Targeting UHRF1-dependent DNA repair selectively sensitizes KRAS mutant lung cancer to chemotherapy

Abstract

Kirsten rat sarcoma virus oncogene homolog (KRAS) mutant lung cancer remains a challenge to cure and chemotherapy is the current standard treatment in the clinic. Hence, understanding molecular mechanisms underlying the sensitivity of KRAS mutant lung cancer to chemotherapy could help uncover unique strategies to treat this disease. Here we report a compound library screen and identification of cardiac glycosides as agents that selectively enhance the in vitro and in vivo effects of chemotherapy on KRAS mutant lung cancer. Quantitative mass spectrometry reveals that cardiac glycosides inhibit DNA double strand break (DSB) repair through suppressing the expression of UHRF1, an important DSB repair factor. Inhibition of UHRF1 by cardiac glycosides was mediated by specific suppression of the oncogenic KRAS pathway. Overexpression of UHRF1 rescued DSB repair inhibited by cardiac glycosides and depletion of UHRF1 mitigated cardiac glycoside-enhanced chemotherapeutic drug sensitivity in KRAS mutant lung cancer cells. Our study reveals a targetable dependency on UHRF1-stimulated DSB repair in KRAS mutant lung cancer in response to chemotherapy.

1. Introduction

KRAS mutant lung adenocarcinoma accounts for ~30% of non-small cell lung cancer [1,2], the most prevalent form of lung cancer. Although pan-ERBB tyrosine kinase inhibitors such as afatinib suppress KRAS mutant lung cancer growth [3,4], activation of the downstream RAF-MEK-ERK pathway usually leads to resistance to tyrosine kinase inhibitors [1,2]. Targeting the downstream signaling by MEK inhibitors often results in compensatory activation of the PI3K pathway or rebound activation of RAF-MEK-ERK [1,2]. Progress has been made in identifying molecules that directly target mutant KRAS, such as allosteric or covalent inhibitors of the KRAS^G12C mutant allele [5–7]. Synthetic lethality screening identified non-RAS pathway genes, such as STK33 [8], TBK1 [9], SHP2 [10–12], etc., whose depletion or inhibition increased the sensitivity of KRAS mutant cancers to conventional or targeted therapies. However, these targets may only apply to a small subset of KRAS mutations in a context dependent manner and clinical validation is still needed [1,2]. On the other hand, anti-PD-L1 based immunotherapy is limited to a small percentage of patients [2,13]. Hence, there remains a clinically unmet need to develop effective strategies that can be translated to treatments for KRAS mutant lung cancer.

Currently, platinum-based chemotherapy remains the standard of care for KRAS mutant lung cancer, although the overall response rate is limited [14]. Therefore, identification of chemo sensitizers and their underlying cellular targets would allow us to selectively target this subtype of cancer and to increase the therapeutic response. Many chemotherapeutic drugs kill cancer cells by causing massive DNA damage, particularly DNA double strand breaks (DSBs). However, cells activate the protective DNA damage response and repair programs to evade the cytotoxic effects of chemotherapy. Hence, genes involved in the DNA damage response and DSB repair pathways may regulate the sensitivity of KRAS mutant lung cancer to chemotherapy, representing potential targets, whose inhibition can selectively sensitize KRAS mutant lung cancer to chemotherapy.

In this study, we first conducted a Western blot-based compound library screen and identified cardiac glycosides as potent DSB repair inhibitors in KRAS mutant lung cancer cells. Using quantitative mass spectrometry, we identified UHRF1 (Ubiquitin-like with PHD and Ring Finger Domains 1) as a DSB repair factor largely responsible for the effects of cardiac glycosides. We further reveal how cardiac glycosides specifically inhibit UHRF1 expression in KRAS mutant lung cancers. Our results illustrate a novel role for UHRF1 in regulating the selective sensitivity of KRAS mutant lung cancer to chemotherapy. They also suggest a unique approach to treat KRAS mutant lung cancer by combining two classes of clinically available drugs, cardiac glycosides and DSB-inducing chemotherapeutic agents.

2. Materials and methods

2.1. Chemical library screen

The chemical library contains 874 compounds from both the Institute of Traditional Chinese Medicine and the DNA damage and repair library from TargetMol (#L3900) (Supplementary Table S1). All compounds were dissolved in DMSO at 10 mM stock concentration and stored at –20 °C. For screening, 6 × 10⁵ A549 cells were plated into 6-well plates for 24 h, pretreated with 5 μM compounds for 2 h, then 500 nM CPT was added for another 4 h. Cell lysates were collected, and pCHK1 and total CHK1 protein levels were assessed by Western blotting. The protein band intensity of pCHK1 and CHK1 was quantitated using the NIH Image J software. The relative level of pCHK1 was normalized to that of the total CHK1 from the same sample. The ratio of pCHK1/CHK1 from compound-treated samples was then normalized to that of CPT-treated cells and was expressed in log2 scale.

