The fine art of vascular wall maintenance. Carriership of XPC, TP53 and APOE polymorphisms may be a risk factor for cerebral vascular accidents in the Bulgarian population

Abstract Efficient maintenance of the integrity of the endothelium of cerebral blood vessels is crucially important, especially when the vessel walls are subjected to greater-than-normal levels of stress. Persistently high levels of genotoxic stress may result from lower capacity to detect and repair DNA damage conferred by carriership of variants of key genes of DNA repair/maintenance of genomic integrity. Adult Bulgarian patients with a history of cerebral vascular accidents (CVAs) and age-matched healthy controls were analyzed for 11 markers, including 7 DNA polymorphisms in genes coding for proteins of DNA repair and maintenance of genomic integrity, 3 hypercoagulability markers and 1 marker for susceptibility for cerebral amyloidosis. Homozygous carriership of the del allele of the polymorphism XPCins83 was associated with decreased risk of CVAs (RR = 0.446, 95% CI:0.225–0.886, p = 0.021). In individuals carrying the 'protective' del/del XPCins83 genotype, carriership of the 'pro-apoptotic' genotype Arg/Arg in the TP53 locus was associated with increased risk of CVAs (RR = 1.845, 95% CI:1.049–3.244, p = 0.034). Carriers of the Lys751Gln genotype at the ERCC2 locus were at increased risk of CVAs (RR = 2.055, 95% CI:1.09–3.876, p = 0.025). Carriership of the E2/E3 genotype at the APOE locus decreased the risk of CVAs in Bulgarian males (RR = 0.279, 95% CI:0.090–0.873, p = 0.028). Male Bulgarian carriers of the APOE4 allele were at increased risk of CVAs. Carriership of the common prothrombotic mutations Factor V Leiden, PT G20210A, MTHFR C677T and PAI1 4G/5G had no significant effect on the risk of CVAs in Bulgarian adults.


