DNA Damage and Repair in Thyroid Physiology and Disease

Abstract

The thyroid is exposed to DNA damage induced by normal physiological processes (eg, oxidative stress resulting from thyroid hormone synthesis or mitochondrial respiration) or through environmental insults (eg, environmental pollutants, ionizing radiation exposure). Robust antioxidative stress defense and DNA repair mechanisms protect thyrocyte genome integrity, but defective or dysregulated DNA repair pathways have been implicated in thyroid pathology, including autoimmune thyroid disease and thyroid malignancy. In thyroid cancer, disturbed antioxidative stress defense, Mismatch Repair, Non-Homologous End-Joining, or DNA damage response pathways contribute to both the onset and progression of the disease. The insight into mechanisms governing thyrocyte genome integrity may help to gain better understanding of the pathology and suggest novel therapeutic regimens, urgently needed in treatment-refractory disease. In the current review, we provide comprehensive description of the exogenous and endogenous factors, as well as DNA repair mechanisms influencing thyrocyte genome integrity. Moreover, we pinpoint major research avenues that should be pursued in future research. This information will be valuable in directing new discoveries to better understand thyroid disease pathomechanisms, as well as aid development of novel diagnostic and therapeutic tools.

Essential Points

• Thyroid DNA integrity is threatened by damaging agents resulting from normal physiological processes (such as ROS/H₂O₂ produced during thyroid hormone synthesis and normal thyrocyte metabolism) or environmental exposure (eg, radiation, pollutant chemicals)

• The key mechanisms protecting thyroid DNA integrity include robust antioxidant defense, compartmentalization of thyroid hormone synthesis and DNA repair

• Cancer genome sequencing data suggest involvement of AID/APOBEC, disturbed mismatch repair or nonhomologous end joining, as well as oxidative stress as causative factors in thyroid carcinogenesis

• Somatic mutations in DNA repair genes (eg, MSH2, MSH6, TP53, ATM) are responsible for evolution, dedifferentiation, and progression of thyroid cancer

• Variations/mutations or disturbed expression of DNA repair genes (eg, XPC, PARP1, MSH2, MSH6, WRN, ATM, TP53, CHEK2, FANCC, FANCG) contribute to thyroid pathology, including hypothyroidism, AITD, or thyroid cancer

• Drugs targeting single-strand (APE1 or PARP1 inhibitors) or double-strand (LIG4, POLQ, TIP60, or WEE1 inhibitors) DNA break repair offer potential treatment strategies for thyroid cancer, including ATC and therapy-refractory WDTC

The thyroid, a butterfly-shaped gland located at the base of the neck, plays a vital function in energy (through thyroid hormones) and calcium (through the peptide hormone, calcitonin) homeostasis. Thyroid health influences function of all cells and organs through thyroid hormone-dependent regulation of gene expression and metabolic rate. Thyroid hormones, 3,3′,5-triiodo-L-thyronine (T₃) and 3,3′,5,5′-tetraiodo-L-thyronine (thyroxine; T₄) are produced in follicular cells (thyrocytes) that surround colloid-filled thyroid follicles. Calcitonin, a central regulator of calcium levels, is produced by neuroendocrine parafollicular cells (C cells) located in the connective tissue next to the follicular cells. Production of thyroid hormones (TH) is controlled by negative feedback regulation involving the hypothalamus (secreting thyrotropin releasing hormone (TRH)) and the pituitary (secreting thyroid-stimulating hormone (TSH)) that together with the thyroid form hypothalamus–pituitary–thyroid axis. TH synthesis is supported by thyroid peroxidase (TPO) that catalyzes iodination of TH precursor thyroglobulin (TG) in the thyroid follicle lumen. The iodide (I⁻), necessary for iodination, is supplied by active transport to thyrocyte by basolateral Na⁺/I⁻ symporter (NIS; SLC5A5), followed by the efflux to follicle lumen through the apical electroneutral anion exchanger pendrin (PDS; SLC26A4) (1). Importantly, the iodination reaction is dependent on TPO-catalyzed oxidation of I⁻ to iodine radical (I⁰). This reaction requires active H₂O₂ synthesis by NADPH oxidases from the NADPH oxidase (NOX) family, namely the dual oxidases 1 (DUOX1) and 2 (DUOX2) (previously known as thyroid oxidases ThOX1 and ThOX2, respectively). The major thyroid NOX producing H₂O₂ for T₃ and T₄ synthesis is DUOX2, that together with dual oxidase maturation factor 2 (DUOXA2), indispensable for DUOX2 activity, is located on the apical membrane of the thyrocyte facing toward thyroid follicle colloid (2) (Fig. 1). H₂O₂, one of the major reactive oxygen species (ROS) involved in the intracellular redox-sensitive signaling pathways, threatens cell integrity when reaching concentrations above physiological levels. The state when ROS formation exceeds their detoxification by antioxidant defense systems is recognized as the oxidative stress (OS) (3). That H₂O₂ is actively produced by the thyroid gland makes this tissue a very good model to study OS-induced intracellular damage, in particular the DNA damage. In this manuscript, we focus on the role of DNA damage and repair processes in thyroid physiology and disease.

Figure 1.

Thyroid hormone synthesis and sources of oxidative stress in thyroid cell. Synthesis of thyroid hormones (TH) 3,5,3′-triiodothyronine (T₃) and 3,5,3′5′-tetraiodothyronine (T₄) requires transport of iodide (I⁻) first to thyrocyte by basolateral Na⁺/I⁻ symporter (NIS), and next to thyroid follicle lumen through the apical electroneutral anion exchanger pendrin (PDS). In the follicle colloid, I⁻ ions are oxidized in a TPO-mediated and H₂O₂-dependent reaction to iodine radicals (I⁰) to result in iodination of tyrosine residues in thyroglobulin (TG). H₂O₂ in the follicle lumen is generated mainly by dual oxidase 2 (DUOX2), whose activity is supported by dual oxidase maturation factor 2 (DUOXA2). Iodinated tyrosines within TG molecule are coupled to generate T₃ and T₄. Iodinated TG is stored within the follicle lumen, but in a situation of TH shortage TG is taken up by the thyrocyte via endocytosis and next degraded by proteases within endolysosome to release T₃ and T₄. TH are next transported from thyrocyte to the blood vessel by MTC8 (SLC16A2) transporter. The major sources of oxidative stress (OS) in the thyroid cell include, in addition to H₂O₂ produced by DUOX2 in the follicle lumen, are intracellular H₂O₂ generated by NOX4 oxidase and ROS originating from mitochondrial respiratory chain. Created in BioRender. Arczewska, K. (2025) https://BioRender.com/k79x007.

Diseases of Thyroid Gland

Both neoplastic and nonneoplastic disorders may affect the thyroid gland. Hyperthyroidism (excessive TH levels) and hypothyroidism (insufficient TH levels) result from various thyroid pathologies including autoimmune thyroid disease, dysregulation of the hypothalamus–pituitary–thyroid axis, or rare genetically determined disorders (eg, thyroid hormone resistance, Allan-Herndon-Dudley syndrome). Hyper-/hypothyroidism may also result from external factors, such as inappropriately high/low iodide levels coming either from dietary ingestion, induced by medication (eg, amiodarone) or diagnostic imaging techniques using iodinated contrast agents. Hypo-/hyperthyroidism may lead to goiter (thyroid enlargement). Goiter is the most common thyroid pathology induced by increased TSH levels that generate growth stimulatory signals leading to thyroid hyperplasia and hypertrophy. Autoimmune thyroid diseases (AITD) include Graves’ disease (GD; usually associated with hyperthyroidism) and Hashimoto thyroiditis (HT; autoimmune thyroid inflammation; leading to hypothyroidism). The prevalence of AITD might reach up to 5% to 10% of the population in areas with adequate supply of iodide. Specific AITD features include the presence of serum autoantibodies directed against TSH receptor (in GD), as well as TPO and/or thyroglobulin in HT (1). The key mechanism leading to AITD development is the loss of immunotolerance to thyroid autoantigens, that is influenced by both genetic predisposition and environmental factors (1). Importantly, thyroidal oxidative stress has been implicated in AITD initiation and progression (4). Additionally, OS in AITD inflammatory disease was suggested to increase the risk of thyroid carcinogenesis (5). On the other hand, concomitant AITD was also proposed to improve papillary thyroid carcinoma (PTC) prognosis through influencing tumor immune microenvironment (6). However, the mechanistic relationships between AITD and TC require further experimental verification.

According to the 5th edition of the World Health Organization Classification, thyroid neoplasms derived from follicular cells are divided into benign tumors, low-risk neoplasms, and malignant neoplasms (7). The most frequent thyroid lesion is benign thyroid follicular nodular disease (FND), involving thyroid nodules, the focal hyperplastic structures different from normal gland architecture. They can be either isolated or manifested as multiple nodule lesions, previously known as multinodular goiter (MNG), with incidence ranging from 5% to above 50% of the population, depending on the iodide supply. Thyroid follicular nodular disease frequency rises considerably in the elderly. Patients with follicular nodular disease frequently develop toxic nodules (ie, producing thyroid hormones) that lead to hyperthyroidism. Toxic nodules may carry activating TSH receptor (TSHR) mutations and thyroids with nodular disease frequently show papillary cancerous lesions (8). The control of TSH levels in nodular or AITD disease is essential because higher TSH concentrations correlate with increased cancer risk (5). Another frequent thyroid neoplasia is thyroid adenoma, which is mostly follicular, with the prevalence of up to 4%, as detected in autopsy tissues. Adenomas are benign solitary nodules with a gland-like structure that are either hypoactive (“cold”), normal, or hyperactive (“hot”—producing thyroid hormones). The latter are termed toxic adenomas (9). Activating TSHR and KRAS mutations, as well as PAX8/PPARgamma rearrangements are detected in thyroid adenomas and their presence correlates with cancerous transformation (10, 11).

Thyroid cancer (TC) is the most frequent endocrine malignancy with increasing incidence rates reaching globally from 3 to more than 10 new cases and 0.5 deaths per 100 000 yearly. TC is 2 to 3 times more frequent in women than in men (12). Most of the thyroid cancers derive from follicular thyroid cells and are sometimes collectively termed non-medullary thyroid cancer (NMTC). Follicular-cell derived TC includes well-differentiated TC (WDTC) with PTC, follicular thyroid carcinoma (FTC), invasive encapsulated follicular variant papillary carcinoma (IEFVPTC), and oncocytic carcinoma of the thyroid (OCA; previously known as Hürthle cell thyroid carcinoma) subtypes. More advanced TCs are high-grade follicular-derived carcinomas including differentiated high-grade thyroid carcinoma (DHGTC) and poorly differentiated thyroid carcinoma (PDTC), whereas anaplastic follicular cell-derived thyroid carcinomas (ATC) are the most aggressive TC subtype (7). The most frequent thyroid malignancy is PTC, reaching 80% to 93% of all TC cases, whereas FTC prevalence reaches 6% to 10% among the TCs. Well-differentiated TCs have favorable prognosis with 5-year survival rates exceeding 95%, but up to 10% to 20% of cases show persistent, recurrent disease, and/or metastasis to distant locations (mainly lung, bones, and brain). PTCs are further subdivided into, among others, classic, tall cell (TC-PTC), columnar cell, and hobnail subtypes (7, 12). PDTC cases constitute 0.3% to 6.7% of all TCs, whereas ATC prevalence is below 2% among TCs, but due to highly aggressive potential and low survival rates ATC is responsible for up to 50% of TC-related deaths. Only about 3% to 5% of TC cases are medullary thyroid carcinomas (MTC) derived from parafollicular C cells (12)

The molecular landscape of PTCs involves, mainly, mutually exclusive activating mutations in the MAPK pathway, of which the BRAF^V600E mutation is most common and seen in 45% to 55% of cases, reaching almost 100% in TC-PTC. The other mutations found frequently in PTC include those affecting RAS (N-RAS, H-RAS, or K-RAS; 10%-20% of cases), RET::PTC rearrangements (seen in 5%-10% of cases), where RET::PTC1 (fusion with coiled-coil domain containing 6; CCDC6) and RET::PTC3 (nuclear receptor co-activator 4; NCOA4) are the most frequent. Rearrangements involving neurotrophic tyrosine receptor kinase 1 and 3 (NTRK1 and 3) genes are seen in less than 5% of PTC cases. Importantly, the frequency of RET::PTC rearrangements is increased in subjects with radiation exposure history. Another frequent change found in PTCs are telomerase reverse transcriptase promoter (TERTp) mutations that are detected in more than 10% cases, but their frequency increases in more aggressive cancers, with prevalence of approximately 30% in TC-PTC and more than 40% in PDTC, DHGTC, and ATC. Importantly, coexistence of BRAF^V600E and TERTp or RAS and TERTp mutations in PTC highly correlates with disease aggressiveness and poor clinical outcome. RAS mutations dominate FTC and IEFVPTC molecular landscape, affecting 15% to 40% and up to 70% of cases, respectively. FTCs and IEFVPTCs carry also PAX8/PPARgamma fusions, that are mutually exclusive with RAS mutations and were detected in 12% to 56% cases. Other frequent changes in FTCs are activating the signaling pathway involving phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K) and AKT serine/threonine kinase signaling pathway mutations, mainly PIK3CA or PTEN in <10% of cases. In FTCs and IEFVPTC, TERTp is also encountered with frequency similar to that observed in PTC (ie, in 15%-20% of cases). PDTC and ATC show altered mutation characteristics, with relatively low levels of BRAF mutations (10%-20% in PDTC and 30%-40% in ATC), elevated levels of RAS mutations (20%-40%), PIK3CA mutations (5%-20%) and, importantly, high levels of TP53 mutations, that appear very rarely in PTCs, but their levels increase to 20% to 30% in PDTC and even to 50% to 80% in ATC. MTC tumors mainly carry mutations activating RET oncogene, with prevalence reaching >95% in familiar MTC syndromes termed multiple endocrine neoplasia (MEN) syndrome type 2A and type 2B. Other frequent changes in MTCs are RAS mutations reaching levels of 10% to 30% cases. Interestingly in the context of the current review, mutations in DNA repair genes were also detected in TC with frequency increasing with tumor aggressiveness (see Table S1 (13)) (10, 14, 15). The functions of these mutations and their impact on the signaling pathways in TC are mostly unknown. Interestingly, BRCA2, FANCD2, FANCA, FANCE, FANCG, ERCC4, NBS1, POLD1, PALB2, and WRN mutations were observed in both benign thyroid goiter and PTC tissues, suggestive of their involvement in progression of the disease from preneoplastic lesions to malignancy (16).

The major treatment regimens in WDTCs include combination of surgery, radioactive iodine (RAI) ablation therapy and long-term TSH suppression. However, because of dedifferentiation, mainly loss of NIS expression, more aggressive variants of WDTCs and ATCs become refractory to RAI. Therefore, there is continuous need for new therapeutic options for TCs resistant to traditional treatment regimens. Apart from the traditional cytotoxic chemotherapy with doxorubicin, taxane, or cisplatin, analysis of molecular characteristics of TCs suggested application of multikinase inhibitors (lenvatinib, sorafenib, cabozantinib, vandetanib, pazopanib), more specific receptor tyrosine kinase inhibitors (larotrectinib, entrectinib, selpercatinib, and pralsetinib) or MAPK pathway inhibitors (dabrafenib and trametinib) that have been approved or are under consideration by the US Food and Drug Administration and the European Medicines Agency for TC therapy (10, 12, 17). Recently, because of the high immune cell infiltration in TC tissues, immune checkpoint inhibitor therapy was proposed for patients not responding to other therapeutic regimens (18). However, it has to be kept in mind that immune checkpoint inhibition may lead to autoimmunity-related adverse events manifested, among others, by hypothyroidism and thyrotoxicosis (19). Moreover, because of limited response rates, development of drug resistance, and high level of adverse events observed for the approved inhibitor therapies, novel treatment approaches are still needed. Targeting DNA damage repair may be one of the options to pursue further. For example, recently, poly(ADP-ribose) polymerase (PARP) inhibitors or combination of non-homologous end joining (NHEJ) inhibition and chemotherapy have been suggested for management of TC refractory to other treatment modalities (20, 21).

Sources of DNA Damage in Thyroid Tissue

One of the major ROS-generating systems in thyroid tissue is H₂O₂ production by DUOX2 NADPH oxidase coupled to TH synthesis. Proper H₂O₂ production is required for thyroid health and inactivating DUOX2 or DUOXA2 mutations are the underlying cause of goiter and congenital hypothyroidism (22). Thyroidal DUOX2 activity and/or expression is upregulated by TSH and H₂O₂. Moreover, excessive iodide inhibits DUOX2, TPO, and NIS activity and/or expression leading to the suppression of TPO-catalyzed TG iodination and TH synthesis, known as the Wolff-Chaikoff effect (2, 23). Additionally, it was observed that TSH-mediated stimulation of thyrocyte proliferation is accompanied by elevated ROS-dependent DNA damage (24). Interestingly, inflammatory cytokines implicated in AITD development stimulate DUOX2 (but not DUOX1) expression leading to increased OS (25). Importantly, ionizing radiation (IR) stimulates DUOX1 expression in thyroid cells, leading to H₂O₂ production and persistent DNA damage (26). Another NOX oxidase studied in the context of thyroid physiology is the NOX4, a unique enzyme among non-DUOX NOX oxidases since it generates H₂O₂, not the superoxide. Importantly, in multiple tissue types, NOX4 localizes to, among others, the nucleus and mitochondria, where it can directly mediate H₂O₂-induced mitochondrial and nuclear DNA damage (27, 28). NOX4 expression is stimulated by TSH in a transforming growth factor beta 1-dependent manner. It is also increased in the experimental goiter model in Wistar rats, which is accompanied by increased H₂O₂ levels in the thyroid tissue. Moreover, NOX4 is implicated in thyroid autoregulation, since it mediates OS induction and expression of thyroid-specific genes, especially upon iodide overload, as observed in FRTL-5 rat thyroid cells (29, 30). Additionally, estrogen induces NOX4 expression and the associated H₂O₂ production in thyroids of female rats, as well as cultured rat thyrocytes, possibly providing mechanistic explanation for higher TC frequency in women (31). Thyroid DNA integrity is also significantly influenced by nutrients and environmental factors, as outlined in our recent review (32).

Oxidative Stress in Thyroid Neoplasia and Autoimmunity

The general intracellular sources of ROS include mitochondrial respiratory chain and several enzymatic systems (eg, xanthine oxidases, cytochrome P450), with membranous NADPH oxidases from NOX family (EC 1.6.3.1) playing pivotal role (33). Functional mitochondria are essential for proper thyroid function, as evidenced, for example, by a mouse model with thyroid-specific mitochondrial respiratory chain dysfunction showing retarded growth and premature death, accompanied by non-autoimmune thyroid dysfunction (34). Mitochondria are implicated in tumor progression through metabolic reprograming of cancer cells leading to the switch from mitochondrial respiration to aerobic glycolysis, known as the Warburg effect, which is required to fulfill growing energy demands and supporting tumor growth. The Warburg effect is inevitably associated with increased oxidative stress (35). Dysfunctional mitochondria are associated with thyroid cancer pathogenesis, as evidenced, for example, by the increased levels of somatic mutations in mitochondrial DNA, mainly in the respiratory chain complex I, detected in TC tissues (36).

Cancer cells, especially RAS-transformed cancer cells, have unbalanced redox status, manifested by elevated ROS generation, possibly resulting from the activity of NOX oxidases (3, 28, 33). NOX oxidases 1-3 and 5 release superoxide anion (O₂^·−), that is converted to H₂O₂ by a separate enzyme superoxide dismutase (SOD) or spontaneously. NOX4 has the intrinsic property to produce mainly H₂O₂ that involves extracellular E-loop DUOX1-2, on the other hand, additionally possess N-terminal peroxidase domain (hence their name “dual oxidases”) that catalyzes superoxide dismutation and thus results in H₂O₂ release. The most extensively studied in connection with thyroid tumorigenesis is the NOX4 NADPH oxidase. NOX4 overexpression was observed in thyroid cancer, as well as in HT tissue. High NOX4 levels correlate with dedifferentiation status and a malignant phenotype of thyroid cancer cells. Importantly, BRAF, HRAS, and TP53 mutations induce NOX4 expression and/or activity in thyroid cells. Moreover, NOX4 expression mediates oncogene-induced ROS-dependent DNA damage and subsequent senescence (28, 37). Among the NOX oxidases related to thyroid physiology, DUOX1 is downregulated through epigenetic mechanisms in several cancer tissues and is suggested to associate with an invasive and metastatic phenotype (2). A genome-wide gene-expression study revealed unaltered DUOX1 expression in thyroid cancer compared to normal tissue (38). However, elevated DUOX1 expression was described in radiation-induced, but not sporadic, cancers (26). DUOX2 expression, on the other hand, is upregulated in most of the analyzed cancer tissues and correlated with malignant phenotype (2). In the thyroid tumor, however, DUOX2 levels are unchanged, but DUOX2, as well as DUOX1 expression is downregulated in less-differentiated BRAF-mutated cancers (39). Conversely, DUOXs expression is unchanged in well differentiated thyroid cancers (ie, those sustaining TPO, NIS, and PDS expression) (40). Moreover, high DUOX expression levels in thyroid tissue correlate with better prognosis of patients with PDTC (41). Therefore, unbalanced DUOXs and NOX4 levels might be implicated in oxidative stress triggering thyroid pathology, whereas OS-dependent malignant traits and thyroid cancer progression are mainly supported by NOX4.

