SuperTAG Methylation-specific Digital Karyotyping Reveals Uremia-induced Epigenetic Dysregulation of Atherosclerosis-Related GenesClinical Perspective
Background—Accelerated atherosclerosis is a hallmark of chronic kidney disease (CKD). Although the role of epigenetic dysregulation in atherosclerosis is increasingly appreciated, only a few studies focused on epigenetics in CKD-associated cardiovascular disease, virtually all of which assessed epigenetic dysregulation globally. We hypothesized that gene-specific epigenetic dysregulation in CKD exists, affecting genes pertinent to inflammation and atherosclerosis.
Methods and Results—Ten clinically stable patients undergoing hemodialysis therapy and 10 healthy age- and sex-matched controls were recruited. Genome-wide analysis of DNA methylation was performed by SuperTAG methylation-specific digital karyotyping, in order to identify genes differentially methylated in CKD. Analysis of 27 043 436 tags revealed 4288 genomic loci with differential DNA methylation (P<10–10) between hemodialysis patients and control subjects. Annotation of UniTags to promoter databases allowed us to identify 52 candidate genes associated with cardiovascular disease and 97 candidate genes associated with immune/infection diseases. These candidate genes could be classified to distinct proatherogenic processes, including lipid metabolism and transport (eg, HMGCR, SREBF1, LRP5, EPHX2, and FDPS), cell proliferation and cell-cycle regulation (eg, MIK67, TP53, and ALOX12), angiogenesis (eg, ANGPT2, ADAMTS10, and FLT4), and inflammation (eg, TNFSF10, LY96, IFNGR1, HSPA1A, and IL12RB1).
Conclusions—We provide a comprehensive analysis of genome-wide epigenetic alterations in CKD, identifying candidate genes associated with proatherogenic and inflammatory processes. These results may spur further research in the field of epigenetics in kidney disease and point to new therapeutic strategies in CKD-associated atherosclerotic disease.
Patients with chronic kidney disease (CKD) suffer from accelerated atherosclerosis. Accordingly, cardiovascular events are the major cause of death in patients with CKD.1
It is widely accepted that nontraditional risk factors are major determinants of the exceedingly high burden of cardiovascular events in CKD.2 Among such nontraditional risk factors, uremia-associated alterations in epigenetic regulation have recently been hypothesized to promote accelerated atherogenesis in patients with CKD.3–8
Clinical Perspective on p 620
Because epigenetic patterns are influenced by a host of environmental factors,9 the unphysiological uremic milieu may trigger substantial alterations of the epigenome. More specifically, hyperhomocysteinemia,10–12 inflammation,13 dyslipidemia,14,15 and oxidative stress16 may promote an unbalanced DNA methylation, which is a major epigenetic modification of the genome. It contributes to transcriptional regulation, imprinting, X-chromosome inactivation, and genomic integrity.3 Consequently, aberrations in genomic DNA methylation are associated with inappropriate gene expression and promotion of disease.17
Although the association among CKD, epigenetic dysregulation, and accelerated atherosclerosis has been proposed in several review articles, only a few pioneering studies analyzed global changes in the epigenome of patients with CKD10–13,18 and yielded conflicting results: Ingrosso et al11 reported global DNA hypomethylation in a small group of patients on hyperhomocysteinaemic hemodialysis (HD), whereas Stenvinkel et al13 found global DNA hypermethylation in CKD, which was associated with both inflammation and poor outcome in patients on HD.
The association between CKD and global DNA methylation thus seems to be complex. Moreover, analyses of global DNA methylation will offer no information on regulation of specific genes, and only scarce data on site-specific epigenetic changes have been reported in patients with CKD so far.10,11,18
To further unravel the impact of epigenetic dysregulation in CKD, we set out to perform whole-genome analysis of DNA methylation in patients on HD. We, therefore, expanded the methylation-specific digital karyotyping (MSDK) method, which has first been described by Hu et al19 and later been modified by Li et al,20 into SuperTAG methylation-specific digital karyotyping (SMSDK). SMSDK uses those longer (26 bp) tags that are used in SuperSAGE,21 and thus allows high-throughput and genome-wide DNA methylation mapping.