Introduction
There's birth, there's death and in between there's maintenance.
Tom Robbins, Fierce Invalids Home from Hot Climates (2000) left ventricular hypertrophy [2]. Other known risk factors for stroke may include hypercoagulability; impaired lipid profile and atherosclerotic and/or amyloid involvement of cerebral blood vessels [4,5]. Cerebral amyloid angiopathy (CAA), already shown to be a major risk factor for Alzheimer's disease (AD), was also identified as a culprit in ischemic CVAs as well as in brain haemorrhages [6,7]. CVAs are typically more common in males than in age-matched females. The chances for survival after CVA are generally lower for males than for females but female survivors are usually left with more disability than males [8,9]. The gender bias in the risk of stroke may result, at least partly, from a variety of potentially modifiable factors that are more likely to be observed in males than in females. Males are more likely to smoke at any age, are less likely to adhere to a strict therapeutic regimen and generally have higher levels of total and LDL cholesterol and lower levels of HDL cholesterol than age-matched females [10], although these differences tend to decrease with age.
Despite the fact that increased risk of stroke may run in families and the detailed risk profiles outlining the modifiable risk factors, it is still difficult to determine the individual risk of stroke. In most cases the risk estimates may safely predict the risk for a group but not for each and every individual in a group.
The capacity of the endothelium to maintain its integrity under increased hydrodynamic stress may constitute yet another type of factor in the assessment of the risk for vascular accidents. Whether the endothelium of the blood vessel would maintain its integrity under adverse conditions (increased levels of oxidative stress, chronic inflammation, accumulation of atherosclerotic plaque and/or amyloid) may be difficult to predict for each and every patient. This factor is stochastic in its nature but it is, nevertheless, determined at least partly by genetic factors, including the above-mentioned inherited defects of lipid metabolism and hypercoagulability. The role of the capacity to manage genotoxic damage has only recently begun to gain the attention of researchers and clinicians as a major factor in human disease.
The sources of oxidative damage in living cells are many and varied, but for the purposes of classification, they fall into two large groups: endogenous (product of normal cell metabolism) and exogenous (resulting from environmental influences such as chemicals, ionizing radiation, etc.). Endogenous oxidative damage is very common, as reactive oxygen species (ROS) are normal by-products of oxidative phosphorylation in living cells. In young and healthy people, genotoxic damage is promptly repaired by the cellular DNA repair machinery. Cells that have sustained damage to their DNA are normally excluded from division until the levels of damage are assessed and DNA is repaired. If the damage is assessed to be too extensive and cannot be repaired, the damaged cell may be routed to programmed cell death. In the latter case, a replacement may be called forth by activating the stem cell niche to produce new differentiated cell/s. With age, however, the capacity to identify and repair DNA damage declines and unrepaired damage begins to accumulate. This may result in increased risk of carcinogenic transformation (if cells with damaged DNA are allowed to divide) or in degenerative disease (if the stem cell niches are unable to provide adequate rates of replacement of damaged cells). The development of degenerative disease may greatly accelerate if the influx or ROS generated by oxidative phosphorylation is too great (e.g. in carriers of mitochondrial haplogroups with ineffective oxygen utilization) and/or when the mechanisms that identify and manage genotoxic (specifically, oxidative) damage are operating with lower-than-normal efficiency.
In sites of vascular ischemia ROS are generated in ample amounts by damaged cells and immune system cells are recruited at the lesion site. Post-mortem studies of post-stroke murine and human brains showed that enzymatic activities of BER (excision of modified bases, excision of apurine/apyrimidine sites) were greatly decreased compared to age-matched individuals with cause of death other than stroke [11].
Several polymorphisms in human genes coding for proteins of DNA repair that are associated with a subtle decrease of the capacity to manage oxidative damage have been described. Polymorphisms in the human OGG1 gene, POLB, the NEIL gene family and XRCC2 and XRCC3 were identified as the potential culprit in common diseases of middle and advanced age such as cancer and type 2 diabetes [12,13]. Thus, a potential link between metabolic syndrome and cancer was proposed [14,15]. Carriership of polymorphic variants of genes coding for proteins functioning in repair of oxidative lesions (hOGG1 and XRCC1) as well as proteins functioning in NER (XPD (ERCC2)) were later demonstrated to be a risk factor for stroke, with variant alleles of the XRCC1 gene being associated also with stroke volume [16][17][18][19][20].
In individuals with genetic backgrounds predisposing to less efficient recognition and repair of DNA damage, persistent genotoxic attack resulting from plaque accumulation in the vascular wall, the accompanying chronic inflammation and, eventually, ischemia may result in increased endothelial attrition and/or delayed cell replacement. This may generate localized areas within the vessel wall where the shear stress becomes critically low and the risk for occurrence of an integrity breach may be greatly increased. The latter may result in platelet activation and formation of thrombi at the vascular injury site, producing acute ischemia. Under ischemic conditions, the levels of production of ROS increase even further, augmented by the oxidative bursts generated by immune cells recruited at the site of the injury. Increased levels of genotoxic damage cause mass cell death at the lesion site and the stroke penumbra, peaking at 24 h post-accident [21]. Cells that are located at significant distance from the injury site (e.g. the Purkinje cells in the cerebellum, the hippocampus and the basal ganglia) may also be susceptible to cell death after a vascular accident [22]. This may occur over several weeks or months after the accident, resulting in post-stroke cognitive decline and/or poor recovery of motor skills. Again, those with less-than-average capacity to manage genotoxic damage may be more prone to delayed and/or long-distance effects than individuals with normal or superior capacity for repair of genotoxic damage.
The primary aim of this study was to assess whether individual capacity to identify and repair DNA damage plays a role in the constitution of risk for CVAs in the Bulgarian population. Seven genetic polymorphisms in genes coding for key proteins of DNA repair and maintenance of genomic integrity have been selected, namely 53 Pro72Arg, XPC 83ins, ERCC1 C8092A, XRCC3 Thr241Met, XRCC1 Arg399Gln, XPD Lys751Gln and MTHFR C677T. Among these polymorphisms there were markers specific for BER (XRCC1 Arg399Gln) and repair of double-strand breaks (XRCC3 Thr241Met) as representatives of the most typical lesions caused by ROS. NER-specific polymorphisms (XPCins83, XPD Lys751Gln and ERCC1 C8092A) were also included, as most types of oxidative damage may potentially be repaired by NER. The panel included the TP53 polymorphism Pro72Arg as a marker of the capacity for maintenance of genomic integrity. In order to differentiate the effects of other genetic factors that may increase the risk of stroke, an analysis of the potential effects of the common prothrombotic markers FV Leiden, PT G20210A, PAI1 4G/5G and the APOE locus (responsible for variations in the lipid profile and the propensity for cerebral amyloidosis) was also carried out. The MTHFR C677T polymorphism has initially been described as a genetic factor for thrombosis and vascular disease but is also a key enzymatic activity in the detoxification of genotoxic agents [23,24]. Thus, we included the MTHFR C677T polymorphism in the study as a marker for individual capacity for management of genotoxic damage as well as a marker for risk of thrombosis.

