Genome‐wide impact of cytosine methylation and DNA sequence context on UV‐induced CPD formation

Exposure to ultraviolet (UV) light is the primary etiological agent for skin cancers because UV damages cellular DNA. The most frequent form of UV damage is the cyclobutane pyrimidine dimer (CPD), which consists of covalent linkages between neighboring pyrimidine bases in DNA. In human cells, the 5′ position of cytosine bases in CG dinucleotides is frequently methylated, and methylated cytosines in the TP53 tumor suppressor are often sites of mutation hotspots in skin cancers. It has been argued that this is because cytosine methylation promotes UV‐induced CPD formation; however, the effects of cytosine methylation on CPD formation are controversial, with conflicting results from previous studies. Here, we use a genome‐wide method known as CPD‐seq to map UVB‐ and UVC‐induced CPDs across the yeast genome in the presence or absence in vitro methylation by the CpG methyltransferase M.SssI. Our data indicate that cytosine methylation increases UVB‐induced CPD formation nearly 2‐fold relative to unmethylated DNA, but the magnitude of induction depends on the flanking sequence context. Sequence contexts with a 5′ guanine base (e.g., GCCG and GTCG) show the strongest induction due to cytosine methylation, potentially because these sequence contexts are less efficient at forming CPD lesions in the absence of methylation. We show that cytosine methylation also modulates UVC‐induced CPD formation, albeit to a lesser extent than UVB. These findings can potentially reconcile previous studies, and define the impact of cytosine methylation on UV damage across a eukaryotic genome.

Many somatic mutations in skin cancers are specifically elevated at dipyrimidine sites flanked by a 3 0 guanine base (i.e., TCG and CCG) (Hayward et al., 2017).For example, mutation hotspots in the TP53 (p53) tumor suppressor gene in non-melanoma skin cancers frequently occurred at CCG sequences (Gonzalgo & Jones, 1997;Tommasi et al., 1997;You et al., 1999).Cytosine bases occurring in CG dinucleotides in the human genome are frequently methylated, which represents an important epigenetic mark that regulates chromatin organization, gene expression, and silencing (Gruenbaum et al., 1981;Lister & Ecker, 2009;Nishiyama & Nakanishi, 2021).The human enzymes DNMT1 and DNMT3A/B methylate cytosines at the 5 0 position of the cytosine base, yielding 5-methylcytosine (5mC), which can be reversed by the action of the ten-eleven translocation dioxygenase and thymine DNA glycosylase (TDG) enzymes (Kohli & Zhang, 2013;Li & Zhang, 2014;Nishiyama & Nakanishi, 2021).Methylation of cytosine bases at CG dinucleotides generally promotes mutagenesis in cancers and other cells in a clock-like fashion (Alexandrov et al., 2013;Alexandrov et al., 2015), since spontaneous deamination of 5mC results in a thymine base, which is likely to be repaired less efficiently than uracil.It has been hypothesized that 5mC also promotes UV mutagenesis by modulating the formation of UV damage (Pfeifer et al., 2005).For example, a recent study discovered that in melanomas, C > T substitutions and UVB-induced CPD lesions are depleted at TCG sites in proximal promoter regions (but enriched elsewhere), likely because these regions are frequently associated with unmethylated CpG 'islands' (Lindberg et al., 2019).
Previous studies tested the effects of cytosine methylation on the formation of UV damage, often with conflicting results.An early study employing ligation-mediated PCR revealed that the presence of 5mC at sites in the TP53 gene induced CPD formation as much as 5-to 15-fold in normal human keratinocytes, particularly following exposure to sunlight (Tommasi et al., 1997).Methylation-dependent induction of CPD lesions was also observed in a plasmid (containing p53 exons) that was methylated in vitro by a CpG methyltransferase (Tommasi et al., 1997).
