Development of a rapid, small-scale DNA repair assay for use on clinical samples

Abstract

Double-strand breaks (DSBs) are the most lethal form of DNA damage. They can be repaired by one of two pathways, homologous recombination and non-homologous end joining (NHEJ). A NHEJ assay has previously been reported which measures joining using cell-free extracts and a linearised plasmid as DNA substrate. This assay was designed for 3 × 10⁹ cells grown in vitro and utilised radioactively labelled substrate. We have scaled down the method to use smaller cell numbers in a variety of cell lines. Altering the cellular extraction procedure decreased background DNA contamination. The cleaner preparations allowed us to use SYBR Green I staining to identify joined products, which was as sensitive as ³²P-end-labelled DNA. NHEJ was found in established tumour cell lines from different originating tissues, though actual levels and fidelity of repair differed. This method also allowed end joining to be assessed in clinical specimens (human blood, brain and bladder tumours) within 24 h of receiving samples. The application of this method will allow investigation of the role of DSB DNA repair pathways in human tumours.

INTRODUCTION

Double-strand breaks (DSBs) are the most lethal type of DNA damage and if not repaired a single DSB can cause cell death. DSBs can be caused by reactive oxygen species generated by oxidative metabolism within a cell and by ionising radiation during radiotherapy. Their repair is by one of two pathways, the homologous recombination (HR) pathway or the non-homologous end joining (NHEJ) pathway (reviewed by 1–3). The NHEJ pathway is thought to play a more dominant role in higher eukaryotes. However, the mechanisms by which the cell selects the pathway to employ are unknown, though it is partly dependent on cell cycle phase (4). The HR repair pathway uses a homologue as a template for repair, so that a true copy of the sequence is produced at the lesion. No template is used in the NHEJ pathway, therefore this process is error prone and can result in small deletions and additions.

The basic NHEJ pathway (reviewed by 1–3) requires the heterodimer Ku70/Ku80 that binds to the DNA ends at a DSB in a sequence-independent manner. DNA-PK_CS is then recruited to the Ku heterodimer to form DNA-PK holoenzyme. The holoenzyme has kinase activity, which may be important in phosphorylating proteins in this pathway. Finally, XRCC4 stimulates DNA ligase IV to join the broken ends. If the ends are ragged, additional processing is performed by proteins such as RAD50, MRE11, NBS1 and Artemis.

Cell lines that are deficient in components of the NHEJ pathway are defective in DSB repair and have an increased sensitivity to radiation. The glioblastoma cell lines M059K and M059J were established from the same primary tumour, but M059J cells have a mutation in the gene for DNA-PK_CS, which makes them less efficient at DSB repair (5,6).

Assays have been described in the literature which assess repair of DSBs. Many of these methods use plasmid linearised by restriction enzyme digestion as a model of the DSB. This substrate may be transfected into live cells, as described in a recent method whereby correct rejoining reconstituted EGFP expression that was then detected by FACS (7). Alternatively, joining is assessed in vitro using cell extracts. Many of the cell extracts are produced from established cell lines in culture, although extracts from fresh primary tissue, such as Xenopus eggs, calf thymus extracts and mouse testicular extracts, have also been used (8). The immediate use of clinical material to make cell extracts has not previously been investigated.

The NHEJ assay previously reported by Baumann and West (9) used radiolabelled linearised plamids as substrates. Linear dimers and trimers produced following joining were separated on agarose gels prior to quantification. However, the method for production of these cell extracts required large amounts of starting material (5 l of cultured cells). Similarly, cell extracts for DNA repair assays have been produced from established cell lines (10–14), or pooled tissue samples, such as rat brains, using this method (15).

Here we describe a modification of the Baumann and West NHEJ assay which permits the study of extracts from much smaller numbers of cultured cells than the original assay, allowing the assay to be performed on a number of cell lines in conventional cell culture facilities, rather than a large-scale cell production unit. We describe a quantitative non-radiolabelled method of measuring end joining using SYBR Green I and we show, for the first time, that the assay can be applied to fresh human tumour and normal tissues, whose end joining abilities can be demonstrated within 24 h of tissue being obtained.

