|Home | About | Journals | Submit | Contact Us | Français|
Tyrosine kinase inhibitors (TKIs) revolutionized the treatment of CML-CP. Unfortunately, 25% of TKI-naive patients and 50–90% of TKI-responding patients carry CML clones expressing TKI resistant BCR-ABL1 kinase mutants. We reported that CML-CP leukemia stem and progenitor cell populations accumulate high amounts of reactive oxygen species (ROS), which may result in accumulation of uracil derivatives in genomic DNA. Unfaithful and/or inefficient repair of these lesions generates TKI resistant point mutations in BCR-ABL1 kinase. Using an array of specific substrates and inhibitors/blocking antibodies we found that uracil-DNA glycosylase UNG2 were inhibited in BCR-ABL1 –transformed cell lines and CD34+ CML cells. The inhibitory effect was not accompanied by downregulation of nuclear expression and/or chromatin association of UNG2. The effect was BCR-ABL1 kinase-specific because several other fusion tyrosine kinases did not reduce UNG2 activity. Using UNG2-specific inhibitor UGI we found that reduction of UNG2 activity increased the number of uracil derivatives in genomic DNA detected by modified comet assay and facilitated accumulation of ouabain-resistant point mutations in reporter gene Na+/K+ATPase. In conclusion, we postulate that BCR-ABL1 kinase-mediated inhibition of UNG2 contributes to accumulation of point mutations responsible for TKI-resistance causing the disease relapse, and perhaps also other point mutations facilitating malignant progression of CML.
Chronic myeloid leukemia in chronic phase (CML-CP) is initiated by t(9;22) encoding for BCR-ABL1 tyrosine kinase that transforms hematopoietic stem cells (HSCs) (1). CML-CP is leukemia stem cells (LSCs) -derived disease, but deregulated growth of LSCs-derived leukemia progenitor cells (LPCs) leads to the manifestation of the disease (2). Most CML-CP patients are currently treated with tyrosine kinase inhibitors (TKIs) such as imatinib, dasatinib and nilotinib, which induce complete cytogenetic response (CCyR) or complete molecular response (CMR) in 60–70% and only 8% of cases, respectively (3, 4). However, it is unlikely that TKIs will “cure” CML because CML-CP cells are elusive targets even for the most advanced therapies employing second- and third-generation TKIs (5). One of the major reasons of TKI-resistance are BCR-ABL1 tyrosine kinase mutations (6, 7). These TKI-resistant BCR-ABL1 forms usually result from point mutations in the fragment encoding the kinase domain. In addition accumulation of point mutations in other genes (e.g., p53, Ras, p16) may lead to malignant progression of the disease to the accelerated phase (CML-AP) and blast phase (CML-BP) (8). Point mutations in BCR-ABL1 and chromosomal aberrations affecting different genes have been detected in Lin-CD34+CD38− LSCs and Lin-CD34+CD38+ LPCs (9) and in freshly established 32D-BCR-ABL1 cells (10). This observation supports the notion that genomic instability in CML stem/progenitor cells is responsible for accumulation of point mutations in BCR-ABL1 kinase causing the initial and persistent resistance to TKIs.
Point mutations may result from enhanced oxidative DNA damage and/or deregulated mechanisms of DNA repair (11, 12). Oxidative DNA damage arises from reactive oxygen species (ROS) (13). The most important primary ROS is superoxide radical anion (•O2−), which may lead to the production of hydrogen peroxide (H2O2) and hydroxyl radical (•OH) (14). ROS can damage DNA bases to produce numerous oxo-derivatives, among the most frequent are 5-hydroxy-cytosine (5-OH-C) generating 5-hydroxy-uracil (5-OH-U) (14). Oxidative deamination of cytosine generates highly mutagenic U:G mismatches which, if not repaired, leads to G:C → A:T transitions, the most frequent point mutations in human tumors (15). The potential role of ROS-induced oxidative DNA damage in accumulation of point mutations in CML-CP cells was highlighted by our previous studies showing that BCR-ABL1-transformed cell lines, and CML-CP LSCs and also LPCs contained 2–6 times more ROS and oxidized bases in comparison to their normal counterparts (16).