A similar study assaying CPD formation at methylated CG dinucleotides in a gene (PGK1) located in the inactive X chromosome in female cells (relative to the unmethylated X chromosome in male cells) found that 5mC caused an average increase in CPD formation of $1.7-fold when the cells were exposed to UVB light (Rochette et al., 2009).Roughly similar levels of CPD induction were observed regardless of the UVB wavelength chosen.In contrast, CPD formation was not induced when cells were exposed to UVC light.These findings are consistent with an in vitro study reporting that CPDs are induced $two-fold more frequently in poly(5mdC):poly(dG) homopolymers as compared to unmethylated poly(dC):poly(dG) DNA following UVB irradiation, but methylation did not induce CPD formation following UVC irradiation (Mitchell, 2000).These studies proposed that a 6 nm red shift in the absorption spectrum of 5mC relative to unmethylated C (You et al., 1999) was responsible for the elevated frequency of CPD formation in response to UVB (but not UVC) irradiation.However, other in vitro studies indicated that the 5mC increases the quantum yield of CPD formation for both UVB and UVC light (Banyasz et al., 2016;Esposito et al., 2014), by promoting the stacking of the 5mC base with the neighboring pyrimidine and suppressing base stacking with the 3 0 guanine base (Banyasz et al., 2016;Esposito et al., 2014), as neighboring guanine bases (5 0 or 3 0 ) can suppress CPD formation (Bryan et al., 2014;Cannistraro & Taylor, 2009;Esposito et al., 2014;Law et al., 2013).Conversely, a more recent in vitro study found that the presence of 5mC decreased CPD formation upon either UVB or UVC irradiation at a very specific set of sequence contexts (e.g., TTTCG[A/G]) (Leung & Murray, 2021); the mechanism responsible for this effect is unclear.
Genome-wide methods of mapping UV damage have proven to be a powerful tool to characterize the effects of DNA-bound proteins and chromatin organization on the frequency of CPD formation (Bohm et al., 2023;Brown et al., 2018;Elliott et al., 2018;Elliott & Larsson, 2021;Hu et al., 2017;Mao et al., 2018;Mao & Wyrick, 2019;Premi et al., 2019;Roberts et al., 2019;Sivapragasam et al., 2021).Here, we use a genome-wide method known as CPD-seq to characterize the impact of cytosine methylation by the CpG methyltransferase M.SssI on UVB-and UVC-induced CPD formation across the yeast genome in vitro.This experimental strategy avoids potential complications due to protein binding and heterogeneous methylation levels that can beset cellular studies, yet analyzes the effects of 5mC on CPD formation in a wide variety of DNA sequence contexts, unlike previous in vitro studies (Banyasz et al., 2016;Esposito et al., 2014;Leung & Murray, 2021;Mitchell, 2000).Our data indicate that cytosine methylation promotes UVB-induced CPD formation nearly two-fold, but the magnitude of induction is strongly dependent on the flanking DNA sequence context.
We also show that 5mC can weakly modulate UVC-induced CPD formation, again in a sequence context-dependent manner.

| Yeast genomic DNA isolation
Wild-type (BY4741) yeast cells were grown overnight until mid-log phase (OD 600 $ 0.8) before genomic DNA was extracted via the PCI (phenol: chloroform:isoamyl alcohol 25:24:1) method.In this, cells from $25 mL of culture were spun down and the resultant cell pellet was mixed with 250 μL of DNA lysis buffer (2% [vol/vol] Triton X-100, 1% SDS, 100 mM NaCl, 10 mM Tris-HCl, pH 8.0, and 1 mM EDTA) and 250 mL acidwashed glass beads.This solution was vortexed on the highest setting in 2 min sets, twice.200 μL of TE (10 mM Tris-HCl, pH 7.5, and 1 mM EDTA) was added and the cell lysates were inverted multiple times to mix before centrifuging at 13,000 rpm for 10 min.The DNA was then precipitated out of the supernatant with 1 mL of ethanol at À20 C for at least 15 min.DNA was pelleted via centrifugation and washed with 70% ethanol.Pellets were then dissolved in 200 μL TE and incubated with RNase A (ThermoFisher Scientific) at 37 C for 1 h.To purify the DNA, a second PCI extraction, ethanol precipitation, and reconstitution of DNA in 100 μL sterile deionized water was performed.
For the optimized experiments (UVB replicate 2 and UVC), the DNA samples were methylated in 75 μL reactions containing 1X NEBuffer2 (50 mM NaCl, 10 mM Tris-HCl, 10 mM MgCl 2 , 1 mM DTT, pH 7.9 @ 25 C), 160 μM S-adenosylmethionine (SAM), 15 μL methyltransferase, and 15 μg previously purified DNA.This optimized reaction was done in quadruplicate for a total of 60 μg of DNA methylated and incubated in a thermocycler at 37 C for 4 h before supplementing with additional 160 μM SAM before the reaction ran overnight.Reactions were then heat-killed at 65 C for 20 min before further processing.Both nonmethylated cohorts of DNA ran through similar reactions in only water and NEBuffer2 conditions for both replicates.

