Nucleotide Excision Repair and Impact of Site-Specific 5’,8-Cyclopurine and Bulky DNA Lesions on the Physical Properties of Nucleosomes

Abstract

The non-bulky 5’,8-cyclopurine DNA lesions (cP) and the bulky, benzo[a]pyrene diol epoxide-derived stereoisomeric cis and trans-N²-guanine adducts (BPDE-dG) are good substrates of the human nucleotide excision repair (NER) mechanism. These DNA lesions were embedded at the In or Out rotational settings near the dyad axis in nucleosome core particles reconstituted either with native histones extracted from HeLa cells (HeLa-NCP), or with recombinant histones (Rec-NCP). The cP lesions are completely resistant to NER in human HeLa cell extracts. The BPDE-dG adducts are also NER-resistant in Rec-NCPs, but are good substrates of NER in HeLa NCPs. The four BPDE-dG adduct samples are excised with different efficiencies in free DNA, but in HeLa-NCPs the efficiencies are reduced by a common factor of 2.2 ± 0.2 relative to the NER efficiencies in free DNA. The NER response of the BPDE-dG adducts in HeLa-NCPs is not directly correlated with the observed differences in the thermodynamic destabilization of HeLa NCPs, the Förster Resonance Energy Transfer (FRET) values, or hydroxyl radical footprint patterns, and are weakly dependent on the rotational settings. These and other observations suggest that NER is initiated by the binding of the DNA damage-sensing NER factor XPC-RAD23B to a transiently opened BPDE-modified DNA sequence that corresponds to the known footprint of XPC-DNA-RAD23B complexes ( ≥ 30 base pairs). These observations are consistent with the hypothesis that post-translation modifications and the dimensions and properties of the DNA lesions are the major factors that have an impact on the dynamics and initiation of NER in nucleosomes.

Table of Contents

INTRODUCTION

The human nucleotide excision repair (NER) apparatus is an important cellular defense system that excises an astounding variety of structurally diverse DNA lesions.¹ Among examples of NER substrates are bulky DNA lesions like those derived from the binding of metabolically activated polycyclic aromatic hydrocarbons (PAH) to DNA.² Certain non-bulky DNA lesions like the oxidatively generated 5’,8-cyclopurine adenine (cdA) and guanine (cdG) lesions (Figure 1) are also substrates of NER,^{3, 4} but not of base excision repair mechanisms (BER).⁵ The relationships between the structural properties of DNA lesions and their susceptibilities to removal by NER mechanisms are of significant interest.²

Figure 1.

Structures of the diastereomeric 5’,8-cyclopurine lesions and the stereoisomeric benzo[a]pyrene diol epoxide-derived BPDE-dG lesions.

The stereoisomeric 5’,8-cyclopurine lesions^6–8 (Figure 1) are formed as a result of chronic infection and inflammation⁹ induced by infectious and environmental factors, and are highly mutagenic.^10–15 Some of these lesions have been detected in tissues of experimental animals,^{16, 11, 17–20} and humans,^{21, 22} and are believed to contribute to human breast and ovarian cancers.^{23, 24}

Benzo[a]pyrene (B[a]P) is a typical PAH that is hydrophobic and practically insoluble in aqueous solutions. However, in mammalian cells it is metabolically activated to the reactive and highly tumorigenic diol epoxide (+)-7R,8,S-dihydrodiol, 9S,10R-epoxy-benzo[a]pyrene (BPDE) intermediate^25–27 to form the covalent and stereoisomeric BPDE-dG DNA adducts²⁸ shown in (Figure 1; The (+)-trans-BPDE-dG adduct shown is positioned in the minor groove of double stranded B-form DNA,²⁹ while the stereoisomeric (+)-cis-adduct adopts a base-displaced intercalated conformation.³⁰) Since these are the only BPDE-dG stereoisomeric adducts discussed here, the (+)-sign will be omitted hereafter. If not repaired by the mammalian nucleotide excision repair (NER) system, BPDE-derived DNA adducts can induce mutagenesis and carcinogenesis in laboratory animal³¹ and in humans.³² In cell-free extracts from human cells, the trans- and cis- BPDE-dG adducts are excised by the NER system³³ with efficiencies that are similar to those of the NER efficiencies of the 5’,8-cyclopurine lesions under the same experimental conditions.³⁴

However, it is unknown whether the same relative NER efficiencies prevail at the level of chromatin or nucleosomes, the primary subunits of chromatin. It is well established that histone variants,³⁵ the post-translational modification (PTM) of histones, and their location on the nucleosome³⁶ can strongly affect the exposure of otherwise occluded DNA sequences.^{37, 38} The accessibilities of DNA sites to other proteins, including Base Excision Repair (BER)³⁹ and Nucleotide Excison Repair (NER) factors⁴⁰ are thus enhanced by these modifications. Both bulky and non-bulky DNA lesions embedded in nucleosomes are less accessible to NER proteins and the efficiencies of NER are therefore reduced relative to free, or naked DNA.^41–43

In this work, we investigated how DNA lesions of different structure, stereochemistry or bulkiness perturb some of the key physical properties of nucleosomes that might affect their processing by the mammalian NER mechanism. In order to assess differences in NER efficiencies in nucleosomes with and without PTM, the same lesions were embedded in nucleosomes reconstituted either from recombinant (Rec) histones without PTMs, or in native histones derived from human HeLa cells with unspecified post-translational modifications or histone variants. The effects of solvent exposure were examined by positioning the same lesions either at the In or Out rotational settings adjacent to the nucleosome dyad axis where the repair of lesions has been shown to be most inhibited.³⁶

As a first step of understanding the NER response of structurally distinct DNA lesions in intact cells, it is of interest to determine the impact of the histone environment at the level of the single nucleosome on the accessibility and repair of the structurally very different bulky BPDE- dG and non-bulky 5’,8-cyclopurine DNA lesions. In this work, we compared the impact of the lesions on the thermodynamic stabilities, hydroxyl radical-induced footprints, and dynamics of nucleosomes measured by Förster Resonance Energy Transfer (FRET) methods. The lesions were positioned at the (In), and the more solvent-accessible (Out) nucleosome rotational settings (Figure 2).