MATERIALS AND METHODS

Cell culture

The MO59K and MO59J glioblastoma cell lines were obtained from Dr S.C. West (Cancer Research UK Clare Hall Laboratories, South Mimms, UK). The HeLa and A549 cell lines were obtained from the ATCC. The MCF-7, HepG2 and RT112 were obtained from Professor M.A. Knowles (Cancer Research UK Clinical Centre, St James’s University Hospital, Leeds, UK). HCT116 colon carcinoma cells were obtained from Professor D.R. Newell (Cancer Research Unit, University of Newcastle-upon-Tyne, UK).

Cells were maintained at 37°C and 5% CO₂ in Hams F12:DMEM (1:1 v/v) (MO59K and M059J cells), RPMI (RT112 and HCT116 cells), DMEM (A549 and HeLa cells) or RPMI:DMEM (1:1 v/v) (HepG2 and MCF-7 cells) culture medium (Gibco BRL Invitrogen, Paisley, UK), supplemented with 10% foetal calf serum (Harlan SeraLab, Loughborough, UK) and 1 mM l-glutamine (Gibco BRL). Medium for MO59K and J cells was further supplemented with non-essential amino acids (0.05 mM) and sodium pyruvate (0.1 mM).

Primary epithelial cells from normal human urothelium were isolated and cultured in keratinocyte serum-free medium supplemented with cholera toxin, epidermal growth factor (EGF) and bovine pituitary extract (Gibco BRL) as previously described (16). All of the cell lines are adherent, with the exception of HeLa cells, which grow in suspension and are loosely adherent, and so were cultured as monolayers. Cell lines were routinely grown in the appropriate culture medium at 0.2 ml/cm² until ∼70% confluence. At this point they were either harvested for the production of cell extract or for further subculture. Cells were harvested by incubation with 0.1% EDTA in phosphate-buffered saline (PBS) for no more than 5 min for M059K and M059J cells and for 5–10 min for the other cell lines. The EDTA solution was removed by aspiration and immediately followed by a 5 min incubation with a trypsin/versene solution [0.05% trypsin in 0.02% versene (Cancer Research UK)]. Cells could then be dislodged by tapping the dish/flask and resuspended in culture medium. For subculture, the cells were routinely split at a 1 in 5 ratio. This was equivalent to ∼1 × 10³ cells/cm² for M059K and M059J cells and 5 × 10³ cells/cm² for the other cell lines.

Cell free extract preparation

A total of 3 × 10⁹ MO59K cells were used for the batch preparation of cell free extract exactly as described (9) with centrifugation of the extract performed for 3 h at 37 000 r.p.m. in a Beckman SW41 rotor. For the micro-scale preparation of extracts ∼5–10 × 10⁷ cells were processed. After washing in hypotonic buffer (10 mM Tris–HCl pH 8.0, 1 mM EDTA, 1 mM DTT) by gently resuspending the cells by pipetting followed by pelleting at 350 g in an MSE Harrier 15/80 centrifuge, the cells were resuspended in 2 ml hypotonic buffer/ml cell pellet. For the cell lines this tended to be ∼1–2 ml of hypotonic buffer. Cells in hypotonic buffer were left on ice for 20 min, with occasional gentle agitation, and were then homogenised in a 1 ml tight-fitting glass Dounce homogeniser (Wheaton Laboratories, USA) in the presence of protease inhibitors (0.17 mg/ml phenylmethylsulphonyl fluoride, 0.01 trypsin inhibitor U/ml aprotinin, 1 µg/ml pepstatin, 1 µg/ml chymostatin, 1 µg/ml leupeptin). Trypan blue staining was used to assess that cell fracture was >95% after homogenisation. The suspension was then left on ice for a further 20 min with occasional gentle agitation, then mixed with 0.5 vol of high salt buffer (50 mM Tris–HCl pH 7.5, 1 M KCl, 2 mM EDTA, 1 mM DTT). Routine ultracentrifugation of homogenate was performed using Beckman thick-walled 3.2 ml polycarbonate tubes (for volumes > 1.2 ml) or Beckman thick-walled polyallomer microfuge tubes and Delrin adaptors (for volumes < 1.2 ml). To maintain the correct sedimentation coefficient the homogenates were centrifuged for 1 h at 60 000 r.p.m. (3.2 ml polycarbonate tubes) or 40 min at 60 000 r.p.m. (microfuge tubes) in a Beckman TL-100 Optima table top ultracentrifuge using a TLA100.4 fixed angle rotor, then dialysed against E buffer (20 mM Tris–HCl pH 8.0, 20% v/v glycerol, 0.1 M KOAc, 0.5 mM EDTA, 1 mM DTT) for 2 h before snap freezing in liquid N₂ and storage at –80°C. Protein concentration was assessed using the Coomassie assay protein reagent kit according to the manufacturer’s instructions (Pierce Biotechnology, Milwaukee, WI) and were ∼5–10 mg/ml.

