G-quadruplex Structures Contribute to Differential Radiosensitivity of the Human Genome.

DNA, the fundamental unit of human cell, generally exists in Watson-Crick base-paired B-DNA form. Often, DNA folds into non-B forms, such as four-stranded G-quadruplexes. It is generally believed that ionizing radiation (IR) induces DNA strand-breaks in a random manner. Here, we show that regions of DNA enriched in G-quadruplex structures are less sensitive to IR compared with B-DNA in vitro and inside cells. Planar G-quartet of G4-DNA is shielded from IR-induced free radicals, unlike single- and double-stranded DNA. Whole-genome sequence analysis and real-time PCR reveal that genomic regions abundant in G4-DNA are protected from radiation-induced breaks and can be modulated by G4 stabilizers. Thus, our results reveal that formation of G4 structures contribute toward differential radiosensitivity of the human genome.

Introduction

Maintaining genomic stability is of utmost importance for survival of any organism. Our genome is constantly challenged by a number of endogenous and exogenous insults, and studies estimate ~1000 DNA lesions in each cell per day (Aguilera and Garcia-Muse, 2013, Burma et al., 2006, Ciccia and Elledge, 2010). Endogenous sources comprise of free radicals generated during metabolic processes, DNA replication, and recombination, whereas the exogenous agents include radiation and DNA-damaging chemicals. Among exogenous agents, radiation is the most significant contributor to DNA damage. Ultraviolet rays (UV), ionizing radiations (IR; X-rays, γ-rays), microwaves etc. are examples of radiations that cause an array of ill effects in cells. Although UV radiation is highly pervasive and causes DNA lesions such as cyclobutane pyrimidine dimers (CPDs) and single-strand breaks (SSBs), majority of DNA damage caused by IR results in SSBs and double-strand breaks (DSBs) (Vignard et al., 2013).

IR damages DNA either directly, wherein DNA breaks occur due to transfer of energy to the phosphodiester bonds, or indirectly by initiating radiolysis of water molecules present in close vicinity of the DNA, subsequently resulting in the production of free radicals. Upon induction of an SSB or DSB, the affected cell manifests a cascade of events, collectively termed as the DNA damage response (DDR), which includes sensing of the DSBs, transduction of the sensed signal, and activation of repair factors, ultimately ensuring the genomic integrity and hence, an intact chromosome. The degree and type of DNA damage dictates whether a cell survives (DNA repair) or undergoes apoptosis (Jackson and Bartek, 2009, Sancar et al., 2004, Symington and Gautier, 2011).

Two schools of thought exist regarding distribution of IR-induced DNA strand-breaks within the genome (Lobrich et al., 1996, Pang et al., 2005, Vignard et al., 2013). Most studies indicate that radiation-induced breaks are random, occur throughout the genome and do not follow a pattern (Lett, 1992, Nikiforov et al., 1999, Pang et al., 2005, Van Der Schans, 1978). On the other hand, a few studies indicate nonrandom pattern of IR-induced DNA strand-breaks (Lobrich et al., 1996, Zhou et al., 2012). It is reported that the four nucleotide bases (adenine, cytosine, guanine, and thymine) vary in their sensitivity to IR-induced DNA damage, guanine being the most readily oxidized among the four (Spotheim-Maurizot and Davidkova, 2011, Steenken, 1997). However, whether this contributes to differential sensitivity of DNA to IR is not clear. Moreover, its implication with respect to the genome remains unexplored.

Organisms differ in the degree of sensitivity to IR-induced DNA damage. This property is attributed to a number of factors such as complexity of genome of the organism, cellular organization, variation in the DNA damage sensing, and repair mechanisms (Daly, 2009). For example, the bacterium Deinococcus radiodurans is extremely resistant to IR-induced DNA damage, whereas other species within the same genus exhibit IR sensitivity (Krisko and Radman, 2013). Interestingly, among the reasons attributed to radiation resistance, skewed GC content forms an important factor, wherein radiation-resistant organisms such as D. radiodurans (66.6%) and Kineococcus radiotolerans (74.2%) possess higher GC content compared with the susceptible ones such as Escherichia coli (50%) or Shewanella oneidensis (45.9%). Discrepancy in radiation sensitivity has also been reported among chromosomes, with some chromosomes affected more often than others, although the reasons behind it remain elusive. Interestingly, distinct clusters of DNA damage, often termed as damage “hot spots,” within the same chromosome have been reported, suggesting variations in radiation sensitivity among genomic regions (Puerto et al., 2001). Although factors such as GC content and chromatin organization were speculated to be responsible for the difference, the mechanistic details remain unexplored.

The human genome, containing 3000 megabases, has a GC content of 42% and largely exists in the form of B-DNA. However, the last decade has witnessed increasing evidence for the formation and regulation of deviant structures, termed as “non-B DNA” forms inside cells (Sinden, 1994). Structures such as G-quadruplex, triplex DNA, R-loops, cruciforms, and Z-DNA have been shown to play key roles in governing several physiological and pathological processes within a cell, such as transcription, replication, telomere maintenance, and generation of chromosomal translocations (Nambiar et al., 2008, Nambiar et al., 2011, Nambiar and Raghavan, 2011, Neidle and Balasubramanian, 2006, Raghavan et al., 2004, Sinden, 1994, Voloshin et al., 1988). G-quadruplex (G4-DNA) is formed in guanine-rich regions of DNA and RNA in the cell (Nambiar et al., 2008, Neidle and Balasubramanian, 2006, Sinden, 1994). It typically consists of four three-guanine repeats, held together by Hoogsteen-hydrogen bonding. The guanines form a planar quartet structure, stabilized by monovalent cations such as K+ that are present in the cellular milieu. Studies have shown several G4-forming motifs (350,000 to 700,000) present throughout the genome in regions including promoters, immunoglobulin switch regions, rDNA, telomeres, and replication origin of several genes (Chambers et al., 2015, Nambiar and Raghavan, 2011). Apart from regulating normal cellular processes, G-quadruplexes have also been implicated in deregulation of oncogenes, tumor suppressors, generation of chromosomal translocations, and hence, oncogenesis (Nambiar and Raghavan, 2011). Thus, whether the human genome is differentially susceptible to radiation-induced DNA damage and if so, the cause behind such a disparity, its mechanism, and relevance is not well understood.

In the present study, we report formation of G-quadruplex DNA structure as an important factor contributing to differential radiosensitivity of genome in human cells. Further, our study establishes that G-quadruplex structures are shielded from radiation-induced DNA breaks in vitro, ex vivo, and in vivo. Thus, our study demonstrates a nonrandom pattern of IR-induced DNA breaks within the human genome due to free radical shielding by DNA G-quadruplex structures.

