Molecular Genetics and Metabolism 108 (2013) 225–231

Contents lists available at SciVerse ScienceDirect

Molecular Genetics and Metabolism

j ourna l homepage: www.e lsev ie r .com/ locate /ymgme

DNA methylome profiling using neonatal dried blood spot samples:A proof-of-principle study

Mads Vilhelm Hollegaard a,⁎, Jonas Grauholm b, Bent Nørgaard-Pedersen a, David Michael Hougaard a

a Section of Neonatal Screening and Hormones, Department of Clinical Biochemistry, Immunology and Genetics, Statens Serum Institut, Artillerivej 5, Copenhagen, Denmarkb AROS Applied Biotechnology A/S, Science Park Skejby, Brendstrupgaardsvej, Aarhus, Denmark

Abbreviations: DBSS, dried blood spot samples;Extract-N-Amp Blood PCR kit; CS, ChargeSwitch Forenquality control.⁎ Corresponding author.

E-mail addresses: mvh@ssi.dk (M.V. Hollegaard), jg@bnp@ssi.dk (B. Nørgaard-Pedersen), dh@ssi.dk (D.M. Ho

1096-7192/$ – see front matter © 2013 Elsevier Inc. Allhttp://dx.doi.org/10.1016/j.ymgme.2013.01.016

a b s t r a c t

a r t i c l e i n f o

Article history:Received 31 December 2012Received in revised form 25 January 2013Accepted 25 January 2013Available online 1 February 2013

Keywords:Neonatal dried blood spotDBSSGuthrieEpigenomeDNA methylationMethylome

DNA methylation is the most common DNA modification and perhaps the best described epigenetic modifi-cation. It is believed to be important for genomic imprinting and gene regulation and has been associatedwith the development of diseases such as schizophrenia and some types of cancer.Neonatal dried blood spot samples, commonly known as Guthrie cards, are routinely collected worldwide toscreen newborns for diseases. Some countries, including Denmark, have been storing the excess neonataldried blood spot samples in biobanks for decades. Representing a high percentage of the population undera certain age, the neonatal dried blood spot samples are a potential alternative to collecting new samplesto study diseases. As such, neonatal dried blood spot samples have previously been used for DNA genotypingstudies with excellent results. However, the amount of material available for research is often limited, chal-lenging researchers to generate the most data from a limited quantity of material.In this proof-of-principle study, we address whether two 3.2 mm disks punched from a neonatal dried bloodspot sample contain enough DNA for genome-wide methylome profiling, measuring 27,578 loci at the sametime. We selected two subjects and carried out the following with each: 1) collected an adult whole-bloodsample as reference, 2) spotted a fraction of the whole-blood sample onto a similar type of filter paper asused in the newborn screening and stored it for 3 years to serve as a dried blood spot reference, and 3) iden-tified the archived neonatal dried blood spot samples, stored for 26–28 years, in the Danish Newborn Screen-ing Biobank as a representative of the archived samples. For comparison, we used two different kits for DNAextraction.The DNA, extracted using the Extract-N-Amp Blood PCR kit, was analyzed, and no statistically significant dif-ferences were observed (Pb0.001) when we compared the methylation profile of the reference whole-bloodsamples to the dried blood spot references. This indicates that two 3.2 mm disks contain enough material forreliable methylome profiling and that storing the whole-blood sample on neonatal dried blood spot filterpaper for 3 years does not interfere with the outcome of the analysis.Furthermore, we compared the adult DNA methylation profile to the neonatal dried blood spot sample pro-file. Approximately 50 sites in the subjects were significantly (Pb0.001) different in the newborn samplecompared with the adult sample. Both being healthy adults and the high quality of the DNA methylationarray led to the conclusion that the archived neonatal dried blood spot samples can be used for methylomeprofiling, despite decades of storage and DNA degradation.In conclusion, we show that reliable methylome data can be obtained from old neonatal dried blood spotsamples, by using a reasonable amount of the limited resource. This further adds to the use of neonatal driedblood spot samples in genetic research and screening and paves the way for unique population-based studiesof epigenetic modifications after birth.

© 2013 Elsevier Inc. All rights reserved.

DS, differential score; ENA,sic DNA Purification kit; QC,

arosab.com (J. Grauholm),ugaard).

rights reserved.

1. Introduction

Epigenetic processes are important for cellular differentiation anddevelopment. DNA methylation is perhaps the best understoodepigenetic adaption and the most common DNA modification. It hasbeen shown to be involved in genomic imprinting (parent-of-originspecific expression), tissue differentiation, gene silencing, chromo-some stability, and the regulation of gene expression [1–5]. Aberrant

226 M.V. Hollegaard et al. / Molecular Genetics and Metabolism 108 (2013) 225–231

DNA methylation profiles have been observed in diseases such asschizophrenia, bipolar disorder [6], and different types of cancer [1].In most cases, it occurs by the addition of a methyl group to the 5′-po-sition of the cytosine residue within a CpG dinucleotide. The CpG sitesare often located in large islands, defined as regions of at least 200 bpwith a GC content greater than 50% and an observed CpG ratio greaterthan 60% [7]. CpG islands are found in the 5′ regulatory (promoter)regions of many genes and housekeeping genes in particular, andthey are not typically methylated. However, hypermethylation ofCpG islands in the promoter regions is seen in many diseases, suchas cancer, and can result in down-regulation of the transcriptionalexpression of the gene or gene silencing. Approximately 45% ofhuman genes do not contain CpG islands, and the role of thesenon-CpG-island promoter regions is unclear and not yet fully under-stood [8,9]. The DNA methylation profile can be determined innumerous ways and on many different platforms. Platforms such asthe Pyrosequencer or the Sequenom are often used if the number oftargets is low. However, if the methylation status of the wholegenome – the methylome – is the target, either arrays (as in thisstudy) or high-throughput sequencing can be used.

