Introduction

Aging leads to an increased risk in the development of several diseases, including cancer. Genetic instability is a common hallmark in cancer and aging, although the mechanisms linking the two remain unclear. Studies have found that repetitive DNA elements occur frequently throughout the genome and can adopt alternative DNA structures that are inherently mutagenic and do not conform to the canonical double-helical B-DNA structure originally described by Watson and Crick1,2,3,4,5. More than ten types of repeat-mediated DNA structures have been characterized to date. The intramolecular triplex H-DNA forms in regions of polypurine-polypyrimidine mirror-repeat symmetry, where one strand folds back and Hoogsteen-hydrogen bonds with the purine-rich strand of the underlying duplex in the major groove6,7. Our lab has previously demonstrated that H-DNA is mutagenic both in vitro and in vivo2,4,8. In addition, bioinformatics analysis has revealed that H-DNA-forming sequences are substantially enriched at translocation breakpoint hotspots in human cancer genomes9.

Our lab and others have demonstrated that repeat-mediated DNA structures are mutagenic via both replication-dependent and replication-independent mechanisms and as such, repeat-mediated mutations are seen in both rapidly dividing cells (e.g., spleen) as well as cells that replicate at low levels (e.g., brain)10,11. During replication, repeat-mediated DNA structures can cause stalling of replication forks leading to the recruitment of DNA replication and repair proteins to resolve the impediment to replication. However, their processing can be error-generating, leading to genetic instability11. We have also found that nucleotide excision repair (NER) proteins can recognize H-DNA in a replication-independent fashion by the distortion it causes in the DNA double helix, leading to error-generating processing4. Relevant to this finding, deficiencies in NER have been shown to result in accelerated aging disorders12,13,14,15. Additionally, studies have reported decreasing expression of NER proteins with age, suggesting a strong relationship between NER and the aging process16.

Although genomic instability is a hallmark common to both aging and cancer17,18, the relationship between aging and DNA structure-induced genetic instability at endogenous mutation hotspots in cancer genomes remains unclear. Here, we utilized our established genetically engineered mutation-reporter mouse model to investigate the impact of aging and tissue type on H-DNA-induced genetic instability in vivo8. Our lab has recently utilized this model to evaluate the effects of obesity on H-DNA-induced genetic instability as well as a pilot study assessing the influence of aging on H-DNA mutagenesis in somatic versus gonadal tissues of male mice19,20. As we have found that H-DNA-induced mutagenesis can occur by both replication-dependent and replication-independent mechanisms4, we analyzed H-DNA-induced mutations in genomic DNA from several tissues with varying cell-turnover rates (i.e., spleen, liver, and brain). Interestingly, we found that aging modulated H-DNA-induced mutagenesis in a tissue-specific fashion, increasing with age in spleen and liver tissues and decreasing in brain tissues. More detailed analysis into the types of mutations induced by H-DNA with age revealed alterations in the frequencies of large deletion mutants mirroring the alterations in overall H-DNA-induced mutation frequencies in the corresponding tissues. Further investigation into the mechanisms of H-DNA-induced mutagenesis revealed tissue-specific alterations in NER and DNA cleavage activities with age congruent with changes in H-DNA-induced genetic instability in corresponding tissues.

Results

Investigating age-dependent H-DNA-induced mutagenesis in a transgenic mutation-reporter mouse model

The transgenic mice that we previously constructed allowed us to investigate the impact of age on DNA structure-induced mutagenesis in vivo. Previous studies of H-DNA-induced genetic instability in mammalian cells showed primarily large deletion mutations, prompting the design and use of a truncated lacZ’ mutation-reporter for the enhanced sensitivity of H-DNA causing mutations in vivo. The transgenic reporter mice contain a chromosomally-integrated mutation-reporter shuttle vector with either a B-DNA or H-DNA-forming sequence from the human c-MYC gene (Fig. 1A, Fig. S1)8,20. In this study, the mutation reporter was recovered from genomic DNA of tissues collected from 2-month and 18-month mice and analyzed via blue-white screening (Fig. 1B). The mutation frequencies were generated by dividing the number of white colonies (mutated mutation-reporter) by the total number of colonies (wild-type blue + mutant white), with at least 15,000 colonies counted per age tissue group. We selected tissues with varying proliferation rates to provide insight into potential replication-dependent and replication-independent mechanisms of H-DNA-induced genetic instability that may be altered by age.