2.2. Cell survival assay

Briefly, 5 × 10³ cells/well were seeded in a 96-well flat-bottomed plate, grown at 37 °C for 24 h, and treated with 5 μM of tested compounds for 48 h. Then 10 μL Cell Counting Kit-8 solution (Beyotime Inst Biotech, China) was added to each well, incubated at 37 °C for 2 h and the absorbance was determined at 450 nm using a microplate reader (Synergy TM HT, BioTEK, USA). For long-term cell survival, 5000 cells per group were seeded in 6-well plates after drug treatment and cultured in drug-free full media for 10–14 days. Cells were fixed in acetic acid/methanol (1:7, vol/vol) at room temperature for 5 min, stained with 0.1% crystal violet dye in methanol at room temperature for 15 min, rinsed with water and air dried. The dried plates were first scanned, then crystal violet was solubilized in 1% SDS and absorbance at 570 nm was read using a microplate reader.

2.3. Western blotting

Total cell lysates were harvested in lysis buffer (Beyotime Inst Biotech, China) containing 1 mM phenylmethylsulfonyl fluoride (Beyotime Inst Biotech, China). The protein concentration was measured using the Pierce® BCA Protein Assay Kit (Pierce, #23225). Equal amounts of proteins were separated by 6%–15% SDS-PAGE and transferred to PVDF membranes (#IPVH00010, Millipore, Billerica, MA, USA). The membranes were blocked with 5% skim milk and probed with primary antibodies overnight at 4 °C followed by horseradish peroxidase conjugated secondary antibodies, reacted with Pierce® ECL Western Blotting Substrate (Thermo Fisher Scientific, Franklin, MA, USA) and detected by an ECL detection imaging system (BioTanon, Shanghai, China).

2.4. Mouse xenograft experiment

5–6 weeks old female BALB/c Nu/Nu mice were purchased from Beijing HFK Bioscience CO. LTD (Beijing, China). All mice were housed in cages with bedding, controlled temperature (23 ± 2 °C), humidity (50 ± 5%) and illumination (12 h light/dark cycle) and adapted to the facility for one week before experiments. All animal experiments were performed in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals (NIH publication No. 80–23, revised in 1996) and were approved by the Animal Ethics Committee of Jinan University (Approval Number: 20130904001). Cisplatin was freshly prepared by dissolving in normal saline buffer. AT2 stock solution was prepared by dissolving the compound in ethylene glycol diethyl ether: cremophor: saline (1:1:8, vol). The combination was freshly diluted 10 times by saline buffer immediately before injection. The vehicle used in the preparation of AT2 was administrated as the control.

For mouse studies, 3–5 × 10⁶ cells suspended in serum free medium were injected subcutaneously into the right flank of mice. When the tumor volume reached ~100–200 mm³, mice were randomly divided into 4 groups with 10 mice in each group: (1) Control; (2) AT2 alone (0.3 mg/kg); (3) Cisplatin alone (1.5 mg/kg); and (4) Cisplatin (1.5 mg/kg) plus AT2 (0.3 mg/kg). AT2 was administered by i.v. injection every two days, whereas Cisplatin was administrated by i.p. injection every four days. Tumor volume and body weight were measured every two days for 20 days. Tumor volume was calculated using formula V = (L × W²) × 0.52 where V = volume, L = length, W = width. At the end of treatment, mice were sacrificed and samples including blood, organ and tumors were collected for further analysis.

2.5. Statistical analysis

All cell culture experiments were performed at least twice. Data are presented as mean ± standard deviation. The statistical analysis was conducted using Prism 5.0 (GraphPad) software. Pairwise comparisons were performed using a two-tailed Student t-test, whereas one-way ANOVA was used for multiple comparisons. P-values of less than at least 0.05 were considered statistically significant.

3. Results

3.1. Compound screen of DNA damage response inhibitors in KRAS mutant lung cancer

To identify chemo sensitizers that selectively increase the sensitivity of KRAS mutant lung cancer to chemotherapy, we carried out a Western blot-based screen to search for small molecules that inhibit chemotherapy-activated DNA damage responses in KRAS mutant lung cancer cells. Briefly, we treated KRAS mutant A549 lung cancer cells with 5 μM of each compound from a library (874 compounds with diverse chemotypes) for 2 h, added 500 nM camptothecin (CPT, a topoisomerase 1 inhibitor that activates the DNA damage response [15–18]) for another 4 h and measured CPT-induced CHK1 phosphorylation (pCHK1), one of the gold standards of DNA damage response activation [19], using specific antibodies (e.g., Fig S1). If a compound inhibited the DNA damage response, it would reduce CPT-induced pCHK1. This strategy is highly reproducible and provides a direct biological readout for the DNA damage response affected by tested compounds. As a validation of the screen, two known ATR inhibitors came out as the top hits (Fig. 1A & B). Interestingly, several cardiac glycosides isolated from the poisonous plants Antiaris toxicaria and Thevetia peruviana (AT and TP series compounds, respectively) [20,21] significantly suppressed CPT-induced CHK1 phosphorylation (Fig. 1B, S1D, S1F, and Supplementary Table S1).

Fig. 1.

