New developments in comet-FISH

Abstract

The comet assay combined with fluorescence in-situ hybridisation (FISH) is a powerful technique for comparative analyses of damage induction and repair in genomes and in specific DNA sequences within single cells. Recent advances in the methodology of comet-FISH will be considered here, with particular attention to the design and generation of fluorescent probes. In general, all the approaches must fulfil a few basic requirements: the probes should be no longer than ~300 nucleotides in length (single or double stranded) to be able to penetrate the gel in which the target genomic DNA is embedded, they should be sequence-specific, and their signal should be detectable and distinct from the background fluorescence and the dye used to stain the DNA.

Introduction

Biochemical processes in living organisms depend on maintaining the integrity of their DNA for coding of essential proteins. Several pathways exist for detection and repair of DNA lesions, including nucleotide excision repair and base excision repair to remove lesions affecting one DNA strand, mismatch excision repair to resolve mispaired bases and non-homologous end joining and homologous recombination to deal with complex structures such as double-strand breaks and interstrand crosslinks. The choice of pathway depends upon factors such as the type of lesion, and which factor or complex arrives first at the damaged site. Of particular concern is the situation in which a replication fork or transcription complex encounters a lesion that may hinder its progression. In addition to the global repair pathways that operate throughout the genome, lesions in the transcribed strands of actively expressed genes may be repaired via a specialised pathway, transcription-coupled repair (TCR), reviewed in refs 1 and 2.

The comet assay has become one of the most commonly utilised approaches for measuring damage in genomic DNA in single cells. There are a handful of methods for detection of TCR in cells from higher eukaryotes, including the ‘Southern blot method’ and ligation-mediated polymerase chain reaction (PCR) that require high frequencies of lesions (one in 10–30kb) (3). However, the application of fluorescence in-situ hybridisation (FISH) to cells processed for the comet assay has provided the means to examine TCR of lesions induced at frequencies of 1–10 per megabase, rendering this approach much more sensitive than other methods and bringing the sensitivity into the physiological range. This feature is important when working with damaging agents whose overall cellular toxicity precludes the use of high doses required for detection of lesions in DNA.

Since Santos et al. (4) first applied fluorescent probes to stretched DNA fibres, several groups have used this highly sensitive approach to detect genomic alterations such as damage, repair and chromosome fragile sites in specific sequences and chromosomal regions and in the genomes of individual cells (reviewed in refs 5–7 and others). Among novel comet developments inspired by the availability of FISH probes, the work of McKenna et al. (8) using comet-FISH to study formation and repair of mitomycin C-induced interstrand DNA crosslinks, and Wasson et al. (9) for analysis of region-specific hypomethylation deserve special mention. The basic comet assay method and its variations have been reviewed extensively (10–14). Here, I will focus upon the application of FISH techniques that facilitate examination of damage and repair in specific DNA sequences together with that in the overall genome in single cells. Recent improvements in this assay will be presented, particularly with respect to the design, preparation and detection of the probes, and the interpretation of the results in the context of the structure of the comets will also be discussed.

Fluorescent probes for comet-FISH

Double-stranded fluorescent probes

Sequence-specific double-stranded probes can be synthesised and coupled to fluorophores in the laboratory using a variety of methods including PCR and nick-translation (reviewed in ref. 15).

Several companies offer probes labelled with various fluorophores that can be used to detect genes and sequences of general interest, such as centromeric regions (which are usually not transcribed), telomeres, chromosomal fragile sites, and genes, such as p53, ATM and others. These probes are comprised of small (<200bp) double-stranded fragments, and they cover rather large (>100kb) genomic regions. We and others (5,10,16) have used these probes to examine strand breaks (frank strand breaks or breaks generated by, e.g., lesion-specific glycosylases/apurinic/apyrimidinic endonucleases) and repair in specific genomic regions, and have reported that probes for actively transcribed genes are visualised in comet tails immediately after damage is induced, and that as repair proceeds the signals appear in the comet heads at a faster rate than probes for silent regions (e.g., a centromere).

According to the generally accepted view, intact DNA is arranged in tightly wound supercoiled domains (loops) tethered to a loosely defined structure known as the nuclear matrix. A single-strand break relaxes the supercoiling of the DNA and causes the strands in the affected loop to unwind and protrude from the nucleus; upon electrophoresis, these loops form the comet tail. The size of the loops has been estimated to be 0.2–1.0×10⁶ bases, and they may contain several genes that are transcribed from either strand. A loop must be completely free of strand breaks to appear in the comet head; thus, it follows that quantitation of signals in comet heads and tails may not accurately reflect repair in any given gene but rather in the entire loop containing that gene.