| UV irradiation
DNA dissolved in water was then spotted onto glass coverslips in 10 μL spots for subsequent UV irradiation with either 500 J/m 2 UVB, using UVP CL-1000 M midrange crosslinker (Analytik Jena) with emission peak at 302 nm, according to the manufacturer's calibration, or 90 J/m 2 UVC (emission peak at 254 nm), according to our previous calibration.
UV irradiation was done on ice to prevent evaporation.An aliquot of DNA was reserved for a "no UV" control that was not irradiated.DNA was then recollected in a new sterile tube for further processing.

| CPD-seq protocol
Naked DNA was irradiated with either 500 J/m 2 UVB or 90 J/m 2 UVC to induce CPD lesions.The genome was then fragmented by sonication (30s ON/OFF, 25 cycles; Diagenode Biorupter 300), and ethanol precipitated.CPD-seq was performed as previously described (Bohm et al., 2021;Mao et al., 2016;Mao & Wyrick, 2020).Briefly, the first adapter was ligated to the ends of each fragment and then any remaining 3 0 hydroxyl groups were blocked via terminal transferase.T4 endonuclease V and APE1 (NEB) were then used to cleave at sites of CPDs across the fragmented genome.The biotinylated second adapter was ligated to the new free 3 0 OH groups.Fragments with both adapters were selected for with streptavidin beads with a high affinity for biotin on the second adapter.These fragments were PCR amplified and sent out for Ion Torrent sequencing.Ampure XP beads (Cytiva) were used for size selection and clean-ups between enzymatic steps.
Alignment of the resultant sequence reads to the yeast reference genome, saccer3 was then performed using Bowtie2 (Langmead & Salzberg, 2012), and the resulting SAM files were converted to BED files using SAMtools (Li et al., 2009).Analysis of CPD formation at different dinucleotide and tetranucleotide sequence contexts was performed using custom Perl scripts and BEDTools (Quinlan & Hall, 2010).The ratio of CPD-seq reads between methylated and unmethylated DNA was normalized using the count of TT CPD-seq reads in each sample, since the TT CPD formation should not be affected by cytosine methylation.Linear regression analysis was performed using GraphPad Prism.

| RESULTS
To characterize the effects of DNA sequence context and cytosine methylation on UV damage, we used our CPD-seq method to map CPD lesions at single-nucleotide resolution across the yeast genome.We irradiated isolated yeast genomic DNA in vitro using roughly equivalent doses of either UVB (500 J/m 2 ) or UVC (90 J/m 2 ) light.Analysis of the resulting CPD-seq reads showed clear enrichment of reads associated with lesions at dipyrimidines (i.e., TT, TC, CT, CC) in the UV-irradiated DNA, but not in the No UV control (Figure 1a, b).Dipyrimidine enrichment was similar between the UVB and UVC-treated samples, and both samples showed similar sequence preferences for CPD formation (i.e., TT > TC > CT > CC; see Figure 1a, b), as expected from previous studies (Mao et al., 2016(Mao et al., , 2020;;Mao & Wyrick, 2020;Selvam et al., 2022).
To investigate how flanking DNA sequences affect CPD formation, we analyzed the frequency of CPD-seq reads in different tetranucleotide sequence contexts, normalizing based on the frequency of each tetranucleotide sequence in the genome.For TT and TC CPD lesions in the UVB-irradiated DNA, CPD formation was elevated if the 5 0 flanking base was a pyrimidine (C or T) and suppressed if the 5 0 flanking base was a guanine (Figure 1c, d).For example, CPD formation was 2.8-fold higher at TT dinucleotides with a flanking 5 0 cytosine than a flanking 5 0 guanine.CPD formation was also suppressed if the 3 0 flanking base was guanine, or extended a string of thymine bases (e.g., ATTT or TTTT).Similar trends were observed at CT and CC CPD lesions (Figure S1a, b), although the effects of flanking bases were not as dramatic.Similar trends were apparent after UVC irradiation (Figures 1e, f and S1c, d), although the presence of a 5 0 or 3 0 flanking adenine more strongly promoted CPD formation upon UVC irradiation, particularly for TT and CT CPD lesions.

