| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Molecular Diagnostics and Genetics |
1 Centre for Research into Circulating Fetal Nucleic Acids, Li Ka Shing Institute of Health Sciences; 2 Department of Obstetrics and Gynaecology; 3 Department of Chemical Pathology; 4 Centre for Emerging Infectious Diseases, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong SAR.
aAddress correspondence to this author at: Department of Chemical Pathology, Rm. 38061, 1/F, Clinical Sciences Building, Prince of Wales Hospital, 30-32 Ngan Shing St., Shatin, Hong Kong SAR. Fax 852-2636-5090; e-mail loym{at}cuhk.edu.hk.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: To develop a similar method for the noninvasive detection of trisomy 21, we used methylation-sensitive single nucleotide primer extension and/or bisulfite sequencing to systematically search 114 CpG islands (CGIs)—76% of the 149 CGIs on chromosome 21 identified by bioinformatic criteria—for differentially methylated DNA patterns. The methylation index (MI) of a CpG site was estimated as the proportion of molecules methylated at that site.
Results: We identified 22 CGIs which were shown to contain CpG sites that were either completely unmethylated (MI = 0.00) in maternal blood cells and methylated in the placenta (MI range, 0.22–0.65), or completely methylated (MI = 1.00) in maternal blood cells and hypomethylated in the placenta (MI range, 0.00–0.75). We detected, for the first time, placental DNA-methylation patterns on chromosome 21 in maternal plasma during pregnancy and observed their postpartum clearance.
Conclusion: Twenty-two (19%) of the 114 studied CGIs on chromosome 21 showed epigenetic differences between samples of placenta and maternal blood cells; these CGIs may provide a rich source of markers for noninvasive prenatal diagnosis.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We have pursued the development of sex- and polymorphism-independent fetal-DNA markers via epigenetic approaches and have used an imprinted locus in selected fetal–maternal pairs to demonstrate the feasibility of this approach for the detection of fetal DNA in maternal plasma (5). To generalize this approach to all pregnancies, we searched for epigenetic differences, namely DNA-methylation differences, between the placenta and maternal blood cells, which have been inferred to be the predominant sources of fetal and maternal nucleic acids, respectively, in maternal plasma (6)(7). Subsequently, we discovered the SERPINB51 gene [serpin peptidase inhibitor, clade B (ovalbumin), member 5; also known as maspin] to be hypomethylated in the placenta but completely methylated in maternal blood cells. We showed that the placental unmethylated form, U-SERPINB5, was released into the maternal plasma during pregnancy and rapidly cleared upon delivery of the fetus. U-SERPINB5 is thus the first sex- and polymorphism-independent fetal-DNA marker to be found in maternal plasma (8). Because SERPINB5 is located on chromosome 18, we can infer the dosage of fetal chromosome 18 by assessing the ratio of the polymorphic alleles of fetus-derived U-SERPINB5 molecules in the maternal plasma. We demonstrated that this epigenetic allelic ratio for U-SERPINB5 molecules in the maternal plasma of pregnancies involving trisomy 18 was distinguishable from the ratios in unaffected pregnancies (9).
Recently, we have found another fetal epigenetic marker, RASSF1 [Ras association (RalGDS/AF-6) domain family 1], which was hypermethylated in the placenta but completely unmethylated in maternal blood cells (10). We have developed the placental methylated form, M-RASSF1, as a universal marker for fetal DNA in maternal plasma (11). Extrapolating from our results with U-SERPINB5 and M-RASSF1, we reasoned that the placenta and maternal blood cells might also show epigenetic differences on chromosome 21 that may allow us to distinguish fetal and maternal chromosome 21 molecules in maternal plasma; however, DNA-methylation patterns on chromosome 21 have not been investigated systematically for both the placenta and maternal blood cells. One study investigated chromosome 21 methylation in blood cells of nonpregnant individuals (12). Another study investigated chromosome 21 methylation in only 1 male placenta, which served as a control for brain tissues (13). Yet another study investigated chromosome 21 methylation in lung and colon cells (14). A report of a study involving 1 placenta and lymphocytes described the methylation status of chromosomes 6, 20, and 22, but not that of chromosome 21 (15). One last relevant study reported chromosome 21 methylation status in placenta and blood cell samples, but it relied on an assay that used a methylation-sensitive restriction enzyme, HpaII (16), that enables analysis of only 3.9% of all CpG sites in the human genome (17).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
participant recruitment, sample collection, and processing
Informed consent was obtained from women who had uncomplicated singleton pregnancies and who were patients in the Department of Obstetrics and Gynaecology at the Prince of Wales Hospital, Hong Kong. The study was approved by the institutional review board. First- and third-trimester participants were recruited from women undergoing pregnancy termination and elective cesarean delivery, respectively. Maternal peripheral blood samples (12 mL) were collected into tubes containing EDTA just before the obstetric procedures. Placental tissues and postdelivery maternal blood were collected immediately and at 24 h after the procedures, respectively. DNA was extracted from processed samples of placental tissues, maternal blood cells, and plasma (see Methods in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol54/issue3).