Fig. 1
figure 1

. Schematic of experimental design. (A) Mouse model and (B) magnetic bead rescue of integrated mutation-reporter vectors from mouse genomic DNA. The mutation reporters were recovered from genomic DNA via LacI/LacZ-conjugated magnetic beads and electroporated into DH10β cells for blue-white mutation screening, with a minimum of 15,000 colonies counted per mouse tissue sample. Figure adapted from D’Amico, et al. Biomolecules. 202419.

Full size image

Aging alters H-DNA-induced genetic instability in a tissue-specific manner

We first determined the impact of age on H-DNA-induced mutation frequencies in genomic DNA from the spleen, which has the greatest cell-turnover rate of the somatic tissues tested in this study. Analysis of spleen tissues from the H-DNA mice demonstrated a significant increase in mutation frequencies in the 18-month age group compared to the 2-month age group (Fig. 2A; p < 0.01). Further, aging demonstrated greater mutagenic effects in H-DNA mice compared to B-DNA mice in spleen tissues (p < 0.0001). These effects were not isolated to spleen: similar results were observed in liver, which has a modest cell-turnover rate and remarkable regenerative capacity in the presence of cell damage, although not as proliferative as the spleen21. Statistical analysis of the liver tissues showed significant increases in H-DNA-induced mutations in 18-month liver tissues compared to 2-month liver tissues (Fig. 2B; p < 0.05), and an increased mutagenic effect of H-DNA compared to B-DNA in the 18-month age group compared to the 2-month age group.

Fig. 2
figure 2

Aging alters H-DNA-induced genetic instability in a tissue-specific manner. Mutation frequencies in young (2-month) versus aged (18-month) female mice from (A) spleen, (A) liver, and (A) brain per 10,000 colonies. Frequencies were calculated from genome-integrated mutation reporters containing either a B-DNA sequence or H-DNA-forming sequence from the human c-MYC gene. Data points are presented on violin plots with the median and quartiles, n > 3 mice, statistical significance determined using one-way ANOVA followed by Sidak’s multiple comparisons *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Full size image

Of the tissues analyzed in this study, the brain has the lowest cell-turnover rate and thus may provide insight into replication-independent mechanisms of DNA structure-induced mutagenesis that may be impacted by age. In contrast to the spleen and liver tissues where aging resulted in increased H-DNA-induced mutations, there was a significant decrease in H-DNA-induced mutation frequencies in the 18-month age group compared to the 2-month age group (Fig. 2C, p < 0.05). Additionally, we observed a significant mutagenic effect of H-DNA compared to B-DNA in both age groups in brain tissues (p < 0.0001), while the effect was only significant at 18-months in liver and spleen tissues (Fig. 2A-B). Together, this data suggests tissue-specific effects of aging on H-DNA-induced mutagenesis.

Interestingly, genomic DNA from the B-DNA mice demonstrated no significant change in mutation frequencies with age, in contrast to our expected results based on previous studies demonstrating increasing mutation frequencies in B-DNA-forming sequences with age22,23,24. Targeted next generation sequencing (NGS) of genomic DNA from 2-month and 18-month B-DNA mouse spleen tissues further demonstrated minimal differences in mutation rates with increasing age in this short genomic region of interest, confirming our results in the context of this model (Fig. S2). Note that the model in this study is designed for detecting H-DNA-induced mutations in mouse genomes, which occur at a higher frequency than spontaneous genomic mutation frequencies previously observed in the literature. As large deletion mutations were the major H-DNA-induced mutation in mammalian cell culture, we designed this model to better prioritize the capture of large deletion mutation events. As a result, both the spontaneous point mutations in B-DNA and H-DNA mice that are unrelated to H-DNA as well as H-DNA-induced point mutations may be underestimated. Within the limited number of mutants analyzed, the spontaneous mutations in H-DNA mice may be overshadowed by the overall higher H-DNA-induced mutation frequencies in the mouse genome. Regardless, we observed significant differences in H-DNA-induced mutagenesis compared to B-DNA amongst tissues and age groups, which was the primary focus of this study and what our transgenic reporter mice were designed to detect (Fig. 2A–C).