Oxidized bases in DNA are repaired by the base excision repair (BER) pathway (17). First step in this process is catalyzed by DNA glycosylases which recognize and excise damaged DNA base leaving an intact abasic site (AP site) followed by its removal by AP endonuclease (APE1) or lyase activity of bifunctional DNA glycosylases. Then, after the cleavage of the phosphodiester bond, BER may be proceed by “short patch” (characterized by the insertion of one nucleotide by polymerase β) or by “long patch” (characterized by the insertion of several nucleotides by polymerase β or the replicative polymerases δ/ε). Finally DNA ends are sealed by DNA ligase.
The following glycosylases are associated with the removal of uracil and its derivatives from DNA: nuclear uracil DNA glycosylase (UNG2) and single-strand selective monofunctional uracil DNA glycosylase (SMUG1) – remove uracil and 5-OH-U (18, 19), a homolog of Escherichia coli endonuclease III (NTH1) and a homolog of Escherichia coli endonuclease VIII-like 1 (NEIL1) – remove 5-OH-U, 5-OH-C, 5,6-OH-U (20); a homolog of Escherichia coli endonuclease VIII-like 2 (NEIL2) – removes 5-OH-U and 5-OH-C (21); finally, thymine glycosylase (TDG) and methyl-CpG-binding domain protein 4 (MBD4) – remove U opposite G (22). Among the above glycosylases, UNG2 has by far the most prevalent activity in mammalian extracts (23).
In some cases, deficiencies in BER pathway can generate DNA mismatches which are repaired by mismatch repair (MMR) mechanisms to prevent point mutations (24). Since we found that MMR is inhibited in CML-CP (25), the activity of BER is critical for prevention of accumulation of point mutations. Here we report that BCR-ABL1 kinase inhibits the activity of uracil DNA glycosylase UNG2 and propose that this reduced activity may contribute to accumulation of point mutations in CML-CP cells.
Murine 32Dcl3 and human M07 parental hematopoietic cell lines, and their counterparts transformed with p210BCR-ABL1 were used before (26–28) and maintained in Iscove’s modified defined medium (IMDM) supplemented with 1 mM glutamine, 10% fetal bovine serum (FBS) and interleukin 3 (IL-3) -conditioned medium (murine cells) or stem cell factor (SCF) –conditioned medium and recombinant granulocyte-macrophage colony-stimulating factor (GM-CSF) (PeproTech Inc., Rocky Hill, NJ) (human cells) in the concentrations necessary to support parental cells proliferation. CML patient cells were obtained from the Stem Cell and Leukemia Core Facility of the University of Pennsylvania, Philadelphia, PA, USA, after receiving informed consent. CD34+ CML cells were obtained using humanCD34+ selection cocktail (StemCell Technologies, Inc., Vancouver, BC, Canada). CD34+ cells from healthy volunteers (NBMC) were obtained from Cambrex Bio Science Walkersville (Walkersville, MD). The research activities involving human samples were approved by the Institutional Review Board. When indicated cells were treated with 1μM imatinib for 24 hrs.
Nuclear cell lysates were obtained as described before (29). In order to obtain nuclear soluble and chromatin-bound protein fractions cells were processed using the Subcellular Protein Fractionation kit (Thermo Scientific, Rockford, IL) following manufacturer recommended procedures. Cell lysates were resolved by SDS-PAGE and blotted with primary antibodies recognizing UNG, SMUG1 and Lamin (Santa Cruz Biotechnology, Santa Cruz, CA), APE-1, OGG1, MBD4, TDG and HDAC II (Abcam, Cambridge, MA), NEIL1 and NEIL2 (Oncogene Research Products, Cambridge, MA), NTH1 (Novus Biologicals, Littleton, CO), following by secondary antibodies conjugated with horseradish peroxidase.
The following oligonucleotides (The Midland Certified Reagent Company, Inc., Midland, TX) were used as substrates:
Oligonucleotides marked by asterisk were radiolabeled with [γ32P]ATP using T4 polynucleotide kinase (Promega, Madison, WI). When indicated radiolabeled oligonucleotides were annealed to a complimentary strand by heating to 95°C followed by slow cooling to 4°C to obtain double-stranded substrates.
Nuclear cell lysates were prepared as described before (30). Briefly, cells were suspended in 10mM Hepes buffer, pH 7.9, containing 1.5mM MgCl2, 10mM KCl, 0.5mM DTT and protease inhibitors on ice and cells were lysed by 10 strokes of Dounce homogenizer. The pellet was suspended in 20mM Hepes, pH 7.9 containing 25% glycerol, 0.42M NaCl, 2mM MgCl2, 0.2mM EDTA, 0.5mM PMSF, 0.5mM DTT and protease inhibitors, incubated on ice for 10 min and sonicated. Supernatants containing nuclear extracts were collected after centrifugation for 15 min, 15,000 × g at 4°C, and stored in −80°C.
