| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Brief Communications |
1 Channing Laboratory, Brigham and Women’s Hospital, Boston, MA;2 Division of Pulmonary and Critical Care Medicine, Department of Medicine, Brigham and Women’s Hospital, Boston, MA; and3 Harvard Medical School, Boston, MA.
aaddress correspondence to this author at: Channing Laboratory, Brigham and Women’s Hospital, Boston, MA 02115. Fax: 617-525-0958; e-mail: michael.cho{at}channing.harvard.edu.
Abstract
Background: High-resolution melting curve analysis is an accurate method for mutation detection in genomic DNA. Few studies have compared the performance of high-resolution DNA melting curve analysis (HRM) in genomic and whole-genome amplified (WGA) DNA.
Methods: In 39 paired genomic and WGA samples, 23 amplicons from 9 genes were PCR amplified and analyzed by high-resolution melting curve analysis using the 96-well LightScanner (Idaho Technology). We used genotyping and bidirectional resequencing to verify melting curve results.
Results: Melting patterns were concordant between the genomic and WGA samples in 823 of 863 (95%) analyzed sample pairs. Of the discordant patterns, there was an overrepresentation of alternate melting curve patterns in the WGA samples, suggesting the presence of a mutation (false positives). Targeted resequencing in 135 genomic and 136 WGA samples revealed 43 single nucleotide polymorphisms (SNPs). All SNPs detected in genomic samples were also detected in WGA. Additional genotyping and sequencing allowed the classification of 628 genomic and 614 WGA amplicon samples. Heterozygous variants were identified by non–wild-type melting pattern in 98% of genomic and 97% of WGA samples (P = 0.11). Wild types were correctly classified in 99% of genomic and 91% of WGA samples (P < 0.001).
Conclusions: In WGA DNA, high-resolution DNA melting curve analysis is a sensitive tool for SNP discovery through detection of heterozygote variants; however, it may misclassify a greater number of wild-type samples.
The comprehensive detection of novel DNA variants has traditionally relied on resequencing. Despite recent advances, however, resequencing is expensive and time consuming. Methods to screen for variants can reduce the amount of sequencing and potentially increase efficiency and reduce cost. High-resolution DNA melting curve analysis is a sensitive and specific method for variant detection (1)(2).
A concern in any genetic or genomic study that includes detection and genotyping of novel variants is the availability of sufficient quantities of DNA. Whole-genome amplified (WGA)1 DNA samples have been reported to have a high reliability in many settings (3), but few studies have specifically compared the performance of high-resolution DNA melting curve analysis (HRM) in genomic and WGA DNA. Our aim was to examine the performance of HRM on WGA DNA as a potential method for high-throughput variant discovery.
We performed whole-genome amplification via the multiple displacement amplification (MDA) technique using phi29 DNA polymerase (REPLI-g; Qiagen) on 39 subjects from the Boston Early-Onset Chronic Obstructive Pulmonary Disease Study (EOCOPD) (4). Participants in EOCOPD gave written informed consent, and the appropriate institutional review boards approved the study.
We chose for analysis 23 amplicons from 9 chronic obstructive pulmonary disease candidate genes with at least 1 known polymorphism based on previous genotyping or sequencing studies (see Supplemental Table S1, which accompanies the online version of this article at www.clinchem.org/content/vol54/issue12), resulting in 897 genomic and 897 WGA amplicon samples (Table 1
). Amplicons varied from 156 to 392 bp, with GC content from 39.4% to 70.9%. We peformed amplification reactions using Idaho Technology Mastermix with primers added to a final concentration of 0.25 µmol/L each and PCR optimization using a 10-µL reaction volume. As an amplification template, we used 10 ng of either genomic or WGA DNA. Cycling conditions were as follows: 2 min at 95 °C; followed by 40–45 cycles of 94 °C for 30 s, 60 °C to 72 °C annealing temperature gradient for 30 s, and 72 °C for 30 s; and a final denaturing and reannealing step (94 °C, 30-s ramping to 25 °C for 30 s, then 4 °C at the end of the cycling protocol). Production PCR was performed using a 5-µL reaction volume; conditions varied by amplicon. If the region amplified was of high GC content (>60%), DMSO (10% vol/vol) was included.
|
We analyzed PCR products by HRM using the 96-well LightScanner (Idaho Technology) by heating from 65 °C to 98 °C at 0.3 °C/s. Melting data were analyzed using the LightScanner Call-It software by fluorescence normalization and difference curve analysis to optimize detection of heterozygous mutants. We used standard sensitivity and autogroup settings for all samples, and a single operator reviewed normalization and default variant calls.