Mononuclear blood cells and platelets were isolated from 60 ml whole blood from healthy subjects using Vacutainer CPT cell preparation tubes (Becton Dickinson Biosciences, Oxford, UK). Platelets were removed by washing three times in PBS and centrifugation for 15 min at 100 g. Brain and bladder tumour tissue biopsies were dissected initially to remove obvious blood clots and stromal tissue, then minced in RPMI medium and washed in PBS three times to remove as many erythrocytes as possible. Mononuclear cells (7 × 10⁷ total) and minced tissue (100–500 mg) were washed and homogenised in hypotonic buffer (final volume 0.2 ml) and mixed with 0.5 vol high salt buffer. Centrifugation was performed using microfuge tubes and adaptors for the Beckman TLA100.4 rotor as detailed above. Informed consent was obtained from all individuals for the use of primary tissue samples in these experiments.

Western blotting

Cell extracts were diluted with buffer (20% v/v glycerol, 4% w/v SDS, 120 mM Tris–HCl pH 6.8, 0.001% w/v bromophenol blue, 100 mM DTT) to 1 µg protein/µl. Samples of 20 µg protein were heated to 95°C, resolved on 7.5 or 12.5% (w/v) polyacrylamide Tris/glycine gels and blotted onto Hybond ECL nitrocellulose membrane (Pharmacia) using Tris/glycine transfer buffer under the recommended conditions. Blots were blocked in Odyssey blocking buffer (Li-cor Biosciences, Cambridge, UK) for 1 h and then incubated overnight in antibody buffer [1:1 v/v Odyssey blocking buffer:phosphate buffered saline containing 0.1% v/v Tween-20 (PBS-T)] containing primary antibody: DNA-PKcs, ab230, 1:2000 (Abcam, Cambridge, UK); actin, clone AC-15, 1:10 000 (Sigma, Gillingham, UK); XRCC4, 1:4000 (Serotec, Oxford, UK); Ku80, Ab-7, 1:1000 (Biocarta, Oxford, UK); ligase IV, ab6145, 1:2000 (Abcam); Ku70, ab87, 1:1000 (Abcam). After three 15 min washes in PBS-T blots were incubated with secondary antibody (1:5000) (AlexaFluor 680-conjugated goat anti-rabbit or goat anti-mouse; Molecular Probes, Leiden, The Netherlands). Following three washes in PBS-T, labelled proteins were detected and quantified using an Odyssey IR Imaging System (Li-cor Biosciences).

