The 147-bp nucleosome positioning sequence 601 was used for NCP reconstitution (31). The 147-mer oligonucleotides were initially prepared from two separate fragments of 52 nts and 95 nts. These oligonucleotides were synthesized separately for fluorescent labeling (52 nts) and UV treatments (95 nts). The 52-mer oligonucleotides, containing internally labeled Cy3 or Cy5, were purified by HPLC (Integrated DNA Technologies). Cy3 was incorporated at position 33, and Cy5 was incorporated at position 113 on opposing strands (Fig. 1A). The 95-nt fragments were dissolved in water and irradiated on ice with (predominantly) 254 nm UV light under two low pressure Hg lamps (Sylvania, Model G30T8). UV doses were measured with a Spectroline DM-254N UV meter (Spectronics Corp., Westbury, NY). (We note that the 95-mer fragments were irradiated separately to avoid photobleaching of the Cy3 and Cy5 chromophores.) The 52- and 95-mer oligonucleotides were mixed at equal molar concentrations and annealed. The annealed products were then treated with ligase to seal the nicks. After ligation, the DNA was gel-purified to remove additional oligonucleotides and unligated products. For DNA containing a single lesion, the structure of the substrate is shown schematically in Fig. 6A. The fragments F1–F4 are 52, 83, 95, and 52 nt in length, respectively. The damage fragment (DF) is a 12-mer oligonucleotide, and the sequence is shown in Fig. 6A. DF1 is the sequence containing a single CPD, and DF2 is the sequence containing a single 6-4PP. These oligonucleotides were the generous gift of Dr. Nicholas Geacintov (Chemistry Department, New York University). The F3 sequence was changed according to the corresponding DF sequence. The DNA was prepared the same as UV-irradiated substrates described above. The 147-bp full-length DNA with both UV damage and internal fluorescent labels was then used for NCP reconstitution.

The four core histones of Xenopus laevis were prepared individually from Escherichia coli as described (32) with some modification. Histones were expressed as inclusion bodies, the cells were disrupted by sonication, and the pellets were collected by centrifugation. The pellets were washed and resuspended in unfolding buffer (7 m guanidinium hydrochloride, 20 mm Tris-HCl, pH 7.5, and 10 mm DTT), and the supernatant was dialyzed against freshly made urea dialysis buffer (7 m urea, 10 mm Tris-HCl, pH 8.0, 1 mm EDTA, 0.1 m NaCl, 5 mm 2-mercaptoethanol, and 0.2 mm PMSF) with three changes. The samples were then separated on DEAE and CM Sepharose columns run in tandem. The DEAE column was removed, and samples were eluted from the CM Sepharose column by salt gradient (urea buffer A, 7 m urea, 10 mm Tris-HCl, pH 8.0, 1 mm EDTA, 1 mm DTT, and 0.2 mm PMSF; urea buffer B, 7 m urea, 10 mm Tris-HCl, pH 8.0, 1 mm EDTA, 1 m NaCl, 1 mm DTT, and 0.2 mm PMSF). Fractions were monitored by absorbance and 15% SDS-PAGE. The purified histones were dialyzed against deionized water containing 5 mm 2-mercaptoethanol and 0.2 mm PMSF and lyophilized to dryness.

Histone octamer was prepared as described (32) and mixed with DNA at different ratios to optimize reconstitution. The molar ratio of octamer:DNA at 1.1:1 was found to yield the highest reconstitution efficiency. Nucleosomes were reconstituted by salt gradient dialysis and verified on 5% native polyacrylamide gels run in 0.25×TBE. The final nucleosome concentration was 50 nm in all FRET measurements.

For qualitative analysis with specific antibodies, UV-irradiated DNA was denatured by boiling for 5 min in 0.1 m NaOH. At each UV dose, equal amounts of DNA (100 ng for 6-4PPs and 2 ng for CPDs) were immobilized on nitrocellulose membranes with a Bio-Dot SF microfiltration apparatus (Bio-Rad). To detect UV lesions, the membrane was probed with antibodies against either 6-4PPs or CPDs (Cosmo Bio Co., Ltd, Japan). The membranes were developed with ECL plus kits (GE Healthcare) and scanned on a STORM 840 FluorImager (GE Healthcare). To determine loading amounts, 601 DNA labeled with ³²P (using random sequence primers and PCR) was used to reprobe the membranes.