| In vitro CpG methylation promotes UVBinduced CPD formation at YCG sequences across the yeast genome
To characterize the effect of cytosine methylation (5mC) on CPD formation across the yeast genome, we used the recombinant CpG methyltransferase derived from Spiroplasma strain MQ1 (M.SssI) to methylate isolated yeast genomic DNA in vitro.This enzyme has been frequently used to methylate DNA at CpG sites in vitro and in vivoincluding in previous studies analyzing the impacts of 5mC on CPD formation (Leung & Murray, 2021;Tommasi et al., 1997)-since the enzyme generates 5mC patterns at CG dinucleotides that mimic those    To characterize the impact of cytosine methylation on CPD formation, we irradiated the methylated yeast genomic DNA with 500 J/ m 2 of UVB light (i.e., the same dose used for the unmethylated genomic DNA control), and mapped the resulting CPD lesions using CPD-seq (Figure 3a).CPD-seq reads were significantly enriched at dipyrimidine sequences in the UVB-irradiated methylated DNA (Figure 3b), similar to the enrichment observed in the UV-irradiated unmethylated DNA (Figure 1a).We compared the number of CPDseq reads at each tetranucleotide sequence context in the methylated DNA relative to the matched unmethylated DNA control (both UVirradiated).While most tetranucleotide contexts showed similar levels of CPD formation regardless of methylation status (Figure 3c), contexts in which a cytosine base was flanked by a 3 0 guanine (e.g., ATCG, TCCG, etc.) generally showed higher CPD formation in the methylated DNA (5mC) sample (see red circles in Figure 3c).Since these sequence contexts are the only ones that would be methylated by the CpG methyltransferase, these findings indicate that cytosine methylation promotes UVB-induced CPD formation.

UVB TC CPDs A T T A A T T C A T T G A T T T C T T A C T T C C T T G C T T T G T T A G T T C G T T G G T T T T T T A T T T C T T T G T T T
To quantify the effect of cytosine methylation on CPD formation, we calculated the ratio of CPD lesions in the UV-irradiated methylated DNA (5mC) relative to the unmethylated DNA for each tetranucleotide sequence context.These ratios were normalized so that the total abundance of TT CPD lesions in both libraries were the same.
This analysis (see Figure 3d, e) indicated that CPD formation was elevated nearly two-fold on average in the methylated DNA specifically at CG-containing dipyrimidine sites (e.g., ATCG, TCCG, etc.).Little to no induction was observed that non-CG sites (Figure 3d), nor in the same sequence contexts at TG dinucleotides (e.g., ATTG, TCTG; sequence context.The strongest induction due to cytosine methylation occurred at GCCG and GTCG sequence contexts, which were elevated 2.7-fold and 2.6-fold, respectively, in the methylated DNA sample (Figure 3d).In contrast, there was relatively little CPD induction in the methylated sample in the TTCG and CTCG sequence contexts (Figure 3c, d).One possible explanation for these lower levels of CPD induction was that the TTCG and CTCG sequence contexts are not efficiently methylated by the M.SssI CpG methyltransferase.To ensure that the sequence biases observed were not a byproduct of incomplete methylation across the genome at those sequence contexts, we optimized our in vitro methylation protocol using a BstBI digest to confirm efficient methylation at TTCG sequences (Figure S2).The optimized methylation protocol blocked DNA cleavage by both the HpaII and BstBI methylation-sensitive restriction enzymes (Figure 4a).
We repeated our CPD-seq analysis using this optimized methylation protocol on both methylated and unmethylated yeast genomic DNA (replicate #2), which showed similar enrichment at dipyrimidine sequences in the UVB-irradiated samples (Figure 4b).We observed a similar pattern of CPD induction specifically at tetranucleotide sequences that contain a 3 0 CG sequence (NYCG; Figure 4c).CPD formation was elevated nearly two-fold on average in the methylated DNA specifically at CG-containing dipyrimidine sites, but the magnitude of induction varied significantly depending on sequence context (Figure 4d, e).Again, the TTCG and CTCG sequence contexts showed the weakest degree of CPD induction in the methylated samples.