DNA end joining assay

DNA substrate with either complementary or blunt ends was prepared by linearizing pGEM3zf+ DNA (Promega, Southampton, UK) with EcoRI or HincII. DNA fragments (3.2 kb) were purified on Qiagen spin columns and resuspended to 5 ng/µl. PstI-linearised plasmid substrate was labelled with ³²P by firstly dephosphorylating the plasmid by incubation with calf intestinal alkaline phosphatase (Promega) according to the manufacturer’s instruction, followed by end-labelling using polynucleotide kinase (New England Biolabs, Hertfordshire, UK) in the presence of [γ-³²P]ATP (Amersham-Pharmacia, Little Chalfont, UK) according to the manufacturer’s instructions. End joining reactions (20 µl) were carried out with 40 µg protein extract and 20 ng DNA substrate in the presence of 50 mM HEPES pH 8.0, 40 mM KOAc, 1 mM Mg(OAc)₂, 1 mM ATP, 1 mM DTT and 0.1 mg/ml BSA at 37°C for 2 h. Where indicated, protein samples were pretreated with wortmannin (30 µM final concentration; Sigma) or XRCC4 antibody (4 µl/sample; Serotec) for 10 min on ice before use in end joining reactions. Control joining by T4 ligase (New England Biolabs) was performed using the manufacturer’s recommended buffer at 37°C for 2 h. Samples were incubated with RNase A (80 µg/ml) for 10 min and then protein was removed by incubation with proteinase K (2 mg/ml) and 0.5% (w/v) SDS for 10 min and extraction with Tris-buffered phenol/chloroform/isoamyl alcohol. For samples which were re-cut with restriction enzymes DNA was precipitated using 0.3 M NaOAc (pH 6.5) and 2 vol ethanol on dry ice for 30 min, resuspended in the appropriate buffer and incubated with EcoRI or HincII for 2 h at 37°C. DNA separation was performed by agarose (0.7%) gel electrophoresis and SYBR Green I staining as previously described (17). Phosphorimage data was collected using a Molecular Imager FX (Bio-Rad) and Quantity One version 4.1.1 software (Bio-Rad). ³²P signal was detected using the Kodak K screen settings and the SYBR Green I signal was detected using the SYBR Green I and II DNA stain settings with an external laser. Analysis was performed using the volume analysis function to quantify signal intensity.

RESULTS

Development of the micro-scale assay

Cell extracts were prepared using the standard large-scale method and our modified micro-scale version using the established tumour cell line M059K. The amount of NHEJ pathway proteins as determined by western blotting and further densitometry did not differ between the two extraction methods (Fig. 1).

Figure 1.

Levels of NHEJ protein components in human MO59K cell-free extracts prepared by the batch (lane 1) and micro-scale (lanes 2–4, three independent preparations) procedures as described in Materials and Methods. Western blots of proteins were quantified relative to the actin loading control and are representative of three independent experiments. Molecular weight markers are shown in kilodaltons.

As the cellular extracts produced from the micro-scale procedure had a tendency to have less background genomic DNA due to less disturbance of the DNA pellet with a fixed angle centrifuge rotor, we were able to use SYBR Green I in assessing NHEJ ability. A comparison of SYBR Green I staining with ³²P-labelled DNA clearly showed that SYBR Green I was as sensitive as radiolabel (Fig. 2).

Figure 2.

Detection of DNA by SYBR Green I staining or ³²P-end-labelling. (A) Concentration range of PstI-linearised pGEM labelled with SYBR Green I or ³²P and detected by phosphorimaging. The relative densities of each band from three experiments are represented in (B). The relative signal from SYBR Green I and ³²P was normalised to the 20 ng DNA concentration. (C) DNA end joining of PstI-linearised pGEM by M059K cell-free extract. Samples were deproteinised immediately (0 h) or after incubation for 2 h at 37°C (2 h). The monomer substrate DNA (1×) is joined to form dimers (2×) and trimers (3×) as indicated.

