBH2^ko3^ko cells were extremely sensitive to treatment with TMZ (Fig. 3C). At 500 μM TMZ, ALKBH2^ko and ALKBH3^ko cells remained viable, but ALKBH2^ko3^ko cells showed extreme sensitivity (Fig. 3D). At 2.0 mM TMZ, all three cell lines were hypersensitive to treatment (Fig. 3E). These results were recapitulated using a colony formation assay in two independent cell lines (Figs. 3F and S4B). ALKBH2^ko3^ko had an IC₅₀ of 162 μM TMZ, as compared to >500 μM TMZ, which was extrapolated from the survival curve (Table S1). Taken together, the ALKBH2 and ALKBH3 are both important in repair, not only for MMS-induced damage (S_N2), but also for repair of TMZ-induced damage (S_N1) at high concentrations.

Open in a separate window

Figure 3

Cell Survival and Colony Formation in HAP1-KO cells after treatment with TMZ. Cell survival of ALKBH2^ko, ALKBH3^ko, MLH1^koALKBH2^ko, MLH1^koALKBH3^ko, and MLH1^koALKBH2^ko3^ko cells was determined after treatment with the S_N1 methylating agent, TMZ, as previously stated. The survival of (A) ALKBH2^ko versus MLH1^koALKBH2^ko, (B) ALKBH3^koversus MLH1^koALKBH3^ko, and (C) ALKBH2^ko3^koversus MLH1^koALKBH2^ko3^ko cells was determined. Survival of all cells after treatment with (D) 500 μM or (E) 2.0 mM TMZ was analyzed using crystal violet staining. F, a colony formation assay was carried out on ALKBH2^ko3^ko and MLH1^koALKBH2^ko3^ko cell lines 7 days after treatment with TMZ. Statistical analysis was carried out using an ordinary one-way ANOVA test. All experiments were performed in triplicate (n = 3). p < 0.05, p < 0.01, p < 0.001, p < 0.0001. ALKBH, alkylation B homolog; MLH1, Mlh homolog 1; TMZ, temozolomide.

Loss of MMR in single and double deficient ALKBH2 and/or ALKBH3 cells leads to resistance to TMZ

The results with MMS indicated a potential role between MLH1 and the ALKBH2 and ALKBH3 proteins (Fig. 2). Therefore, we hypothesized that a similar role would be observed after treatment with an S_N1-methylating agent. To address that hypothesis, cell survival to TMZ was determined, as previously mentioned using crystal violet staining. We found that loss of MLH1 in either ALKBH2^ko (MLH1^koALKBH2^ko) or ALKBH3^ko (MLH1^koALKBH3^ko) cells did not result in reduced cell viability at 500 μM TMZ (Fig. 3, A, B, and D). However, treatment with 2.0 mM TMZ led to resistance in MLH1^koALKBH2^ko and MLH1^koALKBH3^ko cells that was statistically significant when compared to their MLH1-expressing controls (Fig. 3E). The IC₅₀ of MLH1^koALKBH2^ko and MLH1^koALKBH3^ko cells were >500 μM, respectively, which was an increase from the ALKBH2^ko and ALKBH3^ko cells that were previously described (Table S1). Also, we found that the MLH1^koALKBH2^ko3^ko cells showed enhanced resistance to TMZ at 500 μM and 2.0 mM TMZ (Fig. 3, C–E). MLH1^koALKBH2^ko3^ko survival showed a cumulative effect in survival when compared against MLH1^koALKBH2^ko and MLH1^koALKBH3^ko cells (Fig. 3E). Moreover, the IC₅₀ of MLH1^koALKBH2^ko3^ko cells increased to >500 μM TMZ, as compared to 162 μM in ALKBH2^ko3^ko cells (Table S1). These results were recapitulated using another independent MLH1^koALKBH2^ko3^ko cell line (Figs. 3F and S4B). Overall, the findings obtained from the MMS and TMZ treatments highlight a potential role for MLH1 in inducing cell death after recognition of methyl lesions that are repaired by ALKBH2 and ALKBH3 for both S_N1 and S_N2 methylating agents.