Chemical screen of DNA damage response inhibitors in KRAS mutant lung cancer cell. (A) Representative Western blot results. VE822 is a known ATR inhibitor. (B) Screening results. Each dot represents one compound. The value represents pCHK1/CHK1 ratio normalized to that of CPT only and are presented in log2 scale. Zero represents that of CPT. (C) A549 cells were pre-, co- or 1 h post-treated with 500 nM AT2, AT3 or TP4 with 500 nM CPT for 4 h, and protein levels of pCHK1 and CHK1 were analyzed. Ponceau S staining indicates equal protein loading. (D) A549 cells were pre-treated with 500 nM AT2 for 2 h, added 500 nM CPT for 1, 2 and 8 h, and levels of pCHK1 and CHK1 were examined. (E) A549 cells were pre-treated with different concentrations of AT2 for 2 h, added 500 nM CPT for another 4 h, and protein expression was examined using indicated antibodies. Lower: levels of pCHK1 normalized to total CHK1 from 5 independent experiments were used to determine the IC50 of AT2. (F) A549 cells were transfected with siRNA control or targeting the α1 subunit of Na/K-ATPase for 48 h, treated with 500 nM AT2 for 2 h, added 200 nM CPT for another 4 h, and protein expression was examined. Relative pCHK1 levels are shown above.

3.2. Cardiac glycosides inhibit DNA damage response independent of Na⁺/K⁺-ATPase

To confirm the effects of the cardiac glycosides, we chose three structurally representative compounds AT2, AT3 and TP4 (Supplementary Table S1) for further analysis. We found that pre-, co- or even 1 h post-CPT treatment with these compounds inhibited CPT-induced pCHK1 (Fig. 1C), suggesting that the specific order of cardiac glycoside and chemotherapeutic drug treatment is not critical. These agents also inhibited pCHK1 induced by etoposide (ETO) or cisplatin (CIS) (Fig S2A, S2B), two other DNA damaging drugs used clinically to treat KRAS mutant lung cancer. For the following studies, we primarily used AT2 as the tool compound as it is one of the top hits and scale-up purification for in vivo testing was accessible. Various chemotherapy drugs, especially those that cause DSBs, were used to demonstrate the generality of the effects of cardiac glycosides. We found that AT2 inhibited CPT-induced CHK1 phosphorylation at different time points (Fig. 1D) with an IC50 of ~96 nM (Fig. 1E). Treatment with up to 500 nM AT2 did not obviously change cell cycle progression of A549 cells (Fig S2C), indicating that the reduction in pCHK1 level was not due to cell cycle arrest by these compounds.

To date, the only validated cellular target of cardiac glycosides is the membrane Na⁺/K⁺-ATPase, which maintains cellular electrochemical homeostasis [22]. To evaluate if this enzyme was required for the inhibitory effect of AT2 on CHK1 phosphorylation, we depleted the catalytic α1 subunit of the Na⁺/K⁺-ATPase through use of an siRNA and examined CPT-induced pCHK1 levels. We found that greater than 95% depletion of the α1 subunit by two independent siRNAs had almost no effect on the inhibition of pCHK1 by AT2 (Fig. 1F). We observed that one siRNA caused variation in total CHK1 and ATR levels; however, these changes did not affect the normalized pCHK1/CHK1 ratio, and therefore do not alter our conclusion. These results suggest that the inhibitory effects of cardiac glycosides on DNA damage response are largely, if not entirely, independent of the Na⁺/K⁺-ATPase (also confirmed by Fig. 3A below).

Fig. 3.

Cardiac glycosides inhibit DSB 5’ to 3’ end resection. (A) Cells were pre-treated with 500 nM AT2 for 2 h, added different agents (200 nM CPT, 330 nM ETO, 3.3 μM BLE, 1.33 mM HU or 3 μM APH) for another 4 h, and protein levels were determined. Short and long exposures of pCHK1 blots were shown. (B) Representative images of RPA foci in cells treated with 200 nM CPT for 8 h with or without prior AT2 treatment. Scale bar is 10 μm. (C) Cells were pre-treated with 500 nM AT2 for 2 h (–4), added 200 nM CPT for another 2 h (–2), washed off drugs (0) and cultured in drug-free full media for the indicated time points. The X-axis is not scaled. Data represent the number of RPA foci per cell and are presented as mean and standard deviation from at least 50 cells in each group. Two-tailed t-student test was used to determine the statistical significance. *P < 0.001 between CPT + AT2 and CPT alone. (D) Parallel samples from (C) were analyzed for protein expression.

3.3. Cardiac glycosides sensitize KRAS mutant lung cancer to chemotherapy

To further determine the effects of cardiac glycosides on DNA damage response and cell sensitivity to chemotherapeutic agents, we treated KRAS mutant (A549) and wild type (HARA) lung cancer cells with AT2 and ETO and examined pCHK1 levels and long-term cell survival. AT2 inhibited ETO-induced pCHK1 in A549 cells, with a much weaker inhibition in HARA cells (Fig. 2A). Similarly, AT2 significantly enhanced the growth inhibition of ETO in A549 cells compared to HARA cells (Fig. 2B). We observed a low level of pCHK1 in the absence of DNA damage and AT2 reduced this signal (1st and 3rd lanes of Fig. 2A and Fig S2A for A549 cells). This most likely represents a non-specific signal that is frequently seen for this type of phosphor-antibodies. In addition, the signal was not always reduced by AT2 treatment (e.g., 1st and 3rd lanes of Fig. 2A for HARA cells, Fig S2B, Fig. 1F, etc.), further supporting its non-specific nature. Hence, we conclude that AT2 does not affect pCHK1 in the absence of DNA damage.