Figure 2.

(A) Schematic illustrations of the placements of the DNA lesions and the Cy3 donor and Cy5 acceptor molecules at the “Out” and “In” rotational settings in the 147-mer 601 DNA duplexes used in the FRET experiments. (B) The crystal structure of the 601 nucleosome core particles (PDB 3LZ0).⁴⁴ The dyad axis is indicated by the red line. (C) Positions of the Cy3 donor and Cy5 acceptor molecules in the nucleosome FRET experiments. The lesions were positioned at the 66-th or 70-th nt counted from the 5’-end of the 147-mer, corresponding to the Out or In rotational settings. The internal Cy3 and Cy5 labels were positioned at nucleotides 43 and 39 counted from the dyad axis in opposite strands.

While the 5’,8-cyclopurine- and BPDE-derived DNA lesions in free DNA are good NER substrates, the non-bulky cdA and cdG lesions embedded at either In or Out rotational settings of native HeLa histone nucleosomes are completely resistant to NER in human cell extracts. By contrast, the excision of the trans- and cis-BPDE-dG adducts that are excised at different rates in free DNA, is reduced by the same factor of ~ 2.2 in HeLa nucleosomes, and by a factor of ~ 11 in recombinant histone nucleosomes. These findings are compared to the impact of these different DNA lesions on the thermodynamics of formation of the nucleosomes, hydroxyl radical footprints, and nucleosome dynamics evaluated by the FRET method.

MATERIALS AND METHODS

Materials.

The oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA). All oligonucleotides were purified by gel electrophoresis and isolated by ethanol precipitation. The 208 bp DNA fragment from the Lytechinus variegatus 5S RNA gene was purchased from New England Biolabs (Ipswich, MA) and amplified by PCR methods.

Synthesis of Oligonucleotides Containing Single 5’,8-Cyclopurine or trans-and cis-BPDE-dG Lesions.

Phosphoramidites of the four different nucleosides of R and S cdA and cdG (Figure 1) were prepared following the hydroxyl radical-based protocols.^{45, 46} Site-specifically modified 17-mer sequences 5’-d(CCACCAAC[X]CTACCACC) with X = 5’-R- or 5’-S– cdG/cdA were synthesized and purified following published methods.^{47, 48} The cis- and trans-BPDE-dG adducts (X’) embedded in the 11-mer sequence 5’-d(CCTACX’CTACC) were generated by a direct synthesis method and purified as described in detail earlier.⁴⁹

Preparation of 147-mer Duplexes for Reconstitution with Histone Octamers to Form Nucleosome Core Particles (NCP).

The 147-mer ³²P-internally labeled (or 5’-endlabeled) modified and unlabeled fully complementary strands (with the canonical C opposite the BPDE- dG and cdG lesions, and T opposite the cdA lesions) were prepared by the phosphorylation/ligation method³³ as described in Supporting Information (Figure S1). The 147-mer duplexes were equivalent to the Widom 601 strong positioning DNA sequence,⁵⁰ except for the 17- or 11-mer duplexes containing the BPDE-dG adducts or cdG/cdA lesions termed 601^cP or 601^BP, respectively. The 17- or 11-mer duplexes were inserted into the 147-mer by replacing an equal number of 601 nucleotides so that the lesions were positioned at the 66-th or 70-th position counted from the 5’-end of the 147-mer, corresponding to the Out or In rotational settings (Figure 2) at the (superhelical positions SHL = 0.5 and 1.0, respectively. The lesions positioned at “In” rotational settings (the 4^th (modified) nucleotide on the 5’-side of dyad) faced the histone octamer while those at the “Out” position (the 8^th (modified) nucleotide faced the aqueous solution environment).

Refolding of Histone Octamers.

The human recombinant histones H2A (Genbank accession number: AAN59960), H2B (AAN59961), H3.1 (AAN10051) and H4 (AAM83108) were purchased from New England Biolabs (Ipswich, MA); the HeLa cell histones were isolated from HeLa cell lysates and purified into two separated fractions of [H2A-H2B] heterodimers and [H3-H4]₂ tetramers using Histone Purification Kit 40025 from Active Motif (Carlsbad, CA). The histone octamers were assembled following the protocol of Luger et al.^{51, 52} The integrity of histone octamers was confirmed by SDS-PAGE⁵³ and MALDI-TOF/MS in the positive mode.⁵⁴

Nucleosome Reconstitution.

The nucleosomes were assembled from the recombinant, or HeLa cell histone octamers and 147-mer 601^cP and 601^BP duplexes by the gradual dialysis method using the double-dialysis procedure of Lowary and Widom.⁵⁵

Hydroxyl Radical Footprinting of Nucleosomes.

The footprinting experiments were performed with nucleosomes assembled from 5’−³²P-labeled DNA and histone octamers as described previously.^{56, 57}

Impact of Lesions on the Thermodynamics of Nucleosome Assembly.

A competitive DNA–histone octamer reconstitution method was used to evaluate the effects of the DNA lesions on the thermodynamic stabilities of nucleosomes.^{58, 55} Aliquots of histone octamers were added to ~20 pmol unlabeled 208 bp fragment of the 5S RNA gene (competitor DNA) spiked with duplexes containing 5’−³²P-endlabeled 601^cP, 601^BP 147-mer sequences with or without the lesion (the latter constitute the reference DNA) in a 40 μL micro dialysis button system. The nucleosomes were prepared by the double-dialysis method.⁵⁵ The distribution of ³²P between free and nucleosomal DNA was assessed by native 5% PAGE in 0.3×TBE buffer and analyzed by gel autoradiography. The equilibrium constant was defined as K = DNA_NCP/DNA_Free, where DNA_NCP and DNA_Free are quantities of tracer DNA bound to the histone octamers, or free DNA, respectively. The changes in free energies induced by single DNA lesions embedded in 147 bp modified 601 DNA sequences was calculated relative to the unmodified reference 147 bp 601 DNA, ΔΔG° = - RT ln(K^DNA*/K^DNA), where DNA* denotes DNA 147-mer duplexes that contain a single lesion, and DNA denotes the same, but unmodified duplex.^{58, 59, 55} This method was successfully used by Mann et al.⁴⁸ to explore the effects of DNA damage in nucleosomes induced by BPDE or by UV light.