Results

Homopolymeric Guanine Tracts Harbor Minimal DNA Strand-Breaks upon IR Exposure

Efficiency of radiation-induced cleavage on short DNA was investigated by irradiating single-stranded homopolymeric DNA harboring adenine (A), cytosine (C), guanine (G), and thymine (T) nucleotides (Figure 1A). Irradiation using IR (γ-rays) and subsequent DNA damage leading to DNA breaks was analyzed on a denaturing PAGE (Figure 1B). Results showed significant breakage of DNA on every nucleotide in case of single-stranded polyadenine, polycytosine, and polythymine DNA (Figure 1B). To our surprise, DNA with polyguanine exhibited minimal cleavage upon irradiation, and the intensity of cleavage was comparable with that of the unirradiated control (Figures 1B and 1C). Similar results were also observed when irradiated samples were loaded on a native PAGE (Figures S1A and S1B). We chose a relatively higher dose (100 Gy) to ensure breakage at every nucleotide. A dose-dependent increase in cleavage efficiency was observed when the homopolymeric DNA containing A, C, and T were irradiated with increasing doses of IR (Figure S1C). Consistent to above observation, shielding effect was observed for the G-rich sequence even after a broad-range dose titration (Figures S1C and S1D). Importantly, as expected, IR-induced breaks were random leading to breakage at almost every nucleotide when double-stranded DNA containing poly C:G or poly A:T were used (Figure 1B). Similar results were also obtained when oligomers containing AT-rich, GC-rich, or scrambled double-stranded DNA sequences were used (Figures 1D and 1E). Besides generation of DNA breaks, exposure to IR also can result in oxidative damage to individual bases (Cadet and Wagner, 2013, Pouget et al., 2002). To assess whether the polyG DNA under investigation was undergoing IR-mediated oxidative damage, homopolymeric oligomers (A, C, G, and T) were irradiated, followed by piperidine treatment, which can convert a base damage into a strand-break and were subsequently resolved on a denaturing PAGE (Figure 1F). Results revealed that IR-induced cleavage on polyG substrate was significantly less, as compared with other homopolymers, even in the presence of piperidine (Figures 1F and 1G). Importantly, there was no significant difference in induction of strand-breaks even upon piperidine treatment in polyG oligomer, thus ruling out the possibility of extensive base damage in this substrate. Multiple bands observed in case of the polyG oligomer, running lower to those of poly A, C, and T is due to formation of intramolecular G-quadruplexes (Figures 1B, S1A, and S1C).

Figure 1

Evaluation of Radiation Sensitivity on Shorter DNA Fragments when Different Sequences or Structures Are Present

(A) Schematic representation of the assay used for evaluation of IR-induced DNA strand-breaks. Oligomeric DNA substrates of interest were γ-irradiated (100 Gy), resolved on a denaturing (observed as a ladder) or native gel (observed as a smear), and analyzed for the abundance of DNA breaks using PAGE.

(B) Denaturing PAGE profile showing sensitivity of single-stranded homopolymeric (35 mer) adenine, cytosine, guanine, and thymine nucleotides and double-stranded DNA ([A:T]35 and [C:G]35), to γ-radiation (100 Gy).

(C) Bar graph presented shows quantitation of cleavage efficiency for DNA substrates following irradiation. Each experiment was performed a minimum of three times and the intensity of quantified bands is expressed as photo-stimulated luminescence units (PSLU), showing mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001).

(D and E) Denaturing PAGE profile showing sensitivity of AT-rich, GC-rich, and scrambled double-stranded (35 mer) DNA to γ-radiation (100 Gy) (D) and bar diagram representing its quantitation (n = 3) showing mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001) (E).

(F) Denaturing PAGE profile showing the effect of piperidine treatment on irradiated homopolymeric A, C, G, and transfer DNA sequences. In each case, DNA substrates were irradiated (100 Gy), treated with piperidine (1:10 dilution; 90°C for 30 min), followed by resolving on a 15% denaturing PAGE. “Pip” refers to piperidine treatment. The experiment was repeated multiple times (n = 6).

(G) Bar diagram represents IR-induced cleavage intensity based on panel F and other gels (n = 6) showing mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001, “ns” is nonsignificant).

(H) PAGE profile showing radiation sensitivity in three independent heteropolymeric DNA substrates (T23G22, G22T23, and T23C22). Dotted arrow lines indicate position of G, C, or T in the respective DNA.

(I) The intensity of cleaved products was quantified and presented as a bar graph (n = 3) showing mean ± SEM.

(J) PAGE profiles of oligomeric DNA harboring G4-forming motifs (indicated as “G”) derived from HIF1α promoter, VEGF promoter, human telomere sequence upon irradiation (150 Gy). Irradiated complementary strands (marked as “C”) derived from same genes and their corresponding random sequence (marked as “RN”) served as controls. Experiments were repeated a minimum of three times.

(K) Bar graph showing efficiency of IR-induced cleavage shown in panel J and other gels. In each case, the intensity of IR-induced cleavage was determined following subtraction of background from respective unirradiated control and graph depicts mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001).

(L–N) Comparison of IR-induced breaks on double-stranded oligomers derived from three different regions of telomeric DNA (L) named as “Telo A” (3 repeats), “Telo B” (5 repeats), and “Telo C” (7 repeats). PAGE profile shows comparison of cleavage efficiency when increasing dose of IR (0, 50, 100, 200 Gy) was used (M). Bar diagram showing quantification of IR-induced cleavage on duplex DNA containing telomeric repeats is presented (n = 3). Bar graph showing mean ± SEM (*p < 0.05, **p < 0.01, ***p < 0.001, “ns” is nonsignificant) (N).

See also Figure S1.

In order to investigate whether the reduced sensitivity observed on homopolymers of guanines was consistent even in the case of heteropolymeric DNA substrates, we designed oligomers harboring guanine-repeat tracts, in combination with other sequences. Interestingly, IR-induced DNA breaks were consistently observed specifically at the poly thymine end of the oligomer, sparing the end containing guanines, whereas the control oligomer (containing thymines and cytosines) harbored breaks throughout its length (Figures 1H and 1I). In some of the guanine-containing oligomers, we consistently found an elevated inherent level of cleavage, when compared with oligomers with other nucleotide sequences. This could be due to the highest oxidation potential of guanines among the 4 nucleotides (Spotheim-Maurizot and Davidkova, 2011, Steenken and Jovanovic, 1997). However, radiation-induced cleavage observed in the other three polynucleotides was always over and above the basal one, unlike in the case of G-rich oligomer. Normalization of the IR-induced cleavage intensity with that of the basal cleavage for respective oligomers revealed a distinct difference between G-rich and other nucleotide sequences.

The polyguanine sequences can fold into non-B DNA, G4 structures owing to the Hoogsteen hydrogen bonding between the four G residues; in addition to the typical Watson-Crick base pairing, it may have at the loop regions (Bochman et al., 2012, Mirkin, 2013, Nambiar et al., 2011, Sinden, 1994). In order to investigate the G4 structure-forming ability of the polyguanine tracts under study, the G-rich oligomer was subjected to circular dichroism (CD) analyses (Figures S1E and S1F). Results revealed that the guanine homopolymers folded into a parallel G-quadruplex, resulting in a typical absorption spectrum with a peak at 260 nm and a trough at 240 nm, as opposed to the C-rich and duplex C:G-rich controls (Figure S1E) (Nambiar et al., 2013, Neidle and Balasubramanian, 2006). Further, gel mobility shift assay results showed that the G-rich oligomer could fold into an intramolecular (faster migrating species as compared with the corresponding cytosine control), and an intermolecular G-quadruplex (slower migrating species), as compared with its C counterpart (Figure S1A, lanes 5,6). Moreover, CD studies revealed that the G-rich oligomer under study existed in a stable parallel G-quadruplex form even at a denaturing temperature of 95°C (Figure S1F), explaining the observed multiple bands (intra- and intermolecular G-quadruplex forms) even on a denaturing PAGE (Figures 1B and S1C). Thus, our results suggest that oligomeric DNA substrates capable of forming G4 structures in vitro exhibit resistance to IR-induced DNA cleavage.