Many DNA methylation studies of human diseases use DNAextracted from peripheral blood leukocytes, and often the samplematerial is collected after or at the onset of the disease or during aperiod of time in which the individual is being monitored. As analternative or additional possibility, a blood sample collected shortlyafter birth could be a good indicator of a relatively unexposed individ-ual, only influenced by in-utero factors. Interestingly, dried blood spotsamples (DBSS), also known as Guthrie card samples, are routinelycollected in many countries as a part of their neonatal screeningprograms. In Denmark, neonatal screening for metabolic disordersbegan in 1975, and since 1982 the access neonatal DBSS have beenstored in the Danish Newborn Screening Biobank. Today, the biobankcontains almost 2 million neonatal DBSS, representing almost everyDane born since 1982 [10]. In combination with a well-functioningpublic registration system that makes it possible to combine datafrom practically all public databases, researchers can study Denmarkas a cohort [11]. Due to this, the Danish Newborn Screening Biobankhas become an invaluable resource for biological material for researchstudies and is frequently used for genetic and biomarker studies[12–16].

The aim of this study was to evaluate if a realistic amount ofmaterial extracted from stored neonatal DBSS can be used for reliablemethylome profiling, thereby leading to future early-life epigeneticprofiling of whole-blood samples in human diseases.

Venous Blood(Ref.: 2009)

New DBSS(refDBSS: 2011)

Neonatal DBSS (neoDBSS: 1982 and 1984)

DNA Methylation array (2011)

Individual A and B

Fig. 1. Study overview. At birth, the DBSS of the two individuals A (born 1982) and B(born 1984) were collected and archived in the Danish Newborn Screening Biobank(neoDBSS). In 2009, a venous whole-blood sample (Ref.) was drawn from both indi-viduals A and B, aged 28 and 26, respectively. A fraction of the venous blood was spot-ted on filter paper (refDBSS) similar to the type used for the neonatal screening(Whatman 903® Specimen Collection Paper) and stored at −20 °C until analysiswas carried out in 2011.

2. Material and methods

2.1. Sample overview and DNA extraction

Both individuals A and B (siblings) were informed volunteers.Being a potent future newborn screening method, this study wascategorized as a category 2 study (“development of new methodsfor newborn screening analyses,” Nørgaard-Pedersen, et al. 2007),and permission from the ethical committee was not necessary, asper Danish law [10].

Venous blood from each individual was drawn in a heparin collectiontube (Ref.). Fifty milliliters was transferred to Whatman Specimen 903®Collection Paper (refDBSS), dried for 3 hours at room temperature, andstored at −20 °C for 3 years. The neonatal DBSS (neoDBSS) from eachindividual was isolated from the Danish Newborn Screening Biobank.DNA from the venous blood sample was extracted using the Maxwell®16 LEVBloodDNAKit (Promega) according to themanufacturer's instruc-tions. Four 3.2 mmdiskswere punched from the individual's refDBSS andneoDBSS. Two disks were used for a standard protocol DNA extractionwith the Extract-N-Amp Blood PCR kit (ENA) (Sigma), and the othertwo were used with the ChargeSwitch Forensic DNA Purification kit(CS) (Invitrogen). See Fig. 1 for a study overview.

2.2. Bisulphite conversion and Infinium arrays

The DNA samples were bisulphate converted using the EZ-96 DNAMethylation Kit (Zymo Research) according to the manufacturer'sinstructions. For quality control purposes, we profiled positive andnegative control samples from the whole genome amplified DNA ofindividual A using the CpGenome Universal Methylated DNA(Millipore) and REPLIg (QIAGEN) kits, respectively. We used 1 μg ofDNA from the Ref. samples and positive and negative controls. TheDNA extracted from the refDBSS and neoDBSS, approximately 30 ng,was used in its entirety. The bisulphite converted Ref. DNA(input 1 μg) was eluted in 15 μL, and 4 μL was used for InfiniumHumanMethylation27 (Illumina Inc.) labeling. For the refDBSS andneoDBSS, the bisulphite converted elute was vacuum centrifugedbefore using the entire volume for labeling.

2.3. Differential methylation analysis

The methylation level per CpG site was estimated by measuring thesignals of the methylated (signal B, green channel) and unmethylated(signal A, red channel) probes. Each of the intensities was used

Table 1Methylation QC probe evaluation. Probe P-value was set to 0.01. Number: the numberof detected probes with a P-valueb0.01. Percentage: the percentage of detected probes(Pb0.01) of total 27,578 probes on the HumanMethylation27 Beadchip. For positivecontrol we used the CpGenome Universal Methylated DNA (Millipore) and for negativecontrol we whole-genome amplified individual A before bisulphite treatment. For DNAextraction we used the Extract-N-Amp (ENA) and ChargeSwitch (CS) kits. Average isthe average β-value of the detected probes (P-valueb0.01). Minimum/maximum: theminimum and maximum β-value of the sample.