FRET Assays.

The oligonucleotides containing internal Cy3 and Cy5 dyes were purchased from Integrated DNA Technologies (Coralville, IA) and incorporated into 147-mer 601^cP and 601^bp DNA sequences with or without the lesions, were prepared by phosphorylation/ligation methods (the details are described in Figure S1, Supporting Information). The fluorescence of Cy3 and Cy5 was excited with vertically polarized ~200 μW green (520 nm) and ~300 μW red (635 nm) light beams of diode lasers, respectively. The fluorescence emission signals were measured at the magic angle (54.7°) relative to the vertical polarization of the laser light, as recommended by Clegg.⁶⁰ The intensity of FRET emission (F^FRET_670,520) was calculated by the well-established equation:⁶⁰

$$F_{670,520}^{FRET}=F_{670,520}^{DA}\left(1-F_{670,520}^D/F_{565,520}^D-F_{670,520}^A/F_{670,635}^A\right)$$ (1)

where subscripts denote the emission and excitation wavelengths of the Cy3 (donor, D) and Cy5 (acceptor) dyes. The internal Cy3 and Cy5 labels were positioned at nucleotides 43 and 39 counted from the dyad axis in opposite strands depicted in Figure 2. Thus, in the nucleosomes, Cy3 and Cy5 are separated by 81 bp, a distance that provides robust FRET signals.^61–66

The relative stabilities of nucleosomes with and without lesions were determined by measuring the FRET signals as a function of increasing NaCl concentration. The relative FRET efficiencies at particular NaCl concentrations (E) were calculated relative to the E_o value at a low (50 mM) salt concentration, [NaCl]0:

$$E/E_0=\left(F_{670,520}^{FRET}/F_{670,635}^{DA}\right)/[{\left(F_{670,520}^{FRET}\right)}_0/\left[{\left(F_{670,635}^{DA}\right)}_0\right]$$ (2)

where E0 defines the FRET value for fully wrapped NCPs and E for partially unwrapped NCPs.

The relative stabilities of nucleosomes containing single, different lesions were determined by plotting E as a function of salt concentration, [NaCl]. These titration plots were approximated by a sigmoidal function:⁶⁷

$$E/E_0=1/\left(1+\exp \left(\left({[\text{NaCl}]}_{1/2}-[\text{NaCl}]\right)/a\right)\right)$$ (3)

where [NaCl]_½ is the NaCl concentration at the E/E₀ = 0.50 midpoint and a is the slope of the curve at that concentration. The salt concentration at this midpoint is a common measure of the relative stabilities of different nucleosomes.^{68, 69}

Nucleotide Excision Repair Assay.

For the NER experiments, the 17-mer or 11-mer sequences containing the cdA/cdG or BPDE-dG lesions, respectively, were 5’−³²P-endlabeled before the assembly of the 147-mer 601^cP and 601^BP DNA sequences for recovering radioactively labeled NER dual incision products. The HeLa S3 cells were purchased from the American Type Culture Collection (Manassas, VA, USA) and the cell extracts were prepared by standard protocols by cell lysis and ammonium sulfate precipitation following published methods.^70–72 The cell extract DNA repair assays and the recovery of the ~ 24–30 nucleotide long dual incision products were conducted as described in detail elsewhere.⁷² These dual incision products are the hallmark of successful NER activity^{73, 74} and are easily resolved from the 147-mer unreacted oligonucleotides by denaturing 12% polyacrylamide gel electrophoresis. The yield of dual incision products was determined from densitometry tracings of the gel autoradiographs.

RESULTS

Full NER Resistance of 5’,8-Cydopurine Lesions in Nucleosomal DNA.

While all four of the 5’,8-cyclopurine lesions embedded in free 601^cP DNA sequences are excised in HeLa cell extracts, there is no discernable NER activity when the same lesions are embedded at the In or Out rotational settings in nucleosomes reconstituted with either native HeLa or recombinant histones (Figure 3A). The full NER resistance of cyclopurine lesions in nucleosomes indicates that the HeLa-NCPs as well as Rec-NCPs remain intact during the incubation in HeLa cell extracts (Figure 3A). No decomposition (or significant repair activity) is evident in most of the Rec-NCPs with BPDE-G adducts either (Figure 3B). These results indicate that the repair of the nucleosomal DNA in cell free extracts can occur without the full disruption of the nucleosome structure as suggested by previous publications on the repair of DNA lesions embedded in nucleosomes.^{41, 75, 76, 42, 43, 77} Furthermore, it has been shown that key NER factors such as XPC, XPA and RPA form complexes with nucleosomes without dissociating the histone octamers from the DNA molecules.^{41, 78}

Figure 3.

Representative autoradiographs of denaturing gels of nucleotide excision experiments in HeLa cell extracts. (A) The substrates were either free 147-mer 601^cP DNA sequences containing single 5’,8-cyclopurine lesions, or nucleosomes assembled with octamers derived from recombinant (Rec) or native, post-translationally modified histones extracted from HeLa cells. (B) Analogous NER experiments with cis- and trans-BPDE-dG adducts positioned at the same In and Out superhelical locations. Three separate gels are depicted in panels A and B.

The Stereoisomeric BPDE-dG Adducts in Free 601^BP HeLa Nucleosomes Are Excised with Different Rates.

In parallel experiments conducted with BPDE-dG adducts, robust NER excision activities were observed in 601^BPHeLa cell NCPs (Figure 3B).

Kinetics of Excision.