Materials and methods
The initial study groups comprised 73 diabetes-free individuals (41 males, 32 females) with a history of CVAs (ischemic strokes) and 201 clinically healthy volunteers (123 males, 78 females), all of Bulgarian ethnical origin. All patients with CVAs were referred by the Clinic of Neurology at Medical University Hospital 'Alexandrovska' (Sofia). Written informed consent was obtained from all patients and volunteers prior to inclusion in the study. In both study groups, the target male-to-female ratio was 1:1; therefore, the initial groups have been restructured in order to achieve a male-to-female ratio maximally close to 1:1. After initial selection, four patients with CVAs were excluded from the study, two because of distant relatedness to other patients in the group and two because of ages below 20 or above 65 at the time of first CVA. Thus, the patients' group eventually comprised 70 individuals, of which 38 were male and 32 were female. The control group was selected out of the initial group of healthy volunteers so that there was maximal match in the age distribution to the patients' group (in order to minimize the effect of advanced age). Finally, the control group comprised 93 clinically healthy individuals (45 males, 48 females).

Results and discussion
The frequencies of the wild-type and variant alleles for the 11 markers included in the study, the observed and expected heterozygosity and the inbreeding index for each marker were calculated for the two study groups. The results are presented in Table 1 for the control group and in Table 2 for the group of patients.
The panel of seven markers of individual capacity for DNA repair has been specifically selected to reflect the overall state of the basic mechanisms of DNA damage detection and repair: XPCins83, for the early stages of global genomic repair by NER; ERCC1 C 8092 A and XPD Lys751Gln, for the late stages of NER; XRCC1 Arg399Gln, for BER; XRCC3 Thr241Met, repair of double-strand breaks; TP53 Pro72Arg, maintenance of genomic integrity; and MTHFR C677T, detoxification of potentially genotoxic agents. Carriership of APOE variant alleles as a major risk factor for the development of CAA was also studied in conjunction with the capacity to manage genomic damage as an additional risk factor. Our previous works with several of these polymorphisms [37][38][39] have shown that there was some level of association of the carriership of their variant alleles with the risk of stroke in the Bulgarian population but the potential impact of a combination genotype has not been fully assessed until now. In the present study, we also factored in the risk of potential bias due to effects of other common genetic risk factors for thrombosis and vascular disease (FV Leiden, PT G20210A, MTHFR C677T, PAI1 4G/5G). Only results with significance level 0.05 are discussed in detail below.