For quantitative analysis, a T4 DNA polymerase-exonuclease blockage assay was performed as described previously (33). Briefly, samples were incubated with 2.5 units of T4 DNA polymerase-exonuclease (Fermentas) at 37 °C for 2 h. The reaction was stopped by heating at 65 °C for 10 min. The DNA was blotted onto a nylon membrane and probed with ³²P-labeled 601 DNA. The average dimer yield was calculated as described previously (33).

Photoreactivation of UV-damaged DNA was carried out as described (34). The UV-irradiated DNA (40 pmol) was mixed with 1 μl photolyase (70 μm) in repair buffer (50 mm Tris-HCl pH7.5, 100 mm NaCl, 1 mm EDTA, and 10 mm DTT). The mixture was incubated in the dark for 5 min to allow photolyase binding to the CPDs, and the samples were irradiated with 365 nm light (Spectroline, model ENF-240C, Spectronics Corp., Westbury, NY) through a Pyrex filter for 30 min.

Fluorescence experiments were carried out at room temperature (23 °C) on a Photon Technology International Quantamaster UV/Vis Steady State Fluorometer. A quartz microcuvette with cover (105.250 QS, Hellma, Germany) was used to measure the fluorescence intensity. The samples were excited at 515 nm, and emission spectra were recorded from 550 to 700 nm. The Cy5 acceptor was excited at 615 nm directly for recording the acceptor-only emission spectra.

FRET efficiencies were measured from the sensitized emission of the acceptor (A) by the ratio (A) method (35). FRET efficiency (E) was calculated using the following equations (36), where F is the fluorescence signal, at the emission and excitation wavelengths indicated by subscripts, from the dyes indicated by superscripts (with D and A denoting Cy3 and Cy5, respectively); E is the FRET efficiency; ϵ_λ is the extinction coefficient at wavelength λ; and d is the fractional labeling of donor. These values are then used to determine the equilibrium constant (K_eq) for the two states (wrapped and unwrapped), calculated from the relative FRET efficiency [p = (E/E₀)] as K_eq = (1 − P)/P (24).

Naked DNA and nucleosomes were run on native 5% polyacrylamide gels in 0.25× TBE buffer. The gels were scanned on a Typhoon 9400 FluorImager (GE Healthcare), using the green laser (532 nm) and 580 BP 30 or 670 BP 30 bandpass emission filters. For acceptor-only fluorescence, gels were excited with a red laser (633 nm), and emission was captured through a 670 BP 30 bandpass filter. Images were displayed using FluorSep^TM software (GE Healthcare).

The 601 sequence was prepared by EcoRV digestion of the plasmid pLMG601–23, which contains 23 tandem repeats of the 601 sequence (37). The released fragments (149 bp) were exposed to UV irradiation at different doses and used for nucleosome reconstitution, as described above. The reconstitution efficiency was verified on 5% native polyacrylamide gels, and only fully reconstituted NCPs were used for enzyme digestion. In each case, a 0.5-μg aliquot of NCPs was digested with 10 units of HaeIII or RsaI for 2 h at 37 °C, and the reaction was stopped by phenol-chloroform extraction. The digested samples were analyzed on 16% PAGE in TBE, and the gels were stained with SYBR gold and scanned on a STORM 840 FluorImager (GE Healthcare). Quantitative analysis of the digests was performed with ImageQuant software.