rapid methylation analysis by methylation-sensitive single nucleotide primer extension followed by mass spectrometric detection
For the cost-effective and efficient screening of CGIs containing differentially methylated CpG sites in the placenta and maternal blood cells, we used methylation-sensitive single nucleotide primer extension (Ms-SNuPE), a rapid and quantitative method for assessing methylation differences at CpG sites that can be scaled up for high-throughput analysis (18). This method involves a bisulfite treatment that converts unmethylated cytosine residues into uracil residues and leaves methylated cytosine (5-methylcytosine) residues unchanged (19). Methylated and unmethylated CpG molecules can therefore be distinguished by the changes in base sequence. We targeted such changes by primer-extension reactions to yield products of different masses, which we resolved by matrix-assisted laser-desorption and ionization time-of-flight mass spectrometry with the MassARRAY system (Sequenom; see Methods and Tables S1 and S2 in the online Data Supplement) (20). Unlike methods that use methylation-sensitive restriction enzymes (12)(13)(16), this method can screen for CpG sites that fall outside of any restriction enzyme–recognition sequences. As an illustration, >84% of the CpG sites investigated in our study fell outside the recognition site of HpaII, a commonly used methylation-sensitive restriction enzyme.
We screened 1–3 CpG sites by Ms-SNuPE for each CGI. The methylation index (MI) of each CpG site was estimated by dividing the peak height of the methylated extension product by the total peak height for the methylated and unmethylated products. We performed Ms-SNuPE assays on 5 first-trimester placentas (3 male and 2 female fetuses) and 2 samples of maternal blood cells and reported median MI values for these samples. Ms-SNuPE assays that yielded signals for at least 1 sample from each tissue type were reported for interpretation in this rapid screening.
high-resolution methylation analysis by cloning and bisulfite genomic sequencing
To maximize the chance of identifying CpG sites suitable for marker development, we performed bisulfite sequencing at single-CpG resolution to investigate the methylation status of the dense cluster of CpG sites inside selected CGIs and analyzed >2400 CpG sites. Before sequencing, we performed TA cloning and randomly chose at least 8 clones per PCR to minimize biased representation of any molecules or cell type in each tissue sample (see Methods and Tables S1 and S3 in the online Data Supplement). We calculated the MI for each CpG site by dividing the number of methylated clones by the total number of analyzed clones and expressed MI values for samples of the same tissue type obtained from multiple individuals as the median for each CpG site.
noninvasive detection of fetal chromosome 21 sequences in maternal plasma
We then explored differentially methylated CpG sites in the placenta and in maternal blood cells as possible fetus-specific DNA markers in maternal plasma. First, we designed real-time quantitative methylation- specific PCR (qMSP) assays (21) to target the unmethylated form of CGI084, which is linked to PDE9A (phosphodiesterase 9A), in the placenta (U-PDE9A) and the methylated form in maternal blood cells (M-PDE9A). We designed our second approach, the MassEXTEND assay (Sequenom), to detect the unmethylated form of CGI137 (U-CGI137) in the placenta and assayed for the presence of U-PDE9A, M-PDE9A, and U-CGI137 in samples of maternal plasma obtained before and after delivery (see Methods in the online Data Supplement).
statistical analysis
Statistical analyses were performed with SigmaStat 3.0 software (SPSS).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
methylation of analysis of cgiS in maternal blood cells by mS-snUpe
We performed an initial study to exclude the possibility that pregnancy might affect the methylation status of blood cells, because a previous study had reported the methylation status of the CGIs on chromosome 21 only in blood cells of nonpregnant individuals (12). We used Ms-SNuPE to study about half of the 114 analyzable CGIs in maternal blood cells. Of these 53 CGIs, all of which had previously been reported to be unmethylated (12), we confirmed that 47 (88.7%) of the CGIs contained completely unmethylated (i.e., MI = 0.00) CpG sites in maternal blood cells (Fig. 1A
; Fig. S1 in the online Data Supplement). Thus, our data suggest that the methylation status of CGIs in DNA extracted from maternal blood cells is highly concordant with that of CGIs in DNA from blood cells from nonpregnant individuals (12), despite the use of different methods of investigation.
|
identification of cgiS with cPg sites completely unmethylated in maternal blood cells and partially methylated in the placenta
Our experience with M-RASSF1 (10) indicates that loci that are hypermethylated in the placenta would be promising candidates for fetal-DNA markers if they met 2 criteria: (a) criterion 1 [the CpG site is completely unmethylated (MI = 0.00) in maternal blood cells], and (b) criterion 2 [the degree of hypermethylation in the placenta reaches an MI of 
0.20 so as to be detectable in maternal plasma (see Methods in the online Data Supplement)].