G4-forming Human Genomic Regions when Present as Short DNA Fragments Are Protected from IR-induced DNA Breaks

Based on the above results, we were interested in investigating whether regions from the human genome known to form G4 structures, when present on a shorter oligomeric DNA, can exhibit a shielding effect against radiation. To test this, three independent human genomic regions, VEGF promoter, HIF1α promoter, and the telomeric repeats, known to form stable G4 structures were selected and the formation of the structure was confirmed by EMSA and CD analyses (Agrawal et al., 2013, Nambiar et al., 2011, Sun et al., 2011) (Figures S1H–S1J). DNA oligomers were allowed to fold into G-quadruplex and investigated for their radiation sensitivity, as compared with either complementary sequence or random sequence with same length and GC content (Figures 1J, S1H, and S1I). Interestingly, the G4 oligomers showed reduced cleavage intensity, in comparison with their respective complementary or random sequence, in the case of all three regions (Figures 1J, 1K, S1H, and S1I). Further, when double-stranded oligomeric DNA spanning three different regions from telomeric DNA were subjected to increasing doses of IR, a dose-dependent increase in radiosensitivity was observed (Figures 1L–1N). Thus, as a proof of principle, we found that three independent regions from the human genome that can fold into G-quadruplex structures exhibited reduced IR-sensitivity, as compared with the respective control DNA sequences. Although we observed background cleavage at G-rich oligomers even in absence of radiation, the result obtained was significant, because the substrate provided in either of the cases was not limiting, and the irradiation did not lead to an over and above increase in cleavage, unlike C-rich strand.

The Planar Quartet in a G4 Structure Is Resistant to IR-induced DNA Damage

Having studied the in vitro shielding effect of a G-quadruplex structure against IR, we were interested in exploring the mechanistic details of the phenomenon. A typical G4 structure is formed by Hoogsteen hydrogen bonding between the N7 position of guanine residues, resulting in a planar quartet composed of four such guanines (Mirkin, 2013, Nambiar et al., 2011, Sinden, 1994).These quartets are stacked upon each other and connected by a single-stranded intervening loop sequence comprising of the same strand (intramolecular G4) (Figure 2A) or different strands (intermolecular G4) (Bochman et al., 2012, Nambiar et al., 2011). In order to investigate the differential sensitivity between the planar quartet and the single-stranded loop region, we resorted to the DMS protection assay. Quadruplex-forming oligomers derived from G-rich sequence of BU1A gene (Schiavone et al., 2014) or a random DNA were irradiated, treated with dimethyl sulfate (DMS), and resolved on an 18% denaturing PAGE (Figures 2A and 2B). Interestingly, a distinct increase in the cleavage intensity at intervening loop sequences was observed upon irradiation (Figure 2B, lane 2). However, there was no significant change in the intensity of bands due to DMS chemical probing at the guanine residues involved in G4 formation (Figure 2B). In order to investigate the radiosensitivity of a guanine residue, which is not involved in structure formation, a random sequence containing stretch of guanines was examined following DMS protection assay. Results showed enhanced cleavage at the two individual guanines, as compared with those involved in intermolecular G4 formation (Figure 2B, lanes 3,4). Further, when an oligomeric DNA harboring either the human telomere repeat sequence (established to form a G4 structure in vitro) or the known G4 structure-forming sequences from BCL2 and HOX11 regions were employed, a consistent cleavage preference toward the intervening loop sequence was observed (Figures 2C and 2D), unlike the guanine stretches. Interestingly, a loop length of 2 nt was sufficient to result in IR-induced DNA break (Figures 2B and 2D). Thus, our results suggest that the planar quartet formed in a G4 structure is resistant to IR-induced DNA strand-breaks, as opposed to the single-stranded loop region.

Figure 2

Evaluation of Radiation Sensitivity of Different Forms of DNA Structure when Present on an Oligomeric DNA

(A) Strategy for evaluating radiosensitivity of planar quartet region in a G4 structure, as opposed to single-stranded loop region using DMS protection assay.

(B and C) Effect of radiosensitivity of guanine residues when present in the loop region as compared with a Hoogsteen hydrogen-bonded region when a G4 DNA derived from BU1A gene was used. The G4 motif and a random DNA (B) were evaluated for radiosensitivity of the loop region following DMS chemical probing. Guanine residues not involved in planar quartet formation are indicated by blue arrow. Assessment of telomere oligomers for differential sensitivity of the planar quartet (GGG) as compared with the intervening loop sequence (TTA) is also shown (C). Individual guanine position is marked by black arrows, whereas red brackets denote loop in both the panels. 100 mM KCl was used for stabilization of the G4 structures.

(D) Comparison of IR sensitivity at loop regions of various G4 DNA motifs derived from different genes (BU1A, Telomere, BCL2, HOX11a, and HOX11b). Sequences of DNA substrates used are shown. Guanines are indicated in red, whereas loop regions are in black. Radiation sensitive nucleotides are indicated using blue arrow. An oligomer containing random DNA sequence is also used for the study.

(E) Proposed model for shielding effects against radiation-induced damage by G4 planar quartets. Radiation affects water molecules in a cell (radiolysis) to form reactive oxygen species, which in turn causes single- and double-strand DNA breaks in the genome. However, the planar G4-quartet acts as a shield against radiation-induced DNA damage owing to potential reasons such as lower oxidation potential and hole trapping properties of the G4 quartet, thus imparting the property of radioresistance.

(F–I) Evaluation of radiosensitivity on DNA sequences that can fold into different hairpin structures. Schematic representation of evaluation of IR-induced DNA strand-breaks on hairpin DNA substrates, Hp1–Hp5 (F and G). PAGE profile showing radiosensitivity on different hairpin DNA structures. Following γ-irradiation (100 Gy), hairpin-forming oligomers were resolved on a denaturing PAGE, along with their respective unirradiated controls and analyzed for the abundance of DNA breaks (H). In each case, the intensity of IR-induced cleavage and its respective unirradiated control was quantified (n = 3) and plotted (I). Bar graph shows mean ± SEM.

(J) Schematic showing summary of observed sensitivity of different forms of DNA at the IR doses investigated.

See also Figure S2.

Formation of G-Quadruplex Structures Provides Shielding Effect against γ-Radiation-induced DNA Breaks

Previous studies have reported lower oxidation potential of G4 planar quartet, as compared with individual guanine residues (Choi et al., 2013, Lech et al., 2013, Wu et al., 2015). Based on our observations, and the information available in literature, we postulate a mechanistic model for shielding effect of a G-quadruplex structure against radiation-induced DNA damage (Figure 2E). IR induces radiolysis of water, leading to generation of highly reactive free radicals within the cell. Hydroxyl radicals cause breakage of the sugar-phosphate backbone by abstracting an H atom from the deoxyribose sugar unit in the DNA (Balasubramanian et al., 1998, Breen and Murphy, 1995). G4 structures exhibit reduced IR-induced DNA damage, possibly owing to the lower oxidation potential of the planar quartet, than individual guanines, and its free radical trapping property, which needs to be explored further (Figure 2E).

Evaluation of IR-induced DNA Breaks on Other Non-B-DNA Forms

Based on the observed radioprotection seen in G4 DNA, we were interested in studying the effects of irradiation-induced damage on other non-B-DNA structures such as hairpin and triplex DNA (Figures 2F and S2). Results showed that unlike G4-forming regions, five different hairpin-forming oligomers (Hp1–Hp5) were sensitive to IR (100 Gy) (Figures 2G–2I). Preliminary studies suggest that when triplex DNA was subjected to increasing doses of radiation (0, 10, 20, 50, 100, 200 Gy), there was no detectable increase in the DNA cleavage unlike duplex DNA control (Figure S2). Formation of triplex DNA was confirmed by native PAGE following incubation of radiolabeled duplex DNA with a third strand in appropriate incubation buffer (Figure S2B). The mechanism for resistance of triplex toward radiation-induced damage needs to be investigated further.

Therefore, our results reveal that different forms of DNA may provide differential sensitivity to radiation. Although a sequence-independent cleavage was observed when single-stranded DNA, double-stranded DNA, and hairpin DNA were irradiated, formation of G4 DNA and triplex DNA showed reduced or no sensitivity to radiation (Figure 2J).

