High cancer-associated mutational burden in normal blood of xeroderma pigmentosum group C, but not groups A, D, or F

Key Points

• •Patients with XP-C with isolated global genome NER defects show higher somatic mutation rates than patients with other XP groups that lack both NER pathways.
• •A cancer-like mutational burden and signature are already present in normal XP-C PBMCs.

Visual Abstract

Abstract

Xeroderma pigmentosum (XP) is a rare autosomal recessive congenital syndrome characterized by defective nucleotide excision repair (NER), leading to extreme photosensitivity and a strong predisposition to skin cancer. One of the 8 complementation groups, XP complementation group C (XP-C), is defective specifically in the global genome component of NER and presents not only with increased skin cancer but also with hematologic cancers. Using error-corrected single-molecule sequencing, we show a uniquely high spontaneous somatic mutational load in peripheral blood mononuclear cells (PBMCs) of patients with XP-C but not in those of patients from other XP complementation groups (XP-A, XP-D, and XP-F). The hypermutability observed in XP-C was markedly lower in fibroblasts than in PBMCs. The XP-C mutational profile was characterized by elevated single-nucleotide variants (SNVs) associated with mutational signatures SBS5, SBS8, and SBS32, as well as an enrichment of single-nucleotide cytosine deletions, with SNV profiles closely mirroring those found in XP-C leukemias. These findings indicate that a cancer-like mutation burden is already present in normal lymphocytes before malignant transformation, revealing distinct molecular subtypes within XP defined by spontaneous mutational load in normal blood cells.

Introduction

Xeroderma pigmentosum (XP) is a rare autosomal recessive disorder characterized by extreme sensitivity to UV light caused by germ line mutations in 7 DNA repair genes (XPA to XPG) or in the bypass polymerase POLH/XPV. These mutations affect nucleotide excision repair (NER), which consists of 2 subpathways: global genome NER (GG-NER) and transcription-coupled NER (TC-NER).¹ In GG-NER, the XPC and XPE proteins scan the genome for helical distortions and then recruit the repair machinery required for both subpathways, including XPA, XPB, XPD, XPF, and XPG. Consequently, mutations in XPC or XPE specifically impair GG-NER, whereas mutations in the remaining XP genes disrupt both GG-NER and TC-NER. This defect leads to an inability to remove UV-induced DNA adducts and results in a ∼10 000-fold increased incidence of skin cancer in patients with XP.² XP complementation group C (XP-C) is distinct because patients with XP-C also have a ∼1000-fold increased risk of hematologic cancers.³

To investigate the molecular basis for these divergent clinical outcomes, we measured somatic mutation frequency in normal peripheral blood mononuclear cells (PBMCs) and primary skin fibroblasts from 15 patients with XP (aged 5-26 years) across 4 XP complementation groups, 3 unrelated unaffected individuals (aged 40-73 years), and 11 unaffected first-degree relatives, 9 of whom were heterozygous carriers (aged 19-56 years; Table 1). Because somatic mutations are rare and occur stochastically across individual cells, they cannot be accurately quantified using conventional bulk sequencing.⁴ To overcome this limitation, we applied our recently developed single-molecule sequencing (SMM-seq) assay, based on error-corrected next-generation sequencing, enabling highly sensitive, and quantitative detection of low-abundance mutations in normal cells and tissues.⁵

Table 1.

Distribution of analyzed samples

Gene	Participant	No. of participants	Age range, y	Tissue
XPC	Patient	6	8-24	PBMC
	Parent/Sibling	6	19-56
XPA	Patient	4	15-23
	Parent	2	44-45
XPD	Patient	3	5-26
	Parent	2	36-39
XPF	Patient	1	19
	Parent	1	39
XPC	Patient	4	9-20	Fibroblast
XPA	Patient	2	14-15
XPD	Patient	2	5-12
XPF	Patient	1	19
Unaffected	Control	3	40-73

Using this approach, we reveal that GG-NER deficiency in patients with XP-C leads to a pronounced accumulation of somatic mutations in normal PBMCs early in life. Notably, the mutational patterns observed in PBMCs closely mirror those found in XP-C leukemias, providing direct molecular evidence that a cancer-like mutational burden can arise in normal somatic cells before malignant transformation. Our results establish a mechanistic link between GG-NER deficiency, somatic mutagenesis in normal cells, and cancer predisposition, offering a new conceptual framework for understanding how inherited genome instability syndromes drive cancer risk through the gradual accumulation of somatic mutations.

