To the Editor: Recently, Numata et al.1 provided evidence of changes in DNA methylation in the human prefrontal cortex (PFC) through human development and aging. This study reported on three aspects of DNA methylation in 108 fetal, child, and adult PFC samples with the use of the Illumina Infinium HumanMethylation27 BeadChip array. These three aspects were (1) temporal changes in DNA methylation, (2) genetic polymorphisms in association with DNA methylation, and (3) a correlation between mRNA expression and DNA methylation. Significant age- and sex-associated changes in DNA methylation were observed at particular loci and specifically in regions outside of CpG islands. Although this study is unique in that it investigates human DNA methylation in a tissue that is difficult to obtain, there are several technical limitations that need to be considered when interpreting the array data.
Studies examining the relationship between fundamental sample characteristics (including age, sex, and tissue type) and DNA methylation are invaluable as epigenetic investigations permeate many fields of biology. Illumina has been a leader in large-scale epigenetic quantification—it has produced the GoldenGate Methylation array, its successor, the Infinium HumanMethylation27 BeadChip array, and, most recently, the Infinium HumanMethylation450 BeadChip array. However, with the exponential increase in genomic coverage comes much greater complexity to the analysis and interpretation of DNA-methylation microarray data.
Illumina’s Infinium technology is based on 50 bp probes that hybridize to complementary genomic sequences and target specific CpG loci throughout the genome.2 These probes are cloned and incorporated into beads so that any given CpG is interrogated several thousand times. Despite the possible 1.27 × 1030 unique 50 bp sequences, array design is limited to the specific genomic sequences surrounding the CpG site of interest. As a result, the Illumina 27K array is peppered with probes that map to multiple genomic loci, and it is unlikely that signals from these cross-hybridizing probes are exclusively from the intended target CpG. Chen et al.3 characterized nonspecific probes on the Illumina 27K array and found that about 6%–10% of probes mapped to more than one location on the basis of “90% identity, 40–50 matching bases, end-nucleotide match, and gapless sequence alignment.” The majority of these nonspecific probes were designed to target autosomal loci that are in repetitive elements or pseudogenes and thus cross-hybridized to other locations. Specifically, considering X-autosome cross-hybridizing probes is important because of the chromosomal imbalance between sexes.
Chen et al. suggest that autosomal sex-specific DNA-methylation differences identified on the Illumina 27K array might be artifacts of probes that cross-hybridize to the X chromosome. Seven of the ten sexually dimorphic target CpGs identified by Numata et al. were reported to cross-hybridize to the X chromosome.3 We BLASTed the autosomal sexually dimorphic target CpGs identified by Numata et al., and we note the location and affinity of suspected cross-hybridization in Table 1. Six of these probes cross-hybridized to X chromosome CpG-island promoters. At these CpGs, females were reported as hypermethylated in comparison to males, which is consistent with the general hypermethylation of CpG-island promoters on the inactive X chromosome.4 For two probes, females were observed to be hypomethylated in comparison to males. Both of these probes cross-hybridized to the X chromosome; one was in a gene body, and the other was in a non-CpG-island intergenic region. These regions are often hypermethylated on the active X chromosome compared to the inactive X chromosome. The fact that males only possess an active X chromosome most likely accounts for the hypermethylated state of males at these two CpGs.5 Therefore, the top eight target CpGs identified by Numata et al. to have sexually dimorphic DNA methylation are most likely technical artifacts due to X chromosome cross-hybridization.
Table 1.
Top Ten Autosomal CpG Loci with Sex Differences in DNA Methylation
CpG Locus | Target Chromosome | Gene | MIM Number | DNA Methylation Difference |
Cross-hybridization |
|
---|---|---|---|---|---|---|
Chromosome | Affinity | |||||
cg15915418 | 9 | TLE1 | 600189 | female > male | X | 98% |
cg07711515 | 9 | BAG1 | 601497 | female < male | X | 100% |
cg27063525 | 6 | C6orf68 | 610463 | female > male | X | 100% |
cg11673803 | 10 | GLUD1 | 138130 | female > male | X | 100% |
cg21243096 | 1 | POU3F1 | 602479 | female > male | 2, X | 100%, 50% (100% last 25 bp) |
cg04455759 | 11 | SDHD | 602690 | female < male | X | 94% |
cg08284151 | 12 | DPPA3 | 608408 | female > male | 14, X | 98%, 96% |
cg05924191 | 15 | FLJ20582 | N/A | female > male | X | 96% |
cg23758485 | 16 | SMPD3 | 605777 | female > male | none | − |
cg07494248 | 2 | HSPD1 | 118190 | female > male | none | − |
Adapted from Numata et al., 2012. Affinity is based on theoretical base-pair matching by BLAST. The following abbreviation is used: N/A, not available.
Numata et al. identified close to 3,000 correlations between DNA methylation and the presence of a SNP within 1 Mb of the target CpG (such a SNP is termed a cis-SNP), demonstrating that genotype and DNA methylation are associated in many instances. The closer the SNP to the CpG of interest, the greater the association. This pattern holds true for other genomic operators, including enhancers and insulators,6 which are more likely to affect closer targets. Numata et al. also identified 401 trans-SNPs (>1 Mb from the target CpG) associated with DNA methylation. BLASTing the probe sequence of the top ten CpGs associated with a trans-SNP, we noted that seven of these cross-hybridize to other genomic locations, all of which are located within 100 kb of the SNP (Table 2). It is therefore likely that the strong trans-SNP-CpG associations were influenced by cis-SNP-CpG associations at the locus of cross-hybridization.