The relative kinetics of formation of dual incision products are linear in each case up to at least 45 minutes after the start of the incubation in HeLa cell extract (Figure 4B). Each normalized data point reflects the average of three independent experiments (Figure 4). The kinetics of appearance of NER products (relative rates) were evaluated from the slopes of the linear plots shown in Figure 4B and are summarized in Table 1.

Figure 4.

(A) Comparisons of relative rates of NER dual incisions of the cis- and trans-BPDE-dG adducts embedded at the In or Out rotational settings of 147-mer free 601^BP DNA duplex sequences, or embedded in native HeLa histone-nucleosomes. The data points in panel A were determined from the kinetics of formation of the 24–30 nt NER dual incision products as a function of incubation time in HeLa cell extracts shown in panels B, C, D, and E (a - free DNA, b-HeLa NCP, and c–recombinant NCP). These averages and standard deviations were derived from three independent experiments, each with different nucleosome and HeLa cell extract preparations. In each of the three sets of individual experiments, all results were expressed relative to the rate observed in the case of the free DNA cis-In adduct measured at the 45 min time point; the arbitrary value of 100 was assigned to these measured values because the highest NER yields were observed in each individual set of three experiments in the case of each different cis-In BPDE-dG adduct in free DNA (Figure 4). The NER efficiencies were dependent on the particular cell cultures and varied from ~ 4 to 12% in the case of the cis-BPDE-dG adducts in free DNA at the 45 min time points.

Table 1.

Relative values of slopes of NER dual incision product yield formation (%NER/min) as a function of incubation time in He La cell extracts

DNA adduct	NER Free DNA*	NER HeLa-NCP*	NER Ratio Free/HeLa	NER Rec NCP*	NER Ratio Free/Rec
cis-BPDE-dG In	2.53±0.13	1.02±0.13	2.48±0.14	~0
cis-BPDE-dG Out	2.28±0.15	1.11±0.28	2.05±0.26	0.20±0.02	11.3
trans-BPDE-dG In	2.43±0.08	1.02±0.05	2.38±0.06	~0
trans-BPDE-dG Out	2.28±0.15	1.24±0.05	1.84±0.08	0.24±0.01	10.5
Average			2.2±0.2**		11±0.5

^*Averages of three independent NER assays in different cell extracts.

^**P < 0.05 (one-sample t-test).

NER Excision Rates in HeLa Histone NCPs Relative to the Rates in Free DNA.

As shown earlier, the NER activities in free DNA depend on the cis- and trans-stereochemistry^{33, 79} and also on the DNA sequence context.^{72, 80, 81} The relative NER rates observed with HeLa histone NCPs containing cis- and trans-stereoisomeric BPDE-dG adducts at the In or Out rotational setting are reduced by the common factor of 2.2 ±0.3 in HeLa histone NCPs relative to free DNA, regardless of BPDE-dG adduct stereochemistry or rotational setting (Table 1).

Impact of Rotational Setting on NER Activity.

The Out/In NER excision ratios for each adduct are similar in free DNA and in HeLa NCPs with values of ~ 1.10(0.94) in the case of the trans- and ~0.83(0.83) in the case of the cis-BPDE-dG adducts in HeLa NCPs (the numbers in parentheses represent the free DNA values). In recombinant NCPs, the NER efficiencies at the Out position are ~11 times smaller than in free DNA, while no NER activity is observable when the same adducts are positioned at the In rotational setting (Table 1, and Figures 4B - 4E). It is noteworthy that in recombinant histone NCPs, a small NER activity is observable only at the Out, but not at the In rotational setting, while in HeLa NCPs the differences are small and not conclusive (~ 10 – 20% smaller yields at In rotational settings).

DNA Footprinting Results.

Hydroxyl radical footprinting experiments^{56, 57} were performed in order to determine whether the bulky and non-bulky DNA lesions perturb the rotational settings of nucleosomal DNA to different extents. The autoradiographs of hydroxyl radical footprints and the corresponding scans obtained with S cdA lesions embedded in HeLa NCPs shown in Figures 5 demonstrate the same phasing patterns as in the nucleosomal unmodified DNA.⁵⁶, ⁵⁷, ^66. The lack of impact of the non-bulky 5’,8-cyclopurine lesions on the rotational settings of nucleosomes is not unusual since the UV photoproducts CTD⁷⁵ and (6–4) photoproducts⁷⁶ embedded in nucleosomes also do not have a significant impact on the nucleosomal rotational settings.

Figure 5.

Hydroxyl radical footprints obtained with recombinant histone nucleosomes assembled from the 147-mer 601^cP DNA sequences containing S-cdA lesions at the Out and In rotational settings. (A) Autoradiograph of a typical hydroxyl radical footprint experiment. (B) Scan of the results shown (A) in the region containing the lesions (red vertical bar), and (C) Scan of the entire sequence.

The overall hydroxyl radical cleavage patterns of BPDE-dG adducts in HeLa NCPs are not significantly different at the In (Figures 6A - 6D) and the Out (Figures 6E - 6F) rotational settings.

Figure 6.

Hydroxyl radical footprints obtained with HeLa histone nucleosomes assembled from 601 sequences containing cis-or trans-BPDE-dG adducts. (A) Autoradiograph with the BPDE-dG adducts at the Out rotational setting, (B) Autoradiograph obtained with the BPDE-dG adducts at the In setting. (C) Scan of the cis-In and unmodified DNA samples. (D) Scan of the trans-In and unmodified DNA samples. (E) Autoradiograph of the cis-and trans-BPDE-dG adduct samples at the Out rotational setting. (F) Scans of the autoradiograph in panel E.