Another measure of repair is the reduction in the number of signal spots, which, depending on the size of the target and the amount of damage, may be initially higher than the expected number (e.g., two for the two alleles of single-copy genes), thus indicating damage within the probed region; it has been reported that the spot count for actively transcribed sequences decreases faster than that for inactive sequences (10).

Commercial probes are readily available, and in most cases, their signals can be detected without need for amplification. However, these probes generally cover regions that are much larger than a typical gene of interest, and they are also double-stranded, so the sometimes subtle differences in repair between transcribed and non-transcribed strands may not be detected.

Fluorescent oligonucleotide probes complementary to repetitive sequences

Probes for telomeric, centromeric, 5S and 25S rDNA and other genomic regions that contain highly repeated sequences can be labelled in the laboratory by PCR or nick-translation using commercial kits and fluorescently labelled nucleotides such as rhodamine-4-2ʹ-deoxyuridine 5ʹ-triphosphate (dUTP), or antigen-tagged nucleotides such as digoxigenin-11-dUTP, which are detected by using fluorescein isothiocyanate-conjugated anti-digoxigenin antibodies (17). These probes are generally easy to synthesise, and because of the large number of target repeats in tandem, they can generate bright signals, but they are obviously limited to high copy number sequences.

Padlock probes combined with rolling circle amplification

Padlock probes are linear oligonucleotides that are circularised in a strictly target-dependent manner. The ends of the probes contain sequences that are complementary to a target genomic sequence, so that when they are hybridised the ends can be joined by a ligase forming a circular DNA molecule (18). A DNA polymerase such as Φ29 utilises the target molecule as a primer through 3ʹ-5ʹ exonucleolysis of any 3ʹ end in the vicinity of the probe (within a few 100 bases). This process, named target-primed rolling circle amplification (RCA), generates a long single-stranded DNA made out of hundreds of tandem repeats complementary to the padlock probe; these are detected by hybridisation with fluorescent oligonucleotides complementary to a unique sequence within the padlock (Figure 1) (19).

Figure 1.

Synthesis of sequence-specific probes using padlock probes and RCA. The newly synthesised DNA (dashed blue line) is complementary to the padlock DNA and contains the complement of a unique tag sequence (aqua) that is used for hybridisation to fluorescent probes (red).

RCA can only be initiated from a 3ʹ end in or near the target sequence, so the frequency of strand breaks must be high enough to include a nick near the target; otherwise, double-stranded DNA can be fragmented with a restriction enzyme that cuts near the target sequence. The DNA is made single stranded by exonuclease digestion of the non-target strand prior to hybridisation and ligation of the padlock probe (20). When a uniquely located 3ʹ end is desired, the padlock probe is designed such that when hybridised to the target, a site-specific G:A mismatch is generated; the A is then removed by MutY and the resulting AP site is converted to a nick by an AP endonuclease (21).

Henriksson et al. (22) propose that when cells are subject to electrophoresis in alkaline conditions and then neutralised by incubation at pH 8.0, the DNA in comet tails is mostly single stranded and available for hybridisation to padlock probes and RCA, whereas DNA in comet heads reanneals and does not hybridise to the probes. As damage is repaired, the DNA domain or loop containing the target does not migrate to the tail and becomes resilient to hybridisation to probes, so the number of signals should decrease with time. The authors report such findings for three single-copy genes in human lymphocytes.

This method requires multiple steps and optimisations, but its sequence specificity and bright signal renders it a viable option for the examination of repair in defined sequences, and it could in principle be adapted to the approach described in the next section. Shaposhnikov et al. (23) have utilised Alu sequence-specific padlock probes for comet-FISH with a 12-gel format in each slide, an approach with the potential for simultaneous analyses of multiple samples.

Strand-specific fluorescent oligonucleotide probes complementary to the 5ʹ and 3ʹ regions flanking the sequence of interest

Horváthová et al. (24) pioneered the use of two-coloured, single-stranded oligonucleotide probes for the ends of the targets and were first to report differential rates of repair in certain active genes and in total DNA. The 5ʹ ends of their probes were labelled with biotin or fluorescein; development and amplification of the signals required two cycles of antibodies, and repair was measured by quantifying the fraction of the signals in comet heads that are assumed to contain unbroken DNA. However, the sensitivity of this approach is limited by the requirement that all lesions within a loop must be repaired before that DNA appears in comet heads, as mentioned previously, so the damage levels should be adjusted to induce an average of one lesion per loop.