Radiation Causes Minimal Damage at the Telomeres Inside Cells

The observation that telomere repeats when folded into a G-quadruplex structure can protect from radiation encouraged us to investigate whether telomeres on a human chromosome are sensitive to radiation inside cells, as it is well established that telomeric DNA can fold into G-quadruplex structures inside the cells. K562, MCF7, and HeLa cells were irradiated, allowed to recover partially, and assessed for DDR signals at the telomeres using immunofluorescence (green signal) and FISH (red signal) marking either 53BP1 or γH2AX and telomere repeats, respectively. The readout of the assay was based on the degree of colocalization between the two signals, post-irradiation, suggesting either intact telomere (no colocalization) or those harboring DNA strand-breaks (colocalization of signals) (Figure 3A). Each cell was analyzed for the extent of colocalization, independently in case of green over red and red over green signals, using ImageJ software and presented as independent IR (green colored) and IR (red colored) box-and-whisker plots.

Figure 3

Evaluation of Induction of DNA Double-Strand Breaks at Human Telomeric Region within the Cells Following Exposure to IR

(A) Strategy used for assessing IR-induced DNA breaks at telomeres inside cells. Cells were exposed to IR (10 Gy), incubated for 30 min to allow DNA damage marker proteins (53BP1 and γH2AX) to form distinct foci. Post-IF, cells were subjected to telomere FISH to mark the telomeres within cells, followed by colocalization analysis of the two signals.

(B) Representative IF-FISH images of K562 cells following irradiation with IR. K562 cells were irradiated, followed by IF to detect 53BP1 foci (green; FITC), FISH to demarcate telomeres (red; TTAGGG-Cy3), and DAPI staining for nucleus (blue). Multiple “Z stacks” were captured in case of every image, and the resulting images were stacked to give the “merged” image.

(C–E) Box-and-whisker plots depicting Mander's colocalization coefficient (range 0–1; “0” signifying no colocalization and “1” signifying 100% colocalization) as evaluated by JACoP plug-in of ImageJ software. Experiments were performed in two independent cell lines, K562 (C, D) and MCF7 (E) as described above, quantified, and presented. Although 53BP1 was used to detect DNA breaks in panel C, γH2AX was used for panels D and E. A minimum of 100 cells were analyzed for colocalization of red and green signals and plotted as the fraction of red overlapping green (box-and-whisker plot shown in red) and green overlapping red (box-and-whisker plot shown in green). The observed median value is indicated above the respective graph. Control refers to colocalization analysis performed in non-irradiated samples.

(F) Representative IF-FISH images following irradiation in HeLa cells. Irradiated cells were subjected to IF using anti-53BP1 (green; Alexa Fluor 488), FISH to demarcate telomeres (red; TTAGGG-Cy3), and DAPI staining for nucleus (blue). Multiple “Z stacks” were captured and the resulting images were stacked to give the “merged” image as above.

(G) Box-and-whisker plots depicting Mander's colocalization coefficient (range 0–1; “0” signifying no colocalization and “1” signifying 100% colocalization) as evaluated by JACoP plug-in of ImageJ software. Experiment was performed in HeLa cells as described above, quantified, and presented.

(H) Representative IF-FISH images showing localization of 53BP1 foci (green) and centromere (red). A Coste's mask, depicting merge of green and red foci (merged foci shown as white dots depict true colocalization), is also shown in the extreme right panel.

(I) Box-and-whisker plot depicting Mander's colocalization coefficient for green (53BP1) and red (centromere) signals. The median values have been indicated above respective graphs. Each experiment was analyzed by evaluating a minimum of 100 cells.

(J) Representative IF images showing colocalization of 53BP1 and γH2AX in HeLa cells following irradiation (10 Gy). 53BP1 is depicted in green (FITC), γH2AX in red (Alexa Fluor 568), and nucleus in blue (DAPI). Merged image, showing colocalized yellow foci is presented, along with Coste's masked image with white dots indicating true colocalization.

(K) Box-and-whisker plot showing Mander's colocalization coefficient (range: 0–1) for experiments shown in panel J.

(L) Co-IF for TRF2 and γH2AX in HeLa cells. Cells were irradiated (5 Gy), allowed to recover (30 min), and immunostained for TRF2 (FITC; green) and γH2AX (Alexa Fluor 568; red). Nucleus was counterstained using DAPI (blue), and a merged image consisting of all three channels is shown. Images were subjected to colocalization analysis as above and a Coste's mask depicting colocalization between red and green signals was generated.

(M) Box-and-whisker plot showing Mander's colocalization coefficient for experiments shown in panel L.

In panels B, F, H, J, and L, scale bar indicates 2 μm.

See also Figure S3.

Interestingly, we did not find significant colocalization between telomeric FISH signals and γH2AX or 53BP1 foci, when examined in K562 and MCF7 cells following irradiation, suggesting intact telomeres post-IR treatment (Figures 3B–3E, S3A, and S3B). As expected, sham control cells did not show much 53BP1/γH2AX staining, confirming IR-specific foci generation in the experimental samples (Figures 3B, S3A, and S3B). Furthermore, no colocalization of 53BP1 and FISH signals was detected post-irradiation, when an independent cell line, HeLa, was used for the study (Figures 3F and 3G). Colocalization analyses between 53BP1 foci and centromeric FISH signals following irradiation revealed significant colocalization coefficients, suggesting colocalization of both the signals elsewhere, unlike that observed in the case of telomeres (Figures 3H and 3I). Colocalization between γH2AX and 53BP1 served as positive control (Figures 3J and 3K).

We performed a co-immunofluorescence for TRF2, one of the members of shelterin complex (Lazzerini-Denchi and Sfeir, 2016) with γH2AX, following exposure to IR with a 30 min recovery period and found no colocalization between the two signals (Figure 3L). Taken together, our data suggest that radiation was unable to induce DSBs at the telomeric ends. Further, to investigate whether knockdown of TRF2 can lead to disassembly of shelterin complex allowing recruitment of 53BP1 to telomere ends, we performed IF-FISH following TRF2 knockdown in HeLa cells (Figure S3C). Results showed significant colocalization of telomere probes and 53BP1 (Figures S3D and S3E) as reported before (Lackner et al., 2011, Mao et al., 2007). Thus, these results show that there is no colocalization of DSBs with telomere, upon irradiation, when multiple cell lines were used.

Genome-wide Presence of G4 Motifs Explains Variation in Radiation Sensitivity among Chromosomes

In order to investigate variation in IR-sensitivity across the genome, we analyzed human genomic data derived from control and irradiated (5 Gy) samples. Data were downloaded from the online SRA study (Project ID: ERP004219), mapped, and aligned against the hg38 reference genome and viewed in the Integrated Genome Viewer (IGV). DNA breaks induced following exposure to radiation interrupt the continuity of the genome, leading to loss of reads on such templates. Interestingly, results showed that multiple locations across the genome exhibited stretches of at least 1000 base pairs with no reads (0 coverage), whereas corresponding locations in unirradiated controls showed acceptable coverage (≥5 reads) (Figure 4A). The regions with zero reads are termed as “damaged sites” (stretches of broken DNA) in further study.

Figure 4

Evaluation of Human Genome Sequences for Their Sensitivity toward IR

(A) DNA sequences from human genome were evaluated to determine the unprotected regions and the frequency of occurrence of G-quadruplex region in it compared with protected sequences. A minimum length of 1000 bp or greater was analyzed from whole genome. The human genome sequenced files deposited in SRA (Sequence Read Archive) database; sra-id: ERP004219 from control and irradiated human cells (5 Gy) have been downloaded and used for the study. The control input DNA sequence shows aligned reads, whereas the irradiated sample did not have reads in several locations.

(B) A representative set of 1% of unprotected regions in every chromosome. Chromosomes with higher amount of breakage are colored in red and the ones with less breakage are in green. Every bar represents a chromosome with the corresponding length to the scale and each line on the bar represents the location of the unprotected region.

(C) A scatterplot showing inverse correlation between percentage of unprotected length in each chromosome versus the number of G4 motifs per unprotected region.