We now report substantial differences in the epigenome of patients on HD compared with healthy controls, comprising a fundamental dysregulation of atherosclerosis-related genes.
Clinically stable dialysis patients (n=10) undergoing standard HD therapy 3 times a week were recruited from the Department of Internal Medicine IV, Nephrology and Hypertension of the Saarland University Medical Center. In all patients, 20 mL ethylenediaminetetraacetic acid (EDTA)-anticoagulated blood was drawn before a HD session after the long-interdialytic interval. Ten healthy age- and sex-matched hospital employees served as controls.
To circumvent age- and sex-specific impact on our epigenetic analyses, recruitment was confined to men subjects aged between 50 and 60 years. All participants gave written informed consent in accordance with the Declaration of Helsinki. The study protocol was approved by the local Ethics Committee.
Quantification of S-Adenosylmethionine/S-Adenosylhomocysteine and Homocysteine
For the quantification of S-adenosylmethionine (SAM) and S-adenosylhomocysteine (SAH), EDTA samples were directly placed on ice and centrifuged immediately for 10 minutes at 2000g. After centrifugation, 1 mL of EDTA plasma was directly acidified with 100 µL of 1 N acetic acid to prevent SAM degradation, mixed thoroughly, and stored at −70°C until analysis. The high-performance liquid chromatography-mass spectrometry detection of SAH and SAM was carried out by using a Waters 2795 alliance HT, coupled to a Quattro Micro API tandem mass spectrometer (Waters Corporation, Milford, MA) as described by Kirsch et al.22
Homocysteine was measured in plasma with a fluorescence polarization immunoassay on the Abbott AxSYM system (Abbott Laboratories, North Chicago, IL).
DNA Isolation and Construction of SMSDK Libraries
Peripheral blood mononuclear cells (PBMCs) were immediately isolated from anticoagulated blood by Ficoll-Paque (Lymphocyte Separation Medium; PAA, Cölbe, Germany) gradient density centrifugation. DNA was isolated from peripheral blood mononuclear cells using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany).
Genome-wide analysis of DNA methylation was performed by SMSDK at GenXPro GmbH (Frankfurt, Germany). The libraries were generated essentially as described by Li et al20 with modifications as described by Matsumura et al.23 Hinp1I was used as the methylation-sensitive enzyme, recognizing non-CpG-methylated GCGC sites. After digestion by Hinp1I, a biotinylated adapter, containing the recognition site for the restriction enzyme EcoP15I, was ligated to the digested DNA. The resulting product was bound to streptavidin-coated magnetic beads (Dynal). After NlaIII digestion, unbound DNA was discarded, and a second adapter, containing another EcoP15I recognition site adjacent to a CATG overhang and a priming site for Illumina’s p5 primer for high-throughput sequencing on the Illumina Genome Analyzer II, was ligated to the NlaIII site. The streptavidin-bound constructs of DNA fragments, flanked by the two EcoP15I recognition sites arranged in a head-to-head fashion, were digested with EcoP15I to cut off the NlaIII bound adapter and its adjacent 26 to 27 bp long tags from the streptavidin matrix. The resulting adapter-tags were then ligated to p7 adapters from Illumina’s Genome Analyzer system. The constructs were sequenced by synthesis on Illumina’s Genome Analyzer II system.
Quality Assessment and Statistics of SMSDK Data
Quality assessment of generated tags was performed according to Qu et al24 with an in-house software to reduce sequencing errors and artificial tag sequences. Tags were counted using the GenXProgram. Statistical analysis was performed by using an R-script (http://search.cpan.org/~scottzed/Bio-SAGE-Comparison-1.00/lib/Bio/SAGE/Comparison.pm) based on statistics described by Audic and Claverie.25
Gene Ontology Information
Gene ontology (GO) information was obtained from www.GeneOntology.org for the annotated UniTags. P values describing the likelihood for enrichment of GO terms were calculated by the Fisher exact test based on UniTags that were differentially expressed with a P<10–10.26 GO analysis was performed by using an in-house software (http://genxpro.ath.cx).