MMR-induced replication arrest in the first cell cycle is MLH1-dependent

Lesion recognition by the MMR pathway can result in an MMR-dependent cell cycle arrest in the second G₂-phase after methylation damage (32). Based on these findings and the cell survival results, we sought to determine if the resistance to TMZ observed in MLH1^koALKBH2^ko3^ko cells was due to an inability to arrest at the G₂ phase. Actively replicating cells were treated with 100 μM TMZ for 1 h, and replication kinetics were determined by pulse-labeling cells with BrdU, and stained with propidium iodide at 24-, 48-, and 72-h post treatment followed by fluorescent-activated cell sorting (FACS) analysis. If direct damage by TMZ is responsible for sensitivity to damage during the first 24 h, then, we expected a significant decrease of cells present in the G₂/M-phase immediately following treatment. FACS analysis revealed that the progression of ALKBH2^ko3^ko cells, although limited, still occurred at 24 h after TMZ treatment, resulting in an accumulation of cells in G₂-phase (Fig. 4, A and B). Using BrdU-incorporation, we determined that 27% and 41% of cells were present at the S- and G₂/M-phase 24 h after TMZ treatment, respectively (Fig. 4B). These results indicate that cells continued through DNA replication immediately following TMZ treatment. However, at 48 h and 72 h after TMZ treatment, ALKBH2^ko3^ko cells are completely arrested at the G₂-phase (Fig. 4A). However, at 48 and 72 h post treatment, 77.9% and 66.2% of ALKBH2^ko3^ko cells were arrested at the G₂/M-phase, respectively (Fig. 4B).

Open in a separate window

Figure 4

Replication kinetics of HAP1 cells after 100 μM TMZ treatment. FACS analysis of HAP1 cell cultures with 100 μM TMZ carried out at 24 to 72 h post treatment. The indicated KO cells were treated with 100 μM of TMZ for 1 h, then pulse labeled with 10 μM BrdU for 30 min before harvesting cells at the indicated time points. Cells were fixed with 70% ethanol and stained with PI before FACS analysis. A, FACS analysis carried out 24 to 72 post treatment in HAP1, MLH1^ko, ALKBH2^ko3^ko, and MLH1^koALKBH2^ko3^ko cells. B, cell distribution % of KO cells was collected by quantifying BrdU-labeling at 24, 48, and 72 h post TMZ treatment. Over 50,000 events were collected using the BD Accuri C6 Flow Cytometer. ALKBH, alkylation B homolog; FACS, fluorescent-activated cell sorting; MLH1, Mlh homolog 1; TMZ, temozolomide.

In contrast, MLH1^koALKBH2^ko3^ko cells showed a slight attenuation of progression through replication at 24 h post-TMZ treatment. (Fig. 4A). Using BrdU-incorporation to determine replication kinetics, we found that 29.9% of MLH1^koALKBH2^ko3^ko cells were at the G₂/M-phase at 24 h post-TMZ as compared to 20.4% of untreated controls (Fig. 4B). However, replication kinetics returned to levels comparable to parental and MLH1^ko controls at 48-h and 72-h post TMZ treatment (Fig. 4, A and B). Thus, these results indicate that a TMZ-induced checkpoint arrest observed in ALKBH2^ko3^ko cells is dependent on a functional MMR pathway.