5-OH-U incision reaction was performed following the protocol described by Kavli et al. with modifications (31). Briefly, 10μg of the nuclear extract proteins were incubated in 37°C with 50fM of the [32P] radiolabeled single-stranded or double-stranded substrate containing 5-OH-U or control in the reaction buffer (20mM Tris-HCl, pH 8.0, 10mM NaCl, 1mM EDTA, 1mM DTT, 0.5mg/ml BSA, 2mM ATP, 40mM creatine phosphate, 2U/μl creatine phosphokinase). Neutralizing antibodies (2μg) and 2U of uracil glycosylase inhibitor (UGI) (New England Biolabs, Inc., MA, USA) (32) were added when indicated. The reactions were stopped by addition of equal volume of the solution containing 80% formamide, 1×TBE and 10% bromophenol blue followed by heating in 90°C for 2 min. Reactions were separated by electrophoresis in 18% polyacrylamide gel containing 8M urea and visualized by autoradiography. The products were quantified using Quantity One software (Bio-Rad, Hercules, CA, USA).
To assess the level of uracil in DNA, the modified comet assay was conducted as described previously (33). Briefly, cells (approximately 5 × 104 per slide) were centrifuged and pellets were gently mixed with 50 μl of 0.75% low melting point agarose in PBS and spread onto a microscope slides pre-coated with 1% agarose. After solidifying, the slides were treated with chilled lysis buffer containing 10 mM TRIS, 1% Triton X-100, 2.5 M NaCl, and 0.1 M EDTA Na2 (pH 10.0) for 60 min at 4°C, followed by three times washing with enzyme incubation buffer (60 mM Tris-HCl, 1 mM EDTA, 0.1 mg/ml bovine serum albumin; pH 8) at room temperature. A 50 μl aliquot of UNG2 (1U per gel; Roche Applied Science, Mannheim, Germany) or buffer alone, as control was subsequently placed onto gel surface and covered with a cover slip. Enzyme-treated samples and controls were incubated in a moist chamber at 37°C for 60 min. Following incubation and removal of the cover slip, the slides were placed in an electrophoresis tank. DNA was allowed to unwind for 20 minutes in the electrophoresis buffer consisting of 300 mM NaOH and 1 mM EDTA, pH > 13. Electrophoresis was conducted in the same buffer at 4°C for 20 minutes at 0.73 V/cm (300 mA). The slides were then washed in water, drained and stained with 1 μg/ml DAPI and covered with cover slips. The slides were examined at 200 × magnification in an Eclipse fluorescence microscope (Nikon, Tokyo, Japan) attached to COHU 4910 video camera (Cohu, San Diego, CA) equipped with a UV-1 filter block consisting an excitation filter (359 nm) and a barrier filter (461 nm) and connected to a personal computer-based image analysis system, Lucia-Comet v. 5.41 (Laboratory Imaging, Praha, Czech Republic). Fifty images were randomly selected from each sample and the percentage of DNA in the tail of comets (% tail DNA) was measured. The mean value of the % tail DNA in a particular sample was taken as an index of the DNA damage in this sample. All experiments were performed in triplicate. The results obtained for UNG2 were normalized by subtracting the level of DNA damage observed for the enzyme buffer alone.
Na+/K+ ATPase mutagenesis leading to ouabain resistance was examined as described before (34) with modifications. Briefly, 32Dcl3 cells were infected with UGI-expressing retrovirus carrying IRES-GFP (DNA construct kindly provided by Dr. CJ. Jolly, The University of Sydney, Australia) (35) or with IRES-GFP only. GFP+ cells were sorted and colonies growing in the presence of ouabain (concentration) were counted in methylcellulose.
DNA double-strand breaks were examined by immunofluorescent detection of histone γ-H2AX as described before (16) with modifications. Briefly, cells were fixed in 1.5% formaldehyde for 10min at room temperature and permeabilized with cold methanol for 20min at 4°C. Then the cells were washed twice by adding PBS supplemented with 0.5% BSA followed by incubation with anti- γ-H2AX antibody conjugated to Alexa Fluor 647 (Cell Signaling Technology, Danvers, MA) for 1hour. After washing the cells were analyzed by flow cytometry.