We therefore targeted 103 CGIs that had previously been reported to be completely unmethylated in blood cells (12) and performed Ms-SNuPE analyses of 5 first-trimester placentas. Our Ms-SNuPE analyses of 53 CGIs revealed that 4 CGIs (7.5%; CGI009, CGI045, CGI071, and CGI113) contained at least 1 CpG site fulfilling both criteria 1 and 2 (Fig. 1A
; summarized in Table 1
). These data were confirmed by bisulfite sequencing (Fig. 1B
and C; Table 2
; summarized in Table 1
).
|
|
For the remaining 50 CGIs not analyzable by Ms-SNuPE, we attempted bisulfite sequencing with samples from 2 first-trimester placentas. Except for CGI138, which failed to yield any products after systematic optimization, we successfully sequenced all 49 of the other CGIs. Eleven (22.4%) of these CGIs contained at least 1 CpG site with an MI 0.20 in the placenta, fulfilling criterion 2 (Fig. 1B
and C; Table 1
), whereas the rest did not (see Fig. 2
in the online Data Supplement). We therefore performed bisulfite sequencing on 2 samples of first-trimester maternal blood cells for these 11 CGIs. Nine (18.4%) of these 49 CGIs contained at least 1 CpG site with an MI of 0.00 in maternal blood cells and an MI 0.20 (range, 0.22–0.65) in the placentas, fulfilling both criteria 1 and 2 for a potential fetal marker (Tables 1
and 2
).
|
identification of cgiS with cPg sites completely methylated in maternal blood cells and hypomethylated in the placenta
Our experience with U-SERPINB5 (8) suggests that loci hypomethylated in the placenta would be promising candidates for fetal-DNA markers if they fulfilled 2 other criteria: (i) criterion 3 [the CpG site is completely methylated (MI = 1.00) in maternal blood cells], and (ii) criterion 4 [the CpG site is hypomethylated in the placenta with an MI 
0.80, to facilitate PCR detection (see Methods in the online Data Supplement)]. Criteria 3 and 4 are essentially the opposite of criteria 1 and 2, respectively.
We therefore targeted the 31 CGIs previously reported to be methylated in blood cells (12); however, we could not analyze 19 of these CGIs because they contained tandem repeat DNA sequences or were highly similar to regions on other chromosomes. Primers for bisulfite sequencing were successfully designed for the remaining 12 CGIs. We analyzed 5 third-trimester samples of placenta and maternal blood cells. Nine (75.0%) of these 12 CGIs contained at least 1 CpG site with an MI 0.80 in placental tissues and an MI value of 1.00 in maternal blood cells. (Fig. 1D
and E; Tables 1
and 3
). Four of these 9 CGIs were chosen for further investigation of 5 paired samples of first-trimester placentas and maternal blood cells, the supply of which was relatively limited compared with third-trimester tissues. We chose CGI084 and CGI137 because they had the lowest median MI values in the third-trimester placentas for the applicable CpG sites. We also chose CGI040 and CGI043 because they had 20 and 21 applicable CpG sites, respectively, that fulfilled criteria 3 and 4 (just a few less than the 25 applicable CpG sites found in CGI084). For the CGI requiring multiple amplicons for analysis, only the amplicon with the most applicable CpG sites was sequenced for the first-trimester tissues. All 4 CGIs also contained CpG sites with an MI value of 1.00 in the first-trimester samples of maternal blood cells and an MI value 0.80 in the paired samples of placenta (Fig. 1F
and G; Table 3
). Thus, these 4 CGIs were suitable for fetal-marker development for both first- and third-trimester maternal plasma.
|
selection of candidate loci for development as circulating markers of fetal dna
In summary, we identified 22 CGIs (Tables 2
and 3
), encompassing 255 CpG sites, which served as suitable candidates for development as markers of circulating fetal DNA. To find the most promising candidates, we considered the CpG sites with the greatest differences in median MI values between the first-trimester placentas and maternal blood cells and selected CGI113 (0.57 vs 0.00), CGI084 (0.50 vs 1.00), and CGI137 (0.50 vs 1.00). To allow the most flexibility in designing assays with short PCR amplicons for detecting short fetal-DNA fragments in maternal plasma (22), we preferred a relatively short distance between consecutive applicable CpG sites. The median (range) distances between consecutive applicable CpG sites were 26 (22–101) nucleotides, 8 (2–45) nucleotides, and 13 (3–30) nucleotides in CGI113, CGI084, and CGI137, respectively; hence, we chose CGI084 and CGI137 for further investigation in maternal plasma.