Methods

Patient recruitment and sample collection

Patients diagnosed with XP, their unaffected family members, and unrelated unaffected individuals were recruited through the University of Minnesota Medical School. Patients with XP were enrolled based on a confirmed clinical diagnosis, whereas unaffected relatives were identified by the absence of clinical symptoms and, where available, by genetic confirmation (supplemental Table 1). All participants provided written informed consent before enrollment in accordance with the protocol approved by the University of Minnesota Institutional Review Board. Blood samples were collected using standard venipuncture techniques. Samples were processed immediately to separate plasma, serum, and PBMCs, which were stored at −80°C until analysis. Fibroblasts from skin biopsies were cultured using a protocol described previously.⁶ Recruitment strategies included direct physician referrals and outreach through patient advocacy groups to ensure adequate representation of the study population.

DNA extraction, library preparation, and sequencing

DNA was extracted from frozen PBMCs using the Quick-DNA Miniprep Kit (Zymo Research), quantified with a Qubit fluorometer (Thermo Fisher Scientific), and assessed for quality using the TapeStation system (Agilent Technologies). Sequencing libraries for genotyping were prepared using the NEBNext Ultra II DNA Library Prep Kit for Illumina (New England Biolabs) following the manufacturer’s recommendations. SMM-seq libraries were prepared as previously described.⁵ Briefly, genomic DNA was digested with AluI restriction endonuclease (New England Biolabs), and the resulting fragments were end-repaired and ligated to hairpin-shaped adapters to form dumbbell-like structures suitable for rolling circle amplification. To implement a reduced-representation approach, the rolling circle amplification product was subjected to strict size selection using a PippinHT instrument to isolate the 600 to 800 bp fraction, thereby enriching for a consistent subset of the genome (∼7% of total genomic sequence). The selected fraction was then amplified using SD polymerase (Bioron), and indexed polymerase chain reaction was performed to generate the final SMM-seq libraries. Sequencing was conducted by Novogene (US) on a NovaSeq X Plus platform in 150-bp paired-end mode, yielding on average 55 Gb of data per SMM-seq library and 12 Gb per corresponding non–error-corrected reference library.

Variant calling and data analysis

Sequencing reads were aligned to the human reference genome (hg19), and variant calling was performed on an average of 241 million callable bases per sample as previously described.⁵ Briefly, a base was considered callable if represented by at least 7 reads from each DNA strand of a duplex in the SMM-seq library and covered at ≥20× in the corresponding non–error-corrected reference library. A variant was called only when all strand reads reported the same sequence change and was classified as somatic if it was absent from the corresponding reference library and not listed in dbSNP. Somatic single-nucleotide variant (SNV) frequency was calculated as the ratio of the number of somatic mutations to the number of callable bases in each sample. Somatic variants were recorded in Variant Call Format. Mutational spectra for individual samples were subsequently generated using MutationalPatterns (version 3.16.0).⁷ Mutations were then assigned to COSMIC version 3.2 signatures using SigProfilerAssignment (version 1.0.0).⁸ To compare mutational profiles observed in normal XP-C PBMCs with those previously reported in XP-C leukemia, publicly available data from Yurchenko et al³ were analyzed. Mutations from normal XP-C PBMCs and XP-C leukemia samples were pooled, and cosine similarity was calculated to assess how closely XP-C PBMC profiles resembled those of XP-C leukemia. GraphPad Prism version 10 was used for all statistical analyses; statistical significance was determined using a 2-tailed t test.

Results

Samples and clinical characteristics

A total of 42 samples, including 25 PBMC clinical samples and 17 skin fibroblast cultures, from 15 patients, 11 unaffected first-degree relatives, and 3 unrelated unaffected individuals were sequenced using SMM-seq. PBMCs were collected from 6 patients with XP-C (aged 8-24 years), 4 with XP-A (aged 15-23 years), 3 with XP-D (aged 5-26 years), and 1 heterozygous but clinically affected patient with XP-F (age 19 years). Matched control PBMC samples were collected from 5 parents (aged 40-56 years) and 1 sibling (aged 19 years) of the XP-C cohort, 2 parents (aged 44 and 45 years) of the XP-A cohort, 2 parents (aged 36 and 39 years) of the XP-D cohort, and 1 parent (aged 39 years) of the XP-F cohort. All family members were heterozygous for their respective mutations except for 1 noncarrier parent from the XP-D and XP-F cohorts. Fibroblast cultures included 4 XP-C, 2 XP-A, 2 XP-D, and 1 XP-F line. In addition, 5 fibroblast lines from 3 unrelated unaffected individuals were collected to represent the unaffected healthy cohort (Table 1). Information on hematologic malignancies was available for 9 of the 15 patients with XP, and none of these 9 patients had evidence of hematologic malignancies at the time of sample collection (supplemental Table 2).

XP-C PBMCs have uniquely high somatic mutation burden compared with XP-A, XP-D, and XP-F PBMCs

We applied SMM-seq to quantify somatic mutation rates in PBMCs from 4 XP complementation groups (XP-C, XP-A, XP-D, and XP-F) and their respective controls. SMM-seq and non–error-corrected reference libraries were generated using a reduced-representation approach involving AluI digestion and strict size selection, and sequenced on an Illumina NovaSeq platform. The reference libraries were used to exclude germ line variants.