Fig. 2.

Cardiac glycosides specifically target KRAS mutant lung cancers. (A) HARA or A549 cells were pre-treated with 500 nM AT2 for 2h, added 500 nM ETO for 4 h, and protein levels were analyzed. (B) Clonogenic survival of KRAS wild type and mutant lung cancer cells. Data illustrate the survival rate of cells treated with ETO + AT2 normalized to that of ETO alone, and are presented as mean and standard derivation from five replicates. One-way ANOVA analysis was used to determine the statistical significance between pooled KRAS wild type and mutant cancer cells. (C–F) Xenograft tumor growth from KRAS wild type (H292 and H226) and KRAS mutant (EKVX and A549) cancer cells. Tumor growth is presented as mean and standard deviation from 10 mice per group. Two-tailed t-student test was used to determine the statistical significance. *P < 0.05 and#P < 0.001 between CIS + AT2 and CIS alone at indicated days.

AT2 caused a reduction in survival in KRAS mutant (IC₅₀ ~ 209 nM), but not wild type lung cancer cells (IC₅₀ not attained at 500 nM) after treatment with CPT (Fig S3B), correlating with its inhibition on pCHK1 (Fig S3A). These data in combination with results presented elsewhere (e.g., Fig. 1C and Fig S12) lead us to choose a concentration range of 200–500 nM for studying AT2 after considering both the efficacy (on pCHK1 and cell survival) and the toxicity (cell cycle and cell death) of this compound. On the other hand, a specific CHK1 kinase inhibitor (PF-477736) did not sensitize KRAS wild type or mutant lung cancer cells to CPT (Fig S3A, S3C), suggesting that the differential effect is specific to these cardiac glycosides. The elevated pCHK1 levels seen in CHK1 inhibitor treated cells (Fig S3A) is a known feature of CHK1 inhibition [19], suggesting that the CHK1 inhibitor worked in this experimental setting.

Despite having different mutant alleles (KRAS^G12C, KRAS^G12S, etc.), KRAS mutant cancer cells were generally more sensitive to the combination of AT2 plus chemotherapeutic drugs, including irinotecan (the clinical form of CPT), CIS, ETO and gemcitabine (GEM) than KRAS wild type cancer cells (Fig S4–S5). For these colony formation results, we replated the cells after drug treatment. To evaluate if re-plating would affect cell attachment and therefore cell survival, we compared the effects of CPT and AT2 on A549 colony formation with or without cell re-plating. This experiment yielded similar results for AT2 and CPT in inhibiting the survival of A549 cells (Fig S6A–B), suggesting that cell re-plating does not affect cell survival.

To determine the specificity of AT2 in lung cancer with KRAS mutations, we first overexpressed a KRAS^G12D mutant in KRAS wild type HARA cells (Fig S7A), which was confirmed by increased levels of both KRAS and its downstream target pERK. We found that KRAS^G12D expression significantly enhanced the sensitivity of HARA cells to GEM plus AT2 compared to GEM alone (Fig S7B–C). Second, we used a mixture of two siRNAs to knock down KRAS in a mutant H460 cell line and simultaneously expressed the KRAS^G12D mutant in these cells. We found that knockdown of mutant KRAS significantly mitigated both the pCHK1 reduction (Fig S8A) and the sensitivity (Fig S8B) of H460 cells to the combination of AT2 and GEM compared with parental cells. On the other hand, expression of the KRAS^G12D mutant at least partially restored the inhibition of pCHK1 and the sensitivity of cells to AT2 (Fig S8A–B). Even though the effects of siRNA and KRAS^G12D mutant overexpression were relatively weak, the results are consistent with our conclusion, demonstrating the specificity of cardiac glycosides on KRAS mutant lung cancer cells.

To confirm these cell culture results in vivo, we performed xenograft mouse studies to examine the effects of AT2 and CIS on the growth of KRAS wild type (H292 and H226) and mutant (EKVX and A549) tumors in nude mice. A 4-week administration of AT2 at 0.3 mg/kg (every other day, 14 total doses) did not obviously reduce heart weight or alter heart tissues of mice (Fig S9A–C), suggesting that this dose of AT2 is tolerable to mice. We found that AT2 alone had little to no inhibitory effect on the growth of all tumors tested (Fig. 2C–F). CIS alone inhibited the growth of H292 and A549 tumors, weakly on that of EKVX tumors, but had no effect on H226 tumors (Fig. 2C–F). AT2 only marginally enhanced the growth suppressive effect of CIS on KRAS wild type cancers (H292 and H226) (Fig. 2C and E). However, AT2 greatly increased CIS-induced growth inhibition of KRAS mutant A549, and to a lesser degree, EKVX tumors (Fig. 2D and F). The overall tumor inhibitory effect seems to be modest; however, the increase in tumor inhibition induced by AT2 was statistically significant (Fig. 2D and F). Further, AT2 plus CIS greatly increased cleaved Caspase 3 levels in KRAS mutant cancers (Fig S10), indicating increased cell death in KRAS mutant lung tumors. In conclusion, the significant difference in the effect between KRAS mutant and wild type lung cancers suggest that these cardiac glycosides selectively sensitized KRAS mutant lung cancers to chemotherapy, creating an enhanced therapeutic window that allows specific targeting of this disease.