Validation of DNA Methylation
Bisulfite sequencing was performed according to Geisel et al.10 Briefly, 500 ng DNA was treated with sodium bisulfite (EpiTect Bisulfite Kit, Qiagen), and a section of the METTL2B promoter was amplified using the PyroMark polymerase chain reaction (PCR) Kit and the PyroMark CpG Assay (PM00031115, Qiagen, Hilden, Germany). Pyrosequencing was performed on the PSQ 96MA system (Qiagen); percentages of methylated (C) and unmethylated (T) CpGs were subsequently calculated. The mean of 2 CpG sites was used as a marker for METTL2B methylation.
Real-time Quantitative PCR
After isolating total RNA from PBMCs with the QIAamp DNA Blood Mini Kit (Qiagen), 500 ng RNA was reverse transcribed using the DyNAmo cDNA Synthesis Kit (Biozym, Hessisch Oldendorf, Germany). Real-time PCR was performed on the Mx3005P system (Stratagene, Waldbronn, Germany) using the DyNAmo ColorFlash SYBR Green qPCR Kit (Biozym) and predesigned primers (QuantiTect Primer Assay, Qiagen). Relative quantification was performed by the comparative ΔΔCt method. The housekeeping gene GAPDH was used as an internal standard (forward primer: 5′-ctcctccacctttgacgctg-3′; reverse primer: 5′-tcctcttgtgctcttgctgg-3′). Experiments were performed in duplicate.
Flow Cytometric Analyses
Flow cytometric analyses were performed according to our previous report21 using the FACS Canto II with CellQuest Software (BD Biosciences, Heidelberg, Germany). Briefly, antigen expression on CD86-positive monocytic cells was analyzed via a whole blood assay using 100 µL of EDTA anticoagulated blood, quantifying surface expression as median fluorescence intensity standardized against coated fluorescent particles (SPHEROTM; BD Biosciences). Histograms were plotted with FCS Express Software (De Novo Software, Los Angeles, CA). For measurement of reactive oxygen species, 1×106 PBMCs were incubated (15 minutes, 37°C, 5% CO2) with cell-permanent carboxy-H2DFFDA (Invitrogen, Darmstadt, Germany). Afterward, cells were stained with anti-CD14, anti-CD16, and anti-CD86, and intracellular reactive oxygen species levels were determined as median fluorescence intensity in CD14++CD16+ monocytes (the major reactive oxygen species producing subset).
Phagocytosis assay was performed by using Fluoresbrite Yellow Green Carboxylate Microspheres (0.75 µm; Polysciences, Eppelheim, Germany), which were opsonized with heterologous serum (diluted to 50% with Krebs Ringers PBS). About 100 µL of citrate anticoagulated whole blood was incubated with 10 µL of opsonized particles (108 particles/mL) with gentle shaking for 30 minutes at 37°C. Control samples were incubated at 4°C. Afterward, cells were stained with anti-CD86, anti-CD14, and anti-CD16, and counts of FITC-positive CD14++CD16– monocytes (subset with highest phagocytosis potential) were determined flow cytometrically.
The following antibodies were used: CD86-PE (HA5.2B7, Beckman-Coulter, Krefeld, Germany), CD14-PerCP (MΦ9, BD Biosciences), CD16-PeCy7 (3G8, BD Biosciences), and CD43-FITC (eBio84-3C1, eBioscience, Frankfurt, Germany).
We compared levels of SAM, SAH, and homocysteine, as well as data on pyrosequencing, real-time PCR, and flow cytometry between patients on HD and controls using the unpaired Student’s t test. The Kolmogorov–Smirnov test was applied to test normality assumption. Both tests were performed with the IBM SPSS Statistics 18 software.