The end joining abilities of the standard large-scale and the modified micro-scale extracts were comparable and joining was as efficient as described in the original Baumann and West paper (9). In addition, both extracts required DNA-PK, as judged by wortmannin sensitivity, and displayed similar lack of joining in MO59J extracts, which lack DNA-PK_CS. Both types of extract showed a requirement for XRCC4 and both ATP and magnesium, which are needed for DNA ligase IV activity (Fig. 3). Similarly, end joining in both the standard large-scale and the modified micro-scale procedures was inhibited by the addition of anti-Ku70 or anti-ligase IV antibodies (data not shown).

Figure 3.

DNA end joining in human MO59K and MO59J cell-free extracts. (A) The capacity of MO59K extracts prepared by the batch and micro-scale methods to join 5′-overhang (EcoRI) and blunt (HincII) ended DNA substrates. Samples were deproteinised immediately (0 h) or after incubation for 2 h at 37°C (2 h). Joining of monomer substrate DNA (1×) to form dimers, trimers and further multimers (2×, 3× and multi) is indicated. M, molecular weight markers. (B) DNA joined by the MO59K micro-scale extract (lanes 1 and 8) was re-cut with EcoRI (lane 2) or HincII (lane 9) prior to gel electrophoresis. End joining reactions were performed in the absence of ATP (lanes 3 and 10) and magnesium acetate (lanes 4 and 11), with MO59J extract (lanes 5 and 12) or in the presence of 30 µM wortmannin (lanes 6 and 13) and rabbit anti-XRCC4 (lanes 7 and 14). Gels are representative of three independent experiments.

DNA end joining in established cell lines

We then proceeded to assess end joining activity in a number of commonly used human cell lines, established from different tissue types. All the cell lines had the ability to join compatible 5′ overhang ends (Fig. 4A) and blunt ends (Fig. 4B). The joining abilities of cell lines varied, with HepG2 and MCF7 being poorest at joining compatible 5′ overhangs and blunt ends, respectively. We are currently investigating this phenomenon.

Figure 4.

A comparison of cell-free extracts prepared from a panel of human tumour cell lines. End joining of (A) 5′-overhang (EcoRI) and (B) blunt (HincII) ended DNA substrates in the presence of extract from MO59K (glioblastoma), HepG2 (liver), HeLa (cervix), MCF-7 (breast), A549 (lung), HCT116 (colon) and RT112 (bladder) cells. End-joined products were incubated for 2 h in the presence (+) or absence (–) of EcoRI (A) and HincII (B). Gels are representative of three independent experiments.

In general, with the exception of the A549 and RT112 cell lines, extracts were more efficient at joining DNA substrates with compatible 5′ overhang ends than those with blunt ends. The majority of the joined products were re-digestible with the appropriate restriction enzyme, indicating error-free joining. The HCT116 extract had a tendency to degrade DNA substrate.

DNA end joining in tissue samples

Having detected end joining in a range of cell culture types, we hypothesised that end joining activity might be demonstrable in corresponding tissue samples from patients. There was a detectable amount of end joining observed in extracts prepared from mononuclear cells isolated from normal blood samples (Fig. 5A). End joining could also be measured in early primary cultures of normal human urothelium and in tumour samples from brain and bladder; addition of anti-XRCC4 antibody inhibited joining observed in extracts from normal human urothelium and partially inhibited dimer formation in brain and bladder tumour extracts (Fig. 5B).

Figure 5.

DNA end joining of 5′-overhang (EcoRI) ended DNA substrates in extracts derived from human tissues and tumours. End joining using extracts prepared from (A) mononuclear blood cells isolated from three independent subjects and (B) primary cultures of normal human urothelium (lanes 3–5) and brain (lanes 6–8) and bladder (lanes 9–11) tumour tissue. Samples were deproteinised immediately (0 h) or after incubation for 2 h at 37°C (2 h). End joining reactions were performed in the absence (–) or presence (+) of rabbit anti-XRCC4 or wortmannin. Gels are representative of three independent experiments.