3.4. Cardiac glycosides inhibit DSB 5’ to 3’ end resection

To determine the molecular mechanisms underlying the inhibitory effect of these cardiac glycosides on the DNA damage response, we first asked if they inhibited the generation of DNA damage by measuring ATM phosphorylation at Ser-1981, a very early marker of the DNA damage response [23]. The results show that AT2 or TP4 did not affect CPT-induced phosphorylation of ATM or its downstream factor CHK2 (Fig S3A and S11A), suggesting that cardiac glycosides did not suppress the generation of de novo DNA damage. We then examined the effect of AT2 on pCHK1 induced by different DNA damaging agents. The results showed that AT2 inhibited pCHK1 induced by CPT, ETO or bleomycin (BLE), but not by hydroxyurea (HU) or aphidicolin (APH) (Fig. 3A). The differential effect of AT2 on different agent-induced pCHK1 again supports the idea that Na⁺/K⁺-ATPase is unlikely to be involved.

HU and APH block DNA replication by inhibiting nucleotide synthesis and DNA polymerase, respectively, resulting in the generation of long stretches of single strand DNA (ssDNA) to activate ATR, the upstream kinase that phosphorylates CHK1 [24] (Fig S11B). On the other hand, CPT, ETO and BLE are known DSB inducers [18,25,26]. CHK1 phosphorylation induced by DSB-causing agents requires DSB 5’ to 3’ end resection to generate ssDNA [19,24](Fig S11C). These lines of information suggest that it is unlikely that cardiac glycosides are direct ATR inhibitors, which would block CHK1 phosphorylation induced by all agents. To further test this idea and to determine if AT2 could also sensitize cells to HU, we compared the effects of the combination of AT2 with HU or CPT on A549 cell growth. Again, we found that AT2 reduced CPT-induced pCHK1, but had almost no effect on HU-induced CHK1 phosphorylation (Fig S12A). When we examined cell survival using a clonogenic assay, we found that AT2 and CPT further inhibited cancer cell growth compared with either one alone (Fig S12B–C). However, HU at the dose we used did not induce any cell growth inhibition, and its combination with AT2 did not further reduce cell growth compared with AT2 alone. These data suggest that cardiac glycosides did not synergize with HU.

The fact that AT2 specifically inhibited pCHK1 induced by DSB-causing agents, but not by replicative stressors, suggests that it may inhibit DSB 5’ to 3’ end resection. To test this idea, we indirectly measured the generation of ssDNA, the product of DSB 5’ to 3’ end resection. Once generated, ssDNA will be bound by the ssDNA-binding protein RPA (Fig S11C) [24], which when immunostained with anti-RPA antibodies will appear as bright dots (so-called foci, Fig. 3B) under fluorescence microscopy. We found that AT2 significantly reduced CPT-induced RPA foci over time (Fig. 3C), consistent with lower pCHK1 levels (Fig. 3D).

DSB 5’ to 3’ end resection not only promotes high level CHK1 phosphorylation, but also provides a platform for DSB repair by homologous recombination (HR). Inhibition of DSB 5’ to 3’ end resection will promote the non-homologous end joining (NHEJ) route of DBS repair [27,28]. If our hypothesis that cardiac glycosides inhibit DSB 5’ to 3’ end resection is correct, we would then predict that cardiac glycosides should inhibit HR while promoting NHEJ for DSB repair. To this end, we performed fluorescence based DSB repair assays, as we previously reported [29]. We found that indeed treatment with AT2, AT3 or TP4 increased the basal level of NHEJ while inhibiting that of HR (Fig S11D–E). These data support the idea that cardiac glycosides inhibit the DSB 5’ to 3’ end resection for DSB repair.

3.5. UHRF1 as a cellular target inhibited by cardiac glycosides

DSB 5’ to 3’ end resection is controlled by two mutually inhibitory protein complexes centered around 53BP1 and BRCA1 [30]. 53BP1 and its partner proteins including RIF1 inhibit 5’ to 3’ end resection and promote NHEJ in G1 phase, whereas BRCA1 and the associated E3 ligase UHRF1 favor 5’ to 3’ end resection and HR repair in S and G2 phases [30, 31]. Hence, we hypothesize that cardiac glycosides inhibit DSB 5’ to 3’ end resection through regulating expression of genes involved in this process.