G₂-arrest is dependent on ATR activation after TMZ treatment

Recognition of drug-induced DNA adducts by the MMR pathway is followed by recruitment of ATM and/or ATR kinases, which activates a DNA damage response (DDR) that cause cell cycle arrest or induction of apoptosis (49, 50). We hypothesized that the G₂/M-arrest observed in ALKBH2^ko3^ko cells was due to an MLH1-dependent activation of a DDR. To address this hypothesis, we sought to identify the DDR kinase that is required for cell cycle arrest at the G₂/M-phase in response to TMZ. To determine this, cells were treated with TMZ, then cultured in complete media containing an ATR-inhibitor (ATRi, VE-821) or an ATM-inhibitor (ATMi, KU-55933) for the last 18 h before pulse-labeling with BrdU, staining with propidium iodide, and harvesting. The proposed protocol is outlined in Figure 5A (Fig. S5A), previously used to study the role of MMR and MGMT in replication arrest (32). We did not observe any significant changes in cell cycle distribution and replication kinetics under the presence of ATMi, KU-55933 (Fig. S5, B and C). In contrast, using the ATRi, VE-821, we found that ALKBH2^ko3^ko cells were unable to enter the first G₂/M phase of the cell cycle at 24 h after TMZ treatment (Fig. 5B). At 24 h post-TMZ, we found that 84.6% of ALKBH2^ko3^ko cells were arrested at the S-phase (Fig. 5C). In comparison 26.9% of ALKBH2^ko3^ko cells without inhibitors were present in the S-phase at 24 h post TMZ (Fig. 4B). At 48 h, we found that 49.9% of ALKBH2^ko3^ko cells accumulated at the G₂/M phase, but there was still a large population of cells present at the S-phase, 39.2% (Fig. 5C). At 72 h, 68.1% of ALKBH2^ko3^ko cells were present in the S-phase after 18 h incubation with VE-821 (Fig. 5C). In contrast, MLH1^koALKBH2^ko3^ko cells did not show any overt differences in cell cycle distribution in the presence of an ATRi or ATMi (Figs. 5 and S5). These results indicate that ATR-activation plays a significant role in regulating cell cycle progression in the presence of a functional MMR.

Open in a separate window

Figure 5

G₂/M-arrest in ALKBH2^ko3^kocells is dependent on ATR activation, not ATM.A, schematic of protocol used to determine the role of ATR inhibition using VE-821. Parental HAP1, MLH1^ko, ALKBH2^ko3^ko, and MLH1^koALKBH2^ko3^ko cells received a treatment of 100 μM TMZ for 1 h. Cells were harvested at 24, 48, and 72 h post treatment in the presence of 5 μM VE-821 during the 18 h. Prior to harvest, cells were pulse labeled with 10 μM BrdU for 30 min. Cells were then fixed with 70% ethanol and stained with PI before FACS analysis. B, FACS analysis carried out 24 to 72 post treatment in HAP1, MLH1^ko, ALKBH2^ko3^ko, and MLH1^koALKBH2^ko3^ko cells. C, cell distribution % of KO cells was collected by quantifying BrdU-labeling at 24, 48, and 72 h post TMZ treatment. Over 50,000 events were collected using the BD Accuri C6 Flow Cytometer. ALKBH, alkylation B homolog; FACS, fluorescent-activated cell sorting; MLH1, Mlh homolog 1; TMZ, temozolomide.

MLH1^koALKBH2^ko3^ko cells have reduced H2AX signals and lose apoptotic signaling

DNA lesions that are repaired by ALKBH2 and ALKBH3 are significant cytotoxic lesions due to their ability to block DNA replication (51, 52). These lesions, 1meA and 3meC, if left unrepaired can induce DNA damage in the form of DSBs, or other damage as determined by phosphorylated histone variant H2AX (mgmt) (53). Therefore, we hypothesized that MLH1 is required for inducing DNA damage as indicated by the Ser139 phosphorylation of H2AX in the absence of ALKBH DNA repair enzymes. We addressed this hypothesis by assessing the formation of a phosphorylated histone variant H2AX (H2AX), a sensitive marker for DNA damage (54). Cells were treated with 100 μM TMZ for 1 h, and whole-cell protein extracts were subjected to SDS-PAGE. We found a substantial increase in H2AX at 48 h in ALKBH2^ko3^ko cells, but only a small increase (similar to parental cells) in MLH1^koALKBH2^ko3^ko cells (Fig. 6A). As a second measure of DNA damage, we also stained for 53BP1 and then quantified its colocalization with -H2AX. We found that MLH1^ko significantly increased the levels of 53BP1 and -H2AX colocalization. Next, we sought to determine if increased H2AX signaling results in changes in apoptotic signaling. We hypothesized that MLH1-dependent activation of a DDR is required for induction of apoptotic signaling. Apoptotic signaling was assessed by examining the cleavage of PARP1, and cells undergoing cellular apoptosis was assessed by measuring Annexin V after treatment to TMZ (55). Immunoblot analysis revealed cleavage of PARP1 at 48 h post TMZ treatment only in ALKBH2^ko3^ko cells. However, no changes in cleaved PARP1 were observed in parental HAP1, MLH1^ko, or MLH1^koALKBH2^ko3^ko cells (Fig. 6B). Apoptosis and cell death were determined by Annexin V-FITC/PI double-staining and quantified by flow cytometry. We found that after 100 μM TMZ treatment, 16.10% and 33.03% of ALKBH2^ko3^ko cells were undergoing apoptosis at 48 and 72 h, respectively (Fig. 6C). We did not observe an increase in apoptotic cells at 24 h in ALKBH2^ko3^ko cells. In fact, the number of apoptotic cells of ALKBH2^ko3^ko cells at 24 h was comparable to parental and MLH1^ko controls. Furthermore, we did not observe any change in apoptotic cells in MLH1^koALKBH2^ko3^ko cells at 24, 48, and 72 h post-TMZ treatment (Fig. 6C). Analysis of colocalization of 53BP1 and γ-H2AX showed a significant decrease following TMZ treatment for 24 h in MLH1^koALKBH2^ko3^ko compared to WT (Fig. S6). Therefore, our results support the hypothesis that the presence of functional MMR is fundamentally necessary in cells defective in ALKBH DNA repair proteins for inducing DNA damage in the form of DNA breaks and activating cellular apoptosis in response to methylation damage.