DISCUSSION

We have developed a scaled down ‘micro’ method of a NHEJ assay. This method produces extracts which are identical to those produced by the original large-scale method when comparing the levels of DNA repair proteins, DNA-PK-dependent DNA joining activity and requirement for ATP and magnesium (9). By decreasing the number of cells required to make the extracts by ∼30-fold, extracts can be produced within a standard tissue culture laboratory rather than a specialised cell production unit. In addition, the reduced time required for extract preparation (from sample collection to end joining assay the procedure takes 24 h) gives this assay advantages over those predictive DNA repair assays requiring extensive in vitro cell culture, such as the clonogenic assay, micronucleus assay and pulsed field electrophoresis (18). The method for extract preparation has also been modified by altering the centrifugation conditions to use a fixed angle rather than a swing out rotor. This normalises the conditions throughout the extract during centrifugation and decreases mixing of pellet with supernatant when the centrifuge decelerates. The extracts prepared therefore had lower levels of genomic DNA contamination. Due to the reduced genomic DNA contamination in extracts we were able to use SYBR Green I staining to detect NHEJ, which is a novel application of this dye. More recently, another DNA stain, Vistra Green, has been used in a DSB repair assay, which like SYBR Green I is more sensitive than ethidium bromide (19).

This has the advantage of simplifying the original method, which required the production of radiolabelled DNA substrate. In this paper we have described the use of DNA substrates with 5′ overhangs or blunt DNA ends. However, incompatible ends or oxidatively damaged DNA ends could also be used (19,20), which would require processing by nuclease and polymerase action prior to ligation. If such DNA substrates were used radiolabelling DNA ends would be inappropriate as these labels could be removed during the DNA repair process. Incorporation of radiolabel into the entire plasmid substrate would have to be employed, which is a more cumbersome method for clinical application, and the substrate has a limited usefulness due to radioactive decay. In contrast, the use of SYBR Green I staining avoids all these concerns. It is equally effective for all DNA end types used in the assay, as it binds to the entire plasmid DNA sequence. Additionally, large quantities of the DNA substrate with the required ends of interest can be prepared at one time due to the absence of radioactive half-life constraints. Other groups have used ethidium bromide, which is not as sensitive as SYBR Green I (21) and so requires much larger amounts of DNA substrate.

The assay also provided a measure of the fidelity of repair. Joined substrates which can be re-cut (re-digested) with the same restriction enzyme to form monomer substrate will have been repaired accurately, whereas those that are resistant to digestion will not have been joined accurately as a consequence of additional end processing. This resistance to re-digestion was not seen with DNA substrate joined using T4 DNA ligase. Sequence analysis of NHEJ products resistant to digestion showed that additional processing of DNA substrate ends had occurred as the restriction enzyme site was lost (data not shown). The role of other DNA repair factors, including Mre11, NBS1 and RAD50, in the end-processing of DSBs is currently being investigated.

The observed range of end joining efficiency in tumour cell lines from different tumour sites may be a reflection of the different levels of active NHEJ proteins in these cells. Immortalised cell lines in vitro can be used as models of normal and tumour cells and are useful tools. However, there is a concern that due to the accumulation of genetic changes in vitro they may not accurately reflect the tissue of origin. In terms of clinically relevant data we therefore chose to examine fresh tissue. Extracts derived from blood samples from three individuals were capable of end joining DNA substrate with 5′ overhang ends. End joining was also observed in extracts from bladder tumour and brain tumour specimens. In general, the actual levels of end joining tended to be lower in tumour extracts than in the cultured cell lines and end joining in the brain and bladder tumour extracts was not completely inhibited by the addition of anti-XRCC4 antibody. Further investigation is necessary to determine the levels of NHEJ components and involvement of DNA-PK-dependent end joining in the repair of DSBs in these extracts.

We have shown here that it is possible to investigate a panel of cell lines for NHEJ activity without the need for large-scale cell production facilities. This functional assay for NHEJ is quantitative and provides an indication of fidelity of repair and results can be produced rapidly. We have further demonstrated that the method can be applied to primary tissue, including blood samples and tumours, so allowing investigation into the role of DSB DNA repair processes in these clinical samples.