Additional sources of OS in thyroid tissue potentially contributing to carcinogenesis might be infiltrating immune cells that have been detected in many PTC tissues with or without accompanied thyroiditis (42). Accordingly, higher immune cell infiltration in TC tissue correlates with the increased tumor mutation burden (18).

DNA Damage in Thyroid Cancer

H₂O₂ may induce thyrocyte DNA damage and promote cancerous transformation associated with formation of oxidatively modified lesion 8-oxoguanine (8-oxoG), abasic sites, and DNA strand breaks, as well as RET::PTC1 rearrangements (43-45). Moreover, although thyrocytes divide only 5 times during adulthood, spontaneous mutation rates in normal murine thyroid and human thyroid cancers are higher than in other tissues (46, 47). Oxidative stress and the associated oxidative DNA damage, exemplified by 8-oxoG, which is accepted as a biomarker for OS levels, as well as mutagenesis were detected in thyroid pathologies, such as GD, HT, follicular adenoma, PTC, and FTC (48-50). Moreover, higher levels of OS-induced lipid peroxidation, lipid peroxidation products trans-4-hydroxy-2-nonenal (HNE), and/or malondialdehyde (MDA) were also detected in TC, follicular thyroid adenoma, HT, GD, and MNG tissues (51-53). Lipid peroxidation products may react with DNA bases leading to formation of exocyclic etheno, ethano, propano, as well as MDA- and HNE-derived adducts that lead to mutagenesis (54). Reactive nitrogen species (RNS), mainly nitric oxide (NO) produced by inducible NO synthase and the highly reactive peroxynitrite generated from NO and superoxide constitute additional potential sources of DNA damage in the thyroid tissue. RNS compromise DNA integrity by inducing DNA base damage, eg, the rapidly depurinating 8-nitroguanine, deamination and alkylation, DNA strand breakage, as well as inhibition of DNA repair proteins (55). Inducible NO synthase expression is stimulated by inflammatory cytokines and was shown to be upregulated in thyroid cancer tissue and AITD thyroids (56, 57).

Early studies detected C → T, T → C, and G → T base exchanges as the most prevalent mutations in the thyroid cancer tissues and it has been hypothesized that they are generated by oxidative stress (58). Later on, it was suggested that high frequency of A → G (T → C) transitions in thyroid cancers is related to the high metabolic rate, mitochondrion organization, and oxidative stress. Importantly, thyroid tumors with high metabolic rates have increased incidence of NRAS or HRAS A → G mutations at nucleotide position 182 (protein Q61R). BRAF^V600E (1799AT) mutation levels, on the other hand, are reduced in tumors with high metabolic rate (48). Frequent driver mutations at A:T pairs in BRAF and RAS oncogenes have been attributed to error-prone repair of complex oxidative DNA lesions (23). However, elevated mutability of thyroid cancers and oxidative stress as a causative agent in thyroid mutagenesis has been questioned by high-throughput sequencing data, that showed rather moderate mutation load, as well as the prevalence of mutations related to the spontaneous deamination of 5-methylcytosine, activity of cytidine deaminases activation-induced cytidine deaminase/apolipoprotein B mRNA editing catalytic polypeptide-like (AID/APOBEC), defective MMR, NHEJ, or tobacco smoking. Importantly, AID/APOBEC- and MMR-dependent mutations seem to be related with more aggressive thyroid carcinomas and they were suggested to be responsible for dedifferentiation, cancer progression, and high mutation load in these cancers (14, 15, 59-61). Moreover, APOBEC-induced mutations are frequently observed in RAI-refractory PTC patients (62). APOBEC-related TC mutational signatures are dominated by APOBEC3A-dependent pattern (63). However, recent reports revealed also high frequency of SBS18 mutational signature that mainly involves C → A transversions linked to ROS exposure, in PTC tissues in general, as well as in radiation-related PTCs (64, 65). Therefore, it cannot be excluded that oxidative stress inherent to thyroid physiology plays a role in thyroid cancer pathogenesis and promotion.

DNA Damage in Autoimmune Thyroid Disease

ROS were suggested as an initial cause that triggers autoimmune processes in the thyroid (4, 66). Surprisingly, the studies on DNA damage and AITD are scarce. One study reported increased oxidative DNA damage in peripheral blood of patients with HT (67). Increased DNA damage was also reported in lymphocytes of GD patients (68). Urinary excretion of 8-oxo-7,8-dihydroguanosine (8-oxoGuo) and 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxodGuo) is increased in patients with toxic nodular goiter, GD, and HT and decreases with restoration of euthyroidism (69-71). However, a newer study detected no change in urinary 8-oxoGuo and 8-oxodGuo levels in HT patients after reaching euthyroid status (72). On the other hand, the levels of 8-oxodGuo and MDA in GO are increased in orbital fibroblasts (73). Increased DNA fragmentation, most likely resulting from apoptotic cell death, was observed in HT, but not GD, thyroids (74). Moreover, 1 study observed increased 8-oxoGuo levels in HT tissues (75). A source of DNA damage in autoimmunity might be related to inflammatory processes with cytokine-mediated induction of AID/APOBEC expression (76). AID/APOBEC expression in AITD tissues has been shown in high-throughput data that reveal upregulation of several APOBEC3 deaminases, including APOBEC3B, in HT tissues (77).

RAI Treatment-induced Thyroid Disease

Due to the NIS-mediated thyroid cell ability to concentrate iodide, as well as thyroid radioiodide accumulation through TPO-directed iodide oxidation and coupling to the thyroglobulin tyrosine residues, known as iodide organification, diagnostic, or therapeutic procedures involving radioiodinated compounds are the gold standard in TC management, both for locoregional and metastatic lesions. RAI therapy, mainly using I¹³¹, is also prescribed to treat thyroid hyperplasia or hypertrophy, GD, hyperthyroidism, including toxic thyroid nodules or goiter. However, recently, there has been a continuous decline in the use of RAI therapy (78, 79). RAI induces DNA damage, ROS generation, and lipid peroxidation, leading to proliferation inhibition, cell-cycle arrest, apoptosis, and necrosis resulting in thyroid gland destruction and the associated hypothyroidism. DNA repair capacity is an important factor in determining thyroid response to RAI therapy in the diseased, but also healthy thyrocytes (80, 81). Several single nucleotide polymorphisms (SNPs) in DNA repair genes are associated with increased susceptibility to radiation-related TC (82, 83). Radiation may also influence methylation status, and thereby expression of DNA repair genes leading to alterations in DNA repair capacity (84). Thyroid inflammation and release of thyroid autoantigens that trigger autoimmune processes is a secondary consequence of RAI-related DNA damage and cell death (85). Although ¹³¹I has relatively fast clearance rate, with half-time of 5.5 days, radioiodinated compounds may induce long-term consequences, as illustrated by the persistence of genotoxic lesions in thyrocytes and peripheral blood lymphocytes for several months, or even several years, after RAI therapy (86). Importantly, variants in DNA repair genes associate with persistence of these lesions (87). Moreover, RAI therapy in hyperthyroid patients was suggested to increase thyroid cancer incidence and mortality rates (88).

Antioxidative Protection in Thyroid Cells

ROS, including H₂O₂, although detrimental when present in excessive concentrations in the thyroid cells, are essential for TH synthesis and intracellular signaling through redox-sensitive pathways (3, 89). Thyrocytes are well protected against OS resulting from enzymatic activity of NADPH oxidases. First, H₂O₂ synthesis is restricted to the extracellular space by localization of the iodination complex (known as thyroxisome), composed of, at least, TPO and DUOXs, but also DUOX maturation factors (DUOXA1/2) and caveolin-1 (Cav-1), at the apical pole of thyrocyte in the follicular space (53). Interestingly, the protective role of the extracellular localization of the iodination complex is best exemplified by the analysis of the thyroid tissues from patients with Pendred syndrome, an autosomal recessive disorder caused by PDS gene mutations, that is manifested by deafness and goiter, as well as HT thyroid tissues, where intracellularly localized TPO and DUOXs promote oxidative stress and apoptosis (53, 90). The second line of protection against OS is provided by the robust nonenzymatic and enzymatic antioxidative defense including, among others, catalase (CAT), superoxide dismutases (SOD1-SOD3), glutathione peroxidase (GPX1-GPX8)/glutathione (GSH)/glutathione reductase (GSR), and peroxiredoxin (PRDX1/PRX1-PRDX6/PRX6)/thioredoxin (thioredoxin, cytosolic TXN/TRX1 and mitochondrial TXN2/TRX2)/thioredoxin reductase (TXNRD1/TRXR1-TXNRD3/TRXR3) systems. The importance of the antioxidative protection in thyroid physiology and disease is evidenced by the observation, that iodine treatment stimulates expression of Tpo, Sod1, Gpx2, Gpx4, Gpx7, Prdx1, Prdx2, Prdx5, and Txn in control murine thyrocytes, but not in thyrocytes from NOD-H2^h4 mice, that show high incidence of iodine-induced autoimmune thyroiditis (91). In humans, polymorphisms in the genes encoding the above antioxidative system enzymes in general are not correlated with TC risk, with the only exception observed for SOD1 rs1041740 and rs12626475 variants (92). Upregulation of antioxidative stress responses constitutes a part of metabolic reprograming required for cancer progression (35). Accordingly, the net antioxidative capacity of TC tissues seems to be elevated in comparison to normal tissues (93).

CAT is a peroxisomal enzyme catalyzing conversion of H₂O₂ to water and oxygen. Catalase, most probably, is less important in thyroidal H₂O₂ detoxification than other systems, including GPXs and PRDXs (94, 95). CAT expression is in general downregulated in thyroid disorders, including TC, GD, HT, and follicular adenoma tissues (50, 96-99).

Superoxide dismutases catalyze dismutation of O₂^·− to oxygen and H₂O₂. TSH stimulates SOD2 and SOD3, but neither SOD1, CAT, nor GPX1, expression in normal human or rat thyrocytes (100, 101). Several studies that addressed the changes of expression and/or activity of SOD1 (cytosolic Cu/Zn-SOD) and SOD2 (mitochondrial MnSOD) brought mixed results. Some studies reported unaltered SOD1/SOD2 levels in TC and HT, whereas other reports described SOD1 and SOD2 downregulation or even SOD2 upregulation in follicular adenoma, MNG, and/or TC (3, 50, 51, 96, 97, 99, 102). The apparent discrepancy in SOD1 and SOD2 expression measurements in TC in general might be related to the fluctuations in SOD expression during the cell cycle, interindividual differences within the same cancer subtype, differences in multipotent mesenchymal stem/stromal cell (MSC) content in the specimen or variability between malignant potential of the studied tumor tissues. For example, SOD1, SOD2, and SOD3 expression, in contrast to being unaltered or downregulated in tumor thyroid tissue, were observed to be upregulated in PTC MSCs in comparison to normal thyroid MSCs (3, 103). Moreover, Sod2 overexpression in murine follicular adenoma model promoted development of less aggressive tumors, whereas reduction of SOD2 expression was observed in FTCs, as well as RAS-mutated PTCs, and correlated with aggressiveness and poor prognosis in ATCs in humans (104). Similarly, SOD3 (extracellular Cu/Zn-dependent) expression is elevated in benign thyroid tumor model in rats, whereas it was decreased in PTC and ATC tissues. Importantly, lowering of SOD3 expression was shown to be correlated with dedifferentiation (3, 100).

GPXs are selenoproteins reducing H₂O₂ to water, and lipid hydroperoxides to their corresponding alcohols by the expense of converting reduced (GSH) to oxidized glutathione. Oxidized glutathione is reduced back to GSH by glutathione reductase (GSR). Among GPXs, GPX1 is localized in cytosol, mitochondria, as well as peroxisomes, where it controls H₂O₂ levels inhibiting TPO-catalyzed intracellular protein iodination (105). We and others have observed increased GPX1 mRNA and protein levels in the TC tissues, and these observations are corroborated by high-throughput TCGA data (14, 50, 102, 106). However, GPX1 was also described as downregulated in TC (107). Interestingly polymorphism in GPX1, but not in CAT, SOD2, or glutathione S-transferase GSTP1, GSTM1, GSTT1 (involved in GSH-dependent phase II xenobiotic detoxification) genes, is associated with recurrent oncocytic carcinoma of the thyroid (108). GPX2 expression is unaltered in TC (102). GPX3 is the most abundant GPX in the human thyroid and is secreted to follicular space, where it controls H₂O₂ levels. Importantly, induction of TH synthesis by TSH downregulates GPX3 levels in follicular lumen, by retaining it intracellularly, to support the requirement for increased H₂O₂ levels (109). GPX3 expression is decreased in TC and was shown that its silencing is mediated by promoter methylation, where methylation and GPX3 downregulation status is correlated with metastatic potential (102, 110, 111). GPX4 is an intracellular membrane-associated enzyme active on phospholipid hydroperoxides and it is the only GPX embryonically lethal in knockout mouse model. In humans, GPX4 mutations underlie Sedaghatian type spondylometaphyseal dysplasia (SMDS), a genetic disorder manifested by skeletal, heart, and brain abnormalities, as well as premature mortality. Functional GPX4 is essential to control lipid peroxidation and resultant iron-dependent necrotic cell death, ferroptosis. Accordingly, GPX4 inhibitors, as well as siRNA-mediated knockdown of GPX4 expression, were shown to induce ferroptosis and inhibit proliferation, accompanied by increased ROS and DNA damage levels in cultured PTC cells (112). Moreover, GPX4 expression is increased in thyroid cancer tissues and positively correlates with more advanced disease (112). GPX5 and GPX6 are almost undetectable in thyroid (95, 102), whereas GPX7 is unaltered in TC (102). Finally, the analysis of TCGA dataset suggests that GPX8 is expressed in the thyroid but is unaltered in TC (14). It is important to note that GPXs, as all selenoproteins, are regulated on transcriptional, posttranscriptional, translational, and posttranslational levels; therefore, the best estimation of their influence on defense against OS is activity measurement. Accordingly, overall GPX activity in thyroid tissue homogenates seems to be increased in thyroid disease because the majority of studies detected elevation of GPX activity in TC, MNG, and/or GD tissues (51, 52, 113, 114). Moreover, GSH, as well as GSR and glutathione synthetase levels are also increased in TC, but not in goiter or HT tissue samples (98, 114, 115). The importance of glutathione homeostasis in thyroid cell biology is further evidenced by observation that lowering of GSH levels using glutathione synthesis inhibitor L-Buthionine-sulfoximine (BSO) influences TH synthesis by compromising transcriptional activity of thyroglobulin and TPO genes in rat thyroid cells (116). Additionally, dependence of thyroid cell integrity on redox equilibrium maintenance by GSH is exemplified by our studies, where we observed that BSO treatment increased ROS levels, induced DNA damage, and compromised proliferation of normal human SV40-transformed thyroid cells NTHY-ori 3-1 (106). Importantly, GSH-dependent OS defense seems to be crucial in thyrocyte response to H₂O₂, but not to γ-irradiation (43).

PRDXs use active site cysteine residue to reduce H₂O₂ and other peroxides; the thiol-containing PRDX active site is restored by thioredoxin, sulfiredoxin 1 (SRX1), or ascorbate. Apart from being important player in controlling OS through H₂O₂ elimination, PRDXs are involved in preventing OS-induced protein aggregation (molecular chaperone function) and signaling through redox-sensitive pathways (117). Importantly, PRDXs were also implicated in genome maintenance, as evidenced by Prdx1 knockout mice model, where increased tumor incidence is accompanied by elevated genomic 8-oxoG levels, as well as PRDX5 knockdown in human lung adenocarcinoma cells showing increased 8-oxoG formation in DNA (118, 119). Moreover, PRDX1 cooperates with MTH1, pyrophosphohydrolase preventing 8-oxodGTP incorporation during DNA synthesis, in safeguarding telomeres from ROS-induced and PARP1 activity-dependent shortening (120). In FRTL-5 rat thyroid cells Prdx1, but not Prdx2, is stimulated by TSH and H₂O₂ treatment. Moreover, Prdx1 and Prdx2 overexpression protects FRTL-5 cells against H₂O₂-induced apoptosis (121). Prdx3 expression, on the other hand, is stimulated by excessive iodide exposure through enhanced mitochondrial superoxide production in rat thyroids (122). In humans, elevated PRDX1 levels were detected in GD, as well as follicular adenomas, carcinomas, and TC-PTC specimens, but not in HT tissues (99, 123). On the other hand, PRDX2 and PRDX6 are upregulated in cold thyroid nodule tissues, but in follicular adenomas PRDX6 downregulation was observed (124, 125). Moreover, PRDX1-3 and PRDX6 are downregulated in PTCs and ATCs, with even more decreased PRDX1 and PRDX6 levels in more advanced, as well as BRAF-mutated tumors (50, 102, 126). PRDX4 expression is unchanged in PTC specimens but increased in ATC tissues (102). PRDX5 levels, on the other hand, are increased in thyroids of Pendred syndrome patients, as well as in goiter thyroid tissue (90, 127).

TXN/TXNRD system is responsible for reduction of redox-sensitive, thiol-containing signaling proteins and transcription factors, including, among others, PRDXs, phosphatase and tensin homolog (PTEN), ribonucleotide reductase, apurinic/apyrimidinic endodeoxyribonuclease 1 (APE1), ASK1, AP-1, nuclear factor kappa-light-chain-enhancer of activated B cells, p53, NRF2, and hypoxia inducible factor 1 subunit alpha (128, 129). In the context of thyroid physiology, Txn and selenoproteins thioredoxin reductases, cytoplasmic TxnRd1, and mitochondrial TxnRd2, are upregulated by iodide overload-induced OS in rodent thyroids (91, 130). TxnRd1 is required for Ape1 redox activity-dependent upregulation of Pax8-directed Nis expression in response to TSH or TSH+ selenium treatment in rat thyroid cells (131). Murine TxnRd1 and TxnRd2 knockouts are embryonically lethal. In humans, patients with TXNRD2 mutations suffer from familial glucocorticoid deficiency (FGD) and dilated cardiomyopathy (DCM), whereas TXNRD1 and TXN2 mutations predispose to brain abnormalities accompanied by epilepsy (132). TXN is elevated in thyroids of GD patients (133). Moreover, thioredoxin and cytoplasmic/nuclear tioredoxin reductase are elevated in TC, with increasing levels in more in more aggressive cancers (134). Other reports described downregulation of cytoplasmic TXNRD1 and upregulation of mitochondrial TXNRD2 in TC tissues (98, 107).