To test this idea, we performed stable isotope labeling with amino acid (SILAC) analysis (Fig. 4A) and found that the protein level of UHRF1 was reduced the most among known DSB repair genes detected (Supplementary Table S2 and Fig. 4B). UHRF1 was originally identified as a mediator of DNMT1 to restore DNA methylation after replication [32,33]. Recent evidence suggests an important role for UHRF1 in promoting DSB 5’ to 3’ end resection and favoring HR-dependent repair of DNA interstrand crosslinks and DSBs [31,34,35]. Our findings are in line with these reports, suggesting a potential role for UHRF1 in the AT2-inhibited DSB 5’ to 3’ end resection.

Fig. 4.

Cardiac glycosides inhibit UHRF1 expression. (A) Schematic diagram of SILAC analysis to identify AT2 targets. (B) Summary of gene expression by AT2 treatment. Data represent the ratio of heavy/light peptides from control and AT2 in log 2 format. (C) qPCR of UHRF1 mRNA levels from A549 cells treated with 500 nM AT2 for 0, 2, 4, 8 and 12 h. Data represent mean and standard deviation in triplicate. (D) A549 cells were pre-treated with 500 nM AT2 for 2 h, added 200 nM CPT for 0, 1, 2, 4 and 8 h, and protein expression was analyzed using specific antibodies. Arrows indicate the corresponding proteins.

To determine how cardiac glycosides inhibit UHRF1 expression, we first performed quantitative PCR (qPCR) to evaluate its transcription. AT2 gradually reduced the mRNA level of UHRF1 over time (Fig. 4C), indicating transcriptional suppression. Second, we found that AT2, but not CPT, induced a time-dependent reduction in the protein level of UHRF1 with a half-life of ~4 h (Fig. 4D). Importantly, this reduction correlated with the inhibition of pCHK1 and DSB 5’ to 3’ end resection (Fig. 4D). No obvious changes were observed in protein levels of other DSB repair genes including the upstream factors MDC1, RNF8 and RNF168 (Fig. 4D and Fig S13), consistent with the SILAC results. AT2 also reduced the protein levels of P53 and P21 (Fig. 4D and Fig S13), similar to previous reports [36,37]. Because AT2 increased the cytotoxic effect of chemotherapeutic drugs, these results indicate that the reduction in P53 did not provide cancer cells a survival advantage in the presence of cardiac glycosides.

Since the reduction in the protein level of UHRF1 occurred earlier than that in its mRNA (Fig. 4C and D), we examined UHRF1 protein stability by treating cells with the protein synthesis inhibitor cycloheximide (CHX). The results show that AT2 time dependently inhibited UHRF1 levels with the kinetics similar to that of CHX (Fig S14A). Further, co-treatment with the proteasome inhibitor MG132 completely blocked AT2-induced UHRF1 degradation (Fig S14B). These results suggest that AT2 also inhibited UHRF1 through proteasome-dependent protein degradation.

3.6. UHRF1 is critical for the effects of cardiac glycosides

To confirm the importance of UHRF1 in the inhibitory effects of cardiac glycosides on DNA damage response and cell survival, we first asked if overexpressing UHRF1 could rescue AT2-inhibited CHK1 phosphorylation. CPT treatment induced pCHK1 foci and increased cellular pCHK1 intensity (Fig. 5A and B), which were suppressed by AT2 in control cells expressing mCherry (Fig. 5A and B), consistent with Western blot results. However, cells overexpressing mCherry-UHRF1 largely restored both the foci and the level of pCHK1 inhibited by AT2 (Fig. 5A and B), indicating the rescue of the DNA damage response. Similarly, we found that overexpression of UHRF1 at least partially rescued RPA foci inhibited by AT2 (Fig S15A–B).

Fig. 5.

Role of UHRF1 in the inhibitory effects of cardiac glycosides. (A) A549 cells were transfected with mCherry control vector or mCherry-UHRF1 for 48 h, pretreated with 500 nM AT2 for 2 h, added 200 nM CPT for another 8 h, fixed and stained with specific antibodies against pCHK1. Representative images are shown and arrows indicate cells expressing mCherry-UHRF1. Scale bar is 10 μm. (B) The cellular fluorescence intensity of pCHK1 in (A) was quantitated by Photoshop. Each dot represents one cell that expressed mCherry-UHRF1 (low and high expression combined). Two-tailed t-student test was used to determine the statistical significance. (C) KRAS mutant EKVX cells were transfected with siRNA for control or UHFR1 for 48 h, pre-treated with 500 AT2 for 2 h, then added 500 nM gemcitabine (GEM) for another 4 h, and protein expression was analyzed by Western blotting using specific antibodies. (D) Clonogenic survival results of cells in C. Numbers represent mean and standard deviation from 3 replicates. *P < 0.001.

Then we asked if depleting UHRF1 in KRAS mutant cells could mitigate the effect of cardiac glycosides. To this end, we used siRNAs against UHRF1 to reduce its protein level in KRAS mutant EKVX cells and found that depletion of UHRF1 blocked the AT2-induced reduction in pCHK1 in the presence of GEM (Fig. 5C). We then examined cellular sensitivity to cardiac glycosides plus GEM. The results show that AT2 sensitized EKVX cells to the effect of GEM in siRNA control cells (Fig. 5D). Knockdown of UHRF1 alone also increased cellular sensitivity to GEM compared to control siRNA (Fig. 5D); however, AT2 failed to further increase the sensitivity of UHRF1 depleted cells to GEM (Fig. 5D). Together, these biochemical and genetic results confirm UHRF1 as at least one of key cellular targets that contribute to the DNA damage response inhibitory effects of cardiac glycosides.