Probe β-Value

Sample Individual Number Percentage Average Minimum Maximum

Controls Pos. 27,567 100.00% 0.889 0.3386 0.9865Neg. 27,247 98.80% 0.1327 0.018 0.8345

Ref. A 27,557 99.90% 0.287 0.011 0.9823B 27,564 99.90% 0.2889 0.0096 0.9843

refDBSS(ENA)

A 27,359 99.20% 0.2781 0.0115 0.9807B 27,391 99.30% 0.2833 0.011 0.9811

neoDBSS(ENA)

A 27,213 98.70% 0.2824 0.0116 0.9835B 27,509 99.70% 0.2934 0.0111 0.9821

refDBSS(CS)

A 25,214 91.40% 0.2721 0.0096 0.9847B 23,251 84.30% 0.2767 0.0098 0.9842

neoDBSS(CS)

A 24,010 87.10% 0.2701 0.0085 0.9869B 26,365 95.60% 0.2822 0.0091 0.9843

0

1000

2000

3000

4000

5000

6000

7000

8000

Pos. Neg. A B A B A B A B A B Pos. Neg. A B A B A B A B A B

Ctrl Ref refDBSS (ENA) neoDBSS (ENA) refDBSS (CS) neoDBSS (CS) Ctrl Ref refDBSS (ENA) neoDBSS (ENA) refDBSS (CS) neoDBSS (CS)

Green Signal Red Signal

Inte

nsi

ty

Sample by signal

Conv(C2) Conv(U2) Conv(C1) Conv(U1)

0

2000

4000

6000

8000

10000

12000

Pos. Neg. A B A B A B A B A B Pos. Neg. A B A B A B A B A B

Ctrl Ref refDBSS (ENA) neoDBSS (ENA) refDBSS (CS) neoDBSS (CS) Ctrl Ref refDBSS (ENA) neoDBSS (ENA) refDBSS (CS) neoDBSS (CS)

Green Signal Red Signal

Inte

nsi

ty

Samples by signal

NP(G) NP(A) NP(T) NP(C.)

A) Bisulfite Conversion Controls

B) Non-polymorphic Controls

Fig. 2. Quality control probes. Ctrl: control DNA. Pos.: positive control. Neg.: negative control. A and B: individuals A and B. Ref.: reference DNA. refDBSS: reference dried blood spotsample. neoDBSS: neonatal dried blood spot sample. ENA: DNA extracted with the Extract-N-Amp kit. CS: DNA extracted with the ChargeSwitch kit. (A) Bisulphite conversion con-trols: four control probes assessed the efficiency of the bisulphite conversion of the DNA. If the bisulfite conversion was successful the “C” (Conv(C1), and Conv(C2)) probesmatched the converted sequence and obtained extended. If unconverted DNA was present, the “U” (Conv(U1) and Conv(U2)) probes were extended. The performance was onlymonitored in the green channel. (B) Non-polymorphic controls: four non-polymorphic control probes were used to test the overall performance of the assay, from amplificationto detection. One probe was designed for each of the four nucleotides. NP(G) and NP(C) were measured in the green channel while NP(T) and NP(A) were measured in the redchannel.

227M.V. Hollegaard et al. / Molecular Genetics and Metabolism 108 (2013) 225–231

uncorrected and no normalization was applied. The percentage ofmethylation for each CpG site was calculated and represented bythe β-value, theoretically ranging from 0% to 100%, with 0% being notmethylated and 100% being fully methylated. For differential analysis,we chose to use the Illumina custom model, and data were normalizedby the average normalization method. The model operates under theassumption that the methylation value, β, is normally distributedamong the replicates. Furthermore, a differential score (DS) was com-puted for each probe. In the case of a locus with multiple probes,the DS across probes was averaged. The significance level was setto Pb0.001 (−30>DS>30). A positive DS value indicates that themethylation level is lower in the reference sample compared to the“case,” suggesting hypermethylation, while a negative DS value indi-cates the opposite and suggests hypomethylation.

2.4. DBSS evaluation

To evaluate the impact of storing the whole-blood samples asDBSS for 3 years and using only two 3.2 mm disks for methylomeprofiling on the DNA methylation profile, we plotted the β-values of

the Ref. samples against the refDBSS and estimated the square corre-lation coefficient (R2). Moreover, a differential methylation analysisestimated the number of CpG sites that were significantly differentin the compared samples (−30>DS>30; Pb0.001). A similarapproach was used to estimate the difference between the neoDBSSand refDBSS methylation profiles.

3. Results and discussion

3.1. Array performance

A high percentage of detected probes were seen in both the Ref.and refDBSS (Table 1). The positive controls had high β-averagevalues, and the negative controls had low values. The ENA-DNAextracted DBSS had a high percentage of detected probes (>98.7%)and β-average values close to their respective Ref.'s. The workingβ-range of the array was 0.0085–0.9869 (Table 1). The CS-DNAextracted DBSS, however, had low detection probe rates (% probes:84.3%–95.6%) (Table 1). Lower β-average values were also seen, butthe tendency was not as pronounced as that seen for the percentage

Fig. 3. Signal intensity plots of Ref. samples against refDBSS samples. ENA: DNA extracted with the Extract-N-Amp kit. CS: DNA extracted with the ChargeSwitch kit. The signal in-tensities being proportional to the amount of input DNA, the theoretically perfect scenario for the plot, are shown as a red line. Describing the data points as a “cloud,” a cloud belowthe red line indicates that the Ref. DNA input exceeds the refDBSS DNA input, and vice versa. A narrow cloud suggests that the DNA inputs of the two samples are similar, especiallyif the intercept is in origo.