detection of placental u-pde9a dna by real-time Qmsp in maternal plasma and postpartum clearance
We hypothesized that DNA sequences bearing placenta-specific epigenetic signatures would be released into maternal plasma and be cleared rapidly after delivery of the fetus. To test this hypothesis, we examined samples of maternal plasma obtained from 12 third-trimester pregnancies (8 male and 4 female fetuses) before and after delivery for the presence of the unmethylated form of CGI084 (U-PDE9A), which is linked to the PDE9A gene. On the basis of the PDE9A-methylation patterns in the placenta and maternal blood cells, we designed a qMSP assay that targeted U-PDE9A (Fig. 2A
). Before delivery, we detected U-PDE9A in the 12 samples of maternal plasma at a median (interquartile range) concentration of 76 x 103 (38 x 103 – 21 x 104) copies/L (Fig. 2B
). At 24 h postpartum, U-PDE9A concentrations in the maternal plasma rapidly declined to almost undetectable levels (P 0.001, Wilcoxon signed rank test; Fig. 2C
). As a positive control for DNA extraction and bisulfite conversion, M-PDE9A (which is not placenta specific) was detected in all 24 maternal plasma samples, with no significant differences between samples obtained before and after delivery (P = 0.910, Wilcoxon signed rank test).
mASSextend detection of placental u-cgi137 dna in maternal plasma and its postpartum clearance
We next investigated the presence of U-CGI137 in maternal plasma; however, we identified far fewer differentially methylated CpG sites in CGI137 than in CGI084 (Table 3
). This paucity of differentially methylated sites made it more difficult to develop qMSP assays to distinguish U-CGI137 from M-CGI137. We designed a MassEXTEND assay to detect U-CGI137 after MSP that was based on only 5 identified differentially methylated CpG sites within CGI137, in contrast to 10 such sites in PDE9A (cf., Fig. 3A
and Fig. 2A
), and investigated its specificity and sensitivity for detecting U-CGI137 (see Methods in the online Data Supplement; Fig. 3B
and C). In a subsequent application of this assay, we confirmed the presence of U-CGI137 in 8 samples of third-trimester maternal plasma (4 male and 4 female fetuses) obtained before delivery (Fig. 3D
). Postdelivery samples of maternal plasma were also available for 7 of these 8 cases, and no U-CGI137 was detected in these 7 samples (Fig. 3E
). These data demonstrate that loci with few CpG sites differentially methylated in placenta and maternal blood cells can still be used to distinguish fetal DNA in maternal plasma.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our identification strategy involved the use of a rapid Ms-SNuPE approach and a higher-resolution but more labor-intensive approach via bisulfite sequencing. Because both methods are based on bisulfite conversion, we were able to identify CpG sites that were not recognized by any methylation-sensitive restriction enzyme. In fact, only 40 CpG sites applicable for fetal-marker development (16% of the 255 CpG sites; see Table S4 in the online Data Supplement) would be recognized by HpaII, on which a previous chromosome 21 study relied (16). Thus, our bisulfite-based approach has increased the number of applicable CpG sites by 5-fold or more, compared with an HpaII-based approach.
Of the 2440 CpG sites analyzed in this study, 255 CpG sites were found to be suitable for fetal-marker development. Extrapolating this fraction to the >198 000 CpG sites in the nonrepetitive sequence on chromosome 21 (17), we estimate that there are >20 600 CpG sites suitable for fetal-marker development. More recently developed techniques, such as immunoprecipitation of methylated DNA followed by tiling-array analysis (14), are likely to reveal yet more marker candidates.