H-DNA-induced large deletion mutations correlate with overall mutation frequencies

To gain insight into the mechanisms underlying the alterations in mutation frequencies with age in the H-DNA and control B-DNA mice, we subjected mutant white colonies obtained from the blue–white screening assay to DNA sequencing analysis. The mutation spectra were categorized into large deletions (≥ 30-bp deletion) or point mutations, which included base substitutions, insertions, and small deletions, as the majority of these mutations were comprised of a single base-pair.

We observed a significant increase in the frequency of large deletions with age in the spleen genomic DNA of H-DNA mice (Fig. 3A; p < 0.001), consistent with the increase in H-DNA-induced mutation frequencies observed in Fig. 2A. Additionally, there was no significant difference in the frequency of large deletions in B-DNA spleen tissues, similar to the overall mutation frequencies observed in Fig. 2A.

Fig. 3
figure 3

Age-associated alterations in the H-DNA-induced mutation spectra in vivo. Frequencies of large deletion mutations in (A) spleen, (B) liver, and (C) brain and frequencies of point mutations in (D) spleen, (E) liver, and (F) brain of young (2-month) versus aged (18-month) female mice per 10,000 colonies. A random sample of mutants (n > 20) from the results presented in Fig. 2 were sequenced with primers upstream of the lacZ mutation-reporter gene and analyzed via Nucleotide BLAST. Data points are presented on violin plots with the median and quartiles, n > 3 mice, statistical significance determined using one-way ANOVA followed by Sidak’s multiple comparisons *p < 0.05, **p < 0.01, ***p < 0.001.

Full size image

This trend persisted in the livers of the H-DNA mice; however, the increase was not statistically significant as it was in the spleen tissues (Fig. 3B; p > 0.05). In contrast to the spleen and liver, the frequencies of large deletions in brain samples decreased by 2.6-fold with age in H-DNA mice (Fig. 3C; p > 0.05). However, similar to the other tissues, this trend correlated with the change in overall mutation frequencies with age observed in Fig. 2B–C. There were no significant changes in the frequencies of large deletions in liver or brain tissues from B-DNA mice, mirroring the overall B-DNA mutation frequencies (Fig. 3B–C).

Impact of age on H-DNA mutation spectra profile is tissue-specific

Previously, our lab reported that large deletions were the most common type of mutation generated from H-DNA-induced genetic instability in mammalian cells2,4. In this study, however, we observed age- and tissue-specific changes in the H-DNA mutation profile, with the majority mutation changing from point mutations to large deletions with age in spleen tissues and large deletions to point mutations with age in brain tissues (Fig. 3A–F, Fig. S3). There were no significant differences in point mutation frequencies amongst the B-DNA mice with age (Fig. 3D–F, Tables S1–3), partially due to the short mutation reporter not being designed to study the global spontaneous point mutations as we discussed above. Despite this system preference, the frequencies of point mutations in tissues from H-DNA mice are greater than the spontaneous point mutations in B-DNA mice in multiple tissue age groups, suggesting that H-DNA also causes significant amounts of point mutations in the genome. Interestingly, the H-DNA-induced point mutations (Fig. 3D–F) had an inverse relationship with the overall change in mutation frequencies (see Fig. 2A–C) in the spleen and brain and remained unchanged in the liver.

Statistical analysis of the point mutation frequencies in H-DNA spleen tissues demonstrated a significant decrease in aged mice compared to young mice (Fig. 3D; p < 0.01, Table S1). The change in point mutation frequencies was opposite to that seen in the overall H-DNA-induced mutation frequencies in Fig. 2A. Interestingly, this decrease in point mutations and increase in large deletions demonstrates a shift in the overall mutation spectra in the H-DNA mice, wherein the majority mutation type changes from point mutations to large deletions with increasing age (Fig. 3A, D, Fig. S3). Although there was no significant change detected with age in H-DNA-induced point mutations in the liver samples, they did remain the majority mutation type in both the 2-month and 18-month age groups (Fig. 3E, Table S2, Fig. S3). This is in contrast to the spleen, where the mutation type dominance shifted to large deletions with age. Lastly, there was not a statistically significant change in H-DNA-induced point mutations with age in the brain (Fig. 3F, Table S3). However, the modest increase in point mutations coupled with the decrease in large deletions resulted in a shift of the majority mutation from large deletions at 2-months of age to point mutations at 18-months of age (Fig. 3C, F, Fig. S3). Although not statistically significant, this modest trend increase in H-DNA-induced point mutations with age was opposite to the overall H-DNA-induced mutation frequency observed in the brain (see Fig. 2C). These results suggested tissue- and age-specific modulations in the H-DNA-induced mutation spectra rather than a dominance of one mutation type across organs.