228 M.V. Hollegaard et al. / Molecular Genetics and Metabolism 108 (2013) 225–231

of detected probes. In general, the ENA-extracted samples had ahigher performance rate compared to the CS-extracted samples.

3.2. Methylation array quality control

For sample and array quality control, the Illumina Human-Methylation27 BeadChip array included 40 control probes (QC probes):four probes to examine the efficiency of the staining, four extensionprobes to test the extension efficiency, three hybridization probes totest the overall performance of the array, one probe to test the efficiencyof the stripping step after the extension reaction, four probes to assess theefficiency of the bisulphite conversion of the DNA, four G/T mismatchcontrol probes for the non-specific detection of methylation signal overunmethylated background, 16 negative control probes to define thebackground of the system, and four non-polymorphic control probes totest the overall performance of the assay, from amplification to detection.

To rule out array performance or failed bisulphite conversions asreasons for the poor results from the CS samples, we evaluated theperformance of the array using the control probes included withthe methylation array. The bisulphite conversion probe intensities ofthe CS-extracted samples were lower compared to the Ref. andENA-extracted samples, indicating a lower conversion rate, lowerquantity, and/or lower quality of the input DNA (Fig. 2A). An additionalmethod to present the data is to plot the signal intensities of the Ref.samples against the refDBSS (Fig. 3). Whereas the ENA-DNA extractedrefDBSS had similar sample intensities as the Ref. samples, the CS-DNAextracted refDBSS differed substantially, and the “cloud” was wider,suggesting that the amount of DNA was lower in these samplescompared to the Ref. samples (Fig. 3). In addition, the non-polymorphiccontrol test evaluating the overall performance of the assay from ampli-fication to detection revealed a similar pattern, suggesting that DNAextraction of DBSS with the CS approach did not perform as well as if

Table 2Differential methylation analysis of reference samples. Differential methylation analysis of data using the two individuals Ref. samples as References. Reference: individual A (A) orB (B). ENA: DNA extracted with the Extract-N-Amp method. No.: number of hyper- or hypo-methylated genes (Pb0.001). Δβmean: the mean difference in β-values. Δβ range: therange in which these data is presented.

Hypermethylated Hypomethylated

Reference Sample Individual No. Δβ mean Δβ range No. Δβ mean Δβ range

Individual A Ref. B 402 0.4031 (0.2032; 0.7161) 25 −0.328 (−0.5275; −0.2226)refDBSS (ENA) A 0 0 0 0 0 0

B 345 0.3764 (0.2199; 0.6430) 29 −0.3344 (−0.5144; −0.2411)refDBSS (CS) A 128 0.2974 (0.2058; 0.7882) 858 −0.2972 (−0.5178; −0.1585)

B 1080 0.3112 (0.1681; 0.7949) 2513 −0.3119 (−0.7351; −0.1514)Individual B Ref. A 25 0.328 (0.2226; 0.5275) 402 −0.4031 (−0.7161; −0.2033)

refDBSS (ENA) A 25 0.3156 (0.2109; 0.4947) 385 −0.4044 (−0.7059; −0.2112)B 0 0 0 0 0 0

refDBSS (CS) A 234 0.2823 (0.1762; 0.7975) 1439 −0.3205 (−0.7022; −0.1540)B 885 0.2604 (0.1560; 0.6718) 2552 −0.3131 (−0.6519; −0.1535)

229M.V. Hollegaard et al. / Molecular Genetics and Metabolism 108 (2013) 225–231

the DNAwas extractedwith the ENA kit (Fig. 2B). The G/Tmismatch con-trol probe figure indicated that less CS material was available fornon-specific detection of the methylation signal over unmethylatedbackground (Supplementary data 1A). However, there was no indicationof differences in the staining, extension, hybridization, or stripping proce-dures of the different samples or in the definition of the backgroundsignal of the system (Supplementary data 1B–F). Based on the above,we concluded that the differences seen in Table 1 are not assay relatedor due to a failing bisulphite conversion but rather a consequence ofthe amount and/or quality of the extracted DNA.

3.3. Analyzing the refDBSS DNA methylation profiles

Using DBSS for global methylation profiling raises concerns relat-ed to variables such as storage time, storage conditions, the qualityand quantity of the DNA used for bisulphite DNA conversion, andhow these affect the subsequent methylation array results. The idealway to evaluate the effects of these variables would be to comparenumerous whole-blood samples to corresponding DBSS stored forup to 30 years. Unfortunately, we did not have access to such a sam-ple set, so as an alternative, we used two samples that, at the time ofmethylation profiling, had been stored for 3 years.