Mean age was 56.1±3.9 years in patients on HD and 53.5±2.4 years in control subjects (P=0.100). As expected, patients on dialysis had significantly higher C-reactive protein (14.1±12.9 mg/L) compared with controls (2.0±1.3 mg/L; P=0.016). Consistent with the anticipated normal renal function, the average estimated glomerular filtration rate of controls was 86.0±14.3 mL/min/1.73 m2.
Five patients on dialysis had prevalent atherosclerotic disease, defined as the presence of coronary artery disease (prior myocardial infarction or coronary revascularization), cerebrovascular disease (prior stroke with symptoms lasting longer than 24 hours or carotid revascularization), and/or peripheral artery disease (prior revascularization of lower-limb arteries). Controls were free from atherosclerotic disease.
Analysis of central modulators of DNA methylation—homocysteine, SAH, and SAM—demonstrated significantly higher levels of these metabolites in patients on HD (Table 1), which is in accordance with previous studies.10,12 Among these central modulators, SAH levels differed most pronouncedly (30.9-fold increase in patients on HD; P<0.001) compared with homocysteine (2.4-fold increase; P<0.001) and SAM (4.8-fold increase; P<0.001). These exceedingly high SAH levels resulted in a dramatic decrease in the SAM/SAH ratio, an indicator of reduced cellular methylation capacity.
Generation of SMSDK Libraries
DNA from peripheral blood cells was used for the generation of 2 independent SMSDK libraries (control library and HD library). After eliminating low-quality reads (according to Qu et al24 and further elimination of tags with a count <5) and trimming of adapter sequences, the total number of tags was 27 043 436, comprising 11 942 429 from control subjects and 15 101 007 from patients on HD (Table 2 and online-only Data Supplement Figure I). These 27 043 436 tags accounted for 575 744 unique sequences (UniTags), of which 551 002 UniTags were found in both SMSDK libraries, 7250 were uniquely found in the control library, and 17 492 in the HD library. As a result, the control library comprised 558 252 UniTags, and the HD library comprised 568 494 UniTags.
UniTags were classified in abundance groups according to their number of copies (Table 2). Most of the UniTags (>99.9%) were found in low frequency, corresponding to the low abundance group (<100 copies/million). Less than 0.1% UniTags were classified to the mid-abundance group (100–1000 copies/million) or to the high-abundance group (>1000 copies/million), respectively, which is in line with previous reports.27
Uremia-associated Dysregulation of DNA Methylation
Among all 575 744 UniTags, we calculated the likelihoods for different tag frequencies in the control library and HD library according to Audic and Claverie.25 We a priori set a strict level of significance to P<10–10 in order to selectively identify those loci in the genome with very pronounced differences in DNA methylation, avoiding false-positive or biologically irrelevant hits. We thus identified 4288 UniTags differing in their counts between patients on HD and control subjects (Figure 1). Among these 4288 UniTags, 1854 UniTags were found more frequently in control subjects, in line with hypermethylation in patients on dialysis, whereas 2434 UniTags were found more frequently in patients on dialysis, demonstrating hypomethylation in these patients.
Annotation of UniTags
Using an in-house version of the BLAST software (blastn version 2.2.21), we firstly annotated all 575 744 UniTags to different databases in hierarchical order as listed in Table 3 in order to match UniTags to their corresponding genomic loci. An e-value ≤0.001 was defined as a prerequisite for analysis.
Among all 575 744 UniTags, 79 574 (13.8%) UniTags could be annotated to upstream gene regions (databases 1–3) and were used for further analyses. The remaining UniTags matched to genomic loci not located in upstream gene regions (databases 4–6, totaling 265 459 UniTags, 46.1%) or could not be annotated at all (230 711 UniTags, 40.1%).
For further stringency in our next analyses, we excluded those UniTags that did not match perfectly (number of matches <26/26) with sequences in databases 1 to 3. When applying these restrictions, 47 348 (59.5%) out of 79 574 UniTags could be used for further analyses (online-only Data Supplement Table I and Figure I).