Another important player of thyroidal antioxidative protection is nuclear factor erythroid 2-related transcription factor 2, NRF2 (also known as nuclear factor erythroid-derived 2-like 2; NFE2L2), the master regulator of oxidative stress responses. NRF2 availability is controlled by Kelch-like ECH-associated protein 1 (KEAP1)-mediated targeting to polyubiquitination and subsequent proteasomal degradation. Under oxidative stress conditions cytoplasmic NRF2-KEAP1 interaction is disrupted leading to increased nuclear NRF2 levels and the resultant activation of multiple antioxidant defense genes, including, among others, GPXs, PRDXs, TXNRDs, as well as GCLC and GCLM, encoding subunits of glutamate-cysteine ligase involved in GSH synthesis. NRF2 function in thyroid physiology and disease was recently reviewed (135); therefore, here we pinpoint only data most relevant in the context of the current manuscript. NRF2/KEAP1 signaling is essential in controlling oxidative stress in thyroid cell, especially in the context of iodide overload. Moreover, NRF2 influences TH synthesis by positive regulation of thyroglobulin expression, as well as lowering of the excessive TG iodination (122, 136). Interestingly, the importance of NRF2/KEAP1 signaling in thyroid homeostasis is further highlighted by the observation that germline null mutations in KEAP1 predispose to the hereditary disease familial nontoxic multinodular goiter without accompanying extra-thyroidal disease (137). These observations were further corroborated in Keap1 knockdown mice, that show diffuse goiter with subclinical hypothyroidism (135). In thyroid carcinoma, NRF2 levels are elevated and positively correlated with more aggressive TC disease, in agreement with generally increased oxidative stress pressure (138). It has been suggested that in thyroid cancer NRF2 protein levels are upregulated due to KEAP1 inactivation through somatic mutations or promoter hypermethylation (138, 139). Another important factor modulating NRF2 levels in TC might be the regulation by microRNA network and, for example, involvement of miR-200a in KEAP1 expression regulation has been suggested (135).

To sum up OS prevention in thyroid physiology and disease, it has to be mentioned that in general thyrocytes have extremely robust antioxidative defense system, in which selenoproteins play major role (45). However, importantly, selenoproteins are not essential for thyroid function and other antioxidative stress defense enzymes are sufficient to protect thyrocyte integrity against OS-induced damage, as evidenced by murine model with thyroid-specific selenoprotein synthesis defect (140).

The Repair of DNA Damage in the Thyroid

Genomic and mitochondrial DNA is constantly exposed to endogenous and exogenous damaging factors that compromise its integrity. It has been estimated that each human cell experiences about 100 000 DNA lesions every day (141). These lesions must be repaired to avoid deleterious consequences, such as mutagenesis, chromosomal aberrations, blocking of DNA replication, or transcription, that together compromise cell integrity and survival. Loss of genomic integrity is implicated in multiple human diseases, including carcinogenesis and neurodegeneration, as well as aging (54, 142-144). Detailed description of the DNA damage induced by endogenous and exogenous sources is shown in Table 1.

Table 1.

DNA damage induced by endogenous and exogenous insults (for further details, see description in the main manuscript^a)

DNA damage type	Examples of lesionsb	Damage formation/inducing agentsb	Major DNA repair pathwaysc
Replication errors (base mispairs, insertion/deletion loops)	Mismatched bases (eg, G·T, A·C mispairs); small insertion/deletion loops (eg, extrahelical bases from strand slippage)	DNA polymerase errors during replication that escape proofreading. They can result from dNTP pool imbalance or misincorporation of modified dNTPs or NTPs	MMR
Depurination/depyrimidination (Abasic sites)	Apurinic/apyrimidinic sites (AP-sites; abasic sites)-loss of DNA base	Endogenous: spontaneous cleavage of the N-glycosidic bond; BER intermediates. Exogenous: ionizing radiation; chemicals generating rapidly depurinating lesions that destabilize N-glycosidic bond (eg, N7-meG, aflatoxin B1-N7-guanine; 8-nitroguanine)	BER, SSBR, TLS
Base deamination (loss of exocyclic –NH2)	Cytosine → Uracil; 5-methylcytosine → Thymine; Adenine → Hypoxanthine; Guanine → Xanthine. Deamination changes base-pairing properties leading to point mutations and/or mismatches	Endogenous: spontaneous hydrolytic deamination, AID/APOBEC-mediated deamination Exogenous: nitrosative stress (from preservatives or inflammation)	BER, MMR
Oxidative base modifications (ROS damage)	2-oxoA; 5-OHC; 5OHU; 8-oxoA; 8-oxoG; cyPu-cdA and cdG; FaPyA, FaPyG; Tg	Endogenous: reactive oxygen species (ROS) from cellular metabolism (mitochondrial electron transport, peroxisomes) or inflammation Exogenous: ionizing radiation; environmental oxidants (eg, from tobacco smoke) and transition metals (via Fenton chemistry)	BER, MMR, NER
Alkylation damage (methyl/ethyl adducts to bases)	1-meA; 3-meC; N7-meG; O4-meT; O6-meG	Endogenous: S-adenosylmethionine (SAM)-induced Exogenous: alkylating agents—eg, N-nitrosamines (tobacco smoke and cured meats), N-methyl-N-nitrosourea (MNU), methyl methanesulfonate (MMS), ethyl methanesulfonate EMS, chemotherapeutics (eg, temozolomide)	Direct reversal (MGMT, ALKBH2, ALKBH3), BER, MMR
Lesions induced by lipid peroxidation	εA; εC; εG; other etheno; ethano; propano; MDA- and HNE-derived DNA adducts	Endogenous: by-products of lipid peroxidation (due to ROS)—MDA, HNE, acrolein, crotonaldehyde Exogenous: vinyl chloride and ethyl carbamate (industrial exposition), cigarette smoke, acetaldehyde (ethanol metabolite)	BER, NER, Direct reversal (ALKBH2, ALKBH3)
Bulky DNA adducts	Polycyclic aromatic hydrocarbon (PAH) adducts: eg, (+)-anti-benzo[a]pyrene diol epoxide adduct to N2 of guanine (BPDE-dG adduct); Aflatoxin B1-N7-guanine adduct	Exogenous: benzo[a]pyrene (cigarette smoke, charred foods), aflatoxin B1 (from moldy food)	NER, TLS, FA
UV-induced lesions (ultraviolet light photoproducts)	Cyclobutane pyrimidine dimer (CPD); 6-4 photoproduct (6-4PP)	Exogenous: ultraviolet radiation	NER, TLS
DNA single-strand breaks	Single-strand break (SSB): break in the backbone of one of the DNA strands	Endogenous: spontaneous hydrolysis, repair intermediates, spontaneous decay of AP-sites, endogenous ROS/RNS, replication stress Exogenous: ionizing radiation (directly or via ROS), UV (through ROS), cigarette smoke or environmental pollutants (through ROS/RNS), chemotherapeutics (eg, bleomycin, camptothecin)	SSBR
DNA double-strand breaks	Double-strand break (DSB): breaks in both DNA strands, either directly opposite or staggered, separating the DNA into 2 pieces	Endogenous: repair intermediates, replication stress (stalled or collapsed replication forks), endogenous ROS/RNS, programmed DSBs during antibody diversification, Exogenous: ionizing radiation (directly or via ROS), chemotherapeutics (eg, bleomycin, PARP inhibitors), mechanical stresses on chromosomes	NHEJ, HR, Alt-NHEJ, SSA, BIR
DNA crosslinks	DNA interstrand crosslinks (ICLs), intrastrand crosslinks (eg, CPD), DNA-protein crosslinks (eg, topoisomerase-DNA crosslinks)	Endogenous: reactive metabolites (aldehydes, lipid peroxidation products), ROS/RNS, intermediates of topoisomerase activity Exogenous: ionizing radiation, UV, mitomycin C, cisplatin, camptothecin and etoposide (topoisomerase inhibitors), industrial pollutants (eg, epoxides)	DNA crosslinks: FA/HR, NER DNA-protein crosslinks: protein degradation (eg, SPRTN, proteasome) and crosslink removal (eg, TDP1/TDP2, NER), followed by SSBR and/or TLS

^a The table lists the major, or most studied, lesions, sources, and repair pathways but is not meant to be exhaustive.

^b 1-meA, N¹-methyladenine; 2-oxoA, 2-oxoadenine/isoguanine (iG); 3-meC, N³-methylcytosine; 5-OHC, 5-hydroxycytosine; 5-OHU, 5-hydroxyuracil; 8-oxoA, 8-oxo-7,8-dihydroadenine (8-oxoadenine); 8-oxoG, 8-oxo-7,8-dihydroguanine (8-oxoguanine); cdA, 8,5′-cyclo-2′-deoxyadenosine; cdG, 8,5′-cyclo-2′-deoxyguanosine; cyPu, 8,5-cyclopurine; DSB, Double-Strand Break; εA, 1,N⁶-ethenoadenine; εC, 3,N⁴-ethenocytosine; εG, N²,3-ethenoguanine; FaPyA, 4,6-diamino-5-formamidopyrimidine; FaPyG, 2,6-diamino-4-hydroxy-5-formamidopyrimidine; HNE, trans-4-hydroxy-2-nonenal; MDA, malondialdehyde; N⁷-meG, N⁷-methylguanine; O⁴-meT, O⁴-methylthymine; O⁶-meG, O⁶-methylguanine; RNS, Reactive Nitrogen Species; ROS, Reactive Oxygen Species; SSB, Single-Strand Break; Tg, thymine glycol; UV, Ultraviolet Radiation.

^c ALKBH2, Alkylation Repair Homologue 2; ALKBH3, Alkylation Repair Homologue 3; Alt-NHEJ, Alternative Non-Homologous End-Joining; BER, Base Excision repair; BIR, Break-Induced Replication; FA, Fanconi Anemia pathway; MGMT, O⁶-Alkylguanine-DNA Alkyltransferase; HR, Homologous Recombination; MMR, Mismatch Repair; NHEJ, Non-Homologous End-Joining; NER, Nucleotide Excision Repair; SSA, Single-Strand Annealing; SSBR, Single Strand Break Repair; SPTN, DNA-Dependent Metalloendopeptidase Spartan; TDP1, Tyrosyl DNA Phosphodiesterase-1; TDP2, Tyrosyl DNA Phosphodiesterase-2; TLS, Translesion Synthesis.

To avoid detrimental consequences of DNA damage, cells have evolved complex system to detect, repair and/or tolerate, as well as respond to and signal DNA lesions. Historically, it was assumed that a given type of DNA damage is repaired exclusively by a dedicated DNA repair pathway. However, newer evidence suggests that a given DNA lesion might be processed differently depending on which pathways are activated or suppressed and that repair proteins can be shared between the pathways. Next, we briefly describe the major DNA repair pathways in the context of thyroid physiology and disease. However, it has to be mentioned that this subject covers a much broader scope of knowledge than cannot be presented here due to limited space. More comprehensive literature review concerning DNA repair in the thyroid tissue is presented in Table S2 (13).

Direct Reversal of the DNA Damage

One of the simplest ways to correct DNA lesions is through direct reversal, which in humans is mainly relevant for repair of alkylation damage. First, O⁶-alkylguanine-DNA alkyltransferase (AGT/MGMT) reverses O⁶-alkylated guanines, such as O⁶-methylguanine (O⁶-meG), but also longer alkyl and cyclic adducts, such as O⁶-benzylguanine, and even O⁶-G-alkyl-O⁶-G interstrand crosslinks, through transferring the O⁶-linked group to the catalytic site cysteine residue, which leads to irreversible enzyme inactivation. O⁶-meG is not able to block DNA replication, but if left unrepaired, it is mutagenic, leading to G:C → A:T transitions (145). MGMT activity is particularly important in the context of cancer treatment, since O⁶-meG is the major DNA lesion responsible for the cytotoxicity of alkylating chemotherapeutics, such as temozolomide (TMZ). MGMT promoter methylation status is an important determinant for response to TMZ therapy (146). Although alkylating agents are usually not the first-line treatment in TC, TMZ or dacarbazine are sometimes used alone, or in combination, in MTC or metastatic WDTC disease (147). MGMT promoter methylation, and consequently MGMT downregulation observed in TC tissue, is predictive of poor O⁶-meG removal and good therapeutic response (148, 149). However, the potential utility of alkylating agents in TC management warrants evaluation since thyroid cancer cell line TPC-1 appeared to be unresponsive to TMZ treatment (150). Additionally, MGMT SNPs modify WDTC risk after radiation exposure (151, 152).

Another type of direct reversal repair is oxidative demethylation catalyzed by α-ketoglutarate- and Fe²⁺-dependent dioxygenases from the AlkB family, leading to release of methyl group in the form of formaldehyde. The human genome encodes 9 AlkB family proteins, namely alkylation repair homologue 1 to 8 (ALKBH1-8) and fat mass and obesity associated (FTO). Among the human ALKBH proteins only ALKBH2 and ALKBH3 repair DNA alkylation damage, including the simple methylation lesions N¹-methyladenine and N³-methylcytosine, as well as ethyl, propyl, and etheno adducts. ALKBH2 preferably repairs double-stranded DNA (dsDNA), whereas ALKBH3 targets single-stranded DNA (ssDNA) and RNA. The other ALKBHs primarily act on RNA substrates and function in epitranscriptomic regulation (153). In the context of thyroid, several ALKBH3 SNPs are associated with increased TC risk, especially after diagnostic irradiation (151). However, in knockout murine models ALKBHs, deficiency is not inducing thyroid-associated phenotype.

Base Excision Repair and Single-strand Break Repair

Simple, nonbulky DNA base damages that do not significantly distort the DNA helix structure, such as oxidative, alkylation, and etheno lesions, as well as deamination, AP sites and certain base-base mismatches are primarily repaired by base excision repair (BER). The latter (see Fig. 2A) is initiated by N-DNA glycosylases that are monofunctional or bifunctional enzymes with activity toward specific subsets of lesions. Monofunctional N-DNA glycosylases remove damaged bases by hydrolyzing the N-glycosidic bond, leaving intact AP site. They include uracil-DNA glycosylase (mitochondrial UNG1 and nuclear UNG2 isoforms), single-strand selective monofunctional uracil-DNA glycosylase 1 (SMUG1), thymine-DNA glycosylase (TDG), methyl-CpG-domain protein 4 (MBD4), alkyladenine-DNA glycosylase (AAG/MPG/ANPG), and adenine-DNA glycosylase (MUTYH). Monofunctional N-DNA glycosylases process mainly deamination and alkylation base lesions, mismatches, and ethenoadducts, but also oxidative DNA damage. Bifunctional enzymes, possess an additional β-lyase activity that enables incision of the phosphodiester backbone at the 3′-side (leaving SSB with 5′-phosphate and 3′-phospho-α, β-unsaturated aldehyde [3′-PUA] termini) or a β, δ-lyase activity that generates 5′-phosphate and 3′-phosphate ends. Bifunctional N-DNA glycosylases repair oxidized or ring-opened purine lesions, where 8-oxoguanine glycosylase 1 (OGG1), endonuclease III homologue 1 (NTH1/NTHL1), and Nei endonuclease VIII-like glycosylase 3 (NEIL3) show β-lyase activity, whereas NEIL1 and NEIL2 carry out β, δ-elimination reaction. Intact AP sites (generated by monofunctional N-glycosylase or spontaneous/induced depurinations or depyrimidinations) are cleaved by AP endonuclease (APE1, known also as APEX, HAP1, or Ref-1) generating 3′-OH and 5′-deoxyribose phosphate (5′-dRp) termini. APE1 also exhibits 3′-phosphodiesterase activity removing 3′-PUA ends from β-elimination intermediates. APE1, an enzyme essential for embryonic development in mice, has a separate domain involved in redox signaling (hence its alternative name “redox effector factor-1”; Ref-1) stimulating DNA-binding activity of several transcription factors, which makes it a possible target for therapeutic intervention in multiple disease states. 3′-Phosphates generated by NEIL1 or 2 are converted to 3′-OH ends by polynucleotide kinase/phosphatase (PNKP) enabling APE1-independent BER. PNKP, which possesses DNA 3′-phosphatase and DNA 5′-kinase activity, is involved in cleaning both single-stranded and double-stranded DNA breaks. Mutations in PNKP gene lead to several hereditary neurological disorders with symptoms ranging from dystonia, ataxia, and peripheral neuropathy with or without cognitive impairment to neurodevelopmental delay with microcephaly and epileptic encephalopathy. The subsequent BER steps, also collectively termed single-strand break repair (SSBR), may involve trimming of blocked SSB ends, such as aprataxin (APTX)-mediated removal of 5′-AMP terminus (erroneously generated by DNA ligase) or tyrosyl DNA phosphodiesterase-1 (TDP1)- or -2 (TDP2)-mediated removal of 3′-phosphotyrosyl terminus (erroneously generated by DNA topoisomerase 1 [TOP1] or 2 [TOP2]). Defects in APTX, TDP1, and TDP2 also underly neurological disorders characterized by spinocerebellar ataxia, axonal/peripheral neuropathy, oculomotor apraxia, myopathy, and other symptoms. DNA polymerase beta (Pol β; POLB) extends the clean 3′-OH ends by reinserting 1 nucleotide and, after removal of the 5′-dRp-residue by POLB 5′-dRP-lyase activity, DNA ligase III alpha (LIG3α) seals the nick to complete the so-called short-patch BER (SP-BER) pathway. The scaffolding protein X-ray repair cross-complementing protein 1 (XRCC1) is an important player in SP-BER by serving as a platform to assemble the multiprotein repair complex. XRCC1 interacts with, stimulates the activity and maintains stability of, several N-DNA glycosylases, APE1, PNKP, APTX, LIG3α, POLB, the SSB sensor proteins poly(ADP-ribose) polymerase 1 and 2 (PARP1 and PARP2). PARPs decorate SSBs by poly-ADP-ribosylating themselves and/or other acceptor proteins (eg, histones) while being bound to the SSB thus influencing chromatin remodeling, transcription, and recruitment of the XRCC1 repair complex. A major consequence of PARP deficiency is therefore failure to repair genotoxic SSBs. Persistent SSB may be converted to DSBs upon replication that are substrates for homologous recombination (HR). However, upon HR deficiency, such as biallelic BRCA1 loss, PARP inhibition leads to its retention at damaged DNA, which delays processing of the SSBs and results in cell death. This phenomenon is termed synthetic lethality and constitutes the basis for therapeutic effectiveness of PARP inhibitors, such as Olaparib, in biallelic BRAC1-mutated breast cancer tumors. BER may also proceed through long-patch subpathway (LP-BER), where 2 to up to 30 nucleotides are replaced. The factors determining the choice of BER subpathway remain somewhat elusive, but LP-BER seems to be required when the AP-site is oxidized or reduced. LP-BER proceeds through dissociation of POLB, recruitment of replication sliding clamp proliferating cell nuclear antigen (PCNA), replication factor C (RFC), and DNA polymerase delta or epsilon (Pol δ or Pol ε; POLD [heterotetrameric complex composed of catalytic POLD1 and accessory POLD2, POLD3, and POLD4 subunits] or POLE [composed of catalytic POLE and accessory POLE2, POLE3, and POLE4 subunits]) followed by repair synthesis that displaces the 5′ DNA end forming a “flap” structure. Next, the 5′flap is removed by flap endonuclease 1 (FEN1) and the DNA ends are sealed by ligase 1 (LIG1) (142-144, 154). Replication protein A (RPA) is another factor described to be involved in BER. RPA binds to ssDNA stretches to protect them from unscheduled nuclease attack, to coordinate assembly of repair complexes (including its role in HR, see the following section) and recruitment of DNA polymerases, as well as activate ATR kinase signaling (see the following section). RPA is heterotrimeric complex composed of RPA70 (RPA1), RPA32 (RPA2), and RPA14 (RPA3) subunits. In BER, the role of RPA is not well documented, but at least it supports DNA polymerase activity, and is important for the UNG glycosylase function in class switch recombination and pre-replicative uracil repair (155).

Figure 2.

Schematic representation of excision DNA repair pathways. (A) Repair of base damages and single strand brakes by short-patch base excision repair (SP-BER) and long-patch base excision repair (LP-BER). (B) Repair of bulky lesions and crosslinks by global genome nucleotide excision repair (GG-NER) and transcription coupled nucleotide excision repair (TC-NER). (C) Repair of base mismatches by mismatch repair (MMR). For detailed description, please refer to the text. Created in BioRender. Arczewska, K. (2025) https://BioRender.com/x40n381.