Our analysis revealed a ∼10-fold higher burden of somatic SNVs in PBMCs from patients with XP-C (2.08 ± 0.8 SNV/Mb) than in PBMCs from XP-C heterozygous controls (0.21 ± 0.06 SNV/Mb; Figure 1A). Strikingly, the SNV load did not differ significantly among patients with XP-A, XP-D, or XP-F (0.24 ± 0.1 SNV/Mb, 0.12 ± 0.05 SNV/Mb, and 0.10 SNV/Mb, respectively) and their related controls (0.22 ± 0.02 SNV/Mb, 0.22 ± 0.05 SNV/Mb, and 0.15 SNV/Mb, respectively).

Figure 1.

Somatic mutation burden in XP PBMCs and fibroblasts across different complementation groups. PBMCs from patients with XP-C exhibit significantly higher spontaneous somatic SNV (A) and INDEL (B) loads than healthy controls and other XP complementation groups. Mutation frequencies were compared between patients with XP from different complementation groups and unaffected first-degree relative controls. Fibroblasts derived from patients with XP-C also exhibited significantly higher spontaneous somatic SNV (C) but not INDEL (D) loads. Mutation frequencies were compared between patients with XP and healthy controls. Data are presented as mean ± standard deviation. 1M, one million; SNV, single-nucleotide variant; INDEL, insertion/deletion.

PBMCs from patients with XP-C also exhibited a significantly elevated frequency of small INDELs (insertions and deletions; 0.081 ± 0.03 INDELs/Mb) compared with that of those from heterozygous controls (0.012 ± 0.008 INDELs/Mb; Figure 1B). Consistent with the SNV results, no significant difference was observed among patients with XP-A, XP-D, or XP-F (0.017 ± 0.02 INDELs/Mb, 0.0045 ± 0.002 INDELs/Mb, and 0.0051 INDELs/Mb, respectively) and their corresponding controls (0.013 ± 0.008 INDELs/Mb, 0.0081 ± 0.01 INDELs/Mb, and 0.0064 INDELs/Mb, respectively).

SMM-seq analysis of fibroblast cultures from the same participants also revealed a statistically significant but modest (2.2-fold) increase in somatic SNVs in patients with XP-C compared with that in healthy controls (0.46 ± 0.2 SNV/Mb vs 0.21 ± 0.04 SNV/Mb; Figure 1C). In contrast, INDEL frequencies in fibroblasts did not differ significantly between XP-C or any other XP complementation group (Figure 1D).

Normal XP-C PBMCs show a leukemia-like mutational spectrum

The increased SNV frequency in PBMCs from patients with XP-C predominantly reflected mutation signatures SBS5 and SBS8, both associated with DNA replication,⁹ as well as SBS32, a signature not previously linked to XP (Figure 2A). PBMCs from patients with XP-C exhibited a higher proportion of C>A and T>A transversions_than other XP groups, enriched at positions where the mutated base is followed by a thymine or adenine (Figure 2B). Prior studies have reported that C>A substitutions in the cytosine-adenine (CpA) context may indicate bypass by error-prone translesion polymerases.¹⁰ In contrast, XP-C samples exhibited proportionally fewer C>T and T>C transitions than other groups. This reduction did not reflect lower absolute numbers of these transitions, but rather a compositional shift caused by a strong enrichment of other substitution types. The INDELs were almost exclusively single-base deletions of cytosine (Figure 2C).

Figure 2.

Mutational characteristics of PBMCs across different XP complementation groups. (A) Somatic mutation signatures detected in PBMCs from individuals representing multiple XP complementation groups and unaffected relatives. (B) A total of 96 mutational profiles of patients with AML tumors in XP-C and of normal blood from individuals with XP-C, XP-C heterozygous, XP-A, and XP-D. (C) Proportion of INDELs by size category (1 bp to ≥5 bp; left) and the distribution of types among 1-bp INDELs (right). AML, acute myeloid leukemia; XP-C Het, XP-C heterozygous.

In line with our observation of increased mutational load in normal XP-C blood, leukemic cells from patients with XP-C were previously reported to have a ∼25-fold higher somatic mutation burden than sporadic leukemias.³ For comparison, we analyzed published mutation profiles from Yurchenko et al and found that the SNV profiles of normal XP-C PBMCs closely resembled those of XP-C leukemic cells (cosine similarity, 0.981; Figure 2B), with both being enriched for SBS5, SBS8, and SBS32 (Figure 2A; supplemental Figure 1A) as well as frequent single-nucleotide deletions of cytosine (Figure 2C).³ The close correspondence suggests that the same mutagenic process operative in XP-C leukemia is already detectable in normal blood cells of patients with XP-C.^3,11 Notably, the enrichment of SBS5, SBS8, and SBS32 was not consistently observed in XP-C fibroblasts derived from the same individuals or healthy controls (supplemental Figure 1B-C).