These data closely mirrored our previous results, indicating that low levels of CPD induction at TTCG (and CTCG) sequence contexts cannot be attributed merely to incomplete cytosine methylation at these sites, but instead likely reflects the impact of flanking sequence context on 5mC-associated CPD induction (see below).In the original experiment, we also observed lower frequencies of CPDs in CYYN sequence contexts (e.g., CCTN and CTTN in Figure 3e) in the methylated DNA compared to the unmethylated control, even at sequence contexts that are not methylated (e.g., CTTA).However, this effect is largely not recapitulated in the replicate experiment (Figure 4d, e), indicating it may be due to experimental variability or noise.

| In vitro CpG methylation weakly modulates UVC-induced CPD formation
To determine whether cytosine methylation also impacts UVCinduced CPD levels, we irradiated the methylated (and unmethylated) yeast genomic DNA with UVC light (90 J/m 2 ) and mapped the resulting CPD lesions using CPD-seq.We observed significant enrichment of CPD-seq reads associated with lesion-forming dipyrimidine sequences (Figure 5a), similar to the results for the unmethylated UVC-irradiated sample (Figure 1b).Analysis of the CPD-seq data indicated that, in contrast to UVB-irradiation, cytosine methylation caused relatively little CPD induction at NYCG sequences upon UVCirradiation (Figure 5b).
Analysis of the normalized ratio of CPD lesions in the methylated DNA relative to the unmethylated control revealed small increases in CPD formation at certain sequence contexts (Figure 5c), including GTCG ($1.4-fold), GCCG ($1.3-fold) and CCCG ($1.3-fold).These increases in UVC-induced CPD formation were not observed in unmethylated TT and CT lesion-forming sequences (e.g., GTTG, GCTG; see Figure 5d).However, the magnitude of these increases in NYCG sequences was much lower than in UVB-irradiated samples (compare Figures 4d, 5c).In contrast, methylation caused a decrease in UVC-induced CPD formation in the TTCG sequence context ($0.74-fold; Figure 5b, c), but not in the matched TTTG sequence context control ($1.02-fold; Figure 5d).Taken together, these results indicate that cytosine methylation only weakly modulates UVCinduced CPD formation, and can in certain sequence contexts (i.e., TTCG) suppress CPD formation.

| Methylation-induced CPD formation is elevated at poor CPD-forming sequence contexts
We noticed that UVB-induced CPD formation (in the absence of methylation) was lowest for sequence contexts (e.g., GTCG and GCCG) with the greatest degree of CPD induction due to 5mC, and highest for sequence contexts (i.e., TTCG, CTCG) that showed very little CPD induction due to 5mC.To further test this correlation, we plotted CPD formation in unmethylated DNA for each NYCG tetranucleotide context (after normalizing for the tetranucleotide sequence frequency in the yeast genome) relative to the magnitude of CPD induction due to cytosine methylation (5mC).This analysis revealed a strong negative correlation between unmethylated CPD formation and 5mC-dependent CPD induction in both UVB and UVC-irradiated samples (Figure 6a, b).Linear regression analysis of the UVB-irradiated samples indicated a very significant negative correlation ( p < .001; Figure 6a), with an R 2 of 0.91, indicating that most of the variance in 5mC-dependent CPD induction in the different sequence contexts can be explained by the baseline level of CPD formation in unmethylated DNA.The correlation in the UVC data, albeit not as strong as for UVB data, still showed a significant negative correlation ( p < .05; Figure 6b), with an R 2 of 0.70.These findings suggest that methylation-induced CPD formation is strongly influenced by the intrinsic propensity of each sequence context to form CPD lesions in the absence of DNA methylation.