The UNG uracil-DNA glycosylase is the major glycosylase responsible for uracil removal from DNA. It plays pivotal role in antibody maturation, where it initiates processing of U:G pairs generated by activation-induced cytidine deaminase (AID) in activated B lymphocytes. Consequently, UNG mutations predispose to immunodeficiency with hyper-IgM syndrome, type 5, whereas AID defects lead to hyper-IgM syndrome, type 2. AID mutation carriers are prone to develop autoimmunity, but without thyroid involvement. UNG deficiency, on the other hand, is not considered to predispose to autoimmune disease (156). However, UNG defects may also contribute to, or modulate, autoimmunity, as evidenced by increased susceptibility to rheumatoid arthritis in human UNG SNPs carriers, but its role in AITD remains to be revealed (157). Analysis of TCGA datasets suggests downregulation of UNG in PTC tissues (Fig. S1 (13)), which may underlie high frequency of C → T mutations. TDG is another uracil-DNA glycosylase studied in the context of thyroid. Interestingly, TDG, in addition to BER of U:G, 3,N⁴-ethenocytosine (εC), several thymine mismatches, 5-fluorouracil and other lesions, is also involved in active demethylation of the 5-meC epigenetic mark. Active 5-meC demethylation proceeds through ten eleven translocation (TET) dioxygenase-mediated oxidation to 5-hmC, and next to 5-formylcytosine and 5-carboxylcytosine that are next excised by TDG-initiated BER. TDG function is essential for development and is the only N-DNA glycosylase showing embryonic lethality in knockout mouse models (158). TDG interacts with and modulates activity of several transcription factors. The latter include thyroid transcription factor-1 (TTF-1) of which transcriptional activity is repressed by TDG in thyroid and other cell types (159). SMUG1 is another uracil-DNA glycosylase that apart from being able to initiate BER on DNA lesions such as uracil, oxidized uracil residues (eg, 5-hydroxymethyluracil or 5-hydroxyuracil), other oxidized pyrimidines, and εC, was discovered to have additional activity. SMUG1 (similarly to TDG, MBD4, NTH1, and NEIL1) performs active 5-meC demethylation through activity on 5-hydroxymethyluracil generated at CpG sites (158, 160). Moreover, SMUG1 is involved in rRNA quality control mechanisms (161-163). Database search identified SMUG1 overexpression in thyroid cancer tissue and suggested that SMUG1 transcript RNA stability is regulated by N⁶-methyladenosine (N⁶-mA) modification. However experimental data failed to confirm these assumptions (164). MBD4 removes U, T, and oxidized U residues mismatched with guanine at CpG sites. MBD4 was also implicated in active 5-meC demethylation through its weak ability to initiate BER of 5-hydroxymethyluracil at CpG sites. MBD4 is unique among the DNA glycosylases because it has 2 DNA-binding domains: the C-terminal glycosylase domain and the N-terminal methyl-CpG-binding domain (158). In the context of the thyroid, Mbd4 is overexpressed in the livers of male rats after induction of hypothyroidism by 6-n-propyl-2-thiouracil (165). Moreover, MBD4 is upregulated, whereas TDG is downregulated in PTC tissues (Fig. S1 (13)).

Another glycosylase that is of interest in the context of thyroid is MUTYH. In contrast to other BER enzymes, it does not remove damaged DNA bases but rather excises unmodified adenine from A:8-oxoG (and also A:G) pairs, but also, with much lower efficiency, 2-oxoA paired with G. Thyroid cells upregulate MUTYH, as well as OGG1 and APE1, expression under OS conditions dependent on MAPK and AKT signaling (166). Germline MUTYH mutations are identified that predispose to MUTYH-associated polyposis (MAP) syndrome (also termed familial adenomatous polyposis 2; FAP2), an autosomal recessive disease manifested by colorectal adenomas and increased risk of colorectal cancer. Carriers of MUTYH mutation have increased levels of G:C → T:A transversions in the Adenomatous polyposis coli gene that elevate risk of colon carcinogenesis. MAP patients are also prone to develop thyroid nodules that may potentially increase PTC risk, but TC is rare (167, 168).

The major DNA glycosylase initiating BER of oxidized purines is OGG1, that excises 8-oxoG, FaPyG (2,6-diamino-4-hydroxy-5-formamidopyrimidine) and methyl-FaPyG, as well as 8-oxoadenine paired with C. 8-oxoG, which is considered one of the most abundant ROS-induced DNA lesions and therefore used as an OS biomarker, forms a base pair not only with correct cytosine, but also with adenine leading to G:C → T:A transversions. Importantly, apart from its promutagenic potential, 8-oxoG is also involved in transcription regulation, where passive or active 8-oxoG generation and subsequent OGG1 binding might either repress or activate transcription of target genes (158, 169). In the context of the thyroid, the most frequent OGG1 polymorphism is Ser326Cys, which disturbs OGG1 activity and correlates with GD risk, but not TC risk (83, 170). Interestingly, it was reported that OGG1 expression is downregulated in multiple cancer tissues, including FTC. The key mechanism leading to OGG1 downregulation in cancer tissue is loss of heterozygosity (LOH), which is frequently observed in the genomic region coding OGG1 gene. Accordingly, increased OGG1 LOH frequency was observed in PTC and HT tissues (171). However, recent research described OGG1 upregulation in TC in general (50). Additionally, OGG1 and MUTYH, as well as APE1, expression is upregulated by short-term OS in nonproliferating thyroid cells (166).

NTHL1 is a broad substrate specificity enzyme that repairs oxidized pyrimidines, mainly thymine glycol (Tg), 5-hydroxycytosine, 5-hydroxyuracil. The truncating NTHL1 mutation p.Q90* predisposes to colorectal polyposis syndrome associated with colon cancer, known as NTHL1-associated polyposis (NAP) (also termed familial adenomatous polyposis 3; FAP3). NTHL1 mutations predispose to carcinogenesis also in multiple other organs, including rare incidence of TC (172).

Among the downstream BER factors, APE1 was the most frequently studied in the thyroid. APE1 is overexpressed and/or relocalized to nucleus in response to TSH or OS in rat thyroid cells (173). Moreover, upon TSH stimulation, the APE1 redox function modulates transcriptional activity of thyroid specific transcription factors, PAX8 and TTF-1, leading to upregulation of Nis expression in rat thyroid cells (131, 174). Interestingly, APE1 is upregulated in TC, with highest expression detected in BRAF-mutated cancer tissues. Moreover, inhibition of the APE1 redox domain overcomes resistance to the BRAF^V600E inhibitor vemurafenib in BRAF-mutated human thyroid tumor cells in murine xenograft models (175). Application of APE1 inhibition might be beneficial in TC with PTEN mutation, similarly to observations from PTEN-mutated melanoma cells (176). Moreover, inhibition of APE1-mediated SSBR through blocking AP sites induces DNA damage and cell death in cultured ATC cells, as well as compromises ATC tumor growth in murine xenograft model (115). Finally, APE1 SNPs are not associated with TC risk (for detailed discussion see (83)). High-throughput data suggest upregulation of APE1, together with 2 SSBR end-processors TDP1 and PNKP, in HT tissues (77). Moreover, PNKP is upregulated, whereas TDP1 is downregulated in PTC tissues (Fig. S1 (13)).

Concerning the other downstream BER factors, POLB expression is stimulated by TSH and elevated in GD but reduced in TC (177). On the other hand, FTC tissues show elevated POLB expression (178). Elevated POLB levels in GD tissue might suggest higher mutability, since POLB lacks proofreading activity, and therefore shows lower accuracy than replicative polymerases. Interestingly, POLB preferentially incorporates 8-oxodGTP opposite template A, resulting in blocked repair intermediates that cannot be ligated by DNA ligase leading to DNA breakage and consequent cytotoxicity (179). These observations would be worth exploring in the context of GD tissue, where a high level of OS, accompanied by elevated POLB expression, would be expected to lead to accumulation of 8-oxodGTP-induced DNA damage. Regarding XRCC1, multiple studies analyzed associations of SNPs with TC but gave conflicting results (for detailed discussion see (83)). Additionally, XRCC1 is upregulated in PTC tissues (Fig. S1 (13)).

Several studies addressed the role of PARP1 in TC, demonstrating the associations of PARP1 SNPs with TC risk (for detailed discussion see (83)). Previously, PARP1 inhibition was suggested to be beneficial in TC management, since the PARP1 inhibitor PJ34 upregulated NIS expression accompanied by increased radioiodine accumulation in thyroid cancer cells (180). Moreover, Niraparib, another PARP1 inhibitor, was shown to induce DNA damage, disturb cell cycle, compromise translation and survival of ATC and PTC cell lines, as well as inhibit Cal-62 anaplastic TC cell xenograft growth in a murine model (20)

PCNA is a ring-shaped homotrimeric sliding clamp that acts as a processivity factor in DNA replication during S phase and in DNA repair. In BER it is required for repair synthesis by POLB or POLD/E. Defects in PCNA gene predispose to ataxia-telangiectasia-like disorder-2 (ALTD2), a genetic disease with clinical features resembling those observed in Ataxia-telangiectasia (A-T) patients (ie, cerebellar ataxia, immune suppression, premature aging, and telangiectasias) (181). Due to its role in proliferation, high PCNA expression is inevitably connected with carcinogenesis and cancer progression, as observed in multiple solid cancers (182). The majority of the PCNA-interacting proteins bind to PCNA through short peptide sequence known as PIP (PCNA-interacting protein)-box. Interestingly, T₃ interacts with the PIP-box peptide binding site on PCNA. T2AA, a T₃-derived PCNA inhibitor, inhibits DNA replication and induces DNA damage, providing its potential utility in cancer therapy (183). T₃ itself was shown to either stimulate or inhibit proliferation of various cell types, therefore, it cannot be excluded that in certain cellular backgrounds T₃-PCNA interaction can result in inhibition of DNA replication (184). In accordance with elevated proliferation status, PCNA is overexpressed in TC tissues, positively correlates with tumor progression and is upregulated in BRAF^V600E-positive PTC tissues, whereas lower PCNA levels correlate with better prognosis (185). PCNA is loaded on DNA by Replication Factor C (RFC), a protein complex containing 5 subunits RFC1-5 that belong to AAA+ family ATPases. RFCs are frequently overexpressed in cancer tissues and support malignant traits (186). Among analyzed RFC subunits SNPs RFC1 rs1051266 frequency is enriched in TC cases (187). Moreover, high-throughput data suggest RFC4 upregulation in HT tissues (77).

Other BER-related proteins studied in TC include replicative polymerases, FEN1 and LIG1. The replicative polymerases involved in LP-BER include POLD and POLE. Catalytic subunits of POLE and POLD are in general downregulated in PTC tissue, while POLD1 low expression associates with poor prognosis for patients with PTC (188) (Fig. S1 (13)). Moreover, deficiency in the accessory POLE subunit (ie, POLE2) leads to hypothyroidism (189). Mutations in the POLD1 were observed in both benign thyroid goiter and PTC tissues, suggestive of their involvement in progression of the disease from pre-neoplastic lesions to malignancy (16). FEN1 is frequently upregulated in various cancer types and is associated with unfavorable prognosis and therapy resistance. In line with these findings, FEN1 is overexpressed in PTC and ATC tissues (190). In the context of RPA, SNPs in RPA3 subunit modify WDTC risk (152). Finally, SNPs in LIG1 modify TC risk after diagnostic irradiation, whereas LIG1 expression is upregulated and LIG3 expression is downregulated in PTC tissues (150) (Fig. S1 (13)) (151).

Nucleotide Pool Damage Sanitization

Although nucleotide pool damage removal does not constitute BER-type repair, these 2 types of DNA damage protection mechanisms are frequently considered as complementary systems because they process the same type of base damages and therefore act in functional cooperation. The prototype of nucleotide pool damage sanitization enzyme is E. coli MutT, which dephosphorylates 8-oxodGTP to monophosphate and thereby preventing A:T → CG transversions (191). The E. coli mutT mutator strain shows greatly increased level of spontaneous mutations, whereas in animal models such as mice or C. elegans, mutation frequency is only slightly elevated (192, 193). Human MutT homologue 1 (MTH1/nudix hydrolase 1, NUDT1) belongs to the NUDIX hydrolase superfamily of proteins that hydrolyze nucleoside diphosphates linked to x (any moiety) and in humans includes homologues NUDT1-NUDT22 (194). MTH1 is also active on oxidized dATPs, 8-oxo-7,8-dihydro-2′-deoxyadenosine triphosphate (8-oxodATP) and 2-oxo-2′-deoxyadenosine triphosphate (2-oxo-dATP), as well as the corresponding oxidized ribonucleotides, thereby not only preventing oxidative DNA damage, but also transcriptional errors. Furthermore, MTH1 sanitizes N6-methyl-dATP and O6-methyl-dGTP, offering the possibility that it is involved in epigenetic regulation and temozolomide sensitivity, respectively (194, 195). MTH1 is overexpressed under conditions of oxidative stress and inflammatory diseases, as well as, most notably, in cancer tissues. Furthermore, MTH1 supports cancerous transformation, whereas untransformed cells are insensitive to MTH1 deficiency (195, 196). In line with these observations, MTH1 knockout protects OGG1-deficient mice from lung carcinogenesis (197). These findings led to the development of MTH1 inhibitors (MTH1i), such as Karonudib, which is currently being evaluated in phase I clinical studies as a general anticancer agent for management of solid and hematological cancers (NCT03036228 and NCT04077307). Interestingly, Karonudib and other cancer cell-killing MTH1 inhibitors seem to act not only through inhibiting MTH1 8-oxodGTPase activity, but also through perturbing mitotic cell divisions by interfering with MTH1-tubulin protein interaction (196). In agreement with these observations, we detected increased MTH1 level or activity in non-small cell lung cancer, as well as thyroid cancer tissues (106, 198). Moreover, we also found that MTH1 deficiency in TC cells induces DNA damage and suppresses the malignant phenotype (199). Our observation that thyroid cells’ sensitivity to MTH1 deficiency greatly depends on the intracellular glutathione pool suggests an interesting possibility of combination treatment, where MTH1i are supplemented with glutathione pool-depleting agents (106). Interestingly, MTH1i were also suggested in management of autoimmune diseases, due to the selective eradication of MTH1-overexpressing subset of activated T cells in murine models of autoimmune hepatitis and encephalomyelitis (200, 201). However, MTH1i has not been tested in AITD. Conversely, MTH1 is downregulated in high iodide-induced murine HT model (202).

NUDT15 (aka MTH2) is a NUDIX hydrolase with 8-oxodGTPase activity but is not considered as a functional 8-oxodGTPase in vivo because it is much more active on unmodified dNTPs (ie, dGTP, dUTP, and dTTP) (203). However, NUDT15 is active on the thiopurine metabolites 6-thioguanosine triphosphate and 6-thio-deoxyguanosine triphosphate (204). Thiopurines are purine antimetabolites used in management of childhood leukemia, Crohn disease, rheumatoid arthritis, autoimmune hepatitis, patients undergoing organ transplantation, but also in combination with glucocorticoids as second-line treatment in Graves’ orbitopathy (205). NUDT15 polymorphisms result in overt toxicity or therapy resistance. Therefore, analysis of NUDT15 status is currently suggested before thiopurine therapy application (206).

NUDT5 and NUDT18 (aka MTH3) are other NUDIX hydrolases that prevent mutations induced by oxidized nucleotides by dephosphorylating oxidized ribo- and deoxyribonucleoside diphosphates (194). This type of nucleotide pool sanitization is important because 8-oxodGDP inhibits MTH1 8-oxodGTPase activity (207). Importantly, however, the biological role of NUDT5, NUDT15, and NUDT18 in prevention of mutagenesis resulting from oxidative damage to the nucleotide pool warrants further investigation (194, 195). Among these NUDIX proteins, only NUDT5 was studied in the context of thyroid and is overexpressed in HT tissue (208). Other NUDIX proteins analyzed with regard to thyroid are NUDT6 and NUDT16. NUDT6, of which substrate is currently unknown, encodes the antisense transcript of fibroblast growth factor 2 (FGF2) and was proposed to negatively regulate FGF2 expression (194). As we observed, NUDT6 is upregulated in TC cells after siRNA-mediated silencing of the lymphangiogenic factor Prospero homeobox 1 (PROX1) and this is accompanied by stimulation of angiogenesis (209). NUDT16, on the other hand, has multiple enzymatic activities, including removal of RNA caps, hydrolysis of ribo- and deoxyriboinosine di- and triphosphates and removal of ADP-ribose moieties from proteins (210). Inter alia, NUDT16 is involved in DNA repair through de–ADP-ribosylating TP53BP1 and thereby protecting it from degradation (211). Elevated NUDT16 expression in TC associates with poor prognosis (212). TCGA dataset suggests slight upregulation NUDT2, NUDT3, NUDT5, NUDT8, NUDT11, NUDT14, NUDT19, and NUD22 expression, whereas NUDT4, NUDT13, NUDT20 (DCP2), and NUDT21 are downregulated in TC (194).

Other enzymes involved in nucleotide pool sanitation that deserve attention in the context of the thyroid are dUTPase (DUT) and SAM And HD Domain Containing Deoxynucleoside Triphosphate Triphosphohydrolase 1 (SAMHD1). DUT converts dUTP to dUMP, which constitutes one of the steps in dTTP synthesis. DUT activity minimizes genomic uracil incorporation. DUT is essential for cell survival because its deficiency induces DNA breakage through UNG-initiated BER of uracil residues incorporated from the nucleotide pool (213, 214). DUT has also gained attention in the context of cancer therapy, since its inhibitors potentiate anticancer activity of fluoropyrimidine antimetabolites (215). Although application of fluoropyrimidine 5-fluorouracil in combination with other chemotherapeutics was suggested in management of metastatic MTC disease, there was no improvement in clinical outcome (216). DUT expression is downregulated in PTC (Fig. S1 (13)).

SAMHD1 has nucleotide triphosphohydrolase activity able to convert modified and unmodified deoxyribo- and ribonucleoside triphosphates to their corresponding nucleosides thereby being important factor in sensitivity to antimetabolite-based therapies. Defects in SAMHD1 lead to Aicardi-Goutieres syndrome 5, a rare genetic disease characterized by encephalopathy and elevated interferon signaling, which underlies systemic lupus erythematosus induction. Some patients with Aicardi-Goutieres syndrome are afflicted with hypothyroidism (217). In the context of TC, SAMHD1 is unchanged in TC tissues, but its high expression levels are correlated with favorable prognosis, and approximately 4% of thyroid cancer cases carry SAMHD1 mutations, thus potentially contributing to therapeutic intervention response (218).

Nucleotide Excision Repair (NER)

Lesions that distort the DNA double helix include UV-induced damage cyclobutane pyrimidine dimers (CPDs) and 6-4 photoproducts (6-4PP), bulky chemical adducts generated by carcinogens or chemotherapeutic agents, intrastrand crosslinks (eg, induced by cisplatin); as well as 8,5-cyclopurines (cyPus). They are in general removed by the Nucleotide Excision Repair (NER) mechanisms that include transcription-coupled repair (TC-NER), and the global genome repair (GG-NER). GG-NER and TC-NER subpathways share common lesion excision, DNA gap filling and ligation steps, but they use different damage recognition mechanisms (see Fig. 2B). In GG-NER, a complex formed by XPC, RAD23B (UV excision repair protein Radiation sensitive 23B) and CETN2 (Centrin 2) probes DNA for helix distorting lesions; however, it is not able to recognize CPDs. Instead, these damages are recognized by the DDB1-DDB2/XPE heterodimer, which facilitates recruitment of the trimeric XPC-RAD23B-CETN2 complex. TC-NER is initiated by recruitment of the CSB-CSA protein complex and UV-stimulated scaffold protein A (UVSSA) to RNA polymerase II (RNAPII) stalled at the transcription blocking lesion. Subsequently both NER subpathways follow the same route, where initially transcription initiation factor IIH (TFIIH) complex is recruited together with the dissociation of DDB1-DDB2 in GG-NER. TFIIH consists of 10 subunits, including the 2 helicase subunits XPB/ERCC3 and XPD/ERCC2, which unwind the DNA adjacent to the lesion in opposite directions. XPB, XPD, and XPA recruited to the lesion-containing ssDNA serve also as damage verification factors. XPA recruitment is preceded by dissociation of XPC-RAD23B-CETN2 in GG-NER and dissociation or degradation of RNA polymerase II, CSB, CSA and UVSSA in TC-NER. Further repair steps include coating of the undamaged ssDNA strand with RPA and assembling of the preincision complex containing 2 endonucleases: XPF/ERCC4/FANCQ and ERCC1, that bind on the 5′-side of the lesion in an XPA-dependent manner, as well as XPG/ERCC5 that localizes on the opposite side of the repair bubble. Subsequently, XPF-ERCC1 incises the DNA at the 5′-side of the damage, what is followed by loading of the replication machinery including RFC, PCNA sliding clamp and DNA polymerase POLD, POLK (DNA Polymerase Kappa; Pol κ), or POLE that enables XPG-mediated cleavage of the damaged DNA strand at the 3′-side. This results in the removal of 22 to 30 nt DNA fragment and the resulting gap is filled in by the replication machinery followed by ligation of the DNA ends by LIG1 or LIG3α (219, 220).