Open in a separate window

Figure 6

H2AX, and apoptotic signaling in HAP1 cells after TMZ treatment. HAP1 cells were treated with 100 μM TMZ for 1 h and collected at the indicated time point. A, Thirty micrograms of whole cell protein extracts were subjected to electrophoresis on a 10 to 15% SDS-PAGE followed by immunoblotting using anti-H2AX antibodies. Apoptotic signaling was measured by subjecting 30 μg of whole cell protein extracts to SDS-PAGE followed by immunoblotting using (B) anti-PARP1. β-actin was used as the loading control. C, apoptotic cell death was analyzed in HAP1 cells after TMZ treatment using the Annexin-C/PI staining method. Data represent HAP1 cells stained with PI and Annexin-V conjugated with FITC at the indicated time point after treatment with TMZ. Data represented in this figure represents flow cytometry results of 10,000 events. H2AX, phosphorylated histone variant H2AX; TMZ, temozolomide.

After TMZ treatment, MLH1^koALKBH2^ko3^ko cells show reduced replication fork stress as compared to either MLH1^ko or ALKBH2^ko3^ko cells

Because MLH1^koALKBH2^ko3^ko cells did not undergo significant changes in cell cycle progression and H2AX signaling post methylation damage, we investigated the effects of MLH1 and the ALKBH2 and ALKBH3 repair protein deficiencies on the replication fork. We used a DNA fiber combing technique to evaluate possible differences in DNA replication that depend on treatment and genetic background. We hypothesized that the most stress at active replication forks would occur in cells that have functional MLH1 and deficient ALKBH2 and ALKBH3. To investigate this hypothesis, HAP1 cells were pulse-labeled with the nucleotide analog IdU for 30 min, followed by a second pulse-label with CldU in combination with 500 μM TMZ or mock treatment with dimethylsulfoxide (DMSO) (Fig. 7A). Using 500 μM TMZ allowed us to determine replication fork kinetics because we show that at this concentration, there was a higher incidence of DNA breaks and apoptotic signaling at 24 h post treatment in ALKBH2^ko3^ko (Fig. S7). This indicates that substantial DNA damage at this concentration leads to replication stress during the first round of replication. Using DNA fiber analysis, we found that stalled replication forks were observed in MLH1^ko or ALKBH2^ko3^ko cell lines after TMZ treatment. However, no significant changes in fork stalling were observed in MLH1^koALKBH2^ko3^ko cells (Fig. 7B). These results indicate that MLH1, ALKBH2, and ALKBH3 help protect against replication fork stress induced by methylating agents.