While the rotational settings on the 3’-side of the In BPDE-dG adducts are not significantly affected relative to unmodified DNA (Figures 6C - 6D), the modified guanine and 3’-adjacent cytosine-sites are more strongly cleaved than the analogous sites in the unmodified DNA sequence. This observation suggests that the backbone is more solvent-exposed at these two sites in the HeLa histone NCPs. On the 5’-side of the adduct, one nucleotide appears to be missing; this effect arises because the strand break at the site of the modified guanine removes the equivalent of almost two nucleotides (the mass of the BPDE residue (302 Da) is approximately equal to the mass of 2’-deoxyguanosine residue, 283 Da). On the 5’-side of the In adducts, the maxima in the scans (Figure 6C - 6D) are slightly shifted by about 4 – 5 nucleotides. In the case of the same adducts at the solvent-exposed Out position, there is no enhancement in cleavage of the backbone at the site of the BPDE-modified nucleotide and its 3’-adjacent sites (Figures 6E - 6F), since there are no striking enhancements in the cleavage efficiencies as there are in Figures 6C - 6D in the case of the same adducts at the In rotational setting.

Impact of DNA Lesions on the Free Energies of Formation of Nucleosomes.

The free energy change of nucleosome formation containing a DNA lesion is:

$$\Delta \Delta G=RT\left(\ln K^{DNA*}-\ln K^{DNA}\right)=\Delta G(DNA+\text{lesion})-\Delta G(\text{unmodified}DNA)$$ (4)

where DNA* denotes nucleosomal DNA with a single 5’,8-cyclopurine or BPDE-dG adduct. The free energy difference ΔΔG is a positive quantity if the lesions destabilize the DNA (K^DNA* < K^DNA).

5’,8-Cyclopurine Lesions.

The experimental values of AAG(cdG/cdA) indicate that in all cases (except R cdA), the ΔΔG(cdG/cdA) values are ~ 4–5 times greater, and thus more destabilizing at the In than the Out locations of HeLa NCPs (Figure 7A). These ΔΔG values Except for the Out trans-BPDE-dG adduct, the ΔΔG values are in the range of ~ 0.70 – 0.75 kcal/mol in the case of the other three BPDE-dG adducts in HeLa NCPs (Figure 7B). It is noteworthy that the ΔΔG values are rather similar in HeLa-NCPs and Rec-NCPs, although the NER rates are very different (Figures 3 and 4).

Figure 7.

(A) The relative free energies of formation of nucleosome, (ΔΔG). The NCPs were assembled with recombinant histone octamers and 147-mer 601^cP Widom DNA sequences containing single 5’,8-cyclopurine lesions positioned at In or Out rotational settings. (B) The ΔΔG values determined in analogous experiments with nucleosomes assembled from 147-mer 601^BP DNA sequences with single BPDE-dG adducts at the In or Out rotational settings and octamers derived from recombinant or native HeLa cell histones. are close to those reported for another non-bulky DNA lesion, the cyclobutane thymine dimer (CTD, ΔΔG = 1.1 and 0.6 kcal/mol at the In and Out positions, respectively) also positioned near the dyad in another strong positioning sequence (TG) in native chicken erythrocyte NCPs.⁸²

The bulky BPDE-dG adducts are less destabilizing than the 5’,8-cylopurine lesions at the In rotational settings. There is a significant difference between ΔΔG(cdA/cdG) and ΔΔG(BPDE-dG) values at the In and Out rotational settings. There is a ~ 5-fold differences in the case of the 5’,8-cyclopurine lesions in the HeLa NCPs (Figure 7A), while it is significantly smaller in the case of the BPDE-dG adducts (Figure 7B). These differences in thermodynamic free energy of formation do not seem to be correlated with differences in NER dual incision activities.

Nucleosome Stabilities Probed by FRET Methods.

A representative set of fluorescence emission spectra of Cy3-Cy5-labled unmodified DNA NCPs at different NaCl concentrations are shown in Figure 8A; the addition of NaCl decreases the Cy5 FRET signal at 670 nm and is correlated with the growth of the Cy3 fluorescence at 565 nm.

Figure 8.

Unwrapping of nucleosomes assembled with 601^cP 147-mer DNA with S cdA/cdG lesions and HeLa histone octamers as a function of NaCl concentration monitored by the FRET method. (A) Typical fluorescence emission spectra obtained with an unmodified DNA-NCP sample excited at 515 nm and recorded at different NaCl concentrations. (B) Relative FRET signals measured at 670 nm of HeLa histone nucleosomes with S cdG lesions (B), and with S cdA lesions (C) as a function of NaCl concentrations; the sigmoidal curves are plots of Eq. 3. (D) Comparison of the [NaCl]1/2 values.

The FRET ratios obtained with cdG, cdA lesions, and unmodified 601^cP in HeLa histone NCPs are plotted in Figures 8B and 8C. Fits of Eq. 3 to the unmodified DNA-NCP data points yield the [NaCl] 1/2 value of 0.85 M (Figure 8D), in good agreement with published values.⁶⁸ Similar values were obtained with S cdG and S cdA lesions positioned at the Out rotational settings. However, the stabilities ([NaCl]1/2 values) of the same lesions positioned at the In rotational settings are significantly lower (~ 0.6 M, Figure 8D); these results are consistent with the thermodynamic destabilization since the ΔΔG values are ~ 0.8 kcal/mol greater at the In than the Out rotational settings in the case of cdG and S cdA lesions (Figure 7A). In the case of the R cdA lesion this difference is less pronounced (Figure 8D and Supporting Information, Figure S2).

In the case of the cis- and trans-BPDE-dG adducts, ΔΔG^Out > ΔΔG^In in the case of the cis-,as well as the trans-adducts in Rec-NCPs (Figure 7B). The somewhat smaller FRET [NaCl]1/2 values at the In than the Out settings are qualitatively consistent with this trend (Figure 9C). Similar differences are observed in the case of the same adducts embedded in HeLa NCPs, since the differences in the FRET signals at the In and Out settings are more pronounced (Figure 9D). Furthermore, there is no correlation with the ΔΔG^Out and ΔΔG^In values (Figure 7B) as there is in the case of the Rec-NCPs. A clue to these differences are the small, but consistently higher FRET values at the BPDE-dG Out rotational settings than the unmodified DNA NCP [NaCl]1/2 values in Rec- and HeLa NCPs (Figures 9C and 9D). Such differences are not observed in the case of the 5’,8-cyclopurine lesions embedded in HeLa NCPs since the [NaCl]1/2 values in HeLa NCPs at the Out rotational settings are identical to those in unmodified DNA HeLa NCPs (Figure 8D). As discussed below, intermolecular Van der Waals and other electrostatic interactions that depend on the size and shape of the DNA lesions may contribute to the differences in the thermodynamic ΔΔG destabilization of nucleosomes that are also reflected in the measured FRET [NaCl] 1/2 values.