Defects in NER genes cause several genetic disorders, including, among others, Xeroderma Pigmentosum (XP), Cockayne syndrome (CS), and trichothiodystrophy (TTD) (143, 219, 221). Depending on the affected gene, XP is divided into 8 complementation groups (XP-A to XP-G and XP-V). Classical XP is manifested by sunlight sensitivity, ocular problems, increased skin cancer incidence, segmental progeroid features, and fraction of patients is also affected with neurological abnormalities (221, 222). Importantly, in the context of the current review, 1 feature of premature aging in XP patients, especially individuals carrying XPC mutations, is an increased occurrence of thyroid nodules and, consequently, thyroid cancer (223, 224). In line with these observations, XPC rs2228001 SNP increases WDTC susceptibility (225). Moreover, SNPs in other XP genes, such as XPD and XPG, increase TC risk after ionizing radiation exposure (151, 226). XPD SNPs also increase general susceptibility to TC, and downregulation of XPD expression is correlated with BRAF^V600E mutation and more aggressive disease (148, 227). SNPs in XPF and XPD also predispose to familial non-medullary thyroid cancer (FNMTC), a form of thyroid cancer that is estimated to involve up to 15% of all TC cases (228). Patients with Cockayne syndrome, carrying mutations in TC-NER-specific genes (ie, CSA [Cockayne syndrome protein A/ERCC8] and CSB [Cockayne syndrome protein B/ERCC6]), are characterized mainly by growth and developmental retardation, as well as neurological abnormalities, photosensitivity, premature aging (progeria), but without increased incidence of skin cancer. Patients with Cockayne syndrome preserve normal thyroid structure and function (143, 219, 221). Concerning the expression changes, high-throughput sequencing data reveal RAD23B, DDB2, ERCC1, and XPB/ERCC3, upregulation, but CETN2, UVSSA/KIAA1530, XPF/ERCC4/FANCQ, CSA/ERCC8, XPG/ERCC5, XPA, and POLK downregulation in PTC tissues (Fig. S1 (13)).

Mutations in genes encoding TFIIH subunits XPD/ERCC2, XPB/ERCC3, and TTDA/GTF2H5 (general transcription factor IIH subunit 5), which destabilize the TFIIH complex and impair transcription, cause photosensitive forms of trichothiodystrophy (TTD). The main feature of TTD, and common to all forms of the disease, are brittle hair deficient in sulphur-containing proteins. Depending on the genetic background, patients with TTD may be also affected with photosensitivity, eye and skin problems (but without skin carcinogenesis), delayed development, and neurological defects. Interestingly, neurological abnormalities in patients with TTD, and in murine models, resemble those observed in TH deficiency. This feature was suggested to result from defective TFIIH-mediated co-activation of thyroid hormone receptor-responsive genes during brain development (229). Similarly, neurological defects in CS patients were also suggested to be influenced by impaired co-activation of thyroid hormone-induced transcription leading to defective myelination (230). Additionally, livers of progeroid Csb^m/m/Xpa^−/− and Ercc1^−/Δ-7 mice revealed transcriptomic changes resembling those observed in hypothyroidism, that were accompanied by normal thyroid morphology. Moreover, T₃ and T₄ levels were reduced in serum, liver, kidney, heart, and brain tissues of Csb^m/m/Xpa^−/−, but not Ercc1^−/Δ-7 mice. These alterations in TH levels were suggested to result from aging-dependent decrease in activity of iodothyronine deiodinase 1 (DIO1), converting T₄ to T₃, accompanied by increase in activity of iodothyronine deiodinase 3 (DIO3), that inactivates THs, which was observed in the liver and kidney tissue (231). Similar changes in TH levels, as well as DIOs activity were also recently observed in the livers of progeroid Xpg^−/− mice (232). However, neurological abnormalities observed in the brain tissues of murine XP, CS, and TTD models suggest differences in mechanisms underlying neuropathology. Therefore, they cannot be simplified to TH-dependent effects (233).

Mismatch Repair

The major role of mismatch repair (MMR) is the correction of errors generated during replication that escaped polymerase proofreading, such as single base mismatches and insertion/deletion loops (IDLs). MMR shares substrates with other DNA repair pathways, including BER (eg, in 8-oxoG removal), NER (eg, in CPDs repair) recombination, and interstrand crosslink (ICL) repair. MMR is also mechanistically involved in processes that induce genetic variation (eg, antibody diversification) and mediates triplet repeat instability in trinucleotide repeat disorders (eg, myotonic dystrophy, Huntington disease, Friedreich ataxia) (234, 235). MMR (see Fig. 2C) is initiated by heterodimeric protein complex: MutSα (MSH2/MSH6 heterodimer), which recognizes single base mismatches and small insertion/deletion loops, or MutSβ (MSH2/MSH3 heterodimer) recognizing larger IDLs. For proper correction of replication errors, MMR must discriminate between the template (parental) DNA strand and the newly synthetized (daughter) strand. In eukaryotic cells, strand recognition is methyl-independent and most probably governed by DNA nicks present only in the newly synthetized strand. If the break is located at the 3′-side of the damage, MutLα (MLH1/PMS2 heterodimer) is recruited and forms a complex with MutSα/β. Next, PCNA bound to the nick together with RFC, activates the MutLα endonucleolytic activity to cleave the nascent DNA strand at the 5′-side of the mismatch. The 5′-side nick serves as an entry side for Exonuclease 1 (EXO1), which degrades the mismatch containing strand in a 5′ → 3′ direction. The resulting ssDNA gap is protected by RPA and filled in by the high-fidelity POLD assisted by PCNA and RFC. In MMR, DNA ends are ligated by LIG1. MMR initiated at the nick preexisting at the 5′-side of the mismatch requires only EXO1-, RPA-, POLD-, RFC-, PCNA-, and LIG1-mediated nascent strand excision, repair synthesis, and ligation of the DNA ends. Interestingly, repair synthesis in MMR might also be performed by the error-prone polymerases POLB or XPV/POLH (DNA Polymerase Eta; Pol η), which was suggested to be involved in trinucleotide repeat expansion (143, 235, 236). Defects in MMR genes lead to microsatellite DNA instability (MSI) associated with hypermutability, as well as increased cancer incidence, especially colorectal and endometrial cancers, and are the leading cause of hereditary non-polyposis colorectal cancer syndrome, known also as Lynch syndrome (237).

Lynch syndrome patients, especially those with MSH2 and MSH6 defects, reveal high incidence of thyroid nodules (238). However, frequency of TC is not elevated in MMR mutation carriers and therefore TC is not considered to belong to tumor spectrum of the Lynch syndrome (239). Moreover, MSI is rarely observed in TC, with incidence reaching 1.7%, 2.5%, and 7.4% in PTC, FTC, and ATC, respectively (240-242). However, another report described much higher MSI frequency in thyroid tumors (243). Mutations in MSH2 and MSH6 were suggested to predispose to FNMTC (244). Among SNPs in MMR genes, only MSH6 rs1042821 was suggested to increase WDTC incidence (225). MSH2 and MSH6 are upregulated in BRAF^V600E-mutated PTCs (23). MSH5, which is involved in meiosis, is upregulated in HT tissues (77). MLH1 expression is decreased due to the promoter hypermethylation in PTC tissues and low MLH1 expression correlates with BRAF^V600E mutation and more aggressive disease (148). Finally, as described above, acquired somatic MMR gene mutations are responsible for evolution, dedifferentiation and progression of thyroid cancer. In particular, increased MSI, elevated number of somatic mutations, and loss of MMR gene expression, including loss of heterozygosity of MSH2 gene locus, were observed in ATC and PDTC tissue samples (59, 242).

DNA Damage Response and DSBs Repair

The cell reactions to the DNA damage are collectively termed the DNA Damage Response (DDR). DDR is initiated by DNA damage detection that triggers cell signaling that may lead to the delay of cell-cycle progression (to give more time to repair damaged DNA) or to activation of senescence or cell death pathway. DDR is most commonly studied in response to DNA breaks (Fig. 3). The central role in DDR is governed by the 3 serine/threonine kinases from the phosphatidylinositol-3-kinase-like kinase family: ATM (ataxia-telangiectasia mutated; see Fig. 3A), ATR (ATM- and Rad3-related; see Fig. 3B), and DNA-PKcs (DNA-dependent protein kinase catalytic subunit). Activation of ATM, ATR, and DNA-PKcs is induced by their recruitment to sensor proteins that recognize damaged DNA. ATM binds the trimeric MRN complex (consisting of MRE11—Meiotic Recombination 11, RAD50—homolog of S. cerevisiae Rad50 and NBS1—Nijmegen breakage syndrome 1 gene encoding Nibrin protein) at DSBs. DNA-PKcs binds to the Ku70-Ku80 heterodimer recognizing DSB and initiates DNA repair by non-homologous end-joining pathway (see the following section). ATR is recruited to ATRIP (ATR-interacting protein) bound to RPA occupying ssDNA (resulting from various damaging insults or generated as intermediates in DNA metabolic processes and replication or transcription stress). Upon activation, ATM, ATR, and DNA-PKcs phosphorylate multiple proteins facilitating their respective DDR functions (245, 246). The substrates modified by ATM in response to DNA damage described in the context of thyroid disease include checkpoint kinase CHK2/CHEK2, H2AX, DNA-PKcs, mediator of DNA damage checkpoint 1 (MDC1), TP53-binding protein 1 (TP53BP1), tumor suppressor p53 (product of TP53 gene), WRN, BRCA1, BRCA2, CtIP, E3 ubiquitin ligase MDM2 (mouse double minute 2), MDMX, histone acetyltransferase TIP60, and CCDC6 (246). ATM defects lead to Ataxia-Telangiectasia (A-T; also known as Louis-Bar Syndrome), a genetic disorder characterized by cerebellar ataxia, oculocutaneous telangiectasia, cerebellar atrophy, immunodeficiency leading to recurring infections, IR hypersensitivity, premature aging, as well as increased incidence of leukemias and lymphomas. TC is observed only incidentally in A-T patients, and A-T heterozygotes do not show increased TC risk (247, 248). However, children with A-T frequently experience hypothyroidism (249). Moreover, ATM sequence variants were described as one of the factors predisposing to FNMTC (228, 244). Additionally, somatic ATM mutations are frequently detected in TC, with frequency reaching up to 27% of cases in ATCs (see Table S1 (13)). Moreover, ATM SNPs were described to modulate TC risk or disease severity, but several reports excluded such association (for detailed discussion see (83)). ATM pathway activation has been observed in transcriptomic data of TC tissue (250). A single report described ATM inactivation by methylation in TC tissues, but this observation has not been further corroborated (251). Finally, ATM is implicated in NIS downregulation-dependent thyroid stunning after diagnostic radioiodine application (252, 253).

Figure 3.

Schematic representation of DNA damage responses (DDR) activated by DNA single- (A) and double-strand breaks (B). The figure shows only selected ATR and ATM targets. For a detailed description please refer to the text. Created in BioRender. Arczewska, K. (2025) https://BioRender.com/z38w865.

The checkpoint kinase CHK2, encoded by the CHEK2 gene, is a serine/threonine kinase coordinating cellular responses to genotoxic insults. CHK2 activates its downstream targets, that modulate cell-cycle progression, apoptosis, DNA repair, and senescence. Following ATM-mediated phosphorylation and activation, CHK2 phosphorylates CDC25 family phosphatases, which leads to their inactivation, and further hampers dephosphorylation of Cdk/cyclin complexes resulting in cell-cycle arrest at the G₁-S, intra-S, or G₂-M checkpoints. Cell-cycle progression is also controlled by ATM- or CHK2-dependent p53 phosphorylation and the resulting upregulation of Cdk inhibitor p21^WAF1/CIP1. However, several lines of evidence suggest also that CHK2 activity is dispensable for p53-mediated cell-cycle regulation in response to DNA damage, and that CHK2-dependent p53 phosphorylation is mainly involved in triggering apoptosis (254). Other targets modified by CHK2 in response to DNA damage described in the context of thyroid disease include BRCA1, BRCA2, XRCC1, PARP1, MDM2, MDMX, BLM, and FOXM1. Inherited inactivating CHEK2 mutations or SNPs were shown to increase the risk of carcinogenesis in multiple organs, including the thyroid (Tumor Predisposition Syndrome 4) (83, 254, 255). CHEK2 mutations were also suggested to predispose to FNMTC (228, 256). Somatic CHEK2 mutations were detected in 1% to 9% of TC cases in general, with increased incidence in more aggressive variants (see Table S1 (13)). Interestingly, CHK2 overexpression detected by immunohistochemistry is frequently observed in PTC tissues, especially in tissues from CHEK2 mutation carriers, and high CHK2 expression correlates with unfavorable prognosis (257). High expression detected in TC tissues from mutation carriers may result from the fact that at least some CHEK2 genetic variants observed in TC patients result in inappropriate folding of the protein product leading to its aggregation and accumulation (258).

ATM recruited to DSB-bound MRN, phosphorylates histone H2AX to produce γH2AX, one of the main markers used to visualize DSBs by immunocytochemistry. γH2AX modification recruits MDC1, forming a platform for further DDR signaling (245, 246). γH2AX levels are elevated in PTC tissues and correlated with ATM expression in TC tissue and thyroid hormones in serum samples. γH2AX expression is negatively correlated with tumor aggressiveness traits, such as TNM scale and differentiation status (259).

Following recruitment, MDC1 is phosphorylated by ATM, which facilitates binding of the ubiquitin ligase RNF8. RNF8-initiated ubiquitination cascade facilitates accumulation of TP53BP1 protein foci at DSBs. TP53BP1, which is also widely used as immunocytochemical marker to detect DSBs, stimulates repair by NHEJ pathway. On the other hand, TP53BP1 inhibits homologous recombination (HR; see the following) by preventing DNA end resection. BRCA1, BRCA2, and CtIP, which are also phosphorylated by ATM and required for HR (see next), antagonize TP53BP1 leading to NHEJ inhibition (245, 246). The MDC1 gene shows reduced methylation in HT in comparison to GD tissues (260). Moreover, analysis of TCGA dataset suggests its downregulation in PTC (Fig. S1 (13)). MDC1 is more frequently mutated in aggressive PTCs (261). RNF8, similarly, is downregulated in TC (262). TP53BP1 expression is elevated in TC and correlates with malignant potential. Due to the ease of immunohistochemical evaluation of TP53BP1 expression, it has been suggested as a marker facilitating FTC diagnosis (263).

Tumor suppressor p53 is a transcription factor involved in multiple cellular processes including cell-cycle arrest, apoptosis, senescence, and DNA repair. Upon DNA damage, p53 is activated by phosphorylation catalyzed by ATM, ATR, DNA-PKcs, CHK1, and CHK2 (264). Interestingly, DDR controls p53 mRNA translation. Specifically, repair proteins, including ATM, MDM2, MDMX, NBS1, and Ku70/Ku80, bind to the secondary structures of p53 mRNA to support its translocation to the cytoplasm and stimulate or repress translation (265). TP53, encoding p53 protein, is one of the most frequently mutated genes in human cancers. In WDTC TP53 mutation is detected in a small fraction of tumors, whereas it is mutated in up to 80% of more aggressive ATCs (see Table S1 (13)). Accordingly, TP53 mutations were suggested to affect TC dedifferentiation and progression to ATC (59, 61). Interestingly, direct involvement of TP53 mutations in TC dedifferentiation was confirmed by reexpression of the wild-type p53 that upregulated NIS, TPO, and PAX8 and resulted in improved radioiodine uptake in TC cells (266). Moreover, ectopic NIS overexpression alone was sufficient to improve p53-dependent RAI antitumorigenic activity in human ATC cells, as well as murine xenograft model (267). TP53 mutations are correlated with TC unfavorable prognosis (268). Moreover, coexistence of TP53 and BRAF^V600E mutations in ATC patients reduces progression-free survival (269). TP53 mutations were also suggested to help distinguishing neoplastic from benign (ie, mutation-free) thyroid nodules after diagnostic cytology, as well as to assist in estimating the risk of distant metastasis in WDTC (270, 271). Additionally, TP53 SNP rs1042522 (Arg72Pro) increases the risk of sporadic and radiation-induced TC, and was implicated in AITD pathogenesis (83, 272-274). Arg72Pro is one of the most extensively studied TP53 variants and its disease-promoting activity might result from the fact that it influences p53 transcriptional activity and apoptosis induction efficiency (275). In the context of AITD, TP53 is overexpressed in HT tissues (77). Moreover, p53 may suppress the AITD-related autoimmunity through supporting differentiation of regulatory T cells (Tregs) (276). Germline variants in TP53 are an underlying cause of multiorgan cancer predisposition syndrome Li-Fraumeni (LFS). Thyroid tumors, mostly with papillary histology, are also observed in patients with Li-Fraumeni syndrome, with estimated prevalence from 1% to almost 11% in Brazilian TP53 Arg337His mutation carriers (277).

MDM2, an E3 ubiquitin ligase, and its catalytically inactive homologue MDMX (aka MDM4), are also targets of ATM- and CHK2-mediated phosphorylation. The MDM2 homodimer or MDM2-MDMX heterodimer ubiquitinate p53, thus keeping its levels low. DDR-induced phosphorylation of MDM2 or MDMX destabilizes dimeric complexes, which results in p53 upregulation in response to DNA damage (278). Moreover, ATM-phosphorylated MDM2 and MDMX become positive p53 regulators, since they bind TP53 mRNA thus stimulating its translation (265). MDM2 and MDMX also have p53-independent functions, as they were described to bind MRN complex subunit NBS1 and thus inhibit ATM-mediated DDR (279). Moreover, p53 levels are also regulated by circular noncoding RNA CircTP53-miR-1233-3p-MDM2 regulatory axis in TC cells. CircTP53 is upregulated in TC tissues leading to p53 downregulation and resulting enhanced proliferation (280). MDM2 is upregulated in ∼30% of TCs. Moreover, p53 and MDM2 coexpression is also observed in TC tissues and is associated with less aggressive behavior (281). Additionally, MDM2 is regulated by ARF—encoded by the CDKN2A gene, which is frequently deleted in cancers, including TC (15). MDM2 polymorphisms may modify WDTC risk (282). Moreover, somatic MDM2 mutations were also detected in follicular thyroid adenoma tissues (283). Restoring p53 function through inhibition of MDM2-MDMX-mediated ubiquitination has been suggested in TC and AITD management (276, 284). MDMX, on the other hand, is downregulated in TC, reaching the lowest levels in more advanced stages (285).

GADD45 family proteins, notably their most extensively studied member growth arrest and DNA damage inducible alpha (GADD45A), are among the cellular targets upregulated by DDR-mediated p53 activation. GADD45 family proteins are involved in multiple tumorigenesis-related processes, including DNA repair (mainly BER and NER) and cell-cycle progression. Importantly, GADD45A is implicated in ATM-dependent G₂/M checkpoint activation (286). GADD45A upregulation was observed in PTC, but not in benign thyroid lesions (287). However, analysis of TCGA TC cohort revealed inhibition of the ATM-dependent G₂/M checkpoint, mediated by downregulation of GADD45 family proteins accompanied by upregulation of other G₂/M checkpoint components (ATM, CHK2, CDK1, Cyclin-B, CDC25C, and CDKN1A) (250).

Another factor implicated in ATM-mediated DDR is acetyltransferase TIP60/KAT5, which upon phosphorylation binds damaged DNA and stimulates ATM through acetylation, thus leading to potentiation of downstream p53 and CHK2 signaling. TIP60 also inhibits NHEJ and stimulates HR by suppressing TP53BP1 binding and promoting BRCA1 and RAD51 recruitment through chromatin remodeling. Moreover, TIP60-mediated p53 acetylation promotes its apoptotic functions (245, 288). One of the TIP60-mediated acetylation targets is FOXP3—the principal transcription factor involved in the maintenance of the regulatory T-cell population. Acetylation stabilizes FOXP3 and thus suppresses the autoimmune process (289). In line with this, TIP60 is downregulated in HT, but not GD tissues, which may stimulate autoimmunity (290). In ATC and PTC, TIP60 overexpression correlates with unfavorable prognosis and metastatic potential (291, 292). Moreover, TIP60 suppression compromises neoplastic traits in cultured human ATC cells, as well as tumor growth in murine xenograft model (293).