We evaluated whether transferring the adult whole-blood sam-ples to filter paper, storing them for 3 years, and using only a fractionof the recommended amount of DNA for arraying would result in astatistically significantly change in the methylation profile. First, weplotted the β-values of the Ref. samples against the refDBSS, estimat-ed the R2 value (Supplementary file 2), and then carried out a differ-ential methylation analysis (Table 2). As biological replicates, weexpected R2 values of above 98%, equal to what Illumina expectsfrom technical replicates. Plotting the reference sample β-values ofindividual A against those of individual B, we estimated the R2 valueto be below 0.98, indicating that perhaps biology has a higher impacton the R2 value than the handling and storage of the sample

Table 3Differential analysis of the neonatal and adult profiles. Differential methylation analysis of dthe Extract-N-Amp method. No.: number of hyper- or hypo-methylated genes (Pb0.001). Δpresented.

Hypermethylated

Reference Sample Individual No. Δβ mean

refDBSS (A) refDBSS (ENA) A Ref. Ref.B 332 0.3766

neoDBSS (ENA) A 19 0.3328B 421 0.3685

refDBSS (B) refDBSS (ENA) A 22 0.3445B Ref. Ref.

neoDBSS (ENA) A 55 0.3203B 35 0.3456

(Supplementary file 2). Instead of introducing stringent criteria forthe differential methylation analysis, we chose a moderate signifi-cance level (−30>DS>30; Pb0.001) (Table 2). The differentialmethylation analysis, comparing the Ref. DNA methylation profile tothe refDBSS, indicated that there were no statistically significantlydifferences between the two when the DNA was extracted using theENA kit (Table 2). However, the CS-DNA extracted refDBSS DNAmethylation profiles were very different from their Ref.'s, as werethe results when comparing the two subjects (Table 2). Overall, theresults indicate that it is possible to use only two 3.2 mm DBSSdisks for ENA DNA extraction, bisulphite conversion, and globalarray DNA methylation profiling without bias.

3.4. Methylation profile changes from birth to young adulthood

As the methylation profile of the CS refDBSS did not correlate wellwith the Ref. samples, we decided to focus only on the ENA neoDBSSsamples in the following section (Table 3). For a full table, pleaseconsult Supplementary file 3.

As the profiles of the Ref. and refDBSS were shown to be identicalin the previous section, sample A and B refDBSS were chosen torepresent the adult profile. The interval between collecting individualA and B's neoDBSS and refDBSS were 28-, and 26-years, respectively.The correlation between neoDBSS collected 26–28 years ago and theadult refDBSS samples was quite high (R2>0.9770) (Supplementaryfile 2). The difference in the methylation profile from birth to earlyadulthood was estimated as in the previous section, and probeswith a DS>30 (hypermethylated) or a DSb−30 (hypomethylated)(Pb0.001) were considered statistically significant. Of the 27,578CpG array sites, only 50 sites (0.18%; 19 hypermethylated and 31hypomethylated) in individual A and 51 sites (35 hypermethylatedand 16 hypomethylated) in individual B exhibited statistically signif-icant differential methylation from birth to young adulthood (Table 3,Supplementary file 4). Because they were siblings, we compared

ata using the two individual's refDBSS samples as references. ENA: DNA extracted withβ mean: the mean difference in β-values. Δβ range: the range in which these data is

Hypomethylated

Δβ range No. Δβ mean Δβ range

Ref. Ref. Ref. Ref.(0.2007; 0.6328) 22 −0.3445 (−0.4700; −0.2333)(0.2574; 0.4843) 31 −0.3586 (−0.7440; −0.2401)(0.1994; 0.6701) 59 −0.3305 (−0.7304; −0.2203)(0.2333; 0.4700) 332 −0.3766 (−0.6328; −0.2007)Ref. Ref. Ref. Ref.(0.1885; 0.5112) 398 −0.3646 (−0.7063; −0.1976)(0.2400; 0.5036) 16 −0.3731 (−0.6927; −0.2399)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

KCNAB3 SEC31L2 POU3F1 ACTN3 MAMDC2 GP1BB ECEL1 KLHDC7B ZNF710 BTBD3 EDARADD OLFML2A PRKG2 FLJ10945 GBP1

Bet

a-va

lue

Gene Name

beta neoDBSS (A) beta refDBSS (A) beta neoDBSS (B) beta refDBSS (B)

Fig. 4. Genes significantly changed from birth to young adulthood in both individuals A and B. Beta neoDBSS (A) (light blue) represents the β-values of the genes in the neonatalDBSS of individual A; beta refDBSS (A) (dark blue), the adult DBSS of individual A; beta neoDBSS (B) (light red), the neonatal DBSS of individual B; and beta refDBSS (B) (dark red),the adult DBSS of individual B. Only genes in which a statistically significant (Pb0.001) difference was seen from the neonatal to the adult sample in both individuals A and B arepresented. For further details and the full list of significant changes in both individuals, see Supplementary file 4.

230 M.V. Hollegaard et al. / Molecular Genetics and Metabolism 108 (2013) 225–231

the lists of significant genes and found that they shared fivehypermethylated and 10 hypomethylated sites (Fig. 4). To ourknowledge, only one of these 15 genes, EDARADD (Edar associateddeath domain), has previously been associated with changes in DNAmethylation due to ageing [17]. EDARADD is believed to be involvedin the formation of hair follicles, sweat glands, and teeth and in theslowing of wound healing [17]. The results suggest that 26–28 yearsof environmental exposure does not change the global DNA methyla-tion profile of whole-blood samples tremendously (Table 3, Supple-mentary file 4). If we had not included the refDBSS and insteadcompared the whole-blood samples to the “gold-reference” DNAsample, we would most likely have obtained inconclusive results. Inaddition, if the neoDBSS were of poor quality, and if the profile hadbeen influenced by the storage time and low DNA input, we wouldexpect to see more differences between the neoDBSS and referenceprofiles, as well as poor QC probe profiles (Fig. 2, Supplementary file1). However, these differences were not observed, so we find itreasonable to conclude that despite the 26–28 years of storage theneoDBSS in the Danish Newborn Screening Biobank can be used forglobal methylome profiling, further expanding the use of theneoDBSS in studies of human diseases.