Cleavage of H-DNA substrates in vitro

Previously, we have shown that H-DNA is processed by DNA repair proteins in vitro4. Here, we set out to determine how aging impacts the cleavage of H-DNA in whole-cell extracts (WCE) generated from the tissues of young and aged mice. Due to excessive nuclease activity in liver WCE preparations, experiments were performed in WCE preparations from spleen and brain tissues to determine cleavage activity on the H-DNA substrate. To reduce the noise created on the plasmid backbone by WCE, we used a self-assembling H-DNA substrate, R25’, containing a duplex structure on the 5’-end and a 3’-purine rich region that folds and binds into the duplex to form a triplex region, similar to the triplex structure in the H-DNA conformation. We previously validated that the R25’ oligo forms an H-DNA structure in vitro, proving useful for molecular and biochemical assays25. After incubating the R25’ substrate in the WCE, we detected a unique H-DNA cleavage product of ~ 31 nucleotides present in all age groups across both tissues tested (Fig. 4A–D). This cleavage product suggested that processing of the H-DNA substrate in mouse tissue extracts occurred on the loop between the 3rd triplex strand and the duplex; interestingly this cut site was nearer to the loop toward the duplex than the cut site that we observed in human cell extracts4.

Fig. 4
figure 4

(adapted from Del Mundo et al., 2017). The estimated site of cleavage is marked by blue bracket. (B) H-DNA cleavage from young and aged brain WCE. (C) Quantification of cleavage products from brain WCE. (D) H-DNA Cleavage from young and aged spleen WCE. (E) Quantification of cleavage products from spleen WCE. Cleavage products marked by red arrows. M, DNA ladder. NTC, non-extract treated control. Error bars represent mean ± SEM, n = 3 mice, statistical significance determined using student’s t-test

Cleavage of H-DNA conformation with age. (A) R2-5’ radiolabeled oligo forming an H-DNA conformation.

Full size image

There were no changes in either the abundance of the H-DNA cleavage products or cleavage patterns with age in the brain WCE relative to the unprocessed full-length substrate (Fig. 4B–C). This suggests that the processing of an H-DNA structure, at least in WCE, remains unchanged with age in brain tissues. However, there was an ~ 2-fold increase in the cleavage of H-DNA in aged WCE from spleen compared to young WCE (Fig. 4D–E; p > 0.05) relative to the unprocessed full-length substrate, consistent with the significant increase in H-DNA-induced large deletion mutations in the spleen of 18-month-old mice (Fig. 3A). These data indicate tissue-specific alterations in protein activity on an H-DNA structure, even in the absence of additional endogenous factors that could affect H-DNA processing.

UV lesion removal decreased with age

We have previously reported that H-DNA is processed by NER in an error-prone fashion, resulting in DNA double-strand breaks and large deletion mutations. To determine possible changes in NER efficiency with age in our model, we used tissue-derived WCE from young and aged mice. UV-irradiated plasmid DNA, a canonical NER substrate26, was subjected to young and aged WCE and recovered over a time course (from 0 to 60 min) to allow for repair. The remaining UV lesions were detected via immunoslot blot analysis and served as an indicator of UV lesion removal as a function of NER activity (Fig. 5A).