| DISCUSSION
In the human genome, cytosine bases in CG dinucleotides are frequently methylated (5mC).This methylation is thought to modulate UV-induced damage and mutagenesis, but previous reports have yielded conflicting results (Esposito et al., 2014;Leung & Murray, 2021;Mitchell, 2000;Rochette et al., 2009;Tommasi et al., 1997).Here, we used a genome-wide method known as CPD-seq (Mao et al., 2016;Mao & Wyrick, 2020)  CPD-seq analysis of UV-irradiated yeast genomic DNA (unmethylated) confirmed previous reports (Brash & Haseltine, 1982;Bryan et al., 2014;Cannistraro & Taylor, 2009;Law et al., 2013;Lu et al., 2021;Mitchell et al., 1992) that flanking DNA sequence context significantly impacts CPD formation.Our data indicate that CPD formation tends to be stimulated if the 5 0 flanking base is a pyrimidine (C or T) and suppressed if the 5 0 or 3 0 flanking base is guanine.For TT dimers, a 3 0 adenine base also promotes CPD formation, particularly for UVC irradiation, while a 3 0 thymine is often more favorable for TC or CC dimers.These results are largely consistent with previous reports (Bryan et al., 2014;Cannistraro & Taylor, 2009;Law et al., 2013;Lu et al., 2021;Mitchell et al., 1992), although there were some differences (e.g., impact of 3 0 flanking guanine on CPD formation; Lu et al., 2021).Flanking sequence effects on rates of UVCinduced photoreversion of CPD lesions (Law et al., 2013) may also influence CPD levels in the UVC CPD-seq data; however, because we used a relatively low UVC dose in our study (i.e., $90 J/m 2 ), the contributions of photoreversion should be relatively minor.Our recent study of 6-4PP and thymine-adenine (TA) photoproduct formation revealed similar flanking sequence preferences, as a 5 0 flanking pyrimidine base generally promoted 6-4PP and TA-PP formation, while a flanking 5 0 or 3 0 guanine base suppressed photoproduct formation (Bohm et al., 2023).These findings suggest that common biophysical principles dictate the effects of flanking DNA sequences on the formation of different classes of UV photoproducts.For example, a flanking guanine base is thought to suppress CPD formation by quenching photochemistry through an electron transfer mechanism (Cannistraro & Taylor, 2009;Lu et al., 2021;Pan et al., 2011); it is possible that a flanking guanine base may suppress 6-4PP or TA-PP formation through a similar mechanism.However, a 3 0 flanking adenine base (for 6-4PP) or a 3 0 flanking T or A base (for TA-PP) appears to more strongly stimulate 6-4PP and TA photoproduct formation more than CPD formation (Bohm et al., 2023).
While yeast genomic DNA is normally not methylated at cytosine bases, we show that incubation of yeast genomic DNA in vitro with the M.SssI CpG methyltransferase results in efficient methylation, at least at the subset of sequence contexts that we are able to monitor by methylation-sensitive restriction enzyme digestion.CPD-seq analysis indicates that this treatment promotes CPD formation in response to UVB irradiation specifically at CG sequences, which are targeted for methylation.On average, 5-methylcytosine (5mC) causes a nearly two-fold increase in UVB-induced CPD formation relative to unmethylated cytosine, consistent with previous reports of 1.7-fold (Rochette et al., 2009) and $ two-fold induction (Mitchell, 2000).This induction in CPD formation is likely due to both a 6 nm red shift of the 5mC absorption spectra into the UVB range (You et al., 1999) and changes in the 5mC DNA structure which promote the quantum yield of the CPD-forming [2 + 2] cycloaddition reaction (Banyasz et al., 2016;Esposito et al., 2014).
However This sequence-dependent variation in 5mC CPD induction can potentially reconcile conflicting results from previous studies.For example, many of the TP53 mutation hotspots in skin cancers that are associated with methylation-dependent CPD induction (Tommasi et al., 1997;You et al., 1999)  There also have been conflicting reports about the effect of cytosine methylation on UVC (254 nm) damage, with some reports suggesting 5mC promotes UVC-induced CPD lesions (Banyasz et al., 2016;Esposito et al., 2014), whereas others suggest that 5mC has no effect (Rochette et al., 2009) or suppresses its formation (Leung & Murray, 2021).Our CPD-seq data indicate that the effect of 5mC on UVC-induced CPD levels is largely dependent on the sequence context, with some sequence contexts (e.g., GCCG or GTCG) showing slightly higher CPD levels due to cytosine methylation and others showing no effect or even suppressed CPD formation (i.e., TTCG).Again, this dependence on sequence context may explain some of these conflicting results.For example, a previous report that 5mC suppresses UVCinduced CPD formation (0.74-to 0.8-fold) used a TTCG sequence context (see above), which is consistent with our UVC CPD-seq data for TTCG sequences ($0.74-fold lower CPDs).The variation in the effect of 5mC on UVC-induced CPD levels may reflect the relative contributions, in different sequence contexts, of the red shift of the 5mC absorption spectra, which would reduce absorption (and CPD formation) at 254 nm (Banyasz et al., 2016;Esposito et al., 2014;You et al., 1999), and the 5mC-dependent DNA conformational changes, which would promote CPD formation (Banyasz et al., 2016;Esposito et al., 2014).It is possible that the impact of 5mC on UVC-induced photoreversion rates could also affect these results, although the low UVC dose used should minimize its potential impact.