Dysregulated Genes in Patients on Hemodialysis
When applying a strict P value (P<10–10), 1089 out of these 47 348 annotated UniTags differed in their frequencies between patients on HD and control subjects. Thus, about a quarter of those 4288 UniTags that were found in significant different counts between patients on HD and healthy controls in SMSDK analysis could be annotated to a specific upstream gene region, of which the top 100 hits are summarized in online-only Data Supplement Table IIA (hypomethylated genes in patients on HD) and online-only Data Supplement Table IIB (hypermethylated genes in patients on HD). Additionally, all SMSDK data are presented in online-only Data Supplement Table III.
Several of these differentially methylated genes were linked to cell differentiation and cell-cycle regulation (eg, DBF4B, TNFSF10, and PTPRN) and especially to the p53 pathway (eg, TP53, CDC14A, HIPK4, and BAG6).
Moreover, other differentially methylated genes were connected to immune system processes including inflammation (eg, CFB, LY96, SPN, NFKB2, and GPX4), adhesion processes (eg, ICAM2, CD300LG, and CTNNA3), angiogenesis (eg, ANGPT2, ADAMTS10, and FLT4), cholesterol and lipid metabolism/transport (eg, HMGCR, SLC27A1, and PCCA), or other intracellular transport processes (eg, KIF2C, SNX6, and TIMM8A).
Finally, genes that are directly linked to epigenetic control (eg, METTL2B, KDM6A, GADD45A, and JMJD5) and transcriptional regulation (eg, CSRNP2, ZNF382, ZNF251, and ZNF85) were also differentially methylated in patients on HD.
Validation of SMSDK Results
First, we compared our SMSDK results with the only prior genome-wide DNA methylation analysis in CKD. Focusing on renal markers, Sapienza et al18 compared DNA methylation in saliva between patients with diabetes mellitus with and without CKD by the Illumina HumanMethylation 27 BeadChip array. Of 187 differentially methylated genes identified by Sapienza et al,18 70.4% were accordingly differentially methylated (P<0.05) in our analysis (online-only Data Supplement Table IV). Secondly, we aimed to validate our analysis by performing both bisulfite sequencing and real-time PCR of an arbitrary selected gene (METTL2B), which is linked to epigenetic regulation, and thereby confirmed SMSDK results (online-only Data Supplement Table IIB and Figure II). Thirdly, real-time PCR confirmed upregulation of LY96 and TNFSF10 (hypomethylated in SMSDK) and downregulation of EPHX2 and TRPV1 (hypermethylated in SMSDK) in patients on HD (online-only Data Supplement Figure II).
Finally, we aimed to assess the biological relevance of SMSDK results by flow cytometry and functional analyses. Firstly, in line with their CD43 promoter hypomethylation (online-only Data Supplement Table SIII), we demonstrated higher CD43 (SPN) protein expression on monocytic cells of patients on HD (Figure 3). Secondly, in line with their differential methylation of genes linked to inflammation and other immune system processes, patients on dialysis had higher protein expression of the monocytic activation marker CD86, higher cellular production of reactive oxygen species and higher monocytic phagocytosis rate (Figure 3).
Gene Ontology Analysis for Biological and Functional Differences
To investigate whether the above described epigenetic dysregulation may relate to biological and functional alterations of circulating immune cells of patients on dialysis, we next performed GO analysis. Among all 575 744 UniTags, 13 421 UniTags were annotated to the GO term biological process, 13 617 to molecular function, and 14 274 to cellular component. Figure 4 summarizes all GO terms that were assigned to the biological process at level 2 of GO categorization. GO terms showing significant differences (enrichment P<0.05) between control library and HD library are highlighted. These differing GO terms included several central biological processes like immune system processes (P=0.001), response to stimulus (P=0.003), cell proliferation (P=0.006), death (P=0.016), or metabolic processes (P=1.2 × 10–5), among others.
Dysregulation of Atherogenesis-related Genes in Patients on Hemodialysis
Following the postulate that uremia induces dysregulation of both atherosclerosis-protective genes and atherosclerosis-susceptible genes,3–8 we finally analyzed whether the 1089 differentially methylated genes between patients on HD and controls can be directly linked to cardiovascular disease. Using the Genetic Association Database (accessible from the National Institutes of Health; http://geneticassociationdb.nih.gov/), we tested these genes for association with cardiovascular disease, as well as for immune/infection diseases, given that inflammation plays a central role in the pathogenesis of atherosclerosis.