We believe that the current study has opened up an important resource for facilitating noninvasive prenatal diagnosis of trisomy 21 with maternal plasma. Recently, the identification of RNA transcripts that are differentially expressed in the placenta has allowed the specific detection of fetal RNA in maternal plasma (23), and the allelic ratio for one such RNA transcript on chromosome 21 has been demonstrated to facilitate the noninvasive detection of trisomy 21 (24). We have demonstrated in the present study that placental epigenetic signatures offer another class of markers besides placental RNA transcripts for distinguishing chromosome 21–derived fetal and maternal genetic material in maternal plasma. We envision that placenta-specific epigenetic signatures on chromosome 21 would supplement placenta-expressed RNA transcripts for facilitating the noninvasive diagnosis of trisomy 21. Although placenta-expressed RNA transcripts are often detected at higher concentrations in the plasma, epigenetic signatures of placental DNA have the desirable feature of being apparently relatively widespread in the human genome [22 (19.3%) of 114 analyzed CGIs on chromosome 21]. Such epigenetic markers can be used for the noninvasive prenatal detection of trisomy 21, either via analysis of epigenetic allelic ratios (9) or by direct comparison with a placenta-specific DNA-methylation marker on a reference chromosome. The recent development of digital PCR technologies for noninvasive prenatal diagnosis would allow such analyses to be carried out with high precision (25). Furthermore, the discovery of DNA-methylation abnormalities specific for trisomy 21 in future epigenetic studies would ultimately yield a marker for the disorder that is independent of allelic-ratio or chromosomal-dosage analyses. A combination of markers based on both fetal RNA and fetal DNA may pave the way for the development of a noninvasive test for trisomy 21 that would be applicable to almost all fetal–maternal pairs in the general population.
|
Acknowledgments |
|---|
Financial Disclosures: Y.M.D.L., R.W.K.C., C.D., Y.K.T., F.M.F.L., T.Y.H.L., S.J., and S.S.C.C. hold patents for and have filed patent applications on aspects of the use of fetal nucleic acids in maternal plasma for noninvasive prenatal diagnosis, a proportion of which have been licensed to Sequenom, Inc. Y.M.D.L. is a consultant for Sequenom, Inc. C.D. holds equities in Sequenom, Inc.
Acknowledgment: We thank Allen Chan for helpful discussion on data analysis.
|
Footnotes |
|---|
2 Nonstandard abbreviations: CGI, CpG island; MI, methylation index; U-SERPINB5, unmethylated SERPINB5; M-RASSF1, methylated RASSF1; Ms-SNuPE, methylation-sensitive single nucleotide primer extension; qMSP, real-time quantitative methylation-specific PCR.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  T. Chu, K. Bunce, W. A. Hogge, and D. G. Peters Statistical model for whole genome sequencing and its application to minimally invasive diagnosis of fetal genetic disease Bioinformatics, May 15, 2009; 25(10): 1244 - 1250. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. A. Papageorgiou, H. Fiegler, V. Rakyan, S. Beck, M. Hulten, K. Lamnissou, N. P. Carter, and P. C. Patsalis Sites of Differential DNA Methylation between Placenta and Peripheral Blood: Molecular Markers for Noninvasive Prenatal Diagnosis of Aneuploidies Am. J. Pathol., May 1, 2009; 174(5): 1609 - 1618. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E C W Hung, R W K Chiu, and Y M D Lo Detection of circulating fetal nucleic acids: a review of methods and applications J. Clin. Pathol., April 1, 2009; 62(4): 308 - 313. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Beck, H. B. Urnovitz, J. Riggert, M. Clerici, and E. Schutz Profile of the Circulating DNA in Apparently Healthy Individuals Clin. Chem., April 1, 2009; 55(4): 730 - 738. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. F. Wright and H. Burton The use of cell-free fetal nucleic acids in maternal blood for non-invasive prenatal diagnosis Hum. Reprod. Update, January 1, 2009; 15(1): 139 - 151. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. W. K. Chiu, K. C. A. Chan, Y. Gao, V. Y. M. Lau, W. Zheng, T. Y. Leung, C. H. F. Foo, B. Xie, N. B. Y. Tsui, F. M. F. Lun, et al. Noninvasive prenatal diagnosis of fetal chromosomal aneuploidy by massively parallel genomic sequencing of DNA in maternal plasma PNAS, December 23, 2008; 105(51): 20458 - 20463. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. M. F. Lun, N. B. Y. Tsui, K. C. A. Chan, T. Y. Leung, T. K. Lau, P. Charoenkwan, K. C. K. Chow, W. Y. W. Lo, C. Wanapirak, T. Sanguansermsri, et al. Noninvasive prenatal diagnosis of monogenic diseases by digital size selection and relative mutation dosage on DNA in maternal plasma PNAS, December 16, 2008; 105(50): 19920 - 19925. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. F. Smith and Y. Blumenfeld Cell-free Fetal DNA in Maternal Plasma: Progress and Potential NeoReviews, August 1, 2008; 9(8): e332 - e337. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. B. M. Oudejans Noncoding RNA and DNA as Biomarkers: Toward an Epigenetic Fetal Barcode for Use in Maternal Plasma Clin. Chem., March 1, 2008; 54(3): 456 - 457. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