Open in a separate window

Figure 7

MMR, ALKBH2, and ALKBH3 protect against stalled replication forks from methylation damage. DNA fiber combing experiments showing replication fork stability after 500 μM TMZ in HAP1 cells. A, schematic representation of the assay conditions. HAP1 cells were pulse labeled with IdU for 30 min, followed by pulse labeling with CldU for 30 min, in combination with 500 μM TMZ. Fork images show an active fork with IdU and CldU labeling of equal lengths. Stalled forks are observed as reduced CldU length relative to IdU. B, DNA fiber combing experiment shows stalled replication forks in MLH1^ko and ALKBH2^ko3^ko cell lines. CldU and IdU fiber tracts were measured using Zen 2.6 software (Zeiss). The ratio of CldU to IdU tracts lengths is presented. Statistical analysis was carried out using an ordinary one-way ANOVA test. Hundred fibers were counted per group (n = 100). p < 0.05, p < 0.01, p < 0.001, p < 0.0001. ALKBH, alkylation B homolog; MMR, mismatch repair; MLH1, Mlh homolog 1; TMZ, temozolomide.

Antibodies

ALKBH2 and ALKBH3 human purified proteins were provided to Cocalico Biologicals for inoculation of two rabbits under a general IACUC protocol for antibody production. Sera were collected from the rabbits after 3 months. Antibodies were purified from sera using Montage Antibody Purification Kit (Millipore). The following antibodies were purchased from suppliers: rat anti-BrdU (Novus, BU1/75), mouse anti-BrdU (Becton Dickinson, AB_400326), mouse anti-MLH1 (Santa Cruz Biotechnology, sc-582, epitope is the C terminus), rabbit anti-ALKBH3 (Invitrogen, PA5-52672, epitope is amino acids 14–139), rabbit anti-Caspase 3 (Cell Signaling Technology, SA1E), rabbit anti-PARP1 (Santa Cruz Biotechnology, sc-25780), rabbit anti-H2Ax (Cell Signaling Technology, 20E3), mouse anti-β-Actin (Sigma-Aldrich, AC-74), rabbit anti-β-Actin (Sigma-Aldrich, RM112), AlexaFluor 488 chicken anti-rat (Invitrogen, AB_2535873), AlexaFluor 594 rabbit anti-mouse (Invitrogen, AB_2534109), AlexaFluor 488 goat anti-chicken (Invitrogen, AB_2534096), AlexaFluor 594 goat anti-rabbit (Invitrogen, AB_2534079), IRDye-680RD goat anti-rabbit (Li-Cor, AB_2721181), IRDye-800CW goat anti-mouse (Li-Cor, RRID AB_2687825), IRDye-680RD goat anti-mouse (Li-Cor, RRID AB_2651128), and IRDye-800CW goat anti-rabbit (Li-Cor, RRID AB_2651127).

Mass spectrometric analysis of ALKBH2 interacting proteins

HEK293T cells (5 × 150 mm dishes) were exponentially grown in Dulbecco's modified Eagle's medium (DMEM) media containing 10% FBS and lysed in lysis buffer (10 mM Tris–HCl [pH 8.0], 150 mM NaCl, 1 mM EDTA, 0.5% NP-40, and proteinase inhibitor cocktail (Roche)) at 4 °C on rocking platform for 30 min. After centrifugation at 12,000g, the supernatant was incubated with specific antibodies against purified ALKBH2 and Protein A beads (Santa Cruz Biotechnology) for 16 h. Immunoprecipitated samples were loaded on 4 to 12% gradient SDS-PAGE gels, and gels were stained in Coomassie Brilliant Blue solution. Specific bands that were not found in the immunoglobulin G control lane were excised, trypsinized, and identified through the Mass Spectrometry and Proteomics Core at the City of Hope using a Thermo Fisher Scientific LTQ-FT MS that was coupled to an Eksigent NanoLC-2D HPLC system. Potential interactions from MS data were confirmed by immunoprecipitation and Western blotting with specific antibodies.