(D) Stacked bar graphs showing genome-wide analysis of number of G4 motifs harbored by protected and unprotected regions of irradiated CAL51 cells, in the promoter (left panel) and coding DNA sequence (right panel). In both the cases, the white bar represents the percentage of total sequence analyzed, whereas the black bar represents percentage of G4 motifs.

See also Figure S4.

Having defined the broken regions in the human genome upon irradiation, percentage of damaged sites across chromosomes and the number of G4 motifs in the damaged regions (using the online Quadparser database) (Wong et al., 2010) were assessed (Figure 4B). Results showed that although overall GC content of chromosomes was comparable (~42%), chromosome 9, 13–16, 21, and 22 were more sensitive to radiation (Figure 4B). Importantly, these chromosomes harbored less number of G-quadruplex-forming motifs.

The retrieved data was further analyzed using hclust. Results showed two defined clusters of the chromosomes, with a median of 15% (Figures 4B and 4C). Interestingly, among the unprotected regions, chromosomes with higher DNA damage (>15%) harbored less number of G4-forming motifs (8–25/mb), whereas chromosomes with <15% damage harbored higher number of G4 motifs (35–72/mb) (Figure 4C). The regions of higher damage were 6–15 kb in length unlike those with lesser damage (1–2 kb) and harbored less G4 motifs. This suggested a significant inverse correlation between the number of G4-forming motifs, and extent of damage within the genome, upon irradiation (Figure 4C). The above data were statistically significant, as assessed and confirmed using multiple statistical tests such as Student's t-test (p = 0.0001), Mann-Whitney U test (p = 0.0001), and Pearson correlation coefficient (R value of −0.82). Although the GC content of each chromosome was observed to be in the same range, noteworthy variations in IR sensitivity among chromosomes suggest that G4 structures are responsible for the observed differential sensitivity to radiation.

Further, to investigate the distribution of G4 motifs and IR sensitivity in promoter regions, the sequences from unprotected regions were segregated and fragments mapping to the promoters were determined. Interestingly, only 6% of the unprotected fragments mapped to promoters, whereas the rest were at the protected region (Figure 4D). Importantly, 98.3% of the G4-forming motifs were present at the protected regions. When coding regions were analyzed in a similar manner, 4% of the unprotected fragments mapped to the CDS, which harbored only 11% of G4 motifs (Figure 4D). In contrast, 89% G4 motifs were present in the protected CDS (Figure 4D).

Similar study was extended to an irradiated human sample (10 Gy), which was downloaded from the SRA study (Project ID: SRP022845). Irradiated DNA sample was analyzed as described above using same parameters. Results showed that the regions of chromosomes with more DNA damages (>7%) harbored less number of G4-forming motifs (3–47/mb) per damaged region, whereas chromosomes with less damaged region (<7%) harbored more number of G4 motifs (92–320/mb), suggesting an inverse correlation as observed above (Figures S4A and S4B). Statistical evaluation using Student's t-test and Mann-Whitney U-test (p value<0.0001) further confirmed the level of significance. An inverse correlation was confirmed using Pearson correlation analysis (R value −0.80). High-throughput sequence analyses of unprotected regions, which were classified as promoters, and CDS based on their mapping to respective regions on the chromosome revealed ~22% of the unprotected fragments both from promoters and CDS with 11 and 0.4% G4 motifs, respectively (Figure S4C). In contrast, 89 and 99.6% G4 motifs were present in protected regions of promoters and CDS, respectively (Figure S4C). Thus, results suggest that irrespective of the dose used (5 or 10 Gy), presence of G-quadruplexes contributes to reduced radiosensitivity in the genome.

G4-Forming Regions in Cells Remain Intact upon Radiation Exposure

To experimentally confirm the above findings, twelve regions known to form G4 structures and matching control regions (that did not harbor G4 forming motifs) were amplified from genomic DNA isolated from human cells (Nalm6), post-irradiation. Because DNA breaks generated following irradiation can result in reduction in the number of template strands available for amplification, efficiency of PCR will be less compared with an unirradiated control. Thus, the difference in amplification between control (−IR) and irradiated (+IR) genomic DNA can serve as a measure to assess the intactness of the particular genomic DNA region (Figures 5A and S5A). Genomic DNA was irradiated with an increasing doses of IR (1, 1.5, and 2 kGy), and formation of subsequent DNA strand-breaks was confirmed on an agarose gel (Figure S5B). In order to score for the reduced template due to breakage of DNA strands in genomic regions using PCR, we employed a high radiation dose. Interestingly, random regions showed a dose-dependent decrease in amplification, as compared with that of the controls, suggesting IR-induced DNA breaks at the template sequence (Figures 5B, 5C, and S5C). Surprisingly, in case of regions containing G4-forming motifs, nine out of ten regions showed no significant difference upon amplification, suggesting the presence of intact sequence post-irradiation (Figures 5B and 5C). Interestingly, it has been shown that sequences with alterations in the canonical formula (for example, those involving larger loop lengths, DNA bulges, two-plate sequence, GNG motifs etc.) can also fold into G-quadruplex forms (Chambers et al., 2015, Das et al., 2016). Although most of the regions under study possessed canonical G-quadruplex-forming sequences, four regions were capable of forming G4-DNA structures (as analyzed by multiple softwares including Quadbase and QGRS mapper) (Kikin et al., 2006, Yadav et al., 2008) inspite of harboring noncanonical G-quadruplex sequences. KRAS, which exhibited reduction in PCR amplification efficiency even though a G4 motif was present, warrants further investigation (Figures 5B and 5C).

Figure 5

Real Time Analyses of Impact of Various G4 Motifs on DSB Generation in the Human Genome Following Irradiation

(A) Experimental strategy to assess intactness of independent G4-forming motifs inside the cells in comparison with size-matched random sequences, which do not support formation of G4 structures. Cells were irradiated with increasing dose of γ-radiation (1, 1.5 and 2 kGy), genomic DNA was extracted immediately and used for PCR amplification of region of interest. Each set of G4-forming regions and random ones was evaluated by genomic polymerase chain reaction (PCR) and real-time PCR.

(B) Gel profile showing genomic PCR products for control and irradiated samples (1, 1.5, and 2 kGy) for G4-forming regions, VEGF, MYC, CKIT, KRAS, WNT(1), WNT(2), HOX11(1), HOX11(2), PPPC, and NEUROMEDIN (right panel), and regions devoid of G4 motifs, MYOD, PU.1, RAG2, BCL2(1), BCL2(2), VEGF, MYC, GBT, RAG1, and MYC(2) (left panel). The size of each region amplified has also been indicated.

(C) Bar graphs depicting percentage amplification in various regions forming G4 structures (lower panel), as compared with the region devoid of a G4 motif (upper panel). For each region, the amplification of irradiated samples is normalized with respect to that in the control sample (considering it as 100%). Red dotted lines demarcate highest and lowest amplification values on an average for both the datasets, denoting the difference in amplification in each sample.

(D) Representative real-time PCR profiles of G4-forming region of VEGF (upper right panel), its random counterpart (upper left panel), G4-forming MYC (lower right), and its random sequence (lower left) are presented. Individual amplification profiles for each dose have been enlarged, boxed, and indicated with arrow and shown on the right in all 4 profiles.

(E) Box-and-whisker plot showing relative change in Ct value in case of twelve G4-forming motifs (white boxes) and ten random regions (black boxes), as determined by real-time PCR. The height of the bar indicates difference in amplification in control, as compared with that in irradiated samples, indicating a measure of intactness of the respective region.

(F) Bar graph showing change in Ct value across all twenty-two regions tested, normalized with respective unirradiated control value. In each case, G4-forming regions (white bars) and random regions (black bars) are plotted for three independent radiation doses with error bars showing mean ± SEM (ns: not significant, *p < 0.05, **p < 0.005, ***p < 0.0001).