Fig. 5
figure 5

NER is delayed in aging tissues. (A) Experimental schematic of NER UV lesion removal assay. (B) UV lesions detected over a time-course from young and aged brain WCE. (C) 6 − 4 Photoproduct immunoslot blot (top) and SYBR Gold-stained DNA blot (bottom) from NER assay in brain WCE. (D) UV lesions detected over a time-course from young and aged spleen WCE. (E) 6 − 4 Photoproduct immunoslot blot (top) and SYBR Gold-stained DNA blot (bottom) from NER assay in spleen WCE. Error bars represent mean ± SEM, n = 3 mice, statistical significance determined using student’s t-test. *p < 0.05.

Full size image

WCE prepared from aged brain tissues demonstrated a significant delay of lesion removal compared to WCE from young mice (p < 0.05; Fig. 5B). After 30 min, ~ 49% of UV lesions had been removed by young brain WCE, while only ~ 2% were removed in aged WCE (p = 0.02). By 60 min, aged WCE removed ~ 26% of the UV lesions compared to the ~ 60% in young WCE (p = 0.03; Fig. 5C). This delay in UV lesion removal in the brain is consistent with a decrease in H-DNA-induced mutations observed in this tissue (see Fig. 2C).

Aged spleen WCE displayed a similar delay in lesion removal compared to young samples, however this delay was not statistically significant (Fig. 5D). After 30 min, ~ 54% of the UV lesions had been removed by young spleen WCE, while only ~ 5% were removed in aged WCE (p = 0.06). By 60 min, young spleen WCE had removed ~ 64% of UV lesions. In comparison, aged spleen WCE removed ~ 34%, about half as much (p = 0.09; Fig. 5E). This suggests that the reduction in UV lesion removal, an upstream process in the NER pathway, is a common feature amongst aging tissues and supports previous findings that NER activity is reduced with age27,28,29,30. These data suggest that other mechanisms in addition to declining NER activity likely also contribute to the tissue-specific alterations in H-DNA-induced mutations with age.

Discussion

We have found that H-DNA-forming sequences are mutagenic and co-localize at translocation breakpoints in human cancer genomes, suggesting a role for H-DNA in cancer etiology9. In this study, we provide insight, for the first time, into the impact of aging on H-DNA-induced genetic instability across a variety of tissues in mice. In both young and aged mice, the mutagenic effect of H-DNA was greater than B-DNA across all tissues tested. Overall, the effects of age on H-DNA-induced genetic instability were tissue-specific, with mutation frequencies increasing with age in the spleen and liver, and surprisingly decreasing in the brain.

Furthermore, increases or decreases in the frequencies of large deletions in the H-DNA mice mirrored increases or decreases, respectively, in overall mutation frequencies. Interestingly, the primary mutation type of H-DNA-induced mutations appeared to shift with age in the spleen and brain. This is in direct contrast to our previous in vitro studies in mammalian cells that reported large deletions as the primary mutation caused by H-DNA-forming sequences2. This may be due to confounding factors present in vivo that may not be present in cell culture. Although aging demonstrated modest effects on the frequency of point mutations in the spleen and brain, the differences in large deletion mutations with age were more pronounced. This indicates that the changes in H-DNA-induced mutation frequencies with age may largely be attributed to large deletion mutations, caused by changes in DNA double-strand break formation or the subsequent error-free repair. Together, these results suggest that aging has tissue-specific effects on either the mutagenic processing of H-DNA and/or the amount of H-DNA formed in vivo.

Chromosomal translocations and genetic instability are considered genetic hallmarks of cancers such as leukemias and lymphomas31,32,33. Studies have reported an increase in the frequency of the BCL-2 t(14;18) translocations with age, a mutation common in follicular lymphoma34,35. Additionally, there is evidence to suggest that an H-DNA/triplex DNA structure contributes to the BCL-2 translocation event36,37. Our finding that H-DNA-induced large deletions increase with age in spleen provides a potential mechanism for the age-associated increase in BCL-2 translocation frequency. This could also implicate H-DNA/triplexes as a causative factor in additional translocation events that are thought to increase in frequency with age38.

We have previously reported that alternative DNA-forming sequences are processed by multiple DNA replication and repair pathways4,39,40. Independent of replication, H-DNA can be recognized and processed by the NER pathway in a mutagenic fashion, resulting in DNA double-strand breaks and large deletion mutations4. End-joining repair pathways have also been implicated in the repair of double-strand breaks arising from H-DNA-induced genetic instability2. Our observation of increasing or decreasing mutation frequencies in different tissues with age in the H-DNA mice may suggest tissue-specific alterations in one or more of these mechanisms that process H-DNA and H-DNA-induced damage.