Generating KOs in HAP1 cells

Genes were targeted using a dual single-guide RNA CRISPR/Cas9 targeting vector to introduce deletions at adjacent targeting sites (40–80 bp separation) present in a single exon (Fig. 1A). MLH1^ko, ALKBH2^ko, ALKBH3^ko, ALKBH2^koALKBH3^ko, MLH1^koALKBH2^ko, MLH1^koALKBH3^ko, and MLH1^koALKBH2^ko3^ko HAP1 cells were derived using CRISPR/Cas9 gene targeting. Targeting vectors were designed to induce cleavage at exon 2 on ALKBH2, exon 2 of ALKBH3, and exon 12 of MLH1. Single-guide RNAs were designed using the online resource crispor.tefor.net (75), and cloned into the pX333 (plasmid 64073; Addgene) (76). Cells were transiently cotransfected with targeting vector and pmaxGFP vector (Lonza) using Xfect Transfection Reagent (Takara) according to manufacturer’s instructions. After transfection, individual cells were sorted into 96-well plates using flow cytometry (BD FACSAria II, BD Biosciences). Single cell colonies were cultured for up to 10 days following sorting. DNA was extracted from individual clones using Bradley lysis buffer (10 mM Tris–HCl [pH 7.5], 10 mM EDTA, 0.5% SDS, 10 mM NaCl, 1 mg/ml proteinase K, 0.4 mg/ml RNase) by incubating cells overnight at 37 °C. The following day, DNA was precipitated with glycogen, ammonium acetate (7.5 M), and 100% ethanol at −80 °C. Clones were selected for deletions following PCR using primers flanking the target sites on ALKBH2, ALKBH3, and MLH1. Clones identified with a positive PCR screen were further screened using Sanger sequencing to identify frame-shift mutations, and Western blot to verify loss of ALKBH2, ALKBH3, or MLH1 protein levels (Figs. 1, B and C and S4A).

Preparation of whole-cell extract

Total cell extracts were prepared using the rapid immune-filter paper assay lysis buffer system (Santa Cruz Biotechnology). Briefly, harvested cells were lysed using ice-cold rapid immune-filter paper assay lysis buffer containing protease inhibitors for 20 min at 4 °C and centrifuged at 14,000g for 15 min. Protein concentrations were determined using Bradford assay. Equal protein concentrations of protein samples were separated by denaturing SDS-PAGE and transferred onto a nitrocellulose membrane. Membranes were visualized using an Odyssey CLx Imaging System (Li-Cor).

Clonogenic and cell survival assay

For clonogenic assays, 100 cells were seeded in collagen coated 12-well plates. For cell survival assay, 15,000 cells were seeded in collagen coated 24-well plates. Cells were then exposed to serum-free DMEM containing increasing concentrations of either TMZ or MMS for 1 h at 37 °C. Drug treatment was then replaced with complete media, and cells were assessed at 7 or 3 days after treatment for clonogenic and cell survival, respectively. Briefly, cells were fixed with 10% neutral buffered formalin (Thermo Fisher Scientific) and stained with 0.01% (w/v) crystal violet. For clonogenic assays, individual colonies were counted, and relative survival was calculated by setting untreated controls as baseline. For the cell survival assay, crystal violet was solubilized with 10% v/v acetic acid, and the absorbance at 570 nm was determined with a Synergy HTX Multi-Mode Reader (BioTek), and absorbance of untreated controls were set as baseline. All clonogenic and cell survival experiments were carried out in triplicate.

Cell cycle analysis and determination of apoptosis

For BrdU labeling, cells were pulse-labeled with 10 μM BrdU (Sigma-Aldrich) for 30 min before harvesting and fixed in chilled 70% ethanol on ice for 1 h. The cells were processed as described (77). Cells were incubated in 2 N HCl/0.5% Triton X-100 for 30 min, neutralized with 0.1 M Borax for 2 min, and washed with PBS/1% bovine serum albumin. Samples were incubated in anti-BrdU (Novus) for 1 h, washed, treated with secondary antibody anti-rat AlexaFluor-488 (Thermo Fisher Scientific) for 30 min. Cells were washed and resuspended in PBS containing propidium iodide (20 μg/ml) and RNase A (10 μg/ml) for 30 min at room temperature. BrdU incorporation and cell cycle distributions were analyzed using BD Accuri C6 Flow Cytometer (Becton Dickinson) with >50,000 events per determination. Data were analyzed using FCS Express 7 (DeNovo Software, https://denovosoftware.com/).