See also Figure S5.

To verify the above findings, the amplification was performed using real-time PCR and changes in Ct values obtained for control and irradiated samples were analyzed. Interestingly, the G4-forming motif present in VEGF did not show a significant change in the Ct value upon irradiation, whereas an identical sized region devoid of a G4-forming motif, taken from the same gene, showed an IR-dose-dependent increase in the Ct value (Figure 5D, upper panel). Results were reproducible in an independent region of MYC promoter (harboring G4 motif), when compared with a random region from the same gene (Figure 5D, lower panel). Analyses of 22 independent genomic regions showed significant differences between amplification in control and IR-treated samples (Figure 5E). The set of ten random regions showed a significant increase in Ct values when irradiated samples were used, whereas it was minimal for G4-forming regions (Figure 5F). Taken together, real-time analyses of irradiated DNA samples demonstrated noteworthy protection by G4-forming regions, as compared with the rest of the genome. It is known that G4 structures stall polymerases in vitro and in vivo, and hence could lead to a decreased amplification of those regions upon PCR. However, in our analyses, amplification in each irradiated genomic region was compared with its respective unirradiated control to gauge the intactness of that region and thus, this caveat did not bias the observation.

Formation of G4 DNA Contributes to Genome-wide Radioprotection and Can Be Modulated by G4 Resolvase Inside Cells

To evaluate formation of G4 structures in a genome-wide manner, we resorted to the use of a previously characterized G-quadruplex-specific antibody, BG4, which binds to G4 structures inside cells (Biffi et al., 2013, Biffi et al., 2014). Following irradiation (10 Gy), HeLa and MCF7 cells were subjected to IF using purified anti-BG4 (Figures S5D and S5E) and anti-γH2AX and analyzed for colocalization signals (Figures 6A and 6B). Results showed no significant colocalization between the damage-induced foci and the BG4-bound quadruplex regions, suggesting that G4 structures inside cells were unaffected, post-IR (Figures 6A–6D). Interestingly, use of low and high doses of irradiation (5, 20 Gy) did not show any significant colocalization between γH2AX foci and the BG4 foci (Figures 6E and 6F). Further, to test whether stabilization of G4 structures inside cells would impart additional protection against IR-induced DNA damage, we used a known G4 stabilizer, TMPyP4, to stabilize quadruplexes in HeLa and MCF7 cells, followed by exposure to radiation (10 Gy) and assessment of DNA damage by 53BP1 staining (Figures 6G and S6A). Results showed a significant increase in the number of cells devoid of 53BP1 foci, along with a concomitant decrease in cells harboring multiple foci, upon TMPyP4 treatment, as compared with irradiation alone (Figures 6G–6I and S6A). Although DNA damage induction by TMPyP4 is reported previously (Cheng and Cao, 2017), in order to investigate the radiosensitivity upon G-quadruplex stabilization, we employed a lower dose of the stabilizing agent that did not induce any damage by itself (data not shown). Further, we investigated the effect of an independent G-quadruplex stabilizing agent, pyridostatin, on radiation-induced DNA DSBs in cells by employing both immunofluorescence and comet assays. Although pyridostatin by itself has been shown to induce DNA damage (Rodriguez et al., 2012), (data not shown), use of a lower concentration (2 μM) did not result in any DSB induction on its own. 53BP1 staining after pyridostatin treatment showed significant decrease in DNA damage-induced foci in MCF7 cells owing to G-quadruplex stabilization; however, the impact was limited in HeLa cells (Figures 6J, 6K, and S6B). Comet assay results revealed decreased DSB-induction in pyridostatin-treated cells, after radiation, as compared with irradiation alone controls in HeLa cells (Figures S6C and S6D).

Figure 6

Assessment of Genome-Wide Radio-Protection Contributed by G4 Structures Inside Cells

(A and B) Representative immunofluorescence images to determine colocalization of G-quadruplex structures and IR-induced DNA breaks. Irradiated HeLa (A) and MCF7 (B) cells (10 Gy) were incubated for 30 min to allow DNA damage response to initiate and evaluated by IF using anti-BG4 antibody (Alexa Fluor 488; green) and anti-γH2AX antibody (Alexa Fluor 568, red). Nucleus is stained with DAPI (blue). The channels were merged and subjected to colocalization analysis using JACoP plug-in of ImageJ software.

(C and D) Box-and-whisker plot showing quantification of the number of BG4 or γH2AX foci per cell in HeLa (C) and MCF7 (D) cells. Number of cells that show merge of both antibodies are also shown. In each case, a minimum of 100 cells were counted and the resulting population analyses has been shown.

(E) Representative immunofluorescence images showing colocalization of G-quadruplex structures (detected using BG4) and DNA breaks (detected using γH2AX). Irradiated HeLa cells (5, 10, and 20 Gy) were evaluated as described above.

(F) Box-and-whisker plot showing quantification of the number of foci per cell (γH2AX alone, BG4 alone, and merged) following exposure to increasing dose of IR (5, 10 and 20 Gy) in each case, a minimum of 50 cells were counted and presented. “γH2” represents γH2AX.

(G) Evaluation of impact of G4 DNA stabilizers on radiation-induced DNA strand-breaks within HeLa and MCF7 cells. Representative images showing effect of TMPyP4 on radiation-induced breaks in HeLa cells (G). Cells were treated with TMPyP4 (5 μM, for 5 hr), irradiated (10 Gy; 30 min recovery period), and immunofluorescence was performed using anti-53BP1. The images show 53BP1 binding as red foci (Alexa Fluor 594) and nucleus as blue (DAPI). Merged images comprising of both the channels are shown in the lower panel.

(H and I) Bar graphs showing quantification of 53BP1 foci per cell following irradiation in cells treated with TMPyP4 in HeLa (H) and MCF7 (I), as described above.

(J and K) Bar graphs showing quantification of 53BP1 foci per cell following irradiation (10 Gy; 30 min recovery period) after treatment with G4 stabilizer pyridostatin (PDS, 2 μM; 5 h) in HeLa (J) and MCF7 (K).

In panels A, B, E, and G, scale bar indicates 5 μm. In panels H–K, a minimum of 200 cells were counted and in each case bar graph shows mean ± SEM.

See also Figures S5C, S5D, and S6.

Werner (WRN) is one of the RecQ family helicases, known to resolve G-quadruplex structures in the genome (Chu and Hickson, 2009, Mendoza et al., 2016). We investigated the impact of modulation of WRN expression on radiosensitivity in HeLa cells. Interestingly, we observed significant decrease in radiation-induced DSBs upon knockdown of WRN helicase, as compared with scrambled control transfection (Figure 7). In contrast, overexpression of WRN resulted in significant increase in the number of 53BP1 foci (Figures 7B–7D). Further, DSB formation by IR was also investigated after WRN knockdown followed by its overexpression (Figure 7B). Results showed significantly improved number of 53BP1 foci upon irradiation in these cases (Figures 7C and 7D). Similar results were also observed when Nalm6 cells were irradiated following knockdown or overexpression of WRN using comet assay or 53BP1 foci formation assay (Figure S7).

Figure 7

Modulation of Radiosensitivity of Genome Following Downregulation or Upregulation of WRN Helicase Inside Cells

(A and B) Representative Western blot showing level of WRN helicase in HeLa cells following either its knockdown by transfecting shRNA plasmids, upon WRN overexpression or after WRN overexpression following its knockdown. A plasmid with scrambled sequence was used as control during transfection. Ponceau-stained blot served as a loading control.

(C) Representative immunofluorescence images showing 53BP1 foci formation (red, Alexa Fluor 594) following irradiation (10 Gy) after modulation of WRN expression. Nucleus was stained with DAPI (blue), and merged image of both is shown in the lower column. “Sec. control” is secondary antibody control (scale bar, 2 μm).