3.7. UHRF1 expression is controlled by ERK, which is specifically inhibited by AT2

To understand how UHFR1 influences the selective sensitivity of KRAS mutant cells to chemotherapy and cardiac glycosides, we analyzed the Cancer Genome Atlas dataset and found a positive correlation between UHRF1 and the KRAS levels in human tumors with a P value of 0.052323 (Fig. 6A). These results are similar to a previous report showing that UHRF1 levels were nearly 2 times lower in RAS wild type cancers than mutant cancers [38]. Further, analysis of the TCGA Pan-Cancer Atlas lung and colon cancer dataset revealed that while KRAS mutant lung or colon cancers showed normal expression levels of UHRF1, KRAS wild type lung or colon cancers tend to have significantly lower levels of UHRF1 (Fig S16A–B). These analyses indicate that KRAS mutant cancers tend to express higher levels of UHRF1 than KRAS wild type cancers. Interestingly, AT2 reduced the UHRF1 mRNA level to a greater extent in KRAS mutant lung cancer cells than in wild type cancer cells (Fig. 6B), suggesting that mutated KRAS signaling may regulate UHRF1 expression.

Fig. 6.

Elevated ERK regulates UHRF1 levels. (A) Correlation between KRAS and UHRF1 in human tumors from TCGA. Correlation was estimated using proportion log2-cancer/log2-normal fold changes. Pearson correlation was then performed to determine the agreement in gene expression. KRAS and UHRF1 gene expression was evaluated from diverse tumor cancer tissues including Adipose, Bladder, Blood, Brain, Breast, Cervix, Colon, Endometrium, Head Neck, Kidney, Liver, Lung, Ovary, Prostate, stomach, thyroid, Vagin and Vulva. (B) KRAS wild type (H292) or mutant (HOP62) cells were treated with 500 nM AT2 for 0 or 8 h, and UHRF1 mRNA levels were determined by qPCR. Data represent mean and standard deviation in triplicate. Two-tailed t-student test p < 0.001. (C–D) KRAS wild type (HARA and H292) and mutant (EKVX and HOP62) cells were treated with 500 AT2 for 0, 4 and 8 h, protein expression was analyzed by Western blotting using specific antibodies. Numbers represent the relative protein band intensity of UHRF1.

Recent studies demonstrated that the greatest signaling change in both mouse and human KRAS mutant lung cancers is the upregulation of the downstream MEK/ERK pathway [39]. Consistent with this idea, we observed that KRAS mutant lung cancer cells exhibited much higher basal level of pERK than wild type cancer cells (Fig. 6C and D). Remarkably, AT2 time-dependently inhibited the pERK in KRAS mutant lung cancer cells (Fig. 6C and D). In parallel, despite the observation that different cell lines express different levels of UHRF1, AT2 induced downregulation of UHRF1 to a greater extent in KRAS mutant cells than in KRAS wild type cancer cells (Fig. 6C and D). We also observed that the effect of AT2 on pERK and UHRF1 varied slightly among different KRAS mutant cancer cell lines (Fig S17, A549 vs EKVX). This difference in effect is consistent with the response of these cancer cell lines to cardiac glycosides in mouse xenograft studies. However, we confirmed that AT2 selectively inhibited UHRF1 and pERK levels in KRAS mutant lung cancers in mice (Fig S18).

KRAS/ERK signaling often intersects with other pathways, especially the PI3K/mTOR pathway (e.g., in A549 cells), which may compromise the effects of cardiac glycosides. We found that AT2 induced a nearly equal reduction in the level of p-4EBP1, a critical downstream factor of the PI3K/mTOR pathway, in KRAS wild type and mutant cancer cells (Figs. 5C, 6C and 6D). These results indicate that cardiac glycosides did not differentially inhibit the PI3K/mTOR pathway. Together, these data suggest that (1) UHRF1 expression is controlled, at least partially, by elevated ERK signaling, and (2) cardiac glycosides inhibit expression of UHRF1 by selectively suppressing the elevated ERK signaling in KRAS mutant lung cancer cells.

To further determine the role of ERK in UHRF1 expression, we treated KRAS wild type and mutant lung cancer cells with increasing concentrations of a specific ERK inhibitor, SCH77298. The results showed that SCH77298 dramatically inhibited the elevated pERK in KRAS mutant cells while it actually increased pERK in KRAS wild type cells (Fig S19A). Similarly, SCH77298 reduced the protein level of UHRF1 in KRAS mutant, but not wild type cancer cells (Fig S19A). Consistently, SCH77298 significantly inhibited the survival of KRAS mutant cancer cells in a dose-dependent manner (Fig S19B). We then determined that inhibition of ERK selectively reduced both pCHK1 and UHRF1 expression levels (Fig S20A), and increased cell sensitivity to GEM (Fig S20B) in KRAS mutant, but not wild type lung cancer cells.