To study nucleosome dynamics, we used purified recombinant histones, assembled into histone octamers, and the high affinity nucleosome-positioning sequence 601 (31), labeled with donor (D; Cy3) and acceptor (A; Cy5) fluorophores on opposite strands (Fig. 1A, upper panel). The 601 sequence, chosen to optimize the yield of homogeneously positioned DNA on the histone surface (31), was reconstituted into NCPs with histone octamers using stepwise salt dialysis (32). The FRET efficiency (E) relies on the inverse sixth power of the distance (R) between D and A [1 + (R/R₀)⁶]⁻¹ (38) and the Förster distance (R₀), where E is 50%, is ∼6 nm for Cy3 and Cy5 (24). Therefore, in the naked 601 DNA, where R is ∼27 nm, little or no FRET is expected, whereas in NCPs, R is ∼3 nm (39), and efficient FRET is expected (Fig. 1A, lower panel). For these experiments, both naked DNA and NCPs were excited at 515 nm (donor excitation), and the emission spectra were measured from 550 nm to 700 nm. As expected, no FRET was observed with naked DNA (Fig. 1B), whereas robust FRET was observed with the NCP samples (Fig. 1B).

Nucleosomes show increased dissociation with increased salt concentration (24, 37, 40), and we used this property to further test the reliability of our FRET system. Widom and co-workers (24, 29) used 601-NCPs with dyes at the 5′-end of one of the DNA strands and at different positions on the core histones. In our case, both dyes are located in the internal region of the NCP DNA, analogous to the studies of Lohr and co-workers (26, 41) with different DNA sequences. To compare our data with these previous results, we examined the salt-induced dissociation of the 601 NCPs and found a clear decrease in FRET with increasing salt concentrations (Fig. 1C). At 0.75 m NaCl, Li and Widom (24) found that terminally labeled NCPs yield ∼40% FRET efficiency, whereas our internally labeled NCPs yield ∼60% FRET efficiency (Fig. 1C and see Fig. 4). These results highlight the increased dynamics associated with the terminal regions of NCP DNA (41).

We found that, in cells, DNA is damaged within the context of chromatin, and we tried to UV-irradiate the assembled NCPs for FRET measurement. However, a significant fraction of the Cy3- and Cy5-labeling dyes were photobleached following 254 nm UV irradiation in our experiments (see supplemental Fig. S1), confounding the FRET measurement. Therefore, we introduced photoproducts prior to nucleosome assembly, and the focus of this study is on the overall effect of UV lesions on nucleosome unwrapping, regardless of their nucleosome locations. In addition, we have shown that UV damage to the 5 S rDNA positioning sequence does not significantly change the rotational and translational settings in the nucleosome (12).

We synthesized two sets of oligonucleotides, one for fluorescent dye labeling and the other for UV treatment (see “Experimental Procedures”). Complete 147 bp 601 DNA containing both UV lesions and the fluorescent dyes was prepared by ligation and gel purification to ensure that the fragments contain exclusively DNA photoproducts (and essentially no DNA strand breaks). The UV-damaged DNA was reconstituted into NCPs with nearly the same efficiency as undamaged DNA (Fig. 2A), suggesting that incorporation of photolesions in the 601 sequence does not markedly affect the assembly of NCPs. This result is in agreement with our past studies using different sequences (42–44).

Because the two DNA fragments used in these studies are short (95 bases each), we used UV doses in the kJ/m² range to incorporate sufficient levels of UV lesions. Using specific antibodies to detect each photolesion, we observed that CPDs appear to saturate after ∼4 kJ/m², whereas 6-4PPs appeared to increase over the UV dose range used (Fig. 2B). To quantify the level of these photoproducts, we used an exonuclease blockage assay developed previously (33). The 3′ → 5′ exonuclease activity of T4 DNA polymerase, in the absence of dNTPs, is blocked quantitatively by CPDs and 6-4PPs (45), and these phototproducts can be detected by slot blot after T4 DNA polymerase digestion (33). In some samples, CPDs were removed by treatment with UV photolyase, which allowed detection of either both photoproducts together (no photolyase) or just 6-4PPs (with photolyase) following T4 DNA polymerase digestion (Fig. 2C). The average numbers of CPD/fragment and 6-4PPs/fragment (33), show that the level of CPDs starts to saturate at 4 kJ/m², and the maximum level was 0.25 CPDs/fragment for the highest dose tested (Fig. 2D). However, 6-4PPs increased almost linearly up to the highest UV dose used (24 kJ/m²; Fig. 2D). This data agrees well with the antibody data (Fig. 2B) and indicates that up to 40–45% of the fragments contain a UV lesion in these experiments.