To determine the expression status of all uracil-DNA glycosylases in BCR-ABL1 positive cells we performed Western blot analysis of nuclear fractions obtained from 32D-BCR-ABL1 cell line and its parental counterpart. We found that BCR-ABL1 kinase elevates the nuclear expression of UNG2 by approximately 2-fold, whereas expression of APE1 endonuclease, which provides the lyase activity for monofunctional UNG2 was not affected (Figure 1a). Elevated nuclear levels of UNG2 were associated with its enhanced chromatin binding in 32D-BCR-ABL1 cells (Figure 1b). Nuclear expression of other glycosylases capable to recognize uracil derivatives in genomic DNA, such as SMUG1, NTH1, NEIL1, NEIL2, TDG and MBD4 were not changed in BCR-ABL1 (Figure 1c).
To examine the activity of BER responsible for removal of potentially mutagenic uracil derivatives from DNA, we analyzed the excision of 5-OH-U from single-stranded and double-stranded DNA substrates (36). Despite the enhanced expression levels of UNG2 by BCR-ABL1, nuclear protein extracts derived from human CD34+ CML-CP, CML-AP and CML-BP cells displayed 3–8 fold reduction of 5-OH-U removal from a single-stranded DNA substrates when compared to CD34+ cells obtained from healthy donors; the excision capability diminished as the disease progressed (Figure 2a). 32D-BCR- ABL1 murine hematopoietic cell line displayed similar reduction of 5-OH-U excision when compared to parental 32Dcl3 counterpart (Figure 2b). In addition, approximately 2-fold reduction in 5-OH-U removal by 32D-BCR-ABL1 nuclear cell lysates was observed when double-stranded DNA substrates containing 5-OH-U: A or 5-OH-U: G were used (Figure 2cd). The presence of BCR-ABL1 inhibited 5-OH-U incisions in a time-dependent manner (Figure 3).
Interestingly, other FTKs such as TEL-ABL1, TEL-PDGFRβ and NPM-ALK were not associated with deregulation of 5-OH-U removal in BaF3 murine hematopoietic cells (Supplementary Figure 1), suggesting that this phenomenon is unique for CML cells.
Incubation of BCR-ABL1-positive cells with imatinib was able to restore the 5-OH-U incision activity indicating that it depends on the kinase activity of BCR-ABL1 (Figure 4a). To identify the uracil DNA glycosylase(s) inhibited by BCR-ABL1 kinase we applied neutralizing antibodies against a number of DNA glycosylases as described by others (37, 38). We found that only UNG2 neutralizing antibody abrogated 5-OH-U incision activity in parental 32Dcl3 cells (Figure 4b). This observation indicated that UNG2 is the major uracil DNA glycosylase in hematopoietic cells and suggested that UNG2 is affected by BCR-ABL1 kinase. To confirm this speculation we showed that anti-UNG2 neutralizing antibody and UGI, a specific peptide inhibitor of UNG2 (dissociates UNG2 from DNA), abrogated 5-OH-U incision activity, which was restored in imatinib-treated 32D-BCR-ABL1 cells (Figure 4).
The observation that BCR-ABL1 kinase significantly reduce the removal of uracil derivatives from DNA through inhibition of UNG2 led us to hypothesize that accumulation of uracil in genomic DNA could be responsible for genomic instability of CML cells. To test this hypothesis and mimic the situation present in CML cells, we generated 32D-UGI cells using UGI-expressing retrovirus carrying IRES-GFP; control cells expressed GFP only. Modified comet assay revealed that 32D-UGI cells displayed approximately 8-times higher levels of uracil derivatives in genomic DNA; conversely accumulation of DNA double-strand breaks (DSBs) measured by immunofluorescence of histone γ-H2AX was reduced in comparison to control cells (Figure 5a and Figure 5b, respectively).
Since oxidative DNA damage is a primary cause of spontaneous mutations, we tested whether reduced UNG2 activity may contribute to accumulation of mutations. The Na+/K+ ATPase mutagenesis model was employed where specific amino acid substitutions in the α1 subunit of the enzyme cause resistance to ouabain (34). We detected a modest, but statistically significant increase of mutations in 32D-UGI cells when compared to parental cells.