The principle of using probes for sequences flanking the genomic segment of interest has been further developed in our laboratory, as illustrated in Figure 2 (25). The probes are synthesised in-house by PCR with biotin-labelled forward primers, natural reverse primers and a dNTP mixture that contains both aminoallyl-dUTP and dTTP. The products are ~200 to 250bp in length to facilitate their diffusion in the gels, and the incorporation of aminoallyl-dUTP results in exposed amino groups. Upon incubation with streptavidin-coated magnetic beads and mild denaturation, the natural strands are purified from the supernatant. The streptavidin is degraded at 90°C in formamide, and the biotin-labelled strands are purified. Strands are reacted with Alexa fluorophores that bind to the amino groups (Figure 2A). Parallel sets of comet slides are hybridised to probes for the transcribed strand and probes for the non-transcribed strand; each set of probes includes probes for the 5ʹ end of the target genomic segment labelled with Alexa red and probes for the 3ʹ end of the same target labelled with Alexa green; the DNA is uniformly stained with 4',6-diamidino-2-phenylindole. In principle, both strands could be hybridised on the same slide if there were five different fluorescent labels and appropriate filters to permit detection with no significant overlap. As the probes are designed to hybridise with the respective termini of the gene or fragment of interest (Figure 2B), breaks within the strand are scored when the red and green fluorescent signals appear separated; overlapping fluorescent signals or those separated by <2 μm per 100kb of genomic DNA are scored as intact strands (Figure 2C). Global repair is quantified in the standard manner by measuring the percent of DNA in comet tails.

Figure 2.

Synthesis of single-stranded probes for detection of the termini of defined genomic sequences. (A) PCR mixes contain forward primers labelled with biotin (yellow) and aminoallyl-dUTP (NH2). Streptavidin (purple) is used to separate the probe strands. Two sets of primers (pink and orange) are used for probes for the 5ʹ and 3ʹ termini of the sequence of interest. (B) Probes for the transcribed and the non-transcribed strand are hybridised to separate groups of comet slides. The scheme shows one probe per terminus, but 40–50 probes are needed to obtain visible signal. (C) Simplified comet with damaged DNA in relaxed loops.

The principal advantages of this methodology are that repair can be assessed at the gene level independently of whether lesions in the rest of the loop have been repaired, and of course, that repair can be measured in both the transcribed and the non-transcribed strands of the target sequence. Moreover, no signal amplification is needed. However, the approach requires time-consuming syntheses of the multiple probes to obtain visible signals, ≥40 fluorescent probes are needed to hybridise to each target.

The size of the target gene or sequence directly impacts the sensitivity of the assay. For example, as low a dose as 0.1 J/m² of UVC irradiation was sufficient to induce ~1.3 cyclobutane pyrimidine dimers in 10⁶ bases, and after generation of single-strand breaks at the lesion-containing sites by T4 endonuclease V, resulted in an average of 12 damaged strands of 72 analysed (i.e., 17%) in the ~150kb ATM gene (25). Similar levels of 8-oxoGuanine were induced by treatment with 50mM potassium bromate for 1h; incubation with hOGG1 and subsequent alkaline conditions convert the lesions to single-strand breaks (25).

Other methods

Probes for comet-FISH can be prepared using peptide nucleic acids (PNA) (26,27), which can form very stable complexes with DNA with higher affinity than DNA or RNA. The complexity of the solid-phase chemistry required to make PNA precludes its synthesis in the average laboratory; thus, the probes are generally purchased from commercial outfits. Locked nucleic acids can be synthesised on standard automated DNA synthesisers using commercial phosphoramide monomers. Like PNA, these probes form stable hybrids with DNA. Although several groups have reported using locked nucleic acid probes for FISH, to our knowledge the approach has not been used for comet-FISH.

Conclusion

The comet assay is widely used for the assessment of cellular health in academic, pharmaceutical and commercial laboratories, and high-throughput applications and software programs for automated comet analysis have been developed. However, utilisation of the comet-FISH technique has been sparsely reported. It is to be hoped that scientists will apply this exquisitely sensitive approach for examination of the distribution of various features and alterations (e.g., endogenously or exogenously induced lesions, fragile sites, deamination, hyper/hypomethylation, and non-canonical DNA structures) in the genomes of living cells.

Funding

National Institute of Environmental Health Sciences and National Institute of Health (ES018834).