3.5. Comparing previous studies to this study

Wong et al. had previously reported that DBSS can be used forDNA methylation profiling [18]. In their study, they showed that theepigenetic profile of two genes (Vitamin D receptor and DNAmethyltransferase 3-like) was stable when comparing results fromaged and freshly prepared DBSS. However, with the very limitedamounts of neonatal material, it is very important to extract themost data possible from the limited sample material. A recent articlein Genome Research by Beyan et al. has brought focus to the use of

Guthrie cards (neonatal DBSS) for global DNA methylation profiling[19]. The authors suggest that Guthrie cards can be used formethylome profiling. However, they used FTA DMPK-A cards(Whatman FTA® DMPK-A) as filter paper. A significant difference be-tween the filter paper used for neonatal screening and the FTADMPK-A filter paper, is that the latter contains a mixture of chemicalsthat lyse cells and denature degradative enzymes and other proteins,which is beneficial for an optimal DNA storage and extraction process,whereas the filter paper used for neonatal screening (Whatman 903®Specimen Collection Paper) is made of pure cellulose [20]. Moreover,they transferred umbilical cord-blood to filter paper, whereas theblood on neonatal DBSS is drawn from a heel-prick of the newborn.That being said, their study was effectively demonstrated the impor-tance of knowing and adjusting for the genetic profile when analyzingDNA methylation data. They also pointed out the limitations of usingGuthrie cards or other types of filter paper samples for epigenomicanalysis. The most important limitation is the inability to sort cells,which leads to analyzing whole-blood as a heterogeneous cell popu-lation [19].

In comparison to the two previous studies, we first matched theglobal methylation profile of a Ref. sample to the profile made froma fraction (two 3.2 mm disks) of the same sample spotted on neona-tal Whatman 903® Specimen Collection Paper and stored for 3 years(refDBSS). This helped us to evaluate whether spotting whole-bloodsamples, storing them for 3 years, and using a fraction of therecommended amount of DNA introduced bias in the global methyla-tion profile. Second, the methylation profiles of the refDBSS werematched to the profiles of the original neoDBSS that had been storedfor 26 or 28 years showed some changes over time, but the changeswere not overwhelming, suggesting that the original neoDBSS canbe used for reliable DNA methylation profiling despite years of stor-age and suboptimal DNA quality.

231M.V. Hollegaard et al. / Molecular Genetics and Metabolism 108 (2013) 225–231

4. Conclusions

In conclusion, this proof-of-principle study demonstrates thatneonatal DBSS stored for 26 to 28 years can beused for DNAmethylomeprofiling. This further adds to the number of high-throughput tests thatcan be carried out on archived neonatal DBSS and paves the way forfuture unique population-based studies of epigenetic modifications atbirth.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.ymgme.2013.01.016.

Competing interests

The authors declare no competing interests.

Authors' contributions

M.V.H. and D.M.H. initiated the study. M.V.H. designed the study,prepared the samples, participated in the analysis and interpretationof the data, and drafted the article. J.G. carried out the bisulphitetreatment of the samples, the methylation arraying, participated inthe analysis and interpretation of the data, and revised the article.D.M.H. participated in the interpretation of the data and revised thearticle. All authors read and approved the final article.

Acknowledgments

We would like to acknowledge laboratory technician HøgniKallehauge Petersen and senior laboratory technician Lis VestergaardHansen for their work in identifying and handling the DBSS.

References

[1] A. Bird, DNA methylation patterns and epigenetic memory, Genes Dev. 16 (2002)6–21.

[2] S.B. Baylin, S.A. Belinsky, J.G. Herman, Aberrant methylation of gene promoters incancer—concepts, misconcepts, and promise, J. Natl. Cancer Inst. 92 (2000)1460–1461.

[3] P.A. Jones, P.W. Laird, Cancer epigenetics comes of age, Nat. Genet. 21 (1999) 163–167.[4] K.D. Robertson, A.P. Wolffe, DNA methylation in health and disease, Nature

ReviewsGenetics 1 (2000) 11–19.[5] M.E. O'Callaghan, A.H. Maclennan, C.S. Gibson, G.L. McMichael, E.A. Haan, J.L.

Broadbent, P.N. Goldwater, J.N. Painter, G.W. Montgomery, G.A. Dekker, Fetaland maternal candidate single nucleotide polymorphism associations withcerebral palsy: a case–control study, Pediatrics 129 (2012) e414–e423.

[6] E.L. Dempster, R. Pidsley, L.C. Schalkwyk, S. Owens, A. Georgiades, F. Kane, S.Kalidindi, M. Picchioni, E. Kravariti, T. Toulopoulou, R.M. Murray, J. Mill,Disease-associated epigenetic changes in monozygotic twins discordant forschizophrenia and bipolar disorder, Hum. Mol. Genet. 20 (2011) 4786–4796.