TP53 Pro72Arg
The frequencies of the two allelic variants of the TP53 Pro72Arg polymorphism were apparently similar in the  two study groups (Tables 1 and 2) and was following the pattern of clinal distribution described earlier [40,41]. The homozygous 'pro-apoptotic' Arg/Arg genotype was, overall, the most common genotype in both study populations (0.46 in the control group vs. 0.42 in the patients' group), followed by the heterozygous 72Pro/Arg genotype (0.42 in the control group vs. 0.44 in the patients' group). The 'pro-repair' Pro/ Pro genotype was observed with a frequency of 0.12 in the control group and 0.11 in the patients' group, with expected frequency of the Pro/Pro genotype in both groups about 0.11. Thus, from the locus-by-locus analysis, it could be concluded that the TP53 Pro72Arg polymorphism did not have any perceptible effect on the risk for CVAs in our study groups. Nevertheless, analysis of the distribution of compound genotypes indicated that this polymorphism may, in fact, be a weak risk modifier. In patients with CVA carrying the 'protective' del/del XPC genotype (see below), concomitant carriership of the homozygous Arg/Arg genotype of the TP53 locus was quite common (0.35), whereas in the control group concomitant carriership of the del/del:Arg/Arg genotype was less common (0.22). In other words, carriers of the del/del genotype in the control group were more likely to carry also 72Pro allele-containing genotypes, whereas del/del homozygotes in the patients' group were more likely to carry an Arg/Arg genotype. Thus, homozygous carriership of the 'pro-apoptotic' homozygous genotype by the TP53 Pro72Arg locus may be a risk factor for CVAs in the Bulgarian population (RR ¼ 1.845, 95% CI:1.049-3.244), p ¼ 0.034, albeit it confers smaller risk of CVA than carriership of the insertion allele of the XPCins83 polymorphism. The role of the carriership of the Arg allele of the TP53 Pro72Arg polymorphism in the susceptibility to neuronal damage after brain ischemia has already been demonstrated [42]. Recently, it has been shown that carriership of the 72Pro allele was associated with improved outcome after CVAs with regard to functional recovery and that it exerted a beneficial effect in patients with transient ischemic attacks, a CVA-related condition that greatly increases the risk of stroke [43]. It is possible that in the Bulgarian population this polymorphism is inferior to XPCins83 with regard to its significance as a genetic risk factor for CVAs.

XPCins83
In the control group, the allelic frequency for the deletion (del, repair-proficient) allele was 0.657 and that for the insertion (ins, repair-deficient) allele was 0.343. In the patients' group, the del allele was less common (0.587) and the ins allele was more common (0.413) than in the control group. There was a significant difference in the observed and expected distribution of XPCins83 genotypes within groups as well as between the study groups. Specifically, there was a statistically significant (77%) difference between the expected and the observed heterozygosity of the XPCins83 marker in the control group (H exp ¼ 0.454, H obs ¼ 0.257 at p ¼ 0.0007). The proportion of clinically healthy individuals carrying the del/del genotype was 0.52, which was 21% higher than what could be expected given the allelic frequencies (0.43). For comparison, the difference between H exp and H obs in the patients' group was only about 7%. The inbreeding coefficient in the control group was strongly positive (F is ¼ 0.435), whereas the F is in the patients' group was weakly negative (À0.039). The del/del genotype was most common in the control group (0.40), whereas the heterozygous ins/del genotype was most common in the patients' group (0.50). Apparently, in the control group there was an excess of del/del homozygotes at the expense of del/ins heterozygotes. The RR for the carriers of the 'protective' homozygous del/del genotype was 0.446 (95% CI:0.225-0.886), p ¼ 0.021.

XPD Lys751Gln
There was an excess of Lys751Gln heterozygotes in both groups, with the trend being more pronounced in the patients' group. The excess of heterozygotes was 12% more than expected in the control group and 29% in the patients' group than what could be expected by chance, given the allelic frequencies. This trend reflected on the value of F is . The latter was weakly negative in the control group (À0.120) and strongly negative (À0.280) in the patients' group. The RR for the carriers of the heterozygous genotype was 2.055 (95% CI:1.09-3.876), at p ¼ 0.025.
The distribution of the allelic frequencies and the different genotypes for the polymorphisms ERCC1 C8092A, XRCC1 Arg399Gln, XRCC3 Thr241Met and MTHFR C677T in the stiudied groups was unremarkable. Other authors have reported that upregulation of the expression of several proteins of repair by base excision (APE1, OGG1, XRCC1) improved long-term functional recovery after stroke [44]. It is possible that these polymorphisms do not play significant roles in the constitution of the risk for CVAs in the Bulgarian population.