Figure 9.

FRET signals measured as a function of NaCl concentrations of 601^BP 147-mer DNA sequences with BPDE-dG adducts in (A) Rec-NCPs, and (B) HeLa-NCPs as a function of salt concentration. (C) and (D): Comparisons of relative nucleosome stabilities based on FRET [NaCl] 1/2 values.

DISCUSSION

The availability of two families of robust NER substrates with very different physical dimensions allowed us to investigate potential relationships between nucleosome stabilities determined by equilibrium thermodynamic approaches and FRET methods, rotational settings, and the impact of post-translation modification on NER excision efficiencies. All DNA lesions were placed close to the dyad axis where they are expected to exert the greatest impact on repair and thermodynamic stabilities, especially at the In rotational settings. By selecting the 601 strong positioning sequence and 147-mer duplexes, the sliding mechanism of exposure of DNA sequences on NER was eliminated as a possibility.

Previous NER Studies of Defined, Site-Specifically Positioned Lesions Embedded in Nucleosomes.

There are relatively few published investigations of this kind. Studies have been conducted with CTD, (6–4)PP, and N-acetylaminofluorene-guanine (AAF-G) lesions embedded near the dyad in native HeLa histone NCPs. Robust NER response was elicited by the six purified NER factors XPC-RAD23B, XPA, RPA, TFIIH, XPG, and XPF-ERCC1).⁴¹, ⁷⁶, ⁴², ⁴³ Excision of these lesions by in HeLa histone NCPs occurred with lower efficiencies than in free DNA⁴¹, ⁴³ by factors of 10 (CTD), ~ 5 (AAF-G), and ~ 2.5 ((6–4)-PP). With (6–4)PP lesions embedded in different DNA sequences in mono- and dinucleosome model system, a two-fold lower excision efficiency relative to free DNA was observed by other workers.⁷⁶ An important conclusion from studies using purified NER factors is that the presence of chromatin remodeling factors is not essential for observing NER dual incisions of lesions in nucleosomal DNA in vitro. However, the remodeling factors SWI/SNF and ACF did stimulate the removal of AAF-G adducts in nucleosomes by a factor of ~1.5⁴², while the ATP-dependent factor ACF enhanced the removal of (6–4)PP lesions by a factor of ~ 2.⁷⁶ Other mammalian cell extract NER studies of DNA lesions embedded in nucleosomes include CTD in the strong positioning TG DNA motif,⁷⁵ and the cisplatin-derived intra-strand cross-linked GTG-Pt and GG-Pt DNA lesions.^83,77

Structural Features and NER Response of the 5’,8-Cyclopurines in Free DNA.

The cdA and cdG lesions are characterized by intranucleotide cross-links between the C8 purine and the 5’-deoxyribose residue (Figure 1). These oxidatively generated DNA cross-links significantly distort the local DNA duplex structure by constraining the relative orientations of these two residues.^{84–86, 34, 87} Structural changes include an abnormal O4’-sugar pucker, an over twisting of the 5’,8-cyclopurine: dC (or dT) base pair, a widening of the minor groove, a decrease in local base-base stacking interactions, and weakened S-cdG:dC base pairing at the sites of the lesions. Molecular Dynamic Simulation studies suggest that the basic structural and dynamic features of R and S cdG lesion are carried over to the nucleosomal environment.⁸⁷

DNA Conformations and NER Response of BPDE-dG lesion in Free DNA.

The bulky trans-BPDE-dG adducts are positioned in the minor-groove of DNA with all base pairs remaining intact;²⁹ by contrast, the cis-BPDE-dG adduct assumes a base-displaced intercalative conformation that significantly distorts the local base pairs and DNA structure.⁸⁸ The relative NER rates of cis or trans adducts in free DNA depend not only on adduct stereochemistry,^{33, 89} but also on the local and more distant base sequence context.^{90, 72, 81} Therefore, the relative NER excision efficiencies of the cis- and trans-BPDE-dG adducts at the two different rotational settings in free DNA are different from one another (Figure 4).

The 5’,8-Cydopurine Lesions Affect the Physical Characteristics of HeLa Histone NCPs but NER Activity Is Absent.

The cdA and cdG lesions, although small in bulk, cause significant thermodynamic destabilization and lowering of the FRET [NaCl]1/2 NCP stabilities at the In, but not Out rotational settings (Figures 7A, and 8, respectively). However, within the experimental uncertainties, the hydroxyl radical footprints at the In and Out rotational settings are not distinguishable (Figure 6). At the physiologically relevant NaCl concentrations (< 200 mM), the FRET values are up to ~ 20% lower relative to unmodified DNA NCPs at the In rotational settings (Figures 8B,C and Figure S2, Supporting Information). Nevertheless, there is no detectable NER activity (Figure 3A). By contrast, the UV light-induced CTD and (6–4)PP lesions are excised by NER mechanisms in mammalian cell extracts.^{41, 75, 76, 43, 82} Similar FRET values (lowered by 9–16% relative to unmodified DNA NCPs) were observed with these lesions embedded in the 601 DNA sequence near the dyad in native histone NCPs.⁶⁶

Thermodynamic Destabilization of NCPs by DNA Lesions is not Directly Correlated with the NER Response.

In three out of four cases, the ΔΔG^In and ΔΔG^Out values are not significantly different from one another in the case of the cis- and trans-BPDE-dG adducts in Rec-NCPs and in HeLa-NCPs (Figure 5B). However, the NER activities are entirely absent at the In, and very low at the Out rotational settings in recombinant histone NCPs, while relatively strong NER activities are observed in HeLa-NCPs (Figure 4).