Apoptotic frequency was determined using the FITC Annexin V Apoptosis Detection Kit II (Becton Dickinson) following manufacturer’s protocol and quantified by flow cytometry with C6 Accuri Flow Cytometer (Becton Dickinson) with 10,000 events per determination.

Mutant frequency assay

In total, 500,000 cells were treated with 50 μM TMZ for 1 h in serum-free DMEM. After treatment, cells were incubated for 5 days to allow for expression of ouabain-resistant mutations. Subsequently, 1 × 10⁶ cells were incubated in complete media containing 1 μM ouabain for 10 days. Cells were fixed and stained, and the number of ouabain-resistant colonies were counted. In parallel, a cell suspension containing 100 cells was plated in six well plates, and the number of viable cells was counted to determine plating efficiency.

Immunofluorescence

Cells were plated in chamber slides at a density of 5.5 × 10³ cells/cm². After overnight cell adherence, the cells were treated with 100 μM TMZ for 1 h at 37 °C in IMDM media without FBS. After treatment, cells were washed with PBS and IMDM media +10% FBS added. Cells were incubated for 24 h. Following incubation, cells were fixed with 4% paraformaldehyde for 15 min at room temperature. Cells were then washed twice with PBS. Paraformaldehyde was quenched with 0.1 M glycine in PBS for 10 min and then washed twice with PBS. Cells were permeabilized with 0.5% Triton for 15 min and then washed twice with PBS. Blocking solution (2% bovine serum albumin in PBS) was added for 30 min at room temperature with rocking. Primary antibodies (γ-H2AX: Novus NB100-78356, 1:500 and 53BP1: Abcam ab36823, 1:1000) were added overnight in a 37 °C incubator. Cells were then washed three times with PBS. Secondary antibodies (Alexa Fluor 488 donkey anti-mouse: Invitrogen A11029, 1:250 and Alexa Fluor 594 goat anti-rabbit: Invitrogen A11037, 1:250) were added for 1 h in a 37 °C incubator. Cells were then washed three times with PBS. Coverslips were mounted using VECTASHIELD (Vector laboratories) and dried for 2 hours in the dark. Images were obtained on LSM 880 AxioObserver with a Plan-Apochromat 63×/1.4 oil DIC M27 objective. Green and red channels were overlaid and colocalization was determined by the presence of yellow dots with manual counting of the number of colocalization events within a cell.

DNA fiber assays

HAP1 cells were plated 24 h prior to treatment. Cells were labeled with 50 μM IdU (Sigma-Aldrich) for 30 min in complete media. IdU was then replaced with media containing 250 μM CldU (Sigma-Aldrich) with either DMSO or 500 μM TMZ. CldU labeling and treatment was terminated after 30 min. Cells were then washed with ice-cold PBS-DMEM, harvested via trypsinization, and counted. A total of 6.0 × 10⁵ cells were collected and used for DNA fiber analysis. DNA was extracted and spread onto a silanized coverslip in accordance with the FiberPrep: DNA Extraction Kit protocol (Genomic Vision). Coverslips were then air dried, and DNA fibers were fixed onto coverslips with 0.5 M NaOH, 1 M NaCl solution. Fibers were stained with mouse anti BrdU (BD) and rat anti Bardu (Novus) for 2 h. Fibers were then stained with secondary antibodies, rabbit anti-mouse AlexaFluor-594 and chicken anti-rat AlexaFluor-488 for 1 h, followed by tertiary antibodies, goat anti-rabbit AlexaFluor-555 and goat anti-chicken- AlexaFluor-488, for 1 h. DNA fibers were observed under a Zeiss Observer 2 light microscope. The lengths of CldU and IdU tracks were measured using the Zen 2.6 (Blue Edition, https://www.zeiss.com/microscopy/us/products/software/zeiss-zen.html) software. At least 100 forks were analyzed for every condition.

The authors acknowledge the discussions with Drs. Jeremy Stark, Yilun Liu, John Termini, and WenYong Chen.