Common prothrombotic factors
Carriership of the Leiden mutation was quite common in both groups. The frequency of the A allele in the control group was 0.05, which corresponds to the frequency previously reported for the Bulgarian population [32,45]. The frequency of the A allele was slightly higher in the patients' group (0.06). Homozygous carriers of the Leiden mutation were not identified in either of the study populations. Heterozygotes for the Leiden mutation were seen at a rate of 0.10 in the control group and 0.13 in the patients' group.
The frequency of the A allele of the PT G20210A mutation was 0.016 in the control group and 0.043 in the patients' group (2.68 times more frequently observed in patients with CVA than in healthy controls). Homozygotes for the PT G20210A mutation were not identified in either group. Data published elsewhere suggest that the PT G20210A mutation is a stronger risk factor for stroke than Factor V Leiden [46]. Still, as the frequency of the mutated allele is quite low, these results could not be considered significant. Heterozygous carriers were seen at a rate of 0.03 in the control group and 0.085 in the patients' group. Compound heterozygotes (carriers of both Factor V Leiden and PT G20210A) were not observed in the control group, but were identified in the patients' group at a rate of 0.028. The experimental findings about the distribution of the alleles of the PAI1 4G/5G polymorphisms in our study groups were not statistically significant (p ¼ 0.109), although the prevalence of homozygotes for the 5 G allele in the patients' group was unexpectedly low (0.015) at expected 5G/5G prevalence of 0.11.

APOE E2/E3/E4 polymorphic system
The frequency of the 'baseline' E3 allele (i.e. not associated with any perceptible effects on the phenotype) was lower in the control group than in the patients' group (0.812 vs. 0.871). The frequency of the 'low-risk' E2 allele was twice as high in the control group as in the patients' group (0.091 vs. 0.043). The frequency of the 'high-risk' E4 allele was higher in the control group than in the patients' group (0.097 vs. 0.086). Homozygotes by either of the variant alleles were not seen in the control group, whereas a single E2/E2 carrier and two E4/E4 carriers were identified in the group of patients. The latter makes up for 2.8% of the patients' group, whereas the expected percentage of E4/E4 homozygotes was less than 1%, given the allelic frequencies. The rare E2/E4 genotype was not identified in either group. There was a slight excess of heterozygotes in the control group (H obs ¼ 0.376, vs. expected 0.325) and, respectively, F is in the control group was negative (À0.159). In the patients' group, heterozygotes of any type were significantly less common than in the control group (0.171) at expected heterozygocity rates of 0.232, resulting in a positive F is of 0.266. Carriers of a single E4 allele were, overall, more common in the control group than in the patients' group (0.193 vs. 0.114). Carriers of a single E2 allele (E2/E3 genotype) constituted 0.183 in the control group and 0.057 in the control group (3.2-fold difference in the prevalence of E2/E3 genotype in the control group). Carriership of the E2/E3 genotype may be associated with decreased risk for CVAs in the Bulgarian population (RR ¼ 0.279 (95% CI:0.090-0.873), p ¼ 0.028).
The distribution of APOE genotypes in our study groups also showed specificities with regard to maleto-female ratio. The two study groups were specifically selected to have a male-to-female ratio as close as possible to 1:1. In the patients' group, the male-tofemale ratio among the E4 allele carriers was 3:1, whereas in the control group the prevalence of males and females with a single E4 allele was very close to 1:1. The excess of E4 heterozygotes in the control group was apparently due to a larger proportion of E4-carrying clinically healthy females. This may mean that the CVA risk for a male carrier of E4 allele may be higher than the risk for a female E4 carrier.
The sex-dependent effects of carriership of variant APOE alleles have been extensively studied in relation to the risk of AD. Generally, it has been agreed that female E4 carriers are more likely to develop both early-onset and late-onset AD (analyzed in detail in [47,48]). Our results confirm that the effects of the carriership of the 'high-risk' E4 allele may be modulated by biological gender. It could be speculated that the contribution of the carriership of the E4 allele to the risk for CVAs in males is due to modulation of the inherent gender-associated risk for disorders in lipid metabolism.
It was already mentioned that the E2/E3 genotype was >3 times more common in the control group than in the patients' group. There was also a peculiar male-to-female ratio among the E2/E3 carriers, with male E2/E3 carries being 4 times less common in the patients' group than females, whereas in the control group male E2/E3 carriers were two times less common than females (a twofold difference). It could be speculated that in the Bulgarian population the risk for CVAs is lower for a male E2 carrier than for a female carrier. Again, this effect may result from the modulation of the a priori higher risk of vascular disease in males due to impaired metabolism.
The beneficial effects of E2 carriership apparently extend only to carriers of a single allele. Only a single E2/E2 homozygote was identified in both our study groups, a male with early-onset vascular disease (<50 at the age of first stroke).