Discussion

In this study, we show that normal PBMCs from patients with XP-C, but not from patients with XP-A, XP-D, or XP-F, have a greatly elevated burden of base substitution mutations. Mutation burden in XP-C was also elevated in normal fibroblasts, although to a much lesser extent. These findings indicate that XP can be divided into 2 molecular subtypes defined by the spontaneous mutation burden in cells not exposed to sunlight. Patients with XP from complementation groups defective in both GG-NER and TC-NER (eg, XP-A, XP-D, and XP-F) exhibit low spontaneous SNV loads in internal tissues and primarily develop UV-induced skin cancers. In contrast, patients with a defect only in GG-NER, exemplified by XP-C, accumulate a high frequency of spontaneous SNVs even in non–sun-exposed cells and, accordingly, face a much greater risk of internal tumors (chiefly hematologic malignancies) alongside their elevated skin cancer risk.³

A plausible explanation for the low spontaneous mutational load in XP complementation groups lacking both GG-NER and TC-NER is that, when TC-NER is absent, persistently stalled transcription complexes induce apoptosis, thereby eliminating heavily damaged cells before they can generate mutant progeny.¹² This process likely contributes to the progressive neurological degeneration and extreme photosensitivity observed in these groups.¹³ In contrast, when only GG-NER is absent, transcription blockage at unrepaired lesions is less likely to occur because TC-NER can still recognize and repair lesions encountered by RNA polymerase II during transcription by recruiting the intact repair machinery to remove the blockage. Cells therefore survive despite the absence of GG-NER. However, unrepaired DNA adducts in nontranscribed regions persist and eventually become fixed as mutations during DNA replication by error-prone translesion polymerases that also frequently introduce single-nucleotide slippage errors (eg, the 1-base cytosine deletions observed in XP-C cells).¹⁴ This results in an excessive mutational load in GG-NER–deficient cells. Notably, replication itself is not the primary source of these lesions, but it provides the opportunity for these persistent DNA adducts in GG-NER–deficient cells to be converted into permanent mutations through error-prone bypass. Consequently, the particularly high mutation burden observed in XP-C PBMCs compared with syngeneic fibroblasts is likely attributable, at least in part, to the much higher replication rate and turnover of hematopoietic cells.¹⁵

To our knowledge, our study is the first to directly quantify somatic mutation rates in normal human tissues across multiple XP complementation groups and the immediate family members of patients with XP, using XP as a model of inherited DNA repair deficiency. Critically, we demonstrate that the loss of GG-NER establishes a cancer-like mutational burden and characteristic mutational signatures in normal XP-C blood cells, which likely underlies the high incidence of internal, particularly hematologic, malignancies observed in XP-C.³ The mutational profile of normal XP-C lymphocytes closely mirrors that reported in XP-C leukemia,³ sharing SBS5, SBS8, and SBS32 signatures and an enrichment of single-nucleotide cytosine deletions, indicating that the same mutagenic processes are already active long before malignant transformation. In this sense, defective GG-NER directly links to the emergence of a precancerous mutation landscape. In contrast, XP-C fibroblasts do not display SBS5, SBS8, or SBS32 enrichment, suggesting that the manifestation of GG-NER deficiency is highly tissue-dependent and most pronounced in highly mitotically active hematopoietic lineages.

Our work aligns with previous findings that induced pluripotent stem cells (reprogrammed from erythroblasts) derived from patients with XP-C exhibit a mutational profile resembling XP-C leukemia.¹¹ Interestingly, this study reported a much lower somatic mutation frequency than what we observe in primary XP-C blood cells.¹⁶ This discrepancy likely reflects selection against heavily mutated clones during reprogramming, coupled with the limited proliferative history of erythroid-lineage cells. The situation in XP-C leukemia appears to be the opposite: here, the most heavily mutated hematopoietic cell is more likely to undergo malignant transformation, and the tumor mutational burden largely reflects clonal amplification of that founder cell’s genome. In other words, the extremely high mutation load in XP-C leukemia represents the fixed, propagated somatic mutations of a single, already hypermutated progenitor. These observations help explain the continuity between the mutational spectra of XP-C PBMCs and XP-C leukemias.

All our findings provide new insight into how defective GG-NER creates a mutagenic background that links DNA repair deficiency to cancer risk and may inform therapeutic strategies aimed at modulating cell death and DNA repair pathways in XP and other genome instability syndromes.

Conflict-of-interest disclosure: A.Y.M. and J.V. are cofounders of Mutagentech. The remaining authors declare no competing financial interests.