Among all 1089 genes, 52 genes were associated with cardiovascular disease and 97 genes with immune/infection diseases. The most relevant genes are listed in Table 4 (cardiovascular disease) and Table 5 (immune/infection diseases). Of note, various genes connected to inflammation (eg, TNFSF10, LY96, IFNGR1, HSPA1A, and IL12RB1) were found to be hypomethylated in patients on dialysis. Further genes differentially methylated in patients on dialysis were connected to distinct cellular processes, including adhesion processes (eg, PKD1, MADCAM1, and SPN), cell proliferation and cell-cycle regulation (eg, MIK67, TP53, and ALOX12), apoptosis (eg, CASP8, RAD51, and RAD51L1), DNA repair (eg, XRCC1 and DDB2), and lipid metabolism (eg, HMGCR, SREBF1, LRP5, EPHX2, and FDPS). Interestingly, hypermethylation—indicating downregulation of gene expression—affected genes that are supposed to be atherosclerosis-protective (eg, TRPV1 and GPX4).
Failure in epigenetic regulation substantially contributes to the onset and progression of vascular disease.28 Among the 3 cornerstones of epigenetic regulation, namely histone modifications, RNA interference, and DNA methylation, the latter became the prime target for studies on interactions between disturbed gene regulation and promotion of vascular disease.
Accordingly, changes in global DNA methylation have been associated with future development of atherosclerosis in animal studies14 and with prevalent vascular disease in cross-sectional clinical studies.29
Patients with CKD suffer from accelerated atherosclerotic vascular disease, which cannot completely be explained by traditional risk factors.2 Of note, patients on dialysis display aberrations in global DNA methylation,11,13 to which several features of the unphysiological uremic milieu, such as inflammation,13 hyperhomocysteinemia,10,11 oxidative stress,16 and dyslipidemia,14,15 may contribute.
Against this background, it has been speculated that disturbed DNA methylation in CKD may affect atherosclerosis-related genes with consequently higher susceptibility for vascular complications,3–8 although information on site-specific regulation of these genes in CKD is virtually missing so far.
Beyond the field of nephrology, site-specific methylation analyses point to a dysregulation of several atherosclerosis-related genes, such as the estrogen receptor α gene (ESR1), the inducible nitric oxide synthase gene (iNOS), and the extracellular superoxide dismutase gene (SOD3)—in the process of atherogenesis (reviewed in ref. 28).
We now aimed to identify atherosclerosis-related candidate genes in patients with CKD, extending our earlier studies that focused on methylation analysis of a single gene involved in oxidative stress-mediated atherosclerosis (p66Shc [SHC1]).10
Using SMSDK, we sequenced 27 043 436 tags. Despite choosing a very strict level of significance, we found >4000 UniTags differing between control subjects and patients on dialysis. This allowed us to identify disturbed methylation in 52 candidate genes associated with cardiovascular disease and in 97 genes associated with immune and infection diseases according to the NIH Genetic Association Database. These genes could be linked to diverse proatherogenic processes, including lipid metabolism and transport (eg, HMGCR, SREBF1, LRP5, EPHX2, and FDPS), cell proliferation and cell-cycle regulation (eg, MIK67, TP53, and ALOX12), angiogenesis (eg, ANGPT2, ADAMTS10, and FLT4), inflammation (eg, TNFSF10, LY96, IFNGR1, HSPA1A, and IL12RB1), and even epigenetic control (eg, METTL2B, KDM6A, GADD45A, and JMJD5).