The ATM phosphorylation target particularly relevant in the context of thyroid is a proapoptotic protein Coiled-Coil Domain Containing 6 (CCDC6). Functionally, CCDC6 is stabilized upon DNA damage by ATM-mediated phosphorylation to negatively regulate PPA4 phosphatase-dependent γH2AX dephosphorylation. Accordingly, it is involved in maintaining cell-cycle checkpoints and supports DNA repair by homologous recombination (294). As already explained, CCDC6 fusion with receptor tyrosine kinase RET gene, known as RET::PTC1 rearrangement, is one of the most frequently detected translocations in TC, especially in IR-induced cancers. Moreover, the role of CCDC6 in TC pathogenesis is supported by the observation that knock-in mice carrying a deletion in exon 2 of the Ccdc6 gene developed thyroid hyperplasia and hyperthyroidism. CCDC6 Exon 2 encodes coiled-coil domain essential for the pro-oncogenic function of CCDC6 fusions. The coiled-coil domain is required for CCDC6-mediated negative regulation of CREB1 transcription factor, which is an essential player in TSH-simulated thyrocyte growth and differentiation (295). Interestingly, CCDC6 deficiency is synthetically lethal with PARP1 inhibition; therefore, PARP1 inhibitors may prove to be beneficial in management of thyroid tumors carrying RET::PTC1 rearrangements (294).

FOXM1 (or Forkhead Box protein M1) is a transcription factor implicated in cell-cycle progression and DDR. FOXM1 is phosphorylated by CHK2 in response to DNA damage, which results in transcriptional activation of multiple DNA repair factors operating in the BER, NER, MMR, HR, and ICL repair pathways (296). Moreover, the G₂/M transition is dependent on CDK1-mediated phosphorylation of FOXM1, which in turn is inhibited in an ATR-CDC25-CDK-dependent manner in the S phase to prevent transcription of mitotic genes before replication is completed (297). FOXM1 is overexpressed in HT, as well as PTC tissues and supports survival, migration, and invasion of PTC cells (77, 298).

As already mentioned, ATR is DDR apical kinase recruited to RPA-coated ssDNA resulting from replication problems or from end-resection of DSBs during HR, via its interaction partner ATRIP. Furthermore, ATR activation requires allosteric activator proteins, like Ewing tumor-associated antigen 1 (ETAA1) and TOPBP1. TOPBP1 is recruited to dsDNA-ssDNA by interaction with the ring-shaped heterotrimeric RAD9-RAD1-HUS1 (9-1-1) complex, loaded by RAD17-RFC. Upon activation, ATR phosphorylates its downstream targets, of which the most important substrate is CHK1. Similarly to CHK2, a major role of CHK1 is the phosphorylation of CDC25 family phosphatases, resulting in Cdk/Cyclin inactivation and consequent cell-cycle arrest at intra-S or G₂/M checkpoints. Activated ATR and CHK1 regulate cell-cycle checkpoints, with their most important role being in the intra-S checkpoint, where they are involved in the replication stress response by controlling replication origin firing, promoting replication-fork stability and regulating dNTP pools. Among other important substrates of ATR- or CHK1-mediated phosphorylation are CHK2, H2AX, DNA-PKcs, p53, MDM2, BRCA1, NBS1, BLM, WRN, RAD51, SMARCAL1, FANCD2, and WEE1 (245, 299). In contrast to ATM, ATR is essential for survival, probably due to its role in replication. Defects in the ATR gene leading to decreased expression levels are underlying cause of Seckel syndrome 1, a genetic disorder manifested by mental retardation, microcephaly, and dwarfism, but without thyroid function disturbances (300). In TC, ATR mutations are encountered in less than 1% in PTCs but are observed in more than 5% in ATCs (see Table S1 (13)). Interestingly, however, ATR expression is downregulated in PTC tissues (Fig. S1 (13)). ATR inhibition has been proposed in cancer therapy (297). In agreement with this assumption, ATR inhibition induces G₂/M arrest and apoptosis in PTC and FTC cultured cells and murine xenograft models. Accordingly, co-treatment of WDTC cell lines with an ATR inhibitor and lenvatinib or dabrafenib plus trametinib resulted in synergistic effect. Therefore, combination of ATR inhibition with multikinase or MAPK pathway inhibitors was suggested as a promising therapeutic option in WDTC management (301).

ATR interacting protein (ATRIP) is an essential gene in mice (302). In humans, ATRIP mutations result in Seckel syndrome. Genetic variants of ATRIP have been correlated with susceptibility to breast cancer (303). Surprisingly, ATRIP has never been specifically studied in the context of the thyroid. Analysis of TCGA data shows that it is not mutated in TC (see Table S1 (13)). Etaa1 mutations in mice result in partial embryonic lethality and influence proliferation of effector T cells. Therefore, ETAA1 potentially may have implications in autoimmune diseases (304). Similarly to ATRIP, association of ETAA1 with thyroid disease can only be learned from high-throughput data, and accordingly ETAA1 gene sequence is rarely mutated in TC (see Table S1 (13)). However, ETAA1 expression is downregulated in PTCs (Fig. S1 (13)).

DNA topoisomerase II binding protein 1 (TOPBP1) is a scaffold protein that binds multiple protein ligands phosphorylated by activated DDR, including the chromatin remodeling factor SMARCAD1, helicases BLM, FANCJ/BRIP1/BACH1 and RECQ4, RAD9, PARP1, type-II DNA topoisomerases TOP2A and TOP2B, MDC1, BRCA1, TP53BP1, and nucleolar protein Treacle/TCOF1. It is proposed that the major role of the TOPBP1 interactions may be to localize the repair proteins to the damage site (305). SNPs in TOPBP1 increase the risk of TC development after diagnostic irradiation (151). TOPBP1 mutations are very rare in PTCs (0.25%), and in ATCs are reaching 1.5% (see Table S1 (13)).

The Heterotrimeric clamp 9-1-1 complex is specifically targeted by ATP-dependent clamp loader RAD17-RFC to ssDNA/dsDNA junctions with 5′-recessed ends (306). 9-1-1 also interacts with BER and MMR factors to stimulate their activity and coordinate repair complex assembly (307, 308). All subunits of the 9-1-1 complex, as well as RAD17, are essential for survival, which is demonstrated by Rad9, Hus1, Rad1, and Rad17 mice knockouts that are embryonically lethal (308, 309). RAD9, encoded by RAD9A gene, is overexpressed in TC and HT tissues (77, 310). However, low RAD9A expression correlates with PTC recurrence (311). RAD9A sequence is unaltered in PTCs, and very rarely mutated in ATCs, reaching 0.5% (see Table S1 (13)). SNPs in HUS1 are associated with WDTC risk (151, 312). Moreover, HUS1 is downregulated in PTC (Fig. S1 (13)), while its genetic sequence is unaltered in TC in general (analyzed using cBioPortal (313-315)). Based on high-throughput data, RAD1 and RAD17 expression is elevated and decreased, respectively, in PTC (Fig. S1 (13)); however, their genetic sequences are unaltered (see Table S1 (13)). Finally, RAD1 is downregulated in HT tissues (77).

Checkpoint kinase CHK1 is serine/threonine kinase encoded by the CHEK1 gene. CHK1 is implicated in DDR, DNA replication, progression of mitosis, and cell-cycle regulation. CHEK1 is essential, as its loss is embryonic lethal in mice (316). In humans, CHEK1 mutations result in oocyte/zygote/embryo maturation arrest-21 (OZEMA21) syndrome and female infertility. Cultured ATC cells require functional CHK1 to activate G₂/M checkpoint after IR or doxorubicin treatment (21). CHK1 expression positively correlates with malignancy in thyroid nodules (317). Moreover, CHEK1 expression is slightly upregulated in PTCs, more enhanced in FTCs, and even higher in PDTCs (318). CHEK1 gene is very rarely mutated in PTCs (ie, in 0.2% of cases) (see Table S1 (13)).

WEE1 is an essential checkpoint kinase, responsible for phosphorylation-mediated inactivation of CDKs to result in cell-cycle inhibition in the intra-S and G₂/M phases, as well as block helicase- and nuclease-dependent resolution of stalled replication forks (319). WEE1 is extremely rarely mutated in PTCs, and in only 0.5% ATCs (see Table S1 (13)). Moreover, WEE1 expression is upregulated in HT tissues (77). WEE1 is inhibited by AKT kinase-mediated phosphorylation induced by iodine overload, leading to CDK1 activation and enhanced proliferation of PTC and ATC cells in vitro and in vivo (320). Conversely, WEE1 inhibitor adavosertib, applied either alone or in combination with multikinase or MAPK pathway inhibitors, inhibits ATC and WDTC cell growth and induces apoptosis in cell culture and murine xenograft model; therefore, it has been proposed as a promising therapeutic option in patients with ATC and refractory WDTC tumors (321, 322).

Another target of ATR, but also of ATM, is the WRN helicase, whose role in thyroid disease will be described below. WRN, apart its other activities related to NHEJ, is essential in resolving collapsed replication forks (323).

SWI/SNF-related, matrix-associated, actin-dependent regulator of chromatin, subfamily a-like protein 1 (SMARCAL1) is a chromatin remodeling factor that is recruited to stalled replication forks and involved in fork reversal. SMARCAL1 is phosphorylated by ATR and ATM. ATR-mediated phosphorylation inhibits SMARCAL1 activity to prevent fork collapse (299). SMARCAL1 gene defects lead to Schimke immuno-osseous dysplasia (SIOD) multisystem disease, of which most common symptoms include spondyloepiphyseal dysplasia, renal insufficiency, and immunodeficiency. Up to half of the patients are affected by hypothyroidism (324). It has been suggested that SMARCAL1 mutation disrupts TH synthesis and/or secretion pathways (325). SMARCAL1 gene mutations were observed in 0.2% of PTCs and almost 3% of ATCs (see Table S1 (13)).

TCOF1 is involved in multiple processes, including ribosome biogenesis, DDR, mitosis, proliferation, and telomere integrity. One of the major roles of TCOF1 in DDR is regulation of DNA damage induced nucleolar response. Apart from being required for TOPBP1-ATR interaction, TCOF1 is also phosphorylated by ATM and associates with NBS1 in nucleolar DDR. TCOF1 gene defects are the underlying cause of Treacher Collins syndrome manifested by craniofacial dysmorphism. Moreover, TCOF1 has been suggested to be involved in carcinogenesis (326, 327). TCOF1 expression is slightly upregulated in TC (327). TCOF1 genomic sequence is unaltered in PTCs but mutated in 0.5% of ATCs (see Table S1 (13)).

Non-Homologous End-Joining

Double-strand breaks (DSBs) are highly toxic and must be immediately resolved. The major DSBs repair pathways are Non-Homologous End-Joining (NHEJ), also termed classical NHEJ (cNHEJ), and homologous recombination (HR). NHEJ is active during the entire cell cycle, whereas HR is restricted to the S and G₂ phases (328). Importantly, NHEJ requires little or no homology because it relies on rejoining of the DNA ends. Therefore, this repair system is not able to discriminate between corresponding and unrelated DNA ends, which might lead to chromosomal rearrangements. Thus, NHEJ was proposed as the major mechanism mediating genomic rearrangements observed in IR-induced thyroid cancers (329). Interestingly, the origin of rearrangements most frequently observed in TC (ie, RET::PTC1 and RET::PTC3) is postulated to result from spatial colocalization of RET and CCDC6 or NCOA4 genetic regions, respectively, in the interphase nuclei of thyroid cells (330, 331). It is currently proposed that chromosomal rearrangements are primarily generated by alternative NHEJ pathway (Alt-NHEJ, also known as microhomology-mediated end joining (MMEJ) or theta-mediated end joining (TMEJ)) or BIR (break-induced replication), since inactivation of classical NHEJ stimulates DSBs-induced and proximity-dependent translocations (332, 333). Breaks generated by topoisomerases, such as topoisomerase 1 (TOP1) and 2 (TOP2) may also be implicated in generation of RET::PTC rearrangements (334). Classical NHEJ (Fig. 4A) is initiated by binding the 2 ring-shaped Ku70-Ku80 (aka XRCC6-XRCC5) heterodimers to DNA ends at the opposite sides of the DSB, followed by recruitment of DNA-PKcs, DNA end processors, DNA ligase IV (LIG4), as well as the scaffolding proteins XRCC4, XRCC4-like factor (XLF; also known as Cernunnos or NHEJ1), and Paralogue of XRCC4 and XLF (PAXX). This results in bringing and stabilizing the 2 DNA ends together, in a process called synapsis. Long-range synapsis, where DNA ends are >80 Å apart, depends on the Ku heterodimer and DNA-PKcs. Long-range synapsis is quickly transformed in a XRCC4-LIG4- and XLF-dependent manner into short-range synapsis, where DNA ends are in close proximity. Further DNA ends processed by, depending on the nature of the DNA termini, nuclease Artemis, PNKP, TDP1, TDP2, APTX, specialized DNA polymerases DNA Polymerase Mu (POLM; Pol µ), DNA Polymerase Lambda (POLL; Pol λ) or POLB, as well as terminal deoxynucleotidyl transferase generates DNA ends compatible for further ligation. The final step in NHEJ involves sealing the DNA ends by the LIG4-XRCC4 complex. DNA-PKcs acts also as apical DDR kinase phosphorylating and thus activating Artemis, p53, PARP1, XLF, XRCC1, XRCC4, Ku70, Ku80, and other targets (328, 335, 336). Werner syndrome helicase (WRN) contributes to NHEJ by mediating pathway choice, promoting NHEJ over Alt-NHEJ (337). Defects in NHEJ genes are the underlying cause of several human diseases. For example, due to NHEJ involvement in antibody diversification, mutations in PRKDC/XRCC7 (encoding DNA-PKcs), LIG4, XLF, and DCLRE1C (encoding Artemis) genes lead to severe combined immunodeficiency. XRCC4 mutations do not cause immunodeficiency, but result in dwarfism, microcephaly, and endocrine dysfunction including hypogonadism, and rarely multinodular goiter or hypothyroidism (338, 339). Interestingly, Ku70 was identified and cloned as thyroid-specific autoantigen using GD autoimmune serum (340). Anti-Ku70 and anti-Ku80 antibodies are also detected in a fraction of patients with several other autoimmune diseases, including systemic lupus erythematosus. Ku70 is, therefore, also termed lupus autoantigen p70 or thyroid-lupus autoantigen (341). Downregulation of Ku70 mRNA levels correlate with the hypothyroid state after RAI therapy (342). Ku80 rs2440 polymorphism, but not XRCC4 or LIG4 SNPs, is associated with increased risk of PTC, but not FTC, but other studies showed conflicting results (for detailed discussion see (83)). Moreover, Ku80 expression is increased in MTC and PTC tissues (21, 343). Ku80 protein levels are also elevated in TC tissues in general, and Ku80 is required for survival and maintaining malignant phenotype of cultured TC cells (344). However, Ku80 and Ku70 are downregulated in ATC tumors (21). DNA-PKcs expression level is unaltered in TC tissue in comparison to normal tissue, but downregulated in ATC tissues (21, 345). Radiation-resistance of TC cultured cells, on the other hand, is directly correlated with DNA-PKcs activity, and cells with higher DNA-PKcs levels have increased radiation resistance (81). Moreover, one report described association of Ile3434Thr DNA-PKcs gene variant (rs7830743) with WDTC risk, but other reports excluded such possibility (for detailed discussion see (83)). XRCC4, on the other hand, is not overexpressed in TC, but is correlated with immune function markers in TC tissues (346). XLF expression is downregulated in HT tissues (77). Interestingly, BRAF^V600E mutation stimulates NHEJ through upregulation of XLF, what results in increased radiation resistance of TC cells. Therefore, application of BRAF inhibitors was proposed to increase radiation therapy efficacy in BRAF-mutated thyroid cancers (347). Similarly to another end processor, TDP1, Artemis is also downregulated in PTC tissues (Fig. S1 (13)). In the context of LIG4, recent research revealed that combination of LIG4 inhibition with DSBs induction with doxorubicin or γ-radiation stimulates apoptosis of ATC cells and compromises ATC tumor growth in murine xenograft model. Therefore, the combination of NHEJ inhibition with DNA damaging chemotherapeutics was proposed as a promising strategy in management of TCs refractory to current treatment regiments (21). Finally, the NHEJ factor with the best documented association with thyroid cancer is WRN. WRN, also known as RECQL2 or RECQ3, is an ATP-dependent helicase and 3′-5′ exonuclease belonging to RECQ helicase family. It is involved in resolving higher order DNA structures, such as Holliday junctions, DNA duplexes and quadruplexes, and stalled replication forks, as well as participates in DSBs repair through NHEJ, HR, alt-NHEJ, and single-strand annealing (SSA), as well as BER. WRN mutations lead to Werner syndrome, with the highest prevalence in the Japanese population, manifesting with short stature and premature aging, as well as increased cancer incidence, with TC constituting one of the most frequently observed cancers. Among the TCs linked with WRN mutations, FTCs are the most frequent, but PTCs and ATCs are also observed (348, 349). Apart from being mutated, WRN is also frequently inactivated in TC by promoter hypermethylation leading to its downregulation (350, 351). Finally, WRN mutations were observed in both benign thyroid goiter and PTC tissues, suggestive of their involvement in progression of the disease from pre-neoplastic lesions to malignancy (16). Other helicases from RECQ family are RECQL1, RECQL4, RECQL5, and BLM. RECQL1 and RECQL4 are associated with NHEJ and replication stress, REQL5 promotes repair by synthesis-dependent strand annealing (SDSA; see the following section), whereas BLM promotes HR via Holliday junction resolution (see the following section) (349). In TC tissues, RECQL1 and REQL5 are downregulated (351).

Figure 4.

Schematic representation of DNA double-strand break repair pathways (A) non-homologous end-joining (NHEJ) and (B) homologous recombination (HR). For a detailed description, please refer to the text. Created in BioRender. Arczewska, K. (2025) https://BioRender.com/f78i577.

Homologous recombination (HR)

Homologous recombination, also known as Homology-Directed Repair (HDR), is considered an error-free DSBs repair pathway, as it utilizes a long (usually longer than 100 bp) stretch of the homologous sister chromatid DNA strand to serve as a template for repair synthesis. As already mentioned, HR is restricted to S and G₂ phase of the cell cycle. HR is initiated by binding of the trimeric MRN (ie, MRE11-RAD50-NBS1) complex to DSB (Fig. 4B). MRE11, which possesses endonuclease and 3′-5′ exonuclease activity and requires CtIP (C-terminal binding protein-interacting protein) as a cofactor for activity, performs resection of 5′-ends thus generating 3′ ssDNA overhangs. RAD50 facilitates end resection by ATP hydrolysis-dependent DNA unwinding, whereas NBS1 is a scaffold protein interacting with CtIP. DNA end resection displaces Ku70-Ku80 heterodimer thus inhibiting NHEJ. End resection is also facilitated by heterodimer formed by BRCA1 (Breast cancer susceptibility type 1) and BARD1 (BRCA1-associated RING domain 1), that interacts with CtIP and MRN. Next, the 3´-ssDNA tail is further extended, in so called long-range resection, by coordinated action of the 5′-3′ exonuclease EXO1, endonuclease DNA replication helicase/nuclease-2 (DNA2) and Bloom syndrome protein (BLM) helicase, and coated by RPA. Consecutive step involves BRCA1-BARD1-dependent recruitment of the PALB2 (Partner And Localizer of BRCA2)—BRCA2 complex that mediates RPA displacement and loading of BRCA2-associated central HR recombinase RAD51. RAD51 creates a filament on ssDNA and starts homology search to initiate base-pairing with the complementary sister chromatid, in a process termed strand invasion. The homology search and strand invasion are facilitated by the motor protein RAD54 belonging to the SWI2/SNF2 family. The DNA strand not involved in base-pairing is displaced to form a D-loop. Strand invasion depends on RAD51-driven ATP hydrolysis and filament disassembly facilitated by RAD54, which is followed by PCNA/POLD/E (or translesion synthesis polymerase)-mediated 3′-end extension, where the intact sister chromatid serves as a template. Mammalian genomes also encode the RAD51 paralogs, RAD51B/RAD51L1, RAD51C/RAD51L2, RAD51D/RAD51L3, XRCC2, and XRCC3, which facilitate formation and stabilization of the RAD51 nucleoprotein filament. Further scenario of restoring the dsDNA structure might require resolution of double Holliday junctions, generated when both DNA ends are elongated on the templates of the opposite strands of sister chromatid forming the D-loop. This HR subpathway is termed double-strand break repair (DSBR) or gene conversion and requires helicases, nucleases, and topoisomerases, such as BLM, meiotic endonucleases MUS81 and EME1, as well as topoisomerase IIIα (TOP3A). A second scenario, which dominates in somatic cells and does not require Holliday junction resolution, is termed the synthesis-dependent strand annealing (SDSA), and involves displacement of the replicated invading strand from the D-loop, with synthesis of the complementary strand on the template of the displaced DNA strand. Finally, the DNA ends are ligated by LIG1 or LIG3α/XRCC1 (328, 352, 353).