(D) Bar graph showing quantification of number of 53BP1 foci upon irradiation in case of WRN knockdown, overexpression, and knockdown followed by overexpression as compared with transfection control. A minimum of 100 cells were analyzed for each sample, 53BP1 foci counted and plotted as bar graph using GraphPad Prism 5 software depicting mean ± SEM (ns: not significant, *p < 0.05, **p < 0.005, ***p < 0.0001).

See also Figure S7.

Taken together, our findings reveal that stabilization or modulation of G-quadruplex resolving helicases inside cells has a direct impact on radiosensitivity, thus further establishing the role of G-quadruplex structures in shielding the genome against radiation.

Discussion

Our findings establish that the human genome is not uniformly susceptible to ionization radiation and hence DNA strand-breaks. Sequences of the genomic region play a significant role toward its sensitivity to IR. Specifically, our data demonstrate that G-quadruplex structures when formed at GC-rich stretches of genome could reduce sensitivity to radiation-induced DNA damage, as compared with their B-DNA counterparts within cells.

Conformation of the DNA, but Not the Sequence, Governs Sensitivity to Radiation

IR affects DNA in a multitude of ways such as base or sugar damage, single- and DSB formation, DNA inter-strand cross-linking, and DNA-protein cross-linking (Lomax et al., 2013). Oxidative damage to the DNA results in formation of various products such as 8-oxo-guanine; 8-oxo-adenine; 5,6-thymine glycol; 2-hydroxyadenine; and FapyA and FapyG lesions (Cooke et al., 2003). Among the four DNA bases, Guanine is the most sensitive to oxidative damage by IR, 8-oxo-G being the most common and stable oxidation product (Kasai and Nishimura, 1984, Ohno et al., 2006). In line with this, previous studies have reported higher oxidation and base alteration tendencies at telomeric sequences (G-rich), as compared with others (Oikawa and Kawanishi, 1999). Also, higher one-electron oxidation of a single 8-oxo-G when present in a G-quadruplex was observed, as compared with that in a duplex form (Szalai et al., 2002). In our study, experiments using piperidine treatment ruled out the possibility of a G-quadruplex structure harboring excessive modified bases post-irradiation. Thus, although individual guanine bases are the most sensitive to oxidative damage induced by IR, a 35-mer G-quadruplex structure exhibited lower oxidation sensitivity in our study. This is consistent with our data for IR-induced strand-breaks, and thus reiterates a structure-specific, but not sequence-specific, mechanism for the observed reduction in radiosensitivity. Few studies from the literature have shed light on the differential oxidation reactivity of the Guanine moiety when present in a G-quadruplex form, in comparison to that in a duplex form (Fleming and Burrows, 2013). In one of the studies, authors reported an extensive oxidation product profile for the human telomeric G-quadruplex structure, compared with a duplex DNA form, along with differential patterns of oxidation sensitivity within independent G4-DNA forms (hybrid, propeller, basket etc.) (Fleming and Burrows, 2013). Interestingly, they observed reduced base release upon sugar oxidation in G4-DNA (<10%), when compared with a duplex form (~20%), which is consistent with our current findings.

Previous studies have suggested bond strength difference between AT and GC pairs, as one of the plausible reasons underlying variation in radiation sensitivity at GC-rich sequences (Kaplan et al., 1964, Wu et al., 2012). Although GC-richness of the genome alone is not sufficient to influence radiosensitivity of an organism, a GC-rich DNA sequence will possess higher propensity of G4-forming motifs, as compared with an AT-rich one. Thus, our finding that G-quadruplex structures are much less sensitive to radiation provides a rational interpretation to the observed GC content bias in radioresistance. G-quadruplex structures have been shown to exist in several regions of the genome, contributing to a number of vital cellular processes (Hansel-Hertsch et al., 2017). In a genome context, analyses of independent irradiated human genome sequence data revealed significant negative correlation between the number of G4s in each chromosome and the propensity of that chromosome to harbor DNA breaks. Transcriptome analysis coupled with evaluation of G4seq data revealed that in case of highly expressed genes, the unprotected regions harbored significantly lower number of G4-containing promoters (data not shown). Thus, our study reveals a G4-dependent bias in radiation damage of individual chromosomes, irrespective of their similarity in GC content (~42%). Consistent to this, we did not observe colocalization of G-quadruplex-specific antibody (BG4) and radiation-induced DNA breaks (γH2AX or 53BP1 foci). Direct analysis of several genomic regions by both semi-quantitative and real-time PCR further confirmed that most of the genomic regions that form G-quadruplex were insensitive to radiation. Hence, our data establish that radiation sensitivity to human genome is nonrandom and is dictated by DNA conformations.

Radiation Dose and Radiosensitivity Sensitivity of Biological Systems

Exposure to IR is encountered at different instances and at varying intensities. From a daily background radiation exposure of ~5 μSv, a range of doses are encountered during thoracic X-ray (~1 mSv), diagnostic CT (1–10 mSV), or radiotherapy against cancer (up to 100 Sv or Gy) (Lobrich and Jeggo, 2007). Previous studies estimated that exposure to 1 Gy of γ-radiation can induce ~1000 SSBs and 50–100 DSBs in a single cell (Mullenders et al., 2009). Thus, the extent of damage at different instances might vary according to the dose encountered. In the present study, we employed a wide range of radiation doses, from 2 Gy to 2 kGy, depending on the sensitivity of the technique under study. For example, although 2 Gy radiation dose was sufficient for detection of significant DDR repair foci within a cell, a dose of 100 Gy was indispensable in gel-based assays, to ascertain cleavage at every nucleotide, ensuring optimum sensitivity of the assay. Evaluating effect of radiation on genome-wide G4-forming regions using PCR-based approach necessitates the use of higher IR doses (0.5–2 kGy) to ensure strand-breakage at every ~200 nt. Thus, our findings reveal that reduced radiosensitivity of regions harboring G-quadruplex motifs, as assayed using various experimental strategies, will have implications at cellular level even at low radiation doses encountered by an organism.

G4 Structures at Telomeres Exhibit Reduced Sensitivity to Radiation

Among various genomic regions known to harbor G4 structures, telomeres is a well-established example, owing to the elevated GC content and single-stranded nature of the telomeric overhang (Bochman et al., 2012, Bryan and Baumann, 2011). Telomere dysfunction is a complex biological process that has been shown to occur due to a number of factors such as genome instability, senescence, DNA repair defects, improper functioning of telomerase, and aging (Hewitt et al., 2012, Jurk et al., 2014). The role of G-quadruplex structure in DNA end protection at the telomeres is well established (Bochman et al., 2012). However, DNA damage at this site is only one of the several factors that might contribute to overall telomere defects/dysfunction. Although DNA damage at telomeres has been investigated, it is quite challenging to address the end result of an IR-induced DNA break at telomeres in the absence of shelterin complex and compare it to the frequency of damage at a non-telomeric region in the chromosome. In an interesting study, Doksani and Lange reported activation of DDR at telomere-internal DSBs, as assessed using a Fok1 endonuclease-tagged TRF1 mediated system (Doksani and de Lange, 2016). The findings reveal that shelterin-mediated suppression of NHEJ, alt-NHEJ, or the HR pathway is not apparent at a telomere, when a DSB was generated internally, highlighting the possibility of active DDR in case of a strand-break even at telomeres. In our studies, we find that human telomeric repeat sequences are shielded from IR-induced DNA breaks when present on a single-stranded DNA or plasmid DNA. More importantly, absence of colocalization of DNA strand-breaks (γH2AX or p53BP1 foci) and G-quadruplex structures (FISH signals) at telomeric ends following irradiation of the cells confirm the above findings and reveal the role of G-quadruplex in radioprotection. Taken together, our data reveal that G4 structures formed at telomeres play a significant role in safeguarding the chromosome ends in cells against radiation-induced DNA strand-breaks. However, once induced, DNA damage at the telomeres is known to persist owing to the absence of DDR, mediated by the shelterin complex (de Lange, 2005, Hewitt et al., 2012).