Table 2.
Top Ten Most Significant CpG-trans-SNP Associations
CpG Locus | Gene | MIM Number | SNP | SNP Chromosome |
Cross-hybridization |
||
---|---|---|---|---|---|---|---|
Chromosome | Affinity | Distance to SNP | |||||
cg18984499 | RPL26 | 603704 | rs11847580 | 14 | 14 | 84% | 13.3 kb |
cg17704839 | UBL5 | 606849 | rs733675 | 17 | 17 | 78% | 26.6 kb |
cg18634211 | LIN28 | 611043 | rs2288322 | 2 | 2 | 70% | 0.8 kb |
cg18634211 | LIN28 | 611043 | rs10207436 | 2 | 2 | 70% | 27.6 kb |
cg25299176 | YWHAE | 605066 | rs4281963 | 2 | 2 | 92% | 10.2 kb |
cg03923277 | TDG | 601423 | rs326387 | 12 | none | − | 214.9 kb |
cg2599176 | YWHAE | 605066 | rs6716175 | 2 | 2 | 92% | 36.1 kb |
cg18984499 | RPL26 | 603704 | rs4906142 | 14 | 14 | 84% | 94.4 kb |
cg13514129 | MACF1 | 608271 | rs12130070 | 1 | none | − | 205 Mb |
cg13514129 | MACF1 | 608271 | rs2878079 | 1 | none | − | 205 Mb |
Adapted from Numata et al., 2012. Affinity is based on theoretical base-pair matching by BLAST.
Although there are probably sex differences in DNA methylation at some loci on autosomal chromosomes, in addition to trans-SNPs that are associated with DNA methylation, the true number is probably lower than what was initially reported by Numata et al. This is one of several studies that have not omitted cross-hybridizing probes.7,8 Other probes, such as those with a SNP in the target CpG, should also be scrutinized because they might alter assessment of DNA methylation. With the advent of Illumina’s Infinium HumanMethylation450 array, which carries almost 20 times more probes than the Infinium HumanMethylation27 does, there is a far greater potential for probe-related technical artifacts. Given the popularity of the Illumina bead platform, it is important for all users to be aware of these complications. For methylation arrays, we suggest BLASTing candidate CpGs for identifying cross-hybridizing probes with the use of Chen et al.’s conservative criteria. By raising awareness of the complexities of using DNA-methylation microarrays, we hope to increase the likelihood of reporting true biological findings.
Web Resources
The URLs for data presented herein are as follows:
Online Mendelian Inheritance in Man (OMIM), http://omim.org
References
- 1.Numata S., Ye T., Hyde T.M., Guitart-Navarro X., Tao R., Wininger M., Colantuoni C., Weinberger D.R., Kleinman J.E., Lipska B.K. DNA methylation signatures in development and aging of the human prefrontal cortex. Am. J. Hum. Genet. 2012;90:260–272. doi: 10.1016/j.ajhg.2011.12.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Bibikova M., Le J., Barnes B., Saedinia-Melnyk S., Zhou L., Shen R., Gunderson K.L. Genome-wide DNA methylation profiling using Infinium® assay. Epigenomics. 2009;1:177–200. doi: 10.2217/epi.09.14. [DOI] [PubMed] [Google Scholar]
- 3.Chen Y.A., Choufani S., Ferreira J.C., Grafodatskaya D., Butcher D.T., Weksberg R. Sequence overlap between autosomal and sex-linked probes on the Illumina HumanMethylation27 microarray. Genomics. 2011;97:214–222. doi: 10.1016/j.ygeno.2010.12.004. [DOI] [PubMed] [Google Scholar]
- 4.Cotton A.M., Lam L., Affleck J.G., Wilson I.M., Peñaherrera M.S., McFadden D.E., Kobor M.S., Lam W.L., Robinson W.P., Brown C.J. Chromosome-wide DNA methylation analysis predicts human tissue-specific X inactivation. Hum. Genet. 2011;130:187–201. doi: 10.1007/s00439-011-1007-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hellman A., Chess A. Gene body-specific methylation on the active X chromosome. Science. 2007;315:1141–1143. doi: 10.1126/science.1136352. [DOI] [PubMed] [Google Scholar]
- 6.Birney E., Stamatoyannopoulos J.A., Dutta A., Guigó R., Gingeras T.R., Margulies E.H., Weng Z., Snyder M., Dermitzakis E.T., Thurman R.E., ENCODE Project Consortium. NISC Comparative Sequencing Program. Baylor College of Medicine Human Genome Sequencing Center. Washington University Genome Sequencing Center. Broad Institute. Children’s Hospital Oakland Research Institute Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature. 2007;447:799–816. doi: 10.1038/nature05874. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gibbs J.R., van der Brug M.P., Hernandez D.G., Traynor B.J., Nalls M.A., Lai S.L., Arepalli S., Dillman A., Rafferty I.P., Troncoso J. Abundant quantitative trait loci exist for DNA methylation and gene expression in human brain. PLoS Genet. 2010;6:e1000952. doi: 10.1371/journal.pgen.1000952. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Liu J., Morgan M., Hutchison K., Calhoun V.D. A study of the influence of sex on genome wide methylation. PLoS ONE. 2010;5:e10028. doi: 10.1371/journal.pone.0010028. [DOI] [PMC free article] [PubMed] [Google Scholar]