Defects in MRE11 (also known as MRE11A) are the underlying cause of Ataxia-telangiectasia-like disorder 1 (ATLD1), a genetic disease characterized by clinical features resembling those observed in ataxia-telangiectasia (A-T) patients (181). ATLD1 is extremely rare and does not predispose to cancer (354). In murine model, knockout of Mre11 is embryonically lethal (355). In contrast, TCGA dataset suggests MRE11 expression downregulation in PTC (Fig. S1 (13)). Somatic MRE11 mutations are very rare in TC and affect less than 1% in ATCs (see Table S1 (13)). Mutations in NBS1 (also known as nibrin, NBN) predispose to Nijmegen breakage syndrome (NBS) which is characterized by microcephaly, growth deficiency, craniofacial dysmorphism, immunodeficiency, ionizing radiation hypersensitivity, chromosomal instability, and increased predisposition to cancer (primarily of lymphoid origin). NBS1 is an essential gene (355). Heterozygous Nbs1^+/− mice are viable, but prone to increased cancer incidence, including the IR-induced thyroid cancer (356). Similarly to MRE11, NBS1 expression is downregulated in PTC (Fig. S1 (13)). NBS1 mutations were observed in both benign thyroid goiter and PTC tissues, suggestive of their involvement in progression of the disease from pre-neoplastic lesions to malignancy (16). Moreover, somatic NBS1 defects are very rare in TC, with less than 1% mutations observed in PTCs and ATCs (see Table S1 (13)). RAD50 defects are the underlying cause of Nijmegen breakage syndrome-like disorder, an extremely rare disease presenting features similar to NBS, but without immunodeficiency (354). In murine models, Rad50 knockout is embryonically lethal (355). Germline RAD50 mutations were suggested to be associated with TC (357). Somatic RAD50 mutations are rarely observed in TCs, reaching almost 1.6% in ATCs (see Table S1 (13)).

CtIP, encoded by RBBP8 (retinoblastoma-binding protein 8) gene, possesses an endonuclease activity. After activation by ATM-mediated phosphorylation, CtIP stimulates nucleases and helicases that perform end resection. CtIP is essential in mice, whereas heterozygous CtIP^+/− mice are prone to cancer. In humans, defects in CtIP result in Jawad and Seckel 2 syndromes characterized by microcephaly, mental retardation and some body structure malformations, but without increased cancer predisposition (358). High-throughput data reveal RBBP8 upregulation in PTC (Fig. S1 (13)) and lack of somatic mutations in TC (TCGA dataset (14) analyzed using cBioPortal (313-315)).

BRCA1/FANCS is one of the best-known human cancer predisposition genes. BRCA1 mutations are responsible for familial breast and ovarian cancer susceptibility syndrome but also increased pancreatic cancer risk and are encountered in Fanconi anemia (Fanconi anemia, complementation group S [FANCS]). FANCS patients, apart from increased susceptibility to breast and ovarian cancer, are also afflicted with developmental abnormalities, intellectual disability, and anemia. The essential role of HR in genome maintenance is highlighted by the fact that Brca1 and 2 knockouts are embryonically lethal in mice (355). The BRCA1-BARD1 complex possesses ubiquitin ligase activity and BRCA1-BARD1-mediated ubiquitination of H2A histone is required to displace TP53BP1 and thus promote HR (352). The impact of BRCA1 defects on TC is disputed. On the one hand, BRCA1 mutation carriers do not show increased TC incidence (359). On the other hand, BRCA1 mutations were suggested to predispose to FNMTC (244, 256). Moreover, SNPs in BRCA1 were shown to increase TC risk, but other studies excluded such possibility or even showed cancer-protective effect (for detailed discussion see (83)). BRCA1 SNPs were suggested to be associated with RAI therapy resistance in TC (80). BRCA1 expression is downregulated in TC and HT tissues (77, 360). Moreover, high-throughput data suggest that BRCA1 is infrequently mutated in PTC, reaching 6.3% in ATC patients (see Table S1 (13)). In recent years, somatic BRCA1 mutations gained additional interest with the discovery of their synthetic lethality with PARP inhibitors. Further studies revealed that defects in multiple genes involved in DNA repair and DDR are synthetically lethal with PARP inhibitors, and these genes are collectively termed the “BRCAness” genes (361). This opened a new avenue in the research trying to find new combinations between cancer-associated mutations and PARP inhibitors or combination therapies, that may potentially be useful in TC treatment (180, 294, 362).

Similarly to BRCA1, BARD1 defects also predispose to breast cancer and are embryonically lethal in mice (363). Until now, BARD1 has not been analyzed in the context of the thyroid, but high-throughput data reveal upregulation of its expression in PTC (Fig. S1 (13)), lack of somatic mutations in PTC tissues, and 1.6% mutation frequency in ATCs (see Table S1 (13)).

Bloom helicase (BLM, also known as Recq Protein-Like 3; RECQL3) is an ATP-dependent helicase belonging to RECQ helicase family that can dissolve complex DNA structures (349). In murine model Blm is an essential gene (355). Genomic BLM defects predispose to Bloom syndrome (BS), characterized by retarded growth, insulin-resistance, immunodeficiency, photosensitivity, increased diabetes risk, as well as cancer predisposition, with breast, colorectal, and prostate cancer being the most frequent malignancies (349). There are some indications of increased hypothyroidism frequency in patients with BS but this needs to be validated in larger patient cohorts (364). TCGA data suggest that BLM expression is not changed in PTC tissues (351). Moreover, high-throughput data suggest BLM upregulation in HT tissues (77). Finally, somatic mutations are not observed in PTC tissues and detected in approximately 1% of ATCs (see Table S1 (13)).

PALB2/FANCN is another gene, of which defects, similar to BRCA1, predispose to breast and ovarian cancer (familial susceptibility to breast-ovarian cancer 5), pancreatic cancer, and Fanconi anemia. Similar to other HR genes, murine Palb2 knockout is embryonically lethal (355). PALB2 mutation is detected in about 1% of patients with TC. Interestingly, the same study observed similar PALB2 mutation frequency in breast cancer patients (365). Moreover, PALB2 SNPs were suggested to predispose to FNMTC (228). In high-throughput studies somatic mutations were estimated at about 1.1% in ATC tissues and 1.7% of PDTC/ATC tissues and showed lack of mutations in PTCs (see Table S1 (13)). Moreover, PALB2 expression is slightly upregulated in PTC tissues (Fig. S1 (13)).

Germline BRCA2/FANCD1 mutations predispose to multiple cancer types, including breast and ovarian cancer, as well as Fanconi anemia and Wilms tumor. In the context of thyroid, BRCA2 expression is increased in HT tissues and PTC tissues of patients with accompanying HT disease (77, 208). Moreover, BRCA2 expression is also increased in TC in general, and correlates with cancer stage, as well as modulates immune environment of TC tissue. Surprisingly, A2738S mutation carriers have better TC prognosis (366). Moreover, BRCA2 mutations do not predispose to TC and are not associated with malignant potential (359). However, BRCA2 mutations were suggested to predispose to FNMTC (228, 244). Moreover BRCA2 rs15869 SNP increases TC susceptibility (for detailed discussion see (83)).

Inherited RAD51/FANCR mutations, similarly to other HR genes, predispose to breast cancer, Fanconi anemia, and neurological disease. Moreover, Rad51 is essential in murine model (355). RAD51 SNPs do not affect TC susceptibility (for detailed discussion see (83)). Somatic RAD51 mutations are very rare in TC, reaching 0.5% in ATC tissues (see Table S1 (13)). However, RAD51 mutation was observed in benign thyroid nodules (357). Finally, RAD51 is overexpressed in TC and positively correlated with cancer progression (367).

RAD54, encoded by the RAD54L gene, defects were observed in breast, colon, and lymphoma patients (368). RAD54L shows no somatic mutations in PTC, and less than 2% mutations in PDTC/ATC tissues (see Table S1 (13)). Moreover, RAD54L expression is slightly increased in PTC tissues (Fig. S1 (13)).

In relation to RAD51 paralogs, defects in RAD51C and RAD51D predispose to breast and ovarian cancer or Fanconi anemia. Defects in XRCC2 result in fertility issues or Fanconi anemia, whereas in XRCC3 predispose to breast cancer and melanoma. SNPs in RAD51B and XRCC2 do not affect the risk of thyroid neoplasia, while XRCC3 SNPs may increase TC risk (for detailed discussion see (83)). However, in another study RAD51B SNPs were shown to reduce PTC risk (152). RAD51D is downregulated in PTC tissues (Fig. S1 (13)). Moreover, XRCC2, but not XRCC3, expression is upregulated in TC tissues. Additionally low expression of XRCC2 correlates with invasion potential (148). High-throughput data suggest RAD51B and RAD51C upregulation in HT tissues (77). Finally, MUS81 defects may predispose to TC, and are associated with PTC recurrence (369). Moreover, low MUS81 expression correlates with PTC recurrence (311).

Other Pathways Repairing DSBs

DSBs may also be repaired, especially when NHEJ or HR is disabled, through 3 other DNA repair pathways, namely alternative NHEJ (Alt-NHEJ), single-strand annealing (SSA) and Break-induced Replication (BIR) that are more error-prone than HR. In Alt-NHEJ, DSBs are subjected to a short-range resection that is mediated by MRN and CtIP, to reveal microhomologies (usually 3 to 8 bp long). Next, ssDNA gaps are filled in by cooperation of PARP1 and POLQ (DNA Polymerase Theta; Pol θ) and sealed by LIG1 or LIG3α/XRCC1. This pathway is error prone, mainly because of reduced fidelity of POLQ polymerase (328, 335, 370). As already mentioned, Alt-NHEJ is of particular relevance in the context of TC, as it is suggested to be important contributor in generation of IR-induced translocations (332, 333). Expression of POLQ is elevated in TC and HT tissues. Moreover, expression of POLQ (but not of TDG, PMS2, nor POLE), correlates with lymphocyte-infiltration markers in HT thyroids and healthy tissues surrounding PTC tumors. This might suggest involvement of POLQ in thyroid carcinogenesis in autoimmune thyroid disease background (42). Moreover, POLQ expression levels are positively correlated with dedifferentiation and unfavorable prognosis in PTC (371). Currently, POLQ inhibitors are developed as new potential anticancer therapeutics, and due to POLQ overexpression in thyroid cancer tissue there might be a potential application in TC (371). SSA is dependent on end resection and homology range (>50 bp) larger than that required in Alt-NHEJ. Pairing of the homologous sequence stretches is dependent on RAD52 and generates structure, in which noncomplementary 3′ ssDNA tails form flaps that are cleaved by XPF-ERCC1, DNA nicks are filled in by polymerase and ligated by LIG1. SSA generates deletions; therefore, it is potentially mutagenic, and has been associated with cancer-related mutagenesis. Moreover, as SSA connects short complementary sequences, it might lead to translocations through joining distant genomic repeat sequences (328, 335). RAD52 (that is also known as RDM1) SNPs might be associated with TC susceptibility (for detailed discussion see (83)). RAD52 is also overexpressed in TC and required to support survival and proliferation of cultured PTC cells (372). BIR repairs one-ended DSBs through long-range resection, RAD52- and RAD51-dependent strand-invasion and DNA repair synthesis using sister chromatid as a template. However, RAD51 is not absolutely required for BIR. RAD51-independent BIR involves, additionally, RAD54B/RDH54 and MRN (373). RAD54B is upregulated in PTC tissues (Fig. S1 (13)).

Fanconi Anemia Pathway

Fanconi anemia (FA) is a multigene genetic disorder characterized by physical abnormalities, including short stature, skin pigmentation anomalies, skeletal malformations, microcephaly, genitourinary tract anomalies, bone marrow dysfunction, and increased frequency of malignancy, including leukemia and solid tumors. Thirty to 60% of FA patients suffer from hypothyroidism (374). The proteins involved in FA pathway act in a DNA replication dependent-manner (ie, are restricted to S-phase) and are dedicated to resolve ICLs and protein/DNA complexes that are either physiological (eg, generated by lipid peroxidation products or as DNA repair-intermediates), or exogenous (ie, induced by crosslinking agents, such as cisplatin or mitomycin C). FA pathway proteins cooperate with other repair pathways, most notably with HR and translesion synthesis (TLS) and NER, while ICLs can be also resolved by NEIL3 in FA-independent manner. The FA pathway involves at least 23 FANCA-Y genes, as well as the genes encoding FAAPs (FA-associated proteins), in particular FAAP10, FAAP16, FAAP20, FAAP24, and FAAP100. FA pathway is initiated by the crash of the 2 convergent replication forks at ICL and recognition of this structure by the FA core complex, consisting of FANCA, FANCB, FANCC, FANCE, FANCF, FANCG/XRCC9, FANCL, FANCM, FANCT/UBE2T, and FAAPs. Next, FANCL and FANCT ubiquitin ligases ubiquitinate FANCD2-FANCI heterodimer. Monoubiquitinated FANCD2-FANCI complex promotes binding of the nucleases and accessory factors (ie, scaffold FANCP/SLX4 that recruits and regulates activity of structure-specific endonucleases XPF/FANCQ-ERCC1 complex, MUS81/EME1 complex, and SLX1, that is catalytic subunit of the structure-specific endonuclease SLX1-SLX4). Nucleases cleave one strand of the dsDNA at both sides of the ICL and unhook it. The next step involves insertion of the nucleotide opposite the unhooked ICL, and this is mediated by error-prone TLS polymerase, such as POLK, POLH, REV1, or POLZ (DNA Polymerase zeta; Pol ζ; composed of REV3 catalytic subunit and REV7/FANCV/MAD2L2 accessory subunit) and finally, DNA ends are sealed by DNA ligase. This process restores one of the initially replicated dsDNA, while the other, containing DSB resulting from ICL excision is restored by HR (375, 376).

In the context of the thyroid, FA patients frequently show hypothyroidism, which is the most strongly pronounced in FANCC and FANCG mutation carriers (377). Moreover, FANCD1/BRCA2, FANCD2, FANCA, FANCE, FANCG/XRCC9, and FANCQ/XPF/ERCC4 mutations were observed in both benign thyroid goiter and PTC tissues, suggestive of their involvement in progression of the disease from preneoplastic lesions to malignancy (16). Increased frequency of FANCA mutations is also observed in benign thyroid nodules (357). SNPs in FANCD1/BRCA2, FANCD2, FANCA, FANCQ/XPF/ERCC4, PALB2/FANCN, and FANCF were suggested to be associated with FNMTC (228). Cisplatin is one of the therapeutic options proposed in ATC management; therefore, combination treatment with cisplatin and FA pathway inhibitors may improve patient clinical outcome (378). Most of the FA pathway factors are in general upregulated in PTC (Fig. S1 (13)). FANCD2 is upregulated in HT tissues (77). Moreover, importantly, pediatric/adolescent hematologic or brain cancer patients treated with alkylating agents, such as busulfan, cyclophosphamide, or melphalan, inducing DNA crosslinks resolved mainly via NER, FA, and HR pathways, show increased incidence of hypothyroidism and TC later in life (379, 380).

Translesion Synthesis

Translesion synthesis (TLS) polymerases are specialized polymerases capable of DNA synthesis on a DNA template containing lesions. TLS polymerases include the Y-family polymerases POLH, POLI (Polymerase iota; Pol ι), POLK, and REV1, B-family POLZ/REV3L, X-family POLL and POLM, A-family POLQ, primase and DNA-directed polymerase from archaeo-eukaryotic primase family PRIMPOL/CCDC111. TLS polymerases lack stringent proofreading activity and their active sites are larger to give more space for damaged nucleotides or mismatches. Consequently, in most cases their activity leads to mutagenesis; therefore, it is strictly controlled and utilized in emergency situations. TLS is also referred to as DNA damage tolerance pathway (381). Recruitment of TLS polymerases is controlled by the ubiquitin ligase complex RAD18-RAD6 that monoubiquitinates PCNA at stalled replication forks, and this facilitates binding of TLS polymerases, since they exhibit higher affinity to monoubiquitinated PCNA (381).

Expression of TLS polymerases is disturbed in TC tissues. Specifically, low POLM expression correlates with PTC recurrence (311). POLH expression is elevated in BRAF^V600E-mutated PTC tissues (23). POLI is abundantly expressed in the thyroid but strongly downregulated in TC tissues (382) (see also Fig. S1 (13)). REV1 is also downregulated in TC tissue (383). Moreover, analysis of high-throughput data reveals that POLK, POLZ/REV3L, and PRIMPOL/CCDC111 are also downregulated, whereas REV7/FANCV/MAD2L2 is upregulated in PTC tissues (Fig. S1 (13)). Moreover, RAD18 is upregulated in BRAF^V600E-mutated PTC tissues, and elevated RAD18 expression correlates with unfavorable prognosis (384). Unbalanced levels of TLS polymerases were suggested to promote mutagenesis in cancer tissues. For example, RAD18 is overexpressed and positively correlated with mutation load in most of the analyzed cancers. Therefore, it was hypothesized that RAD18 supports low-fidelity TLS synthesis leading to elevated mutagenesis (381, 385). To our knowledge, there are no published studies focusing on TLS polymerases in AITD.

Concluding Remarks

Dysfunctional DNA repair is often detected in thyroid malignancies and autoimmune thyroid disease. However, despite extensive studies, multiple questions pertaining to the clinical significance of these findings remain to be answered. One of the issues not addressed previously is which DNA repair pathways are active in the healthy thyroid tissue. Surprisingly, there are no studies evaluating activities of distinct DNA repair pathways in normal thyroid cells. As already described, cell-doubling time of normal thyrocytes is extremely low, therefore it might be expected that they would behave as stationary terminally differentiated cells (46, 386). Neurons, myotubes, and adipocytes (that are terminally differentiated postmitotic cells) have attenuated HR, MMR, TLS, and ATR pathways, whereas ATM activation followed by NHEJ is the major way to resolve DSBs in those cells. On the other hand, BER, SSBR, and NER activities are either reduced or elevated depending on the cellular context (387-390). Based on data available from quiescent terminally differentiated cell models, we can speculate that normal adult thyrocytes would show diminished MMR, TLS, HR, and ATR pathway activity and respond to DSBs mainly via ATM and NHEJ. Accordingly, thyrocytes in stationary state respond to DSBs in ATM-dependent manner and are more resistant to DNA damage compared to cycling cells (252, 391). Increased proliferation rates are evident in most of the thyroid pathologies, including AITD, thyroid nodular disease, adenoma, and TC (386, 392, 393). This entails activation of repair pathways attenuated in normal thyrocytes, that are essential in dividing cells. Indeed, cultured ATC and PTC cell lines repair DSBs equally well via HR and NHEJ (21). Importantly, HR seems to be preserved in TC tissues because LOH (that is suggested to reflect homologous recombination deficiency) is observed with the lowest frequency in thyroid carcinoma among 33 analyzed tumor types in TCGA cohort (394). Moreover, interestingly, high-throughput transcriptomic data of HT tissues suggest activation of G₂/M and mitotic cell-cycle checkpoints with upregulation of HR-specific genes (eg, BRCA1, BLM, WEE1) (77). These considerations are particularly relevant in the context of the emerging therapeutic opportunities in thyroid pathology by application of DNA repair inhibitors.

Another subject not covered in this review is the interconnection between DNA repair and immune system. Developing immune cells utilize AID and DNA repair systems, most notably BER, MMR, DDR, and NHEJ, during maturation of antibodies and T-cell receptors. This underlies immunodeficiency disorders in individuals carrying inactivating mutations in the implicated genes, including UNG, MRE11, XLF, LIG4, and ATM (395). Several lines of evidence suggested that DDR might shape the immune landscape to promote autoreactive responses leading to autoimmunity (396). In consequence, application of DDR inhibitors was suggested in management of autoimmune diseases, but the potential clinical efficacy of DDR inhibitors in AITD remains to be analyzed (397). Moreover, in cancer cells, damaged DNA, resulting either from inborn or carcinogenesis-related acquired somatic DNA repair defects or DNA-damaging therapy (eg, induced by radiation, chemotherapeutics, DNA repair inhibitors), activates the cGAS-STING pathway-dependent antitumor immune responses and immune checkpoints (398, 399).