In summary, our data suggest that cytosine methylation modulates UV-induced CPD formation in a manner that is dependent on the flanking DNA sequence context.Since many CG dinucleotides are methylated in the human genome, and many mutation hotspots in skin cancer are associated with these methylated dinucleotides (Pfeifer et al., 2005;Tommasi et al., 1997;You et al., 1999), these findings have potentially important ramifications to our understanding of the mechanism of skin carcinogenesis.
Methylation states were validated by enzymatic digests of a subset of the DNA with HpaII and McrBC, and in a subset of samples BstBI (NEB), according to manufacturer's specifications.For each cohort of DNA in each enzymatic reaction, $5 μg of DNA was digested and run out on a 0.8% agarose gel via gel electrophoresis.In parallel to each enzymatic digest, a "no enzyme" control was run.Digestion and cleavage products on a gel with BstBI and HpaII represent the presence of non-methylated DNA, and digestion and cleavage products on a gel with McrBC represents the presence of methylated DNA.After validation of respective methylation states, DNA was immediately UV irradiated or stored at À20 C until irradiation.
U R E 1 CPD Formation in Non-Methylated Yeast DNA.(a) CPD-seq read counts for non-methylated wild-type yeast genomic DNA irradiated with 500 J/m 2 of UVB light or No UV control.(b) Same as (a), but for DNA irradiated with 90 J/m 2 of UVC light.(c), (d) Normalized CPDs at different tetranucleotide contexts flanking CPD lesions at (c) TT or (d) TC dinucleotides in UVB-irradiated non-methylated yeast genomic DNA.CPD-seq read density across both UVB replicates was normalized to the tetranucleotide sequence context frequencies across the yeast genome.(e), (f) Same as (c), (d), but for UVC-irradiated samples.
U R E 2 Validation of Methylated Yeast DNA.(a) Agarose gel electrophoresis of yeast genomic DNA samples digested with McrBC, which specifically cleaves methylcytosine-containing DNA.M.SssI is a CpG methyltransferase (MT).(b) Same as panel (a), except using HpaII to digest the DNA.HpaII will only cleave sites that are not methylated (i.e., no 5-methylcytosine).Effect of cytosine methylation on UVB-induced CPD formation.(a) Schematic showing protocol of how the effects of CpG methylation by M.SssI CpG methyltransferase on UVB-or UVC-induced CPD formation was analyzed across the yeast genome using CPD-seq.CPD-seq schematic adapted from Mao et al. (2016).(b) Number of CPD-seq reads associated with putative lesions at the indicated dinucleotides in UVB-irradiated methylated DNA (5mC) versus UVB-irradiated unmethylated DNA and non-UV (and unmethylated) DNA controls.UVBirradiated unmethylated DNA data is from Figure 1A.(c) Plot showing number of CPD-seq reads in each tetranucleotide sequence context (centered on a CPD-forming dipyrimidine) in the UVB-irradiated 5-methylcytosine (5mC) DNA relative to the UVB-irradiated unmethylated (No Methyl) DNA control.Red dots represent tetranucleotide sequences that match a NYCG pattern, as these are targets for 5mC methylation by M.SssI methyltransferase.(d), (e) Normalized ratio of CPD-seq reads in UVB-irradiated 5-methylcytosine (5mC) DNA relative to the UVBirradiated unmethylated (No Methyl) DNA control.Ratio was normalized so that the number of CPD-seq reads at TT dinucleotides would be the same between the two samples.(d) Shows normalized ratios for lesions at TC and CC dipyrimidines, which when flanked by a 3 0 guanine are targeted for methylation; (e) shows normalized ratios for lesions at TT and CT dipyrimidines, which would not be methylated.The color of the bar indicates the 3 0 flanking base.found in human cells.To verify the methylation of yeast genomic DNA at CG dinucleotides, we digested the methylated DNA with McrBC, which only cleaves at methylated 5mC sites, and HpaII, whose cleavage is blocked by CpG methylation (Figure S2).Unlike the unmethylated DNA control, the methylated yeast genomic DNA was cleaved by McrBC (Figure 2a) and protected from cleavage by HpaII (Figure 2b), confirming CpG methylation in the yeast genomic DNA incubated with M.SssI.