DNA repair protein levels have been reported to decrease with age in several organisms, although the literature is inconsistent across models and tissue samples tested27. However, numerous studies have corroborated the finding that the efficiency of several DNA repair pathways decreases with age28,29,41,42,43. Concurrent with previous reports of a decline in error-free repair of DNA double-stranded breaks, we observed an increase in the frequency of large deletion mutations in spleen and liver tissues with age in H-DNA mice compared to tissues from the B-DNA mice. We also observed an increase in the cleavage of H-DNA in spleen tissues from aged mice compared to younger mice, congruent with increasing large deletion mutations. Furthermore, the transition from B-DNA to alternative DNA structures are induced by the energy generated during negative supercoiling that occurs during DNA metabolic processes, including replication, transcription, and repair44,45,46. As a consequence, tissues with higher proliferation rates may have more H-DNA structures forming compared to tissues with lower proliferation rates. Consistent with this, we observed increases in H-DNA-induced genetic instability in spleen and liver tissues, both of which are proliferative tissues.

In contrast, the decline in H-DNA-induced genetic instability in the brain may be related to decreases in DNA replication and/or NER activity, where proteins that either process or recognize H-DNA as “damage” may have impaired function28,29,30. Here, we observed a decrease in the removal of a canonical NER lesion with age in the brain, suggesting decreased NER activity. This correlated with the decrease in H-DNA-induced mutation and large deletion frequencies with age, suggesting that NER may contribute to changes in H-DNA-induced genetic instability in mice with age in brain tissues. Additionally, cells with lower metabolic activity or proliferation rates, such as those in aged brain tissue samples, could have less H-DNA formation due to decreases in overall DNA metabolism47,48. The decline in H-DNA-induced genetic instability in brain tissues is congruent with its low proliferative potential and declining metabolic activity. Further, we have observed increased levels of oxidative DNA damage in genomic DNA from aged brain mouse tissues, and that increasing oxidative stress destabilizes H-DNA structure formation49. This may in part explain the decreases in both H-DNA-induced mutation frequency and the frequency of large deletions with age in brain tissues outside of declining NER efficiency.

It is important to recognize that the effect(s) of aging on H-DNA-induced genetic instability is the result of several contributing factors, including H-DNA structure formation, recognition and processing, and the processing and repair of H-DNA-induced DNA strand breaks. Due to the varying roles and functions of organs, aging may impact one or all of these mechanisms differently. As a result, we observed significant but diverse effects of aging on H-DNA-indued mutations in mouse tissues. Further studies are warranted to better understand the complex relationship between aging and the tissue-specific alterations in mechanisms involved in DNA structure-induced genetic instability.

Understanding the link between aging and genetic instability is crucial for the development of novel therapeutic strategies for the prevention and/or treatment of cancer. Here, we provide insights into a possible mechanism of age-induced genetic instability. The occurrence of H-DNA-forming sequences at DNA breakpoints in human cancers and the increase of H-DNA-induced mutagenesis with age in multiple tissues underscores its plausible role in age-associated cancer etiology. Additionally, that H-DNA-induced mutagenesis differs with age in tissues of varying proliferative and metabolic rates may be of interest in the context of tissue-specific diseases that arise with age.

Materials and methods

Mice

Transgenic mice containing a recoverable p2RT mutation-reporter vector were rederived from cryogenically stored mouse embryos that we have previously constructed8. Female FVB/N mice containing a B-DNA or H-DNA mutation reporter were aged to 2- and 18-months and euthanized by carbon dioxide inhalation followed by cervical dislocation in accordance with IACUC guidelines. Spleen, liver, and brain tissues were harvested and washed with PBS, snap-frozen in liquid nitrogen, and stored at − 80 °C until ready for processing.