Of note, epigenetic dysregulation not only affects genes associated with the promotion of atherosclerosis. Instead, we found hypermethylation of genes that have been characterized as atheroprotective in animal studies, namely TRPV1 and GPX4.30,31 Hypermethylation of these 2 genes points to a reduced transcription level of these protective factors in patients on dialysis.32
DNA methylation is centrally modulated by C1 metabolism, which itself is skewed in CKD. In C1 metabolism, the amino acid methionine is converted to SAM, which serves as a universal methyl group donor for methyltransferases. After transfer of its methyl group, SAM becomes SAH, which in turn binds to the active site of methyltransferases and thus strongly inhibits further methylation reactions. SAH may next be hydrolyzed to homocysteine. Of note, SAH and homocysteine are in equilibrium, so that hyperhomocysteinemia is inevitably associated with elevated SAH levels and with subsequent DNA hypomethylation.11,33
Although increased homocysteine levels are a common finding in CKD,10,11,34 and although cohort studies associated hyperhomocysteinemia with subsequent cardiovascular events in CKD,34 surprisingly none of the interventional studies targeting the high homocysteine levels in CKD proved a benefit,35–39 which is in line with trials in the general population.40,41
Nonetheless, we reckon that those negative interventional trials neither rule out a causative role of a disturbed C1 metabolism in atherogenesis nor preclude epigenetic dysregulation from being a promising future therapeutic target in atherosclerosis.
Of note, epigenetic dysregulation has predominantly been addressed by measurement of the surrogate marker homocysteine in most clinical trials; this homocysteine-centered approach may result from the relatively convenient homocysteine analysis in contrast with the more cumbersome measurement of SAH. Based on our pathophysiological understanding of SAH as the direct inhibitor of methylation reactions, its measurement as a biomarker for epigenetic dysregulation in cardiovascular disease seems to be more meaningful than measuring its derivate homocysteine.42,43
Interestingly, all trials aiming at lowering homocysteine for primary or secondary prevention of cardiovascular disease used folate, vitamin B6, and/or vitamin B12, all of which efficiently lower plasma homocysteine levels, but unfortunately do not affect SAH levels.44 Furthermore, in patients with CKD, SAH accumulates more compared with homocysteine because the kidneys are the major site of SAH disposal in humans.45 In accordance, in this study, we observed higher differences in SAH levels (30.9-fold increase in patients on dialysis) than in homocysteine levels (2.4-fold) and SAM levels (4.8-fold).
Importantly, chronic inflammation is commonly observed in patients on dialysis and associated with increased cardiovascular morbidity and mortality in this patient population (reviewed in ref. 46). As such, inflammation is a central uremic feature that—opposed to homocysteine—rather seems to trigger DNA hypermethylation13 and thus may induce further aberrations in epigenetic regulation.
In line, Stenvinkel and coworkers13 reported that global DNA hypermethylation was associated with both inflammation and poor outcome in CKD. Moreover, it was shown that the inflammatory cytokine IL-6 regulates a DNA methyltransferase gene47 that may result in epigenetic dysregulation.
In this study, we demonstrated that epigenetic dysregulation in CKD comprises both DNA hypomethylation and DNA hypermethylation at different genomic loci. Of note, our data neither refute the finding of DNA hypomethylation nor challenge the reported DNA hypermethylation in CKD. Instead, our results underscore the importance of site-specific methylation analyses to deepen our knowledge of epigenetic regulation in CKD.
As a limitation, our present analysis neither allows to distinguish whether renal replacement therapy or uremia per se induces changes in DNA methylation in patients on HD nor to characterize the individual contribution of specific causative factors to epigenetic dysregulation.
Furthermore, no data on the prognostic impact of site-specific methylation in patients with CKD exist so far. We are, therefore, presently initiating a prospective study that shall confirm our hypothesis that dysregulation of predefined candidate genes may serve as early markers that predict future cardiovascular events in a large cohort of CKD.
A better understanding of the underlying causes of disease burden in CKD is desperately needed as conventional therapies failed to demonstrate a definite survival benefit in patients with CKD.48–50 Against this background, we present genome-wide data on DNA methylation in patients on dialysis and characterize epigenetic dysregulation of candidate genes for accelerated atherosclerosis in CKD. We are hopeful that our findings may reveal relevant pathophysiological pathways that contribute to cardiovascular disease in patients with CKD, thus pointing toward potential new avenues for prevention and therapy.