5’,8-Cyclopurine Lesions can be More Destabilizing than BPDE-dG Adducts.

At the In rotational setting the non-bulky cdA and cdG lesions are more destabilizing than the bulky BPDE-dG adducts. However, at the Out setting, the BPDE-dG adducts are more destabilizing. These observations suggest that non-covalent Van der Waals and other electrostatic interactions between the bulky hydrophobic aromatic ring systems of BPDE and adjacent histone residues might counteract the destabilizing impact of structural distortions caused by the bulky DNA adducts, thus accounting for the smaller ΔΔG^In values. This conclusion is consistent with the findings of Mann et al.⁵⁴ who found earlier that randomly BPDE-modified (with an average of one randomly positioned adduct per nucleosome particle) 146-mer Xenopus borealis 5S rRNA gene fragment DNA give rise to a small, overall nucleosome-stabilizing effect with ΔΔG ≈ - 0.30 kcal/mol.

NER Dual Excision Ratios at In and Out Settings are Different from Normal Patterns Observed in Base Excision Repair Assays.

The removal of non-bulky lesions (e.g., uracil, thymine glycol, 8-oxoG, and abasic sites) by BER mechanisms has received significant attention.^{91, 36, 92, 93, 39} While exceptions have been noted,^{91, 94–96} BER substrates are excised more efficiently when positioned at the Out than the In rotational settings.^{97, 91, 98–100, 39} However, there are few analogous studies of NER activities. In the case of the non-bulky CTD UV photoproducts, an Out/In ratio of ~ 1.5 was reported for the NER in mammalian cell extracts embedded in the strong positioning TG motif DNA near the dyad in native chicken erythrocyte histone nucleosomes.⁸²

The very slow rate of NER observed in the case of cis-and trans-BPDE-dG adducts in Rec-NCPs, and the absence of NER signals at the In settings, is qualitatively consistent with an Out > In trend of excision efficiencies (Figure 4B). In the case of the cis-and trans-BPDE-dG adducts in HeLa-NCPS, the Out/In ratios are 1.2 and 1.1, respectively (Table 1). The intrinsic Out/In ratios in free DNA are ~ 0.92 ± 0.02 (cis-and trans-BPDE-dG), and thus the difference in the Out/In ratios observed in HeLa-NCPs are considered to be small and not definitive in the absence of further examples.

Comparing NER rates of all four different BPDE-dG samples in HeLa histone NCPs, the NER rates are reduced in each case by the common factor of 2.2 ±0.3 (Table 1). Thus, within the approximate ±10 % experimental range, the relative probabilities of excision are similar in free DNA and in HeLa histone NCPs, although the NER efficiencies of the stereoisomeric BPDE-dG in free DNA vary by as much as a factor of 4 (Figure 4A).

The approximately constant proportionalities between NER efficiencies observed in the cases of free and nucleosomal DNA suggest that the nucleosomal environment reduces the probability of NER by a similar factor for all of the different DNA adducts that are repaired with different efficiencies in free DNA. There are no obvious quantitative correlations between NER efficiencies, thermodynamic destabilization and rotational setting. These observations are consistent with a similar extent of exposure of each of the different BPDE-dG adducts to the NER factors that initiate and catalyze the dual incision NER pathway. Since XPC-RAD23B is the first NER factor that recognizes NER substrates and its DNA footprint is ~ 30 base pairs,¹⁰¹ the lengths of the DNA sequences transiently exposed must be longer than this number of base pairs for efficient XPC-RAD23B binding to occur. Once the lesion-containing DNA sequences are exposed, these observations suggest that the relative NER efficiencies of the different DNA adducts parallel those observed in free DNA.

The footprint of XPC-RAD23B (molar mass ~ 180 kDa) is ~ 30 base pairs,¹⁰² and efficient binding requires DNA sequences at least 45 base pairs in length.¹⁰³ Therefore, the complete unwrapping of the nucleosome may not be necessary for NER to occur. Indeed, it has been demonstrated that XPC binds to nucleosomal DNA with a lower probability than to free DNA without dissociating the DNA from the histone core.^{41, 78} We have reproduced these results and verified that XPC-RAD23B also binds to HeLa histone NCPs with BPDE-dG lesions without the full dissociation of the nucleosomes (Supporting Information)‥

Relationships between NER Rates and Nucleosome Unwrapping Phenomena: a Kinetic Model.

The higher rates of NER of BPDE-dG adducts in HeLa histone than in recombinant histone NCPs indicate that the post-translational modification of histones and nucleosome unwrapping dynamics play an important role in the repair of the BPDE-dG but not the 5’,8-cyclopurine lesions. With XPC-RAD23B (abbreviated as XPC), the other abbreviations are NCP _Closed , NCP_open , and XPC-NCP, represent the protein-accessible, protein-inaccessible NCPs, and XPC-NCP complexes, respectively.

The following sequence of steps is proposed:

$$\left[{\text{NCP}}_{\text{Closed}}\right]\underset{k_2}{\overset{k_1}{\rightleftarrows }}\left[{\text{NCP}}_{\text{Open}}\right]\stackrel{k_2[\text{XPC}]}{\to }[\text{XPC}-\text{NCP}]\to \text{Excision}$$ (5)

The binding of XPC to the nucleosomal DNA depends on the rate constant k2. The local unwrapping and rewrapping occur with rate constants k₁ and k_-1, respectively, that define the equilibrium constant K_eq = k₁/k_-1 = [NCP_open]/[NCP_Closed]. The formation of the XPC-NCP complex leads to the subsequent recruitment of the other NER factors that ultimately results in the formation of the NER excision products (Figure 3). The rate of formation of the [XPC-NCP] complex can be described by the following equation:

$$\frac{d[\text{XPC}-\text{NCP}]}{dt}=\frac{k_1{k}_2[\text{XPC}]\left[{\text{NCP}}_{\text{Closed}}\right]}{k_{-1}+k_2[\text{XPC}]}$$ (6)

Previous results indicate that the access of XPC to the DNA lesions in HeLa histone NCPs is sterically hindered;^{41, 78} however, once the initial XPC-NCP complex is formed, the unwrapped state of the NCP is stabilized by the XPC-DNA complex, thus facilitating the assembly of the other NER factors. The relationship between the equilibrium constant K_eq and experimental FRET values is the following

$$K_{eq}=k_1/k_2=\left(1-E/E_0\right)/\left(E/E_0\right)$$ (7)

This equation was originally developed for describing the unwrapping-rewrapping NCP equilibria at different salt concentrations.¹⁰⁴ In our case, E0 is the reference FRET value for unmodified DNA (K_eq < 0.1),¹⁰⁴ while E is the value for the same NCP but containing a single DNA lesion. We consider three limiting cases of Eq. 6 that are significant.