Genomic DNA extraction from mouse tissues

Approximately 50–100 mg of frozen tissue was finely cut with a sterile blade and incubated in DNA extraction buffer (0.5 mM EDTA, 20 mM Tris–HCl pH 8.0, 400 mM NaCl, 1% SDS, 1 mg/mL proteinase K) overnight at 50 °C. Nucleic acids were isolated by phenol:chloroform:isoamyl alcohol extraction and treated with 50 μg RNAseA for 3 h at 37 °C. Genomic DNA was isolated and purified by phenol:chloroform:isoamyl alcohol extraction and ethanol precipitation.

Magnetic bead p2RT plasmid rescue from genomic DNA

Approximately 50–70 μg of genomic DNA extracted from mouse tissues (n ≥ 3 mice per tissue age group) underwent two rounds of restriction digestion with Spe1-HF restriction enzyme (NEB) at 37 °C for 4–5 h. The DNA was purified via phenol:chloroform:isoamyl alcohol extraction and ethanol precipitation. The DNA was resuspended in 56 μL of nuclease-free water followed by 14 μL of 5 × binding buffer (50 mM Tri-HCl, 5 mM EDTA, 50 mM MgCl2, 25% glycerol, pH adjusted to 6.8). The DNA sample was added to an equal volume of LacI/LacZ-conjugated magnetic beads and incubated at 37 °C for 1 h while rotating. The LacI protein on the beads bind specifically to the LacO DNA sequence on the mutation-reporter fragment released from the restriction digested genomic DNA. The beads were washed with 1 × binding buffer using a magnetic stand to remove unbound genomic DNA. The mutation reporter was eluted from the beads in DNA Elution Buffer (0.5% IPTG, NEBuffer2, 3.75 mM Tris–HCl pH 7.5, 0.37 mM EDTA, 46 mM NaCl) at 37 °C for 30 min followed by incubation at 65 °C for 20 min while rotating. The recovered mutation reporter was re-circularized by 400 units of NEB T4 DNA Ligase plus 0.05 mM ATP for 1 h at room temperature, followed by ethanol precipitation.

Blue-white screening analysis

The recovered p2RT mutation reporter was electroporated into the cytosine methylation restriction-deficient and alpha-complementary compatible DH10β cells and grown on LB agar plates containing X-Gal, IPTG, and carbenicillin (XIC) at 37 °C for 18–20 h for blue-white screening analysis. A blue colony indicated a functional lacZ reporter and a white colony indicated a mutated lacZ reporter. The mutation frequency was determined by dividing the number of mutant (white) colonies by the total number of colonies (blue + white). A minimum of 15,000 colonies were counted for each mouse tissue sample. White colonies were streaked onto additional XIC plates to confirm mutants.

Characterization of mutants

To determine the mutation spectra, randomly selected white mutant colonies from each mouse tissue sample were grown in LB broth plus ampicillin overnight at 37 °C. The plasmids were extracted using the Qiagen Miniprep kit and subjected to Sanger sequencing via our on-site DNA Sequencing Core facility. Identification of mutations was determined by NCBI BLAST alignment comparison with the wild-type sequence. Deletions were separated into large deletions (≥ 30 base pairs deleted) and small deletions (< 30 base pairs). For mutation spectra determination, one mutation per mutant was counted toward the total mutants for that group. If multiple point mutations were present in the mutant sequence, then the point mutation closest to the multiple cloning site (i.e., H-DNA or control B-DNA sequence) was determined to be the primary contributing factor to the mutated lacZ reporter. Mutation spectra were normalized to the total mutation frequency to identify the relative amount of mutations attributed per mutation type.

Next generation sequencing

For targeted next generation sequencing analysis (NGS) of the chromosomally integrated mutation-reporter gene, purified genomic DNA from spleen tissues of 2- and 18-month B-DNA mice was amplified using KAPA HF polymerase kit (Roche) and primers NGS_Forward: 5’-CGGCATCAGAGCAGATTGTA-3’ and NGS_Reverse 5’-AGTGAGCGCAACGCAATTA-3’ under the following PCR parameters: 98 °C 30 s, [98 °C 20 s, 60 °C 30 s, 72 °C 30 s, repeat 14x], 72 °C 1 min, and 4 °C incubation. PCR products were cleaned up and size selected using SPRIselect Beads (Beckman Coulter) following manufacturer’s protocol and concentrated via ethanol precipitation. Purified amplicons were subjected to MiSeq PE300 NGS via our onsite genomics facility. After assessing quality with FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/), reads were aligned to the reference sequences using BBMap, an aligner with particularly good performance for detection of deletions, using default settings50. Read counts by reference sequence position were quantified using bam-readcount51. Further processing of the resulting data was done using custom Python and R scripts, with visualizations generated using ggplot252.