Potential genotype-phenotype correlations
It could be speculated that homozygous carriership of the del allele of the XPCins83 polymorphism exerts some protective effect against CVAs in the Bulgarian population. XPC, as part of the XPC-hHR23ffl complex is one of the first proteins to arrive at the site of lesions in DNA, signalling to the nucleotide excision repair machinery that there is unrepaired damage [48]. This mechanism of activation of NER, however, plays a role only in the repair of untranscribed DNA. There are cell types (long-lived cells such as differentiated neurons and memory B-cells)) that selectively suppress the repair of untranscribed DNA, focusing all repair capacity on transcribed DNA [49,50]. Therefore, unrepaired damage in DNA would matter significantly only if it happened to be regions undergoing active transcription. Repair of damage in untranscribed regions of the genome, however, may matter significantly in cell types that are naturally subjected to a rapid turnover. The endothelium of the vascular wall is normally replaced rapidly and is, therefore, strongly dependent on prompt repair of damage in untranscribed DNA. Subtle deficiencies in the recognition and/or repair of damage in untranscribed regions may not have significant effects on the phenotype in younger individuals but may result in accumulation of unrepaired lesions as age advances, when the capacity for repair of genotoxic damage naturally declines. Presence of unrepaired damage in the genome of a rapidly cycling cell may result in induction of mechanisms of programmed cell death and, potentially, activation of the endothelial stem cell niche in order to provide replacement cells. Adult stem cells, however, are also subject to aging and may be unable to keep up the pace of cellular replacement, especially if there is additional genotoxic load and/or subtle but lifelong deficiency in the mechanisms for management of genotoxic damage. Thus, the vascular wall in individuals with subtle deficiencies of DNA repair may be prone to destabilization. Unrecognised/untreated or poorly treated hypertension, increased oxidative stress due to persistent hyperglycaemia and/or presence of atherosclerotic or amyloid plaque may provide additional stress on a vascular wall that is already failing in its integrity.
The heterozygous ins/del genotype in the XPC locus was significantly more frequent in the patients' group than in the control group. No statistically significant effects were observed for the distribution of the homozygous ins/ins genotype. Similarly, statistically significant effects of heterozygous carriership of variant alleles on the risk of CVAs have also been identified for the polymorphism XPD Lys751Gln. Notably, association between heterozygosity for the latter polymorphism and senile cataract has already been demonstrated [51]. Presumably, higher-grade deficiency of NER (conferred by carriership of two variant allele copies) may be perceived and compensated for (e.g. by means of upregulation of expression of key repair proteins, etc.) or effects similar to negative allele complementation may also play a role.
Carriership of a single E2 allele of the APOE polymorphism may confer increased protection from CVA in the Bulgarian population, the effects being specifically relevant to male carriers. Carriership of the A4 in the APOE locus may modulate the risk of CVAs in male carriers of APOE4 in the Bulgarian population. Considering that carriership of variant APOE alleles is associated with modulation of the lipid metabolism profile (E4, unfavourable; E2, favourable), it is likely that APOE exerts its effects in the Bulgarian population by modulating the effects of some of the major factors for vascular disease that exhibit sex-dependent differences.

Conclusions
Carriership of homozygous del/del XPCins83 genotype may be associated with decreased risk of CVAs in the Bulgarian population. Carriership of variant 'repairdeficient' genotypes (heterozygous XPCins83 ins/del, XPD Lys751Gln) or 'pro-apoptotic' (TP53 Arg/Arg) genotypes may increase the risk for vascular disease and vascular accidents, as they contribute to destabilization of the endothelial layer of the vascular wall by interfering with normal cellular turnover. This effect may potentially increase as age advances and/or when the individual develops diseases and conditions associated with increased level of endogenous oxidative damage (e.g. metabolic syndrome/type 2 diabetes). Carriership of a single E2 allele in the APOE locus may protect from CVAs, whereas carriership of a single E4 allele may increase the risk of CVAs in the Bulgarian population, both effects pertaining predominantly to male carriers. No significant association between carrier status and the risk of CVAs in the Bulgarian population was found for the common prothrombotic mutations Factor V Leiden, PT G20210A, MTHFR C677T and PAI1 4G/5G.