Case I.

In the first limit k−1 >> k2[XPC] and the rate of XPC-NCP formation is proportional to K_eqk2[XPC]. In this case the rate of repair is directly proportional to the equilibrium constant K_eq, and thus to the relative FRET values E/E0, the bimolecular rate constant k2 and the [XPC] concentration.

Case II.

In the second limit, k−1 ≈ k2[XPC], the dependence of the XPC binding rate on the measured E/E₀ FRET value will be smaller than in Case I, and will diminish as a function of increasing k₋₁/k2[XPC] ratio.

Case III.

In the third limit, k2[XPC] >> k_-1, and the XPC capture rate (Eq. 6) is proportional to the unwrapping rate constant k1, and thus not directly proportional to K_eq and the FRET values.

The lack of NER activity of the 5’,8-cyclopurine lesions in HeLa NCPs is surprising since, like BPDE-dG adducts, it is an excellent NER substrate in free DNA. This lack of NER activity suggests that the binding of NER factors to these physically small lesions is sterically hindered by local histone-DNA configurations that interfere with the proper binding and alignment of XPC to the cdA and cdG lesions. Such a mechanism is consistent with Case I, with k-1 >> k2[XPC], suggesting that the rate constant k2 is small in comparison to the same rate constant in the case of BPDE-dG adducts in the same HeLa histone NCPs.

In the case of BPDE-dG adducts in recombinant histone NCPs, the impact of the adducts on FRET and [NaCl] 1/2 salt destabilization parameters is significantly smaller (Figures 9A,C) than in the case of HeLa histone NCPs (Figure 9B,D). A weak NER activity is observed in the case of the Out trans- and cis-BPDE-dG adducts in recombinant histone NCPs (Figured 4B,D). However, there is no discernable NER at the In rotational setting (Figures 4A,C) in the case of the BPDE-dG adducts. These observations suggest that Case I is applicable to BPDE-dG adducts in Rec-NCPs at the In position with k-1 >> k2[XPC] with no XPC binding and observable NER activity. At the Out position, however, weak NER activity is observable and, k-1 ≤ k2[XPC] (case II).

In the case of the cis-and trans-BPDE-dG adducts positioned either at the In or the Out rotational settings in HeLa-NCPs, Case III is applicable with k2[XPC] > k-1. In this case, the rate of XPC binding is proportional to the unwrapping rate constant k₁, but is not proportional to the equilibrium constant K_eq, or the FRET efficiency.

The results obtained by the FRET method (Figure 9D) indicate that on average, the nucleosome dynamics are more enhanced when the BPDE-dG adducts are positioned at the In than the Out rotational settings. On the other hand, the bulky polycyclic aromatic residues are less exposed to the solvent environment at the In than the Out rotational setting. However, since robust NER activities are observed in all four cases (Figure 4), the solvent exposure factor seems to play a minor role in the case of the bulky BPDE-dG adducts. The unwrapping/rewrapping phenomena that expose sufficiently long DNA sequences that allow the XPC factor to gain a toehold on the exposed sequences leads to irreversible XPC-DNA complex formation that initiates the recruitment of subsequent NER factors.

CONCLUSIONS

The strong differences in NER activities between post-translationally modified and recombinant histone nucleosomes are consistent with the notion that the NER activities depend on the transient unwrapping and partial exposure of nucleosomal DNA sequences to the NER factor XPC-RAD23B. This notion is consistent with earlier studies that suggest that enhancements in NER activities may be associated with the acetylation of NCPs.¹⁰⁵ Electrophoretic mobility shift assays indicate that the octamer histone core particles are not evicted in the process.^{41, 78} The complete NER resistance of the cyclopurine lesions is attributed to the relatively small physical size of the cdG and cdA lesions that is not significantly different from natural nucleotides. In contrast to the larger-sized and bulky BPDE-dG adducts, the 5’,8-cyclopurine lesions do not appear to sufficiently perturb the local DNA-histone interactions, thus hindering the successful recruitment of the NER factor XPC-RAD23B that initiates the NER process in human cell extracts. This hypothesis is consistent with the structural features deduced from and NMR and molecular dynamics analysis of an S-cdG lesion opposite dC in a double-stranded oligonucleotide duplex.⁸⁴ While the cdG lesion perturbed the helical twist and base pair stacking at the lesion site and adjacent base pairs, Watson-Crick base pairing and the helical structure of the duplex, though altered, were maintained. This suggests that 8,5’-cyclopurine-containing DNA duplexes might not cause sufficient local perturbations of the histone fold of nucleosomes to elicit the NER response. The robust NER activity observed with the same lesions embedded in free double-stranded DNA (Figure 3A) is attributed to the significant thermodynamic destabilization of double-stranded DNA caused by the 8,5’-cyclopurine lesions.

ABBREVIATIONS

BER

base excision repair

NER

nucleotide excision repair

PAH

polycyclic aromatic hydrocarbon

NCP

nucleosome particle

SHL

superhelical location

FRET

Fluorescence Resonance Energy Transfer

CTD

cyclobutane thymine dimer

cdA

5’,8-cycloadenine

cdG

5’,8-cycloguanine

BPDE-dG

benzo[a]pyrene diol epoxide-derived N²-guanine adduct

bp

base pair

nt

nucleotide