[7] M. Gardiner-Garden, M. Frommer, CpG islands in vertebrate genomes, J. Mol. Biol.196 (1987) 261–282.

[8] H. Han, C.C. Cortez, X. Yang, P.W. Nichols, P.A. Jones, G. Liang, DNA methylationdirectly silences genes with non-CpG island promoters and establishes anucleosome occupied promoter, Hum. Mol. Genet. 20 (2011) 4299–4310.

[9] M. Weber, I. Hellmann, M.B. Stadler, L. Ramos, S. Paabo, M. Rebhan, D. Schubeler,Distribution, silencing potential and evolutionary impact of promoter DNAmethylation in the human genome, Nat. Genet. 39 (2007) 457–466.

[10] B. Norgaard-Pedersen, D.M. Hougaard, Storage policies and use of the DanishNewborn Screening Biobank, J. Inherit. Metab. Dis. 30 (2007) 530–536.

[11] L. Frank, Epidemiology. When an entire country is a cohort, Science 287 (2000)2398–2399.

[12] B. Feenstra, F. Geller, C. Krogh, M.V. Hollegaard, S. Gortz, H.A. Boyd, J.C. Murray,D.M. Hougaard, M. Melbye, Common variants near MBNL1 and NKX2-5 areassociatedwith infantile hypertrophic pyloric stenosis, Nat. Genet. 44 (2012) 334–337.

[13] S. Ripke, A.R. Sanders, K.S. Kendler, D.F. Levinson, P. Sklar, P.A. Holmans, D.Y. Lin, J.Duan, R.A. Ophoff, O.A. Andreassen, E. Scolnick, S. Cichon, D. St Clair, A. Corvin, H.

Gurling, T. Werge, D. Rujescu, D.H. Blackwood, C.N. Pato, A.K. Malhotra, S. Purcell,F. Dudbridge, B.M. Neale, L. Rossin, P.M. Visscher, D. Posthuma, D.M. Ruderfer, A.Fanous, H. Stefansson, S. Steinberg, B.J. Mowry, V. Golimbet, M. De Hert, E.G.Jonsson, I. Bitter, O.P. Pietilainen, D.A. Collier, S. Tosato, I. Agartz, M. Albus, M.Alexander, R.L. Amdur, F. Amin, N. Bass, S.E. Bergen, D.W. Black, A.D. Borglum,M.A. Brown, R. Bruggeman, N.G. Buccola, W.F. Byerley, W. Cahn, R.M. Cantor,V.J. Carr, S.V. Catts, K. Choudhury, C.R. Cloninger, P. Cormican, N. Craddock, P.A.Danoy, S. Datta, L. de Haan, D. Demontis, D. Dikeos, S. Djurovic, P. Donnelly, G.Donohoe, L. Duong, S. Dwyer, A. Fink-Jensen, R. Freedman, N.B. Freimer, M.Friedl, L. Georgieva, I. Giegling, M. Gill, B. Glenthoj, S. Godard, M. Hamshere, M.Hansen, T. Hansen, A.M. Hartmann, F.A. Henskens, D.M. Hougaard, C.M.Hultman, A. Ingason, A.V. Jablensky, K.D. Jakobsen, M. Jay, G. Jurgens, R.S. Kahn,M.C. Keller, G. Kenis, E. Kenny, Y. Kim, G.K. Kirov, H. Konnerth, B. Konte, L.Krabbendam, R. Krasucki, V.K. Lasseter, C. Laurent, J. Lawrence, T. Lencz, F.B.Lerer, K.Y. Liang, P. Lichtenstein, J.A. Lieberman, D.H. Linszen, J. Lonnqvist, C.M.Loughland, A.W. Maclean, B.S. Maher, W. Maier, J. Mallet, P. Malloy, M.Mattheisen, M. Mattingsdal, K.A. McGhee, J.J. McGrath, A. McIntosh, D.E.McLean, A. McQuillin, I. Melle, P.T. Michie, V. Milanova, D.W. Morris, O. Mors,P.B. Mortensen, V. Moskvina, P. Muglia, I. Myin-Germeys, D.A. Nertney, G.Nestadt, J. Nielsen, I. Nikolov, M. Nordentoft, N. Norton, M.M. Nothen, C.T.O'Dushlaine, A. Olincy, L. Olsen, F.A. O'Neill, T.F. Orntoft, M.J. Owen, C. Pantelis,G. Papadimitriou, M.T. Pato, L. Peltonen, H. Petursson, B. Pickard, J. Pimm, A.E.Pulver, V. Puri, D. Quested, E.M. Quinn, H.B. Rasmussen, J.M. Rethelyi, R. Ribble,M. Rietschel, B.P. Riley, M. Ruggeri, U. Schall, T.G. Schulze, S.G. Schwab, R.J.Scott, J. Shi, E. Sigurdsson, J.M. Silverman, C.C. Spencer, K. Stefansson, A.Strange, E. Strengman, T.S. Stroup, J. Suvisaari, L. Terenius, S. Thirumalai, J.H.Thygesen, S. Timm, D. Toncheva, E. van den Oord, J. van Os, R. van Winkel, J.Veldink, D. Walsh, A.G. Wang, D. Wiersma, D.B. Wildenauer, H.J. Williams, N.M.Williams, B. Wormley, S. Zammit, P.F. Sullivan, M.C. O'Donovan, M.J. Daly, P.V.Gejman, Genome-wide association study identifies five new schizophrenia loci,Nat. Genet. 43 (2011) 969–976.