The skillful technical assistance of M Bodis and M Wagner is greatly appreciated.
Sources of Funding
This work was partly funded by an intramural research grant (HOMFOR 2010) and a grant from the Else Kröner-Fresenius-Stiftung.
The online-only Data Supplement is available at http://circgenetics.ahajournals.org/lookup/suppl/doi:10.1161/CIRCGENETICS.112.963207/-/DC1.
- Received March 7, 2012.
- Accepted September 21, 2012.
- © 2012 American Heart Association, Inc.
- Sarnak MJ,
- Levey AS,
- Schoolwerth AC,
- Coresh J,
- Culleton B,
- Hamm LL,
- et al
- Stenvinkel P,
- Ekström TJ
- Loehrer FM,
- Angst CP,
- Brunner FP,
- Haefeli WE,
- Fowler B
- Lund G,
- Andersson L,
- Lauria M,
- Lindholm M,
- Fraga MF,
- Villar-Garea A,
- et al
- Zaina S,
- Lindholm MW,
- Lund G
- Valinluck V,
- Tsai HH,
- Rogstad DK,
- Burdzy A,
- Bird A,
- Sowers LC
- Zawada AM,
- Rogacev KS,
- Rotter B,
- Winter P,
- Marell RR,
- Fliser D,
- et al
- Kirsch SH,
- Knapp JP,
- Geisel J,
- Herrmann W,
- Obeid R
- Qu W,
- Hashimoto S,
- Morishita S
- Audic S,
- Claverie JM
- Huang da W,
- Sherman BT,
- Lempicki RA
- Castro R,
- Rivera I,
- Struys EA,
- Jansen EE,
- Ravasco P,
- Camilo ME,
- et al
- Ma L,
- Zhong J,
- Zhao Z,
- Luo Z,
- Ma S,
- Sun J,
- et al
- Rakyan VK,
- Down TA,
- Thorne NP,
- Flicek P,
- Kulesha E,
- Gräf S,
- et al
- Bostom AG,
- Carpenter MA,
- Kusek JW,
- Levey AS,
- Hunsicker L,
- Pfeffer MA,
- et al
- Heinz J,
- Kropf S,
- Domröse U,
- Westphal S,
- Borucki K,
- Luley C,
- et al
- Wrone EM,
- Hornberger JM,
- Zehnder JL,
- McCann LM,
- Coplon NS,
- Fortmann SP
- Zoungas S,
- McGrath BP,
- Branley P,
- Kerr PG,
- Muske C,
- Wolfe R,
- et al
- Liu C,
- Wang Q,
- Guo H,
- Xia M,
- Yuan Q,
- Hu Y,
- et al
- Hodge DR,
- Xiao W,
- Clausen PA,
- Heidecker G,
- Szyf M,
- Farrar WL
Patients with chronic kidney disease (CKD) suffer from a tremendous burden of cardiovascular (CV) disease. Therapies focusing on classical CV risk factors, such as hypercholesterolemia, failed to substantially reduce this high CV morbidity. Therefore, a better pathophysiological understanding of CKD-associated CV disease is mandatory to define new therapeutic strategies. Against this background, we tested the hypothesis that epigenetic dysregulation of genes linked to CV disease occurs in CKD patients. Using SuperTAG methylation-specific digital karyotyping (SMSDK), we compared genome wide DNA methylation between 10 hemodialysis patients and 10 age- and gender-matched controls. We identified 52 genes linked to CV disease and 97 genes linked to immune/infection diseases to be differentially methylated in hemodialysis patients. These results point for the first time towards epigenetic dysregulation of atherosclerosis-related genes in hemodialysis patients, indicating a potential contribution of changes in DNA methylation to accelerated CV disease in CKD. Future studies should first analyse how far pre-defined candidate genes may serve as early markers for future CV events in CKD, and subsequently explore preventive and therapeutic strategies against CKD-associated epigenetic dysregulation.