Pre-existing and acquired point mutations in BCR-ABL1 kinase are detected in approximately 23% of imatinib-naive and 50–90% of imatinib-treated CML-CP patients (6). Moreover, BCR-ABL1 kinase mutants resistant to second and third generation TKIs emerged, for example due to new and/or compound mutations (39). Thus, CML cells are elusive targets due to continuous accumulation of point mutations in BCR-ABL1 kinase, which encode the resistance to next generations of TKIs (5). The mechanisms responsible for the appearance of TKI-resistant BCR-ABL1 kinase mutants have not been fully characterized.
Point mutations are usually acquired due to the activity of unfaithful DNA polymerases during DNA replication and/or as a result of unfaithful/inefficient repair of oxidized DNA bases. We, and others have shown that CML-CP LSCs (including quiescent LSCs) and LPCs and also BCR-ABL1 –positive cell lines accumulate high levels of ROS-induced oxidized DNA bases, which are responsible for accumulation of TKI-resistant mutations (16, 34, 40, 41). Oxidized bases are excised during BER and, if not removed, MMR corrects the mismatches to avoid point mutations (24). Since we reported that MMR is inhibited in CML (25), BER become an essential mechanism to prevent accumulation of point mutations.
We found that the activity of UNG2, a major uracil DNA glycosylase is inhibited by BCR-ABL1 kinase in CML primary cells and BCR-ABL1 –transformed cell lines. The effect is highly specific for BCR-ABL1 kinase, because other fusion tyrosine kinases including TEL-ABL1 did not affect 5-OH-U incision (Supplementary Figure 1). Downregulation of UNG2 activity resulted in elevation of uracil derivatives in genomic DNA and accumulation of point mutations.
The molecular mechanism of BCR-ABL1 kinase –mediated inhibition of UNG2 is not known. Nuclear expression and chromatin localization of UNG2 is even increased in BCR-ABL1 –transformed cells, the latter observation may be due to prolonged interaction of an “inactive” UNG2 with DNA substrate. Although mass-spec analysis revealed that ABL1 kinase phosphorylates UNG2 on tyrosine 8 in in vitro kinase assay, the phosphorylation was not reproducibly detectable in vivo in CML-CP cells (data not shown). Thus, is it rather unlikely that BCR-ABL1 –mediated tyrosine phosphorylation of UNG2 inhibits its glycosylase activity. In conclusion, BCR-ABL1 kinase-dependent inhibitory mechanism on UNG2 remains to be elucidated.
Inhibition of UNG2 activity is usually associated with G:C→A:T transitions (42), often detected in TKI-resistant BCR-ABL1 kinase mutants (43–45), thus supporting the role of insufficient UNG2 activity in accumulation of TKI-resistant BCR-ABL1 kinase mutations. At this time we cannot exclude the possibility that other glycosylases may be also affected in CML cells. For example, the activity of 8-oxoguanine glycosylase (OGG1) responsible for removal of 8-oxoG lesions from genomic DNA appears inhibited in CML cells and BCR-ABL1 –transformed cells in the kinase-dependent manner (Supplementary Figure 2) leading to G:C→T:A transversions. Since the inhibitory effect on UNG2 and OGG1 depends on BCR-ABL1 kinase, inefficient BER may contribute to acquisition of point mutations in TKI-naive CML cells (46), in cells carrying TKI-resistant BCR-ABL1 kinase mutations (47), and in TKI-refractory CML cells, in which BCR-ABL1 kinase remains active (48, 49). Genomic instability in CML cells resistant/refractory to TKIs may cause CML-CP relapse and malignant progression to CML-BP (50).
Besides G:C→A:T transitions and G:C→T:A transversions other mutator phenotypes of TKI-resistant BCR-ABL1 kinase mutants have been reported, implicating the involvement of additional DNA repair mechanisms (51). For example, the role of DNA polymerase β in unfaithful nucleotide excision repair (NER) and/or homologous recombination repair (HRR) in accumulation of point mutations in BCR-ABL1 –transformed cell lines was suggested (40, 52–55).
In conclusion, we postulate that inhibition of BER activity mediated by UNG2 and perhaps also by OGG1 glycosylases contributes to accumulation of TKI-resistant BCR-ABL1 kinase mutations.
This work was supported by NCI R01 CA123014 from the National Cancer Institute.
CONFLICT OF INTEREST
The authors have no conflicts of interest to report.