The interconnection between DNA repair and thyroid pathology leaves some unanswered questions; there is no consensus on the impact of genetic variation in several DNA repair genes, including XRCC1, ATM, MDM2, Ku80, DNA-PKcs, and BRCA1. Moreover, promising in vitro and in vivo data on the efficiency of PARP1, WEE1, POLQ, and LIG4 inhibitors, either alone or in combination with DNA-damaging chemotherapeutics, in TC therapy require evaluation in preclinical and clinical settings. So far, only 1 clinical trial evaluates PARP1 inhibitor olaparib in TC (NCT03162627). Additionally, association of genetic variation in POLE2, SAMHD1, CSB, XPA, XPG, ATM, SMARCAL1, XRCC4, BLM, FANCC, and FANCG with hypothyroidism observed in affected patients or murine models should be mechanistically analyzed. It should be also clarified if altered expression of DDR genes, including APOBEC3B, POLB, MTH1, TIP60, BRCA2, and POLQ, contributes to the thyroid immune infiltration in AITD, or is only a reflection of the intense ongoing inflammation. Finally, several DNA repair pathway genes (eg, RAD17, CtIP) have not been explored in thyroid pathologies, despite data from large-scale studies showing their altered expression in TC tissues. Inevitably, further exploration of DNA repair in thyroid physiology and pathology will bring new valuable data that will support current diagnostics and treatment of thyroid hyperplasia, autoimmunity, and malignancy (Fig. 5).

Figure 5.

Role of DNA damage and repair in thyroid physiology and disease. Thyroid DNA integrity is threatened by damaging agents resulting from normal physiological processes (such as ROS/H₂O₂ produced during thyroid hormone synthesis and normal thyrocyte metabolism), thyroid disease (thyroid cancer or autoimmunity), or environmental exposure (eg, radiation, pollutant chemicals). Robust antioxidative stress defense and DNA repair mechanisms protect thyrocyte genome integrity, but defective or dysregulated DNA repair pathways have been implicated in thyroid pathology. The insight into mechanisms governing thyrocyte genome integrity may help to gain better understanding of the pathology and suggest novel therapeutic regimens, urgently needed in treatment-refractory disease. Created in BioRender. Arczewska, K. (2025) https://BioRender.com/o09t899.

Abbreviations

2-oxoA

2-oxoadenine/isoguanine (iG)

3′-PUA

3′-phospho-α,β-unsaturated aldehyde

5′-dRp

5′-deoxyribose phosphate

6-4PP

6-4 photoproducts

8-oxoA

8-oxo-7,8-dihydroadenine (8-oxoadenine)

8-oxodG

8-oxo-7,8-dihydro-2′-deoxyguanine

8-oxodGTP

8-oxo-7,8-dihydro-2′-deoxyguanosine triphosphate

8-oxodGuo

8-oxo-7,8-dihydro-2′-deoxyguanosine

8-oxoG

8-oxo-7,8-dihydroguanine (8-oxoguanine)

8-oxoGuo

8-oxo-7,8-dihydroguanosine

AID

activation-induced cytidine deaminase

AITD

autoimmune thyroid disease

ALKBH2

AlkB Homolog 2, Alpha-Ketoglutarate Dependent Dioxygenase

ALKBH3

AlkB Homolog 3, alpha-ketoglutarate dependent dioxygenase

APE1/APEX1/HAP1/Ref-1

apurinic/apyrimidinic endodeoxyribonuclease-1, redox effector factor-1

APOBEC

apolipoprotein B mRNA editing catalytic polypeptide-like

APOBEC3A

apolipoprotein B mRNA editing enzyme catalytic subunit 3A

APOBEC3B

apolipoprotein B mRNA editing enzyme catalytic subunit 3B

AP-site

apurinic/apyrimidinic site

APTX

aprataxin

Artemis/DCLRE1C

DNA cross-link repair-1C

ASK1/MAP3K5

mitogen-activated protein kinase kinase kinase 5

ATC

anaplastic follicular cell-derived thyroid carcinoma

ATLD1

ataxia-telangiectasia-like disorder 1

ATM

ataxia telangiectasia mutated

ATR

ataxia telangiectasia and Rad3-related protein

ATRIP

ATR interacting protein

BARD1

BRCA1 associated RING domain-1

BER

base excision repair

BIR

break-induced replication

BLM/RECQL3/RECQ2

Bloom Syndrome RecQ Like Helicase

BRAF

B-Raf proto-oncogene, serine/threonine protein kinase

BRAF^V600E

B-Raf Proto-Oncogene carrying 600 valine to glutamic acid mutation

BRCA1/FANCS

BRCA1 DNA repair associated/breast cancer type 1 susceptibility protein

BRCA2/FANCD1

Breast Cancer Type 2 Susceptibility Protein/BRCA2 DNA repair associated

BS

Bloom syndrome

BSO

L-buthionine-sulfoximine

CAT

catalase

CAV-1

caveolin-1

CCDC6/H4

coiled-coil domain containing-6

cdA

8,5′-cyclo-2′-deoxyadenosine

cdG

8,5′-cyclo-2′-deoxyguanosine

CETN2

centrin-2

CHK1/CHEK1

checkpoint kinase-1

CHK2/CHEK2

checkpoint kinase-2

CPD

cyclobutane pyrimidine dimer

C-PTC

classic papillary thyroid carcinoma

CS

Cockayne syndrome

CSA/ERCC8

Cockayne syndrome protein CSA

CSB/ERCC6

Cockayne syndrome protein CSB

CtIP/RBBP8

C-terminal binding protein (CtBP)-interacting protein/retinoblastoma-binding protein 8

cyPu

8,5-cyclopurine

DDB1

damage specific DNA binding protein-1

DDB2/XPE

damage specific DNA binding protein-2

DDR

DNA damage response

DHGTC

differentiated high-grade thyroid carcinoma

DNA-PKc/PRKDC/XRCC7

DNA-dependent protein kinase catalytic subunit

DSB

double-strand break

dsDNA

double-stranded DNA

DUOX1

dual oxidase 1; ThOX1

DUOX2

dual oxidase 2; ThOX2

DUOXA2

dual oxidase maturation factor 2

DUT

deoxyuridine triphosphatase

EME1

essential meiotic structure-specific endonuclease-1

ERCC1

excision repair cross-complementation group-1

ETAA1

Ewing tumor associated antigen-1

EXO1

exonuclease-1

εA

1,N⁶-ethenoadenine

εC

3,N⁴-ethenocytosine

εG

N ²,3-ethenoguanine

FA

Fanconi anemia pathway

FAAP10/CENPX

Fanconi anemia-associated polypeptide of 10 KDa, centromere protein X

FAAP100

Fanconi anemia core complex associated protein 100

FANCA/FANCH

Fanconi anemia complementation group A

FANCC

Fanconi anemia complementation group C

FANCD2/FANCD

Fanconi anemia complementation group D2

FANCE

Fanconi anemia complementation group E

FANCF

Fanconi anemia complementation group F

FANCG/XRCC9

Fanconi anemia complementation group G

FANCI/KIAA1794

Fanconi anemia complementation group I

FANCJ/BRIP1/BACH1

Fanconi anemia complementation group J/BRCA1 interacting helicase-1

FANCL

Fanconi anemia complementation group L

FANCP/SLX4/BTBD12

Fanconi anemia complementation group P/SLX4 structure-specific endonuclease subunit

FANCT/UBE2T

Fanconi anemia complementation group T/ubiquitin conjugating enzyme E2 T

FaPyG

2,6-diamino-4-hydroxy-5-formamidopyrimidine

FEN1

flap structure-specific endonuclease-1

FNMTC

familial non-medullary thyroid cancer

FOXM1

Forkhead Box protein M1

FTC

follicular thyroid carcinoma

FTO

fat mass and obesity associated

GADD45A

growth arrest and DNA damage inducible alpha

GCLC

glutamate-cysteine ligase catalytic subunit

GD

Graves’ disease

GG-NER

global genome repair

GPX1

glutathione peroxidase 1

GPX2

glutathione peroxidase 2

GPX3

glutathione peroxidase 3

GPX4

glutathione peroxidase 4

GPX5

glutathione peroxidase 5

GPX6

glutathione peroxidase 6

GPX7

glutathione peroxidase 7

GPX8

glutathione peroxidase 8

GSH

reduced glutathione

GSR

glutathione reductase

GSSG

oxidized glutathione

GSTM1

glutathione S-transferase Mu 1

H2AX

H2A histone family member X

H₂O₂

hydrogen peroxide

HR

homologous recombination

H-RAS

Harvey rat sarcoma virus oncogene

HT

Hashimoto thyroiditis

HUS1

checkpoint clamp component HUS1

I⁻

iodide

I⁰

iodine radical

ICLs

interstrand crosslinks

IDLs

insertion/deletion loops

IEFVPTC

invasive encapsulated follicular variant papillary carcinoma

IR

ionizing radiation

KEAP1

Kelch like ECH associated protein 1

KRAS, K-RAS

Kirsten Rat Sarcoma Virus Proto-Oncogene

Ku70/XRCC6

X-ray repair cross complementing-6; thyroid autoantigen 70 kD (Ku antigen)

Ku80/XRCC5

X-ray repair cross complementing-5

LIG1

DNA ligase 1

LIG3α/LIG3

DNA Ligase 3alpha; DNA ligase-3

LIG4

DNA ligase-4

LOH

loss of heterozygosity

LP-BER

long-patch base excision repair

MAP

MUTYH-associated polyposis; familial adenomatous polyposis 2 (FAP2)

MBD4

methyl-CpG binding domain-4

DNA

glycosylase

MDA

malondialdehyde

MDC1

mediator of DNA damage checkpoint-1

MDM2

mouse double minute 2, human homolog

MDMX/MDM4

mouse double minute 4, human homolog

MGMT/AGT

O-6-methylguanine-DNA methyltransferase

MLH1

MutL homolog-1

MMR

mismatch repair

MNG

multinodular goiter

MRE11/MRE11A

meiotic recombination 11 homolog 1

MRN

trimeric MRE11-RAD50-NBS1 complex

MSC

mesenchymal stem/stromal cells

MSH2

MutS homolog-2

MSH3

MutS homolog-3

MSH5

MutS homolog-5

MSH6

MutS homolog-6

MSI

microsatellite DNA instability

MTC

medullary thyroid carcinoma

MTH1/NUDT1

MutT homologue 1; nudix hydrolase 1; 8-Oxo-dGTP triphosphatase

MUS81

MUS81 structure-specific endonuclease subunit

MUTYH

A/G-specific adenine DNA glycosylase

N ⁷ -meG

N ⁷ -methylguanine

NAP

NTHL1-associated polyposis; familial adenomatous polyposis 3 (FAP3)

NBS1/NBN

Nijmegen breakage syndrome-1/Nibrin

NEIL2

Nei-Like DNA Glycosylase-2

NEIL3

Nei-Like DNA Glycosylase-3

NER

nucleotide excision repair

NHEJ

non-homologous end-joining

NIS

SLC5A5; sodium-iodide symporter

NMTC

non-medullary thyroid cancer

NO

nitric oxide

NOX

NADPH oxidase

NOX1

NADPH oxidase 1

NOX2

NADPH oxidase 2

NOX3

NADPH oxidase 3

NOX4

NADPH oxidase 4

N-RAS

neuroblastoma RAS virus oncogene

NRF2/NFE2L2

nuclear factor erythroid 2-related transcription factor 2

NTH1/NTHL1

endonuclease III-like protein-1

NTRK1

neurotrophic tyrosine receptor kinase 1

NTRK3

neurotrophic tyrosine receptor kinase 3

NUDT11

Nudix hydrolase-11

NUDT13

Nudix hydrolase-13

NUDT14

Nudix hydrolase-14

NUDT15/MTH2

Nudix hydrolase-15; MutT homologue 2

NUDT16

Nudix hydrolase-16

NUDT18/MTH3

Nudix hydrolase-18; MutT homologue 3

NUDT19

Nudix hydrolase-19

NUDT2

Nudix hydrolase-2

NUDT20/DCP2

Nudix hydrolase-20

NUDT21

Nudix hydrolase-21

NUDT22

Nudix hydrolase-22

NUDT3

Nudix hydrolase-3

NUDT4

Nudix hydrolase-4

NUDT5

Nudix hydrolase-5

NUDT6

Nudix hydrolase-6; FGF-AS

NUDT8

Nudix hydrolase-8

O₂^·−

superoxide anion

O⁴-meT

O ⁴-methylthymine

O⁶-meG

O ⁶-methylguanine

OGG1

8-oxoguanine DNA glycosylase-1

OS

oxidative stress

p53

tumor protein P53 protein

PALB2/FANCN

partner and localizer of BRCA2

PARP1

poly(ADP-ribose) polymerase-1

PARP2

poly(ADP-ribose) polymerase-2

PAX8

paired box 8 transcription factor

PCNA

proliferating cell nuclear antigen

PDS

SLC26A4; pendrin

PDTC

poorly differentiated thyroid carcinoma

PIK3CA

phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha

PMS2

postmeiotic segregation increased (PMS) homolog-2

PNKP

polynucleotide kinase 3′-phosphatase

POLB

DNA polymerase beta; Pol β

POLD

DNA polymerase delta; Pol δ

POLD1

DNA polymerase delta 1, catalytic subunit

POLE

DNA polymerase epsilon; catalytic subunit; Pol ε

POLE2

DNA polymerase epsilon 2, accessory subunit

POLE3

DNA polymerase epsilon 3, accessory subunit

POLE4

DNA polymerase epsilon 4, accessory subunit

POLH

DNA polymerase eta; Pol η; Pol ι

POLK/DINB1

DNA polymerase kappa; Pol κ

POLL

DNA polymerase lambda; Pol λ

POLM

DNA polymerase mu; Pol µ

POLQ

DNA polymerase theta; Pol θ

POLZ

DNA polymerase zeta; Pol ζ

POLZ/REV3L

REV3 like, DNA directed polymerase zeta catalytic subunit

PRDX1/PRX1

peroxiredoxin 1

PRDX2/PRX2

peroxiredoxin 2

PRDX3/PRX3

peroxiredoxin 3

PRDX4/PRX4

peroxiredoxin 4

PRDX5/PRX5

peroxiredoxin 5

PRDX6/PRX6

peroxiredoxin 6

PRIMPOL/CCDC111

primase and DNA directed polymerase

PTC

papillary thyroid carcinoma

PTEN

phosphatase and tensin homolog

RAD1

checkpoint DNA exonuclease RAD1

RAD17

checkpoint clamp loader component RAD17

RAD18

RAD18 E3 ubiquitin protein ligase

RAD23B

UV excision repair protein RAD23 homolog-B

RAD50

Homolog of S. cerevisiae Rad50

RAD51/FANCR

RAD51 recombinase

RAD51B/RAD51L1

RAD51 paralog B

RAD51C/RAD51L2/FANCO

RAD51 paralog C

RAD51D/RAD51L3

RAD51 paralog D

RAD52/RDM1

RAD52 homolog, DNA repair protein

RAD54/RAD54L

DNA repair and recombination protein RAD54-like

RAD54B/RDH54

DNA repair and recombination protein RAD54B

RAD9A/RAD9

cell-cycle checkpoint control protein RAD9A

RAI

radioactive iodine ablation therapy

RECQL1/RECQL

DNA helicase, RecQ-like type-1

RECQL4

RecQ like helicase-4

RECQL5

RecQ like helicase-5

RET::PTC

rearrangement of the RET proto-oncogene gene with its fusion partner

RET::PTC1

fusion of the RET proto-oncogene with coiled-coil domain containing 6; CCDC6

RET::PTC3

fusion of the RET proto-oncogene nuclear receptor co-activator 4; NCOA4

REV1/REV1L

REV1 DNA directed polymerase

REV7/FANCV/MAD2L2

polymerase (DNA-directed), zeta 2, accessory subunit

RFC

replication factor C

RFC1

replication factor C subunit 1

RFC2

replication factor C subunit

RFC3

replication factor C subunit 3

RFC4

replication factor C subunit 4

RFC5

replication factor C subunit 5

RNF8

ring finger protein-8

RNS

reactive nitrogen species

ROS

reactive oxygen species

RPA1/RPA70

replication protein A 70-kDa DNA-binding subunit, replication factor A protein-1

RPA2/RPA32

replication protein A 32-kDa subunit; replication factor A protein-2

RPA3/RPA14

replication protein A 14-kDa subunit; replication factor A protein-3

SAMHD1

SAM and HD domain containing deoxynucleoside triphosphate triphosphohydrolase-1

SMARCAL1

SWI/SNF-related, matrix-associated, actin-dependent regulator of chromatin, subfamily a-like protein-1

SMUG1

single-strand-selective monofunctional uracil-DNA glycosylase-1

SNP

single nucleotide polymorphism

SOD

superoxide dismutase

SOD1

superoxide dismutase 1

Cu/Zn superoxide dismutase

cytosolic

SOD2

superoxide dismutase 2; Mn superoxide dismutase, mitochondrial

SOD3/EC-SOD

superoxide dismutase 3, extracellular Cu/Zn-dependent

SP-BER

short-patch base excision repair

SSA

single-strand annealing

SSB

single-strand break

SSBR

single-strand break repair

ssDNA

single-stranded DNA

T₃

3,3′,5-triiodo-L-thyronine

T₄

3,3′,5,5′-tetraiodo-L-thyronine, thyroxine

TC

thyroid cancer

TCGA

The Cancer Genome Atlas Program

TC-NER

transcription-coupled repair

TCOF1

Treacher Collins-Franceschetti syndrome-1

TC-PTC

tall cell papillary thyroid carcinoma

TDG

G/T mismatch-specific thymine DNA glycosylase

TDP1

tyrosyl DNA phosphodiesterase-1

TDP2

tyrosyl DNA phosphodiesterase-2

TERTp

telomerase reverse transcriptase promoter

TFIIH

transcription initiation factor IIH

Tg

thymine glycol

TG

thyroglobulin

TH

thyroid hormone

TIP60/KAT5

lysine acetyltransferase-5

TLS

translesion synthesis

TMZ

temozolomide

TOP1

DNA topoisomerase 1

TOP2

DNA topoisomerase 2

TOPBP1

DNA topoisomerase II binding protein-1

TP53

tumor protein P53 gene

TP53BP1/53BP1

tumor protein P53 binding protein-1

TPO

thyroid peroxidase

TSHR

TSH receptor

TTD

trichothiodystrophy

TTF-1

thyroid transcription factor-1

TXN2/TRX2

thioredoxin 2, mitochondrial

TXNRD1/TRXR1

thioredoxin reductase 1

TXNRD2/TRXR2

thioredoxin reductase 2

TXNRD3/TRXR3

thioredoxin reductase 3

UNG

uracil DNA glycosylase

UV

ultraviolet radiation

UVSSA/KIAA1530

UV-stimulated scaffold protein A

WDTC

well-differentiated thyroid cancer

WEE1

G2 checkpoint kinase WEE1

WRN/RECQL2/RECQ3

Werner syndrome helicase; WRN RecQ like helicase

XLF/Cernunnos/NHEJ1

XRCC4-like factor; non-homologous end joining factor-1

XP

xeroderma pigmentosum

XPA

xeroderma pigmentosum, complementation group A

XPB/ERCC3

xeroderma pigmentosum, complementation group B

XPC

xeroderma pigmentosum, complementation group-C

XPD/ERCC2

xeroderma pigmentosum, complementation group D

XPF/ERCC4/FANCQ

xeroderma pigmentosum, complementation group F

XPG/ERCC5

xeroderma pigmentosum, complementation group G

XPV/POLH

DNA Polymerase Eta; Pol η

XRCC1

X-ray repair cross-complementing protein-1

XRCC2/FANCU

X-ray repair cross complementing-2

XRCC3/CMM6

X-ray repair cross complementing-3

XRCC4

X-ray repair cross complementing-4

Funding

This work was supported by the Centre of Postgraduate Medical Education grant nos. 501-1-025-01-23 and 501-1-025-01-24.

Author Contributions

K.D.A.: conceptualization, writing, researching the literature, preparation of drawings, substantial editing, and revising draft manuscript critically for important intellectual content; D.S.: data analysis, preparation of drawings, and revising manuscript critically for important intellectual content; H.L.N.: substantial editing and revising draft manuscript critically for important intellectual content; A.P.W.: supervision, conceptualization, writing, substantial editing, and revising draft manuscript critically for important intellectual content. All authors approved the final manuscript.

Disclosures

The authors have nothing to disclose.