[14] S. Steinberg, S. de Jong, O.A. Andreassen, T. Werge, A.D. Borglum, O. Mors, P.B.Mortensen, O. Gustafsson, J. Costas, O.P. Pietilainen, D. Demontis, S. Papiol, J.Huttenlocher, M. Mattheisen, R. Breuer, E. Vassos, I. Giegling, G. Fraser, N.Walker, A. Tuulio-Henriksson, J. Suvisaari, J. Lonnqvist, T. Paunio, I. Agartz, I.Melle, S. Djurovic, E. Strengman, G. Jurgens, B. Glenthoj, L. Terenius, D.M.Hougaard, T. Orntoft, C. Wiuf, M. Didriksen, M.V. Hollegaard, M. Nordentoft, R.van Winkel, G. Kenis, L. Abramova, V. Kaleda, M. Arrojo, J. Sanjuan, C. Arango, S.Sperling, M. Rossner, M. Ribolsi, V. Magni, A. Siracusano, C. Christiansen, L.A.Kiemeney, J. Veldink, L. van den Berg, A. Ingason, P. Muglia, R. Murray, M.M.Nothen, E. Sigurdsson, H. Petursson, U. Thorsteinsdottir, A. Kong, I.A. Rubino, M.De Hert, J.M. Rethelyi, I. Bitter, E.G. Jonsson, V. Golimbet, A. Carracedo, H.Ehrenreich, N. Craddock, M.J. Owen, M.C. O'Donovan, M. Ruggeri, S. Tosato, L.Peltonen, R.A. Ophoff, D.A. Collier, D. St Clair, M. Rietschel, S. Cichon, H.Stefansson, D. Rujescu, K. Stefansson, Common variants at VRK2 and TCF4conferring risk of schizophrenia, Hum. Mol. Genet. 20 (2011) 4076–4081.

[15] H. Stefansson, R.A. Ophoff, S. Steinberg, O.A. Andreassen, S. Cichon, D. Rujescu, T.Werge, O.P. Pietilainen, O. Mors, P.B. Mortensen, E. Sigurdsson, O. Gustafsson, M.Nyegaard, A. Tuulio-Henriksson, A. Ingason, T. Hansen, J. Suvisaari, J. Lonnqvist, T.Paunio, A.D. Borglum, A. Hartmann, A. Fink-Jensen, M. Nordentoft, D. Hougaard, B.Norgaard-Pedersen, Y. Bottcher, J. Olesen, R. Breuer, H.J. Moller, I. Giegling, H.B.Rasmussen, S. Timm, M. Mattheisen, I. Bitter, J.M. Rethelyi, B.B. Magnusdottir, T.Sigmundsson, P. Olason, G. Masson, J.R. Gulcher, M. Haraldsson, R. Fossdal, T.E.Thorgeirsson, U. Thorsteinsdottir, M. Ruggeri, S. Tosato, B. Franke, E. Strengman,L.A. Kiemeney, I. Melle, S. Djurovic, L. Abramova, V. Kaleda, J. Sanjuan, R. deFrutos, E. Bramon, E. Vassos, G. Fraser, U. Ettinger, M. Picchioni, N. Walker, T.Toulopoulou, A.C. Need, D. Ge, J.L. Yoon, K.V. Shianna, N.B. Freimer, R.M. Cantor,R. Murray, A. Kong, V. Golimbet, A. Carracedo, C. Arango, J. Costas, E.G. Jonsson,L. Terenius, I. Agartz, H. Petursson, M.M. Nothen, M. Rietschel, P.M. Matthews, P.Muglia, L. Peltonen, D. St Clair, D.B. Goldstein, K. Stefansson, D.A. Collier, Commonvariants conferring risk of schizophrenia, Nature 460 (2009) 744–747.

[16] M.V. Hollegaard, K. Skogstrand, P. Thorsen, B. Norgaard-Pedersen, D.M. Hougaard, J.Grove, Joint analysis of SNPs and proteins identifies regulatory IL18 gene variationsdecreasing the chance of spastic cerebral palsy, Hum. Mutat. 34 (2013) 143–148.

[17] S. Bocklandt, W. Lin, M.E. Sehl, F.J. Sanchez, J.S. Sinsheimer, S. Horvath, E. Vilain,Epigenetic predictor of age, PLoS One 6 (2011) e14821.

[18] N. Wong, R. Morley, R. Saffery, J. Craig, Archived Guthrie blood spots as a novelsource for quantitative DNA methylation analysis, Biotechniques 45 (2008)423–424, (426, 428 passim).

[19] H. Beyan, T.A. Down, S.V. Ramagopalan, K. Uvebrant, A. Nilsson, M.L. Holland, C.Gemma, G. Giovannoni, B.O. Boehm, G.C. Ebers, A. Lernmark, C.M. Cilio, R.D.Leslie, V.K. Rakyan, Guthrie card methylomics identifies temporally stableepialleles that are present at birth in humans, Genome Res. 22 (2012)2138–2145.

[20] J.V. Mei, J.R. Alexander, B.W. Adam, W.H. Hannon, Use of filter paper for the collectionand analysis of human whole blood specimens, J. Nutr. 131 (2001) 1631S–1636S.