To examine nucleosome unwrapping in damaged and undamaged NCPs, we initially used gel-based FRET, which enabled direct visualization of energy transfer on multiple samples (46). The labeled naked DNA and NCPs were resolved on native polyacrylamide gels and examined through different emission filters to visualize FRET in each sample. As shown in Fig. 3A (top panel), scanning gels with the green laser (532 nm; donor excitation) yields the unquenched Cy3 signals in both naked DNA and NCPs. In the second panel of Fig. 3A, excitation of the donor (green laser) results in Cy5 fluorescence (shown in red by using the 670 nm emission filter) only from the NCP bands. No FRET signal is detected in the naked DNA bands. Furthermore, the Cy5 signal from NCP bands becomes weaker with increased UV dose to the DNA when visualized either directly (Fig. 3A, second panel) or when the signals are merged (third panel). Quantifications of the fluorescence intensities show that the FRET efficiency decreases with increased UV irradiation (Fig. 3A, numbers between panels). Finally, scanning the gels by direct excitation of the acceptor with a red laser (633 nm) yields comparisons of labeling efficiency and loading, which are approximately the same for the gel shown (Fig. 3A, bottom panel). Therefore, the gel-based FRET results indicate that there is reduced energy transfer efficiency in the UV damaged NCPs.

To quantify the effect of UV dose on the FRET signal, we measured the FRET efficiency in solution. The emission spectra clearly demonstrate that the FRET efficiency decreases with increased UV dose (Fig. 3B). For comparison, the FRET efficiency from the sensitized emission of the acceptor was normalized to the FRET efficiency of the nondamaged, fully reconstituted NCPs. As seen in Fig. 3C, the efficiency decreased more over the first 8 kJ/m² UV dose than at higher UV doses. At the highest UV dose used (24 kJ/m²), the efficiency was reduced by at least 30% compared with nondamaged NCPs (Fig. 3C).

At a low salt concentration, NCPs exhibited a dynamic conformational equilibrium between fully wrapped and partially unwrapped states, with equilibrium constant, K_eq (24). We used this constant to describe the change in conformational equilibrium induced by UV DNA lesions. This analysis clearly shows that the equilibrium shifts from wrapped to unwrapped states with increased DNA damage (Fig. 3D), suggesting that a given UV-damaged nucleosome spends more time in an unwrapped state compared with that of an undamaged nucleosome. At the highest dose tested (24 kJ/m²), we calculate that the NCP population in our samples spends almost 50% more time in an unwrapped state. This is the case, even though <50% of the NCPs in our samples contain a UV-induced lesion (Fig. 2D). Thus, the change in equilibrium between wrapped and unwrapped states of just the UV-damaged NCPs in these samples is substantial.

The electrostatic interaction between histone octamers and DNA is weakened by increased ionic strength of the solvent (47), and consequently, the dynamic properties of NCPs are affected (24). To determine whether UV damage changes the salt-induced dynamic properties of NCPs, we titrated NCPs with increasing salt and compared the FRET efficiency with different levels of UV lesions. For NCPs damaged at a given UV dose, the FRET efficiency (E) was normalized to the efficiency at the lowest salt concentration (50 mm NaCl). At low salt concentrations (below 250 mm NaCl), where changes are associated with DNA “breathing” (37, 40), there was little change in FRET (Fig. 4). However, we observed a small, yet significant, variation in FRET between 0.25 m and 1 m NaCl salt (Fig. 4). The undamaged NCPs yield a higher FRET in this salt concentration range compared with UV-damaged NCPs. This indicates that, on average, the distance between the two dyes is larger in the NCPs containing UV lesions, most likely due to increased unwrapping of the DNA in damaged NCPs. At a 1.5 m NaCl concentration, the FRET efficiency decreases to <10% of the low salt value, indicating the DNA is dissociated from histone octamers.