Figure 3e )
Figure3e), since these sequences are not methylated.Closer inspection revealed that the magnitude of CPD induction at different tetranucleotide sequences containing a terminal CG sequence (i.e., 5 0 -NYCG-3 0 ) significantly varied depending on the U R E 5 CPD-seq analysis of methylated DNA following UVC irradiation.(a) Plot of dinucleotide counts of putative lesions giving rise to CPD-seq reads for methylated yeast genomic DNA (5mC) irradiated with 90 J/m 2 of UVC light relative to unmethylated DNA that is not UV irradiated (No UV).(b) Plot of CPD-seq reads associated with each tetranucleotide sequence context (centered on a dipyrimidine) for UVCirradiated methylated DNA (5mC) relative to UVC-irradiated unmethylated (No Methyl) DNA.Red dots represent tetranucleotide sequences containing a NYCG sequence, which are targets for methylation by M.SssI methyltransferase.(c), (d) Normalized ratio of CPD-seq reads in UVCirradiated methylated DNA (5mC) relative to UVC-irradiated unmethylated DNA.Normalization was performed using the number of TT CPD-seq reads in each CPD-seq library, since these should not be affected by methylation.Color of the bar indicates the flanking 3 0 base.
Magnitude of CPD induction due to methylation depends on efficiency of CPD formation in unmethylated DNA.(a), (b) Plot of methylationdependent CPD induction (i.e., normalized ratio of CPDs in 5mC DNA relative to No methyl control) relative to frequency of CPD-seq reads in unmethylated DNA for NYCG sequence contexts in (a) UVB-irradiated (both replicates combined) or (b) UVC-irradiated yeast genomic DNA.R 2 value calculated by linear regression analysis: **p < .001;*p < .05.
to measure UV-induced CPD formation across the yeast genome in the presence or absence of in vitro cytosine methylation by the CpG methyltransferase M.SssI.Our results indicate that 5mC promotes UVB-induced CPD formation at methylated sites on average nearly two-fold, but the magnitude of induction strongly depends on the flanking DNA sequence context.Cytosine methylation also weakly modulates UVC-induced CPD levels, promoting CPD formation in some sequence contexts (e.g., GTCG and GCCG), but suppressing it in others (i.e., TTCG).Our analysis indicates that the magnitude of the effect of 5mC on CPD formation in different sequence contexts is dependent upon the baseline level of CPD formation in the absence of methylation.These findings elucidate the impact of sequence context and cytosine methylation on CPD formation across a model eukaryotic genome, and can potentially reconcile conflicting findings from previous studies.
occur in sequence contexts in which methylation should strongly promote UVB-induced CPD formation (e.g., TCCG [R196, R248], GCCG [G245], and ACCG [R248, R282]; underline indicates mutated base, mutated p53 residues are indicated in brackets).In contrast, a recent report suggesting that 5mC does not promote UVB-induced CPD formation in vitro (but instead weakly suppresses it; Leung & Murray, 2021) analyzed CPD formation at TTCG sequence contexts, which our data indicate show the weakest CPD induction ($1.1-fold) of any sequence context.The fact that the previous report saw a weak, 5mC-dependent CPD suppression ($0.9-fold) may be due to the extended sequence context (i.e., TTTCG[A/G]) used in this study or the fact that their linear amplification method detects both CPDs and 6-4PPs (Leung & Murray, 2021).
, our data indicate that the magnitude of CPD induction (e.g., GCCG or GTCG) showing the greatest degree of induction.This finding suggests that either 5mC has an additive instead of multiplicative effect on CPD formation, so that weak CPD-forming sequences have the greatest relative increase in 5mC-dependent CPD levels, or that 5mC is unable to efficiently promote CPD formation in sequence contexts (e.g., TTCG or CTCG) that already have a strong intrinsic propensity to form CPDs.