Preparation of whole-cell extract for in vitro assays

Approximately 100 mg of tissue was crushed under liquid nitrogen and treated with mild extraction buffer [50 mM Tris–HCl pH 8.0, 150 mM NaCl, 1 mM EDTA, 0.1 mM DTT, 1% NP-40, supplemented with protease and phosphatase inhibitor cocktail (Roche)] for 1–2 h at 4 °C. Cell debris was removed by centrifugation at 13,000 rpm for 10 min at 4 °C. Protein concentrations were determined using Pierce BCA Protein Assay Kit (ThermoFisher) and whole-cell extracts (WCE) flash frozen and stored at − 80 °C.

DNA cleavage assay

DNA substrates were prepared and annealed for formation of H-DNA self-foldback structure or control duplex as previously described4. Approximately 6 × 10−9 M DNA substrate was treated with 50 μg of tissue-derived brain or spleen WCE (n = 3 mice per tissue age group) in reaction buffer (40 mM Tris–HCl pH 7.5, 10 μM MgCl2, 5 mM DTT, 200 μg/mL bovine serum albumin) for 10 min at 30 °C. Reactions were stopped with 20 mM EDTA. Proteins were removed with 40 μg proteinase K and 0.5% SDS followed by phenol:chloroform:isoamyl alcohol extraction and ethanol precipitation. Cleavage products were separated on a 12% denaturing Urea-PAGE gel and detected by Typhoon Phosphor Imaging system. Cleavage products were detected and compared to the relative amount of full-length DNA substrates using the ImageQuant analysis software. The ratio of cleaved:full length oligo was compared between age groups.

UV lesion removal assay

Purified supercoiled plasmid DNA (~ 7 kb) was irradiated with 5 J/m2 ultraviolet light (UV, 254 nm) and treated with 150–200 μg tissue-derived brain or spleen WCE (n = 3 mice per tissue age group) in DNA repair buffer (40 mM HEPES, 5 mM MgCl2, 0.5 mM DTT, 2 mM ATP, 0.1 mg/mL creatine phosphokinase, 100 mM phosphocreatine, 0.04 mg/mL bovine serum albumin) at 30 °C for up to 1 h. Reactions were stopped at 0, 30, and 60 min with 20 mM EDTA. Proteins were removed with 40 μg proteinase K and 0.5% SDS followed by phenol:chloroform:isoamyl alcohol extraction. DNA was purified using the Qiagen PCR purification kit and blotted onto nitrocellulose membranes using the BioRad slot blot apparatus per manufacturer’s protocol. Membranes were air dried for 20 min then baked at 80 °C for 2 h to fix DNA. The membranes were rehydrated in TBS-T (Tris-buffered saline with 0.1% tween) and probed with anti-6–4 Photoproduct antibody from Cosmo-bio (1:5000) overnight at 4 °C, followed by secondary anti-mouse IgG (1:5000) at room temperature for 1 h. Blots were developed with chemiluminescent reagents then stained with 1:10,000 SYBR gold for total DNA to serve as a loading control. Densitometric measurements of 6–4 Photoproduct and total DNA bands were calculated via NIH ImageJ software. 6–4 Photoproduct signals were normalized to total DNA signals and compared between age groups.

Statistical analyses

Statistical analyses were performed using GraphPad Prism v9 statistical software. Plotted values included mean ± S.E.M. Any biological outliers were detected and removed using the interquartile method. For all assays, each group (e.g., 2-month H-DNA spleen, etc.) had a minimum of n ≥ 3 mice as biological replicates for scientific rigor and reproducibility, even after removal of outliers. One-way ANOVA analysis followed by Sidak’s multiple comparisons test was used to detect significant differences between mutation frequencies and mutation spectra with age. Cleavage assays and UV lesion removal assays were analyzed using unpaired student’s t-test.