The FRET data indicate that UV-damaged NCPs exist in open structures more often than nondamaged NCPs. Therefore, UV-damaged nucleosomes may be more accessible to site-specific DNA-binding proteins. REA is a method that has been successfully used to measure spontaneous nucleosomal DNA exposure (23, 25, 48). We tested the REA of nucleosome DNA using two restriction sites in the 601 sequence to examine digestion at different locations from the DNA ends in NCPs. A HaeIII site is located near an end of DNA within the NCP, whereas an RsaI site is close to the dyad center (Fig. 5A). For these experiments, we decreased the UV dose to adjust for the exposure of the full-length 601 fragment (149 bp) versus the shorter single-strand fragment (95-mer) used in the FRET experiments. As a control, naked DNA (± UV damage) was digested by the enzymes and showed no differences in activity (supplemental Fig. S2), highlighting the fact that any UV-associated changes in REA should reflect nucleosome dynamic variations rather than DNA structural changes.

As shown in Fig. 5, these two restriction sites are only slightly accessible in NCP DNA (Fig. 5, B and C, P1 and P2 fragments in lane 0). Furthermore, there are no obvious differences in RsaI REA among UV-treated NCPs at different UV doses (Fig. 5B, lanes 2.5–15, kJ/m²), where only ∼10% of the DNA is cleaved. This result suggests that UV damage does not lead to long range opening events of nucleosome DNA, near the dyad center of NCPs. However, NCPs bearing UV lesions show enhanced REA to HaeIII (Fig. 5C), reflected by enhanced yields of long (P1) and short (P2) restriction fragments with increased UV dose (Fig. 5C, lanes 2.5–15 kJ/m²). Quantification of these results shows that up to ∼30% more HaeIII sites in NCPs are rendered accessible over this UV dose range (Fig. 5D). (We note that a minor fraction of NCPs can dissociate during mixing with REA buffer (23, 49), giving rise to a small drop in the fraction of undigested DNA (Fig. 5D, lane 0.)) These results indicate that UV lesions promote greater NCP unwrapping within at least the first ∼30 bp from the ends of NCP DNA. Moreover, the UV-dose dependence suggests that the magnitude and rate of the unwrapping-rewrapping fluctuations are influenced by the extent of UV damage. These data support the “site-exposure” model, where NCP associated DNA peels off from the end, facilitating invasion of nucleosomes by DNA binding proteins (24).

Our analysis with UV-irradiated and nonirradiated NCPs demonstrates that UV lesions enhance nucleosome dynamics (Fig. 3). However, as “global” UV irradiation incorporates both 6-4PPs and CPDs at variable nucleosome positions and in only a fraction of the NCPs (Fig. 2), we generated site-specific UV-damaged 601 fragments by incorporating a single CPD (601.DF1 CPD) or 6-4PP (601.DF2 6-4PP) into the same position of 601 NCPs (58 nt from the 5′-end; Fig. 6A and supplemental Fig. S3). As observed with other DNA sequences (43, 44, 50), these UV-damaged 601 fragments can be reconstituted successfully with histone octamers (supplemental Fig. S4). In addition, we observed significant retardation of the naked 601.DF2 6-4PP fragments on gels, possibly the result of enhanced DNA bending of these fragments, although this did not prevent NCP formation (Fig. S4B). Finally, the presence of a single CPD in the 601.DF1 CPD fragments was confirmed by complete digestion with T4 endoV (data not shown).

We then examined the effect of each of these lesions on nucleosome dynamics. As can be seen in Fig. 6, B and C, the FRET efficiency was reduced by ∼16% in CPD-containing NCPs, and ∼9% in 6-4PP-containing NCPs compared with undamaged NCPs. Given the fact that the donor-acceptor distance (R) for undamaged NCPs is 3 nm (39), R for NCPs containing a single CPD or 6-4PP lesion is ∼4.6 nm and ∼4.2 nm, respectively. Furthermore, in each case, the FRET efficiency is reduced, indicating that the introduction of a single UV lesion in nucleosome DNA can increase NCP unwrapping. This supports the UV dose-dependent FRET change we observed with UV-irradiated NCP DNA (Fig. 3). Moreover, NCPs with either a single CPD or 6-4PP are more sensitive to salt (Fig. 6, D and E). In the range of 0.25–1 m salt concentrations, the FRET efficiency in NCPs with single UV lesions is significantly lower than that of undamaged NCPs, indicating that NCPs containing just one DNA photoproduct are significantly less stable with increasing salt.