4. Discussion

KRAS mutant non-small cell lung cancer is a subtype of lung cancer in which fewer targeted therapies are available. Although a number of strategies, for instance anti-PD-L1 based immunotherapy and specific molecules targeting the mutagenic KRAS^G12C, have been pursued or are under development, the long-term response rate is poor or limited to a small percentage of patients [2,13]. Therefore, there remains an unmet clinical need to develop effective strategies to treat KRAS mutant lung cancer. Through compound library screening, we identified cardiac glycosides as potent DSB repair inhibitors that specifically sensitized KRAS mutant lung cancer cells to chemotherapeutic drugs. We reveal UHRF1 as a critical target that regulates the effects of cardiac glycosides. UHRF1 has been implicated in repairing both DNA interstrand crosslinks caused by cisplatin and DSBs induced by agents such as ETO and CPT [31,34,35]. We demonstrate that UHRF1 expression is controlled by the elevated ERK signaling driven by oncogenic KRAS mutation.

Both UHRF1 and ERK regulate cell growth and proliferation. However, we showed that cardiac glycosides, which inhibited both UHRF1 and pERK, did not affect cell cycle progression at least under the particular treatment conditions in this study. We reason that this is likely because cardiac glycosides never completely inhibited UHRF1 and pERK; therefore, the remaining levels of UHRF1 and pERK might be sufficient to promote cell proliferation. However, given the specific role of UHRF1 in promoting DSB repair, inhibition of UHRF1, even partially, could block faithful DSB repair, selectively sensitizing KRAS mutant lung cancer cells to chemotherapy. In contrast, because ERK is involved in a wide range of cellular activities including normal signaling, targeting ERK should have more effects than just on DSB repair, which could consequently produce unintended effects. Further, it has been shown that long-term ERK inhibition eventually causes rebound of RAF-MEK-ERK signaling, counteracting the effects of ERK inhibitors. Therefore, targeting UHRF1 could be more specific than targeting ERK in sensitizing KRAS mutant lung cancers to chemotherapy (see model in Fig S21). Since KRAS mutations are also found in other types of cancer, we expect that other types of cancers with KRAS mutations might respond similarly to cardiac glycosides.

Our model and our data suggest that ERK inhibitors should also sensitize KRAS mutant lung cancers to chemotherapy, although the effect of the ERK inhibitor was weaker than cardiac glycosides in terms of UHRF1 reduction and survival inhibition (e.g., Fig S19 and S20), indicating that other pathways such as topoisomerase 2 [40] and HIF1α [41] might also be involved. A number of MEK/ERK inhibitors have been developed and tested in KRAS mutant cancers as single agents or in combination with other agents, such as EGFR inhibitors [14]. Surprisingly, few clinical trials have been designed to test the combination of MEK/ERK inhibitors with DSB-inducing chemotherapy to treat KRAS mutant lung cancer patients. A clinical trial that tests the combination of MEK inhibitors, such as trametinib, with gemcitabine or radiation has not reached any conclusion yet [14](clinicaltrial.gov). Our findings may lead to the design of clinical trials that combine cardiac glycosides (like digoxin) or MEK/ERK inhibitors with DSB-inducing chemotherapies to treat KRAS mutant lung cancers in the near term.

Cardiac glycosides, best known for treating cardiac arrhythmias, elicit various biological effects including anticancer activity [22]. However, the lack of detailed mechanistic understanding of their anticancer activity and associated cardiac toxicity have limited their development as anticancer drugs. A recent study identified several cardiac glycosides as DSB repair inhibitors [42], similar to our observation. However, no specific mechanisms were revealed as to how cardiac glycosides inhibit DSB repair. Here, we demonstrate that cardiac glycosides inhibit the DSB 5’ to 3’ end resection process independent of the canonical cellular target, the Na⁺/K⁺-ATPase. We reveal UHRF1 as an important mediator for the effects of cardiac glycosides. The specific inhibition of UHRF1 by cardiac glycosides in KRAS mutant lung cancer may create a therapeutic window to target this subtype of cancer. Further, these results support for future reverse engineering to synthesize cardiac glycoside analogs with augmented inhibitory effect on DSB repair while reducing their internal cardiac toxicity. These compounds should greatly sensitize tumors to chemotherapy while reducing side effects. In addition, this study also lays the groundwork for future development of more potent and specific UHRF1 inhibitors to selectively treat KRAS mutant lung cancer.

Abbreviations:

APH

Aphidicolin

AT

Antiaris toxicaria

ATR

ATM and Rad 3 related

BLE

Bleomycin

CHK1

Checkpoint kinase 1

CIS

Cisplatin

CPT

Camp-tothecin

DSB

Double strand break

ERK

Extracellular signal-regulated kinase

ETO

Etoposide

GEM

Gemcitabine

HR

homologous recombination

HU

Hydroxyurea

KRAS

Kirsten rat sarcoma virus oncogene homolog

NHEJ

non-homologous end joining

SILAC

Stable isotope labeling with amino acid

ssDNA

Single strand DNA

TP

Thevetia peruviana

UHRF1

Ubiquitin like with PHD and Ring Finger Domains 1

WT

Wild type