Levels of G-Quadruplex Resolving Helicase; WRN Modulates Radiosensitivity in Cells

G-quadruplexes are dynamic structures that are well regulated inside cells by means of various stabilizing factors and resolving helicases. WRN, one of the RECQ family helicases, possesses 3′-5′ exonuclease, 3′-5′ helicase and single-strand DNA annealing activities (Huang et al., 2000, Orren et al., 1999). Previous studies have observed ATP-dependent and -independent helicase activities, thus unwinding secondary structures such as G-quadruplexes throughout the genome, including those at the telomeres, in vivo (Mendoza et al., 2016). Our study revealed that WRN-mediated unwinding of these structures in cells resulted in elevated radiosensitivity, whereas knockdown of WRN or stabilization of G-quadruplexes led to a reduction in radiosensitivity. Thus, perturbation of G-quadruplex structures inside cells, either by stabilizing factors or by resolving helicases, modulates their susceptibility to radiation in vivo.

Further, WRN also plays important roles in various cellular processes such as DNA repair and replication, telomere maintenance, and transcription (Bernstein et al., 2010, Croteau et al., 2014). These functions are regulated by several factors such as post-translational modifications, protein-protein interactions, preferential activity of particular protein domains (helicase, in resolving G4-DNA vs. helicase and exonuclease, both, for DNA repair activity). Previous studies have observed modest increase in radiation sensitivity in the absence of WRN owing to reduced DSB repair (Saintigny et al., 2002). It is important to delineate the current observations of reduced damage induction upon WRN knockdown, immediately post-IR, as compared with other studies in literature, which assess DNA DSB repair efficiency and cell death after a significant time period post-IR (e.g., 9–15 days in this study) (Yannone et al., 2001). Decreased repair efficiency upon irradiation and subsequent cell death in the absence of WRN is well attributed to its role in DNA DSB repair and could be potentially distinct from its G-quadruplex resolving abilities in the cell. Our findings reveal that perturbation of G-quadruplex resolving helicases inside a cell can modulate the extent of IR-induced DNA damage induction. However, its effect on DNA repair and overall survival could be variable depending on the multi-faceted functions of the independent helicases.

Other Factors Contributing toward Radioprotection of Genome

It is presumed that sensitivity of a cell toward radiation is affected by a number of factors such as genome organization, nuclear microenvironment, hypoxia, DNA repair capacity, cell type, proliferation rate, and cell cycle phase. Polyamines such as spermine and spermidine, essential for growth and survival of mammalian cells, are well-studied scavengers of oxygen-radical species, which protect the genome against oxidative damage, and thus, are key mediators of cellular redox balance in a cell (Douki et al., 2000, Murray Stewart et al., 2018).

Influence of genome compartmentalization (euchromatin/heterochromatin) on sensitivity to IR-induced DNA damage and repair has been long debated in the field. Although some groups believe that heterochromatin is indeed less susceptible to IR-induced DNA damage as compared with euchromatin owing to DNA compaction, others report distinct differences in DDR and downstream repair between the two compartments (Chiolo et al., 2013, Falk et al., 2008, Lorkovic et al., 2017). Fascinating studies in eukaryotes have revealed differential DDR pattern in heterochromatin upon induction of IR-induced DNA damage, wherein induced DSBs are shifted away from heterochromatic regions before processing for repair (Chiolo et al., 2011, Ryu et al., 2015). Preliminary studies suggest that both heterochromatin and euchromatin regions showed susceptibility to IR, although cells that showed colocalization of DSB marker (53BP1) and euchromatin marker (AcH3K9) or heterochromatin marker (TIF1β) were seen less often (only ~7% of the cells; Figure S8). Although multiple factors affecting radiation sensitivity in a cell have been addressed previously, to our knowledge, this is the first report of a DNA secondary structure modulating IR-induced DNA damage.

Mechanism of G4-mediated Radioprotection and Its Impact on DNA Rearrangements

The findings presented here help to comprehend the plausible mechanism by which a quadruplex structure is protected by the action of radiation. Elegant experiments using flash photolysis and pulse radiolysis techniques, have revealed “hole trapping” property of a G-quartet structure (Figure 2E) (Choi et al., 2013, Lech et al., 2013, Wu et al., 2015). A single guanine residue possesses highest oxidation potential among the four nucleotides, whereas a G4 planar quartet displays low oxidation potential (Choi et al., 2013). Consistent with these observations, we observed selective susceptibility of the single-stranded loop sequence, as opposed to the resistant planar quartet in a G4 structure. Thus, the observed radioresistance of G-quartet structures can be well attributed to Hoogsteen hydrogen bonding between the guanine residues, as it may contribute to the differential radiosensitivity observed between a single guanine and that of a G-quadruplex structure (Figure 2E). Breakage at guanine in the single-stranded loop region reiterates the finding that structure of the DNA plays a vital role in imparting radioprotection to a G-quadruplex form of DNA. DNA strand-breaks are the primary requisites for genome rearrangements such as chromosomal translocations, inversions, and deletions. Our study establishes a GC-rich bias in protection of genomic sequences against radiation-induced breaks, making AT-rich sequences more prone to breakage, and subsequent rearrangements. For example, AT-rich Alu repeats have been shown to be involved in MLL rearrangements (Mani and Chinnaiyan, 2010). Thus, our findings substantiate the observed sensitivity of AT-rich sequences toward radiation-induced DNA damage and rearrangements in the genome.

Differential Sensitivity of the Genome to IR: implications in Cellular Functions and Evolution

In addition to the skewed GC content in many radioresistant bacterial genomes, studies have also predicted a high number of G4-forming motifs in Deinococcus radiodurans (Kota et al., 2015). Considering this, it is easy to envisage a survival advantage gained by these bacteria, by virtue of G-quadruplexes in the genome, highlighting an important role of these structures in the evolution of radioresistance. Nonetheless, other factors such as elevated DNA repair, compaction of the genome, protection against oxidative stress etc., might contribute toward radioresistance in general, as is hypothesized in the famously AT-rich radioresistant organism, Dictyostelium discoideum (Deering, 1968).

Type and nature of DNA damage often dictates the repair pathway choice operating at the damaged site. Since our findings indicate an increased sensitivity of AT-rich sequences as opposed to GC-rich regions, we predict an AT sequence-biased evolution of DNA repair proteins operating in the cell, in order to maintain genomic integrity. Artemis, a versatile exonuclease and endonuclease involved in NHEJ has been shown to specifically resect AT-rich sequences but not those harboring GC (Chang et al., 2015). However, this needs to be investigated further. The findings presented here could also improve our understanding on the potential mechanism of radioresistance in cancer cells, which has important clinical implications. This information could be effectively harnessed for developing novel cancer treatment modalities, especially against radioresistant cancers.

Overall, this study establishes a paradigm shift in our understanding of the distribution of radiation-induced DNA breaks within the genome. We anticipate that our findings will help shed light on the evolution of GC content of an organism, regulation of multiple cellular processes, and differential radiation sensitivity among organisms.

Limitations of the Study

Since G-quadruplexes are highly dynamic and regulated structures, it is important to note that all the G-quadruplex-forming motifs analyzed in our study will not exist as secondary structures at a given time inside the cell. Since several experiments are based on population analysis, this caveat may not affect our interpretations. Considering that G4-forming motifs account for about 2% of whole genome, it may be interesting to investigate the role of flanking sequences and other non-B DNA structures such as Triplex and sticky DNA in conferring radioresistance. Further, detailed studies are required to understand the contribution of chromatin organization toward radiosensitivity of an organism.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.