We selected at least 1 representative sample from each melting curve, as well as samples with discordant genomic vs WGA melting patterns, for resequencing. Samples were purified using the MinElute PCR Purification Kit (Qiagen), then bidirectionally resequenced on an ABI 3730xl Genetic Analyzer (Applied Biosystems); the resulting data were analyzed using Phred/Phrap/Consed and Polyphred software (5)(6)(7)(8). In cases where the DNA melting sample was not resequenced as part of this project, we used data from prior genotyping and sequencing reactions on genomic samples as a reference. For the purposes of analysis, we classified a sample as a heterozygous variant if any single nucleotide polymorphisms (SNPs) in the amplicon were genotyped or sequenced as heterozygous variant and as wild type if sequencing or genotyping data demonstrated that the subject was homozygous wild type for all known SNPs in that amplicon. Homozygous variants were analyzed separately. Statistical analysis was performed using SAS 9.1 (SAS Institute). We compared WGA and genomic samples using McNemar’s test for paired samples or the exact binomial as appropriate.
Of 1794 high-resolution DNA melting attempts, 1759 melting reactions (resulting in 164 unique melting patterns) were successful, for an overall completion rate of 98%. There were a greater number of failures in the WGA samples (25 failures, 2.8%) than in the genomic (10 failures, 1.1%; P = 0.009; see online Supplemental Table S2). We analyzed 863 melting pattern pairs in both genomic and WGA samples. Agreement on the presence of a wild-type vs heterozygote variant was seen in 95% of sample pairs. Of the discordant patterns, there were a greater number of melting curve patterns in the WGA samples, suggesting the presence of a mutation (P = 0.004; see online Supplemental Table S3).
We performed targeted resequencing in 135 genomic and 136 WGA samples (Table 1
). Overall, we found 42 different variants; the number of variants per amplicon ranged from 1 to 5. All variants found in the genomic samples were also identified in the WGA samples. A single SNP present in 1 amplicon sample identified and confirmed through resequencing was not identified by DNA melting pattern in either the WGA or genomic sample. In 132 cases, we were able to compare sequences in both WGA and genomic samples; 3% had discordant results, of which the majority were loss of heterozygosity in the WGA samples.
In addition to resequencing performed specifically for this project, we used genotyping and sequencing data from other projects to assess the accuracy of high-resolution melting, allowing classification of 628 genomic and 614 WGA samples (Table 1
). Excluding homozygous variants, the sensitivity for heterozygous variant detection was similar in WGA and genomic samples, (257 of 266, 96.6%, vs 269 of 274, 98.2%; P = 0.11), whereas the specificity (percentage of wild types correctly assigned) was lower in WGA samples (260 of 287, 90.6%, vs 289 of 293, 98.6%, P < 0.001). The positive predictive value and negative predictive value were 90.5% and 96.7% for WGA and 98.5% and 98.3% for genomic samples, respectively, although it should be noted that these values reflect the fact that nearly half the amplicon samples harbored heterozygous variants. Of the homozygous variants, 31 of 61 (50.8%) genomic and 32 of 61 (52.5%) WGA samples had variant melting curves. Low sensitivity (approximately 30%) for homozygotes has been reported (9); techniques to optimize detection of homozygotes (use of smaller amplicons, mixing with known wild type) were not used for this study (10).
The majority of missed heterozygous variants (7 of 10; see online Supplemental Table S4) occurred in an amplicon in LTBP4 (latent transforming growth factor-β–binding protein 4).2
This was the longest amplicon tested (392 bp) and had 3 SNPs. In addition, the sample selected as representative of a wild type was in fact a heterozygous variant, and the difference curve a priori did not appear to allow clear discrimination of samples. In all cases, a tendency for the WGA melting patterns to have more dispersion was noted, though in general proper classification was possible. A representative run from an amplicon in SFTPB (surfactant protein b) is shown in Fig. 1
. All variants in all samples in this amplicon were called correctly.
|
The multiple displacement amplification technique using the phi29 polymerase (11) has been shown to have a very low error rate, with minimal amplification bias. In general, previous studies have demonstrated a very high rate of genotyping concordance between genomic and WGA DNA with a variety of methods (3), although others demonstrate evidence of a higher failure rate in WGA samples; loss, or more rarely, gain of heterozygosity; or bias (12)(13)(14).
We are aware of 2 previous studies that used WGA DNA in high-resolution melting curve analysis. Margraf et al. (15) identified mutations in RET. All known mutations were detected, without false positives; however, the WGA samples were not directly compared via high-resolution melting to genomic samples, and wild-type controls and smaller amplicons were used. Bastien et al. (16) examined TP53 mutations in breast cancer samples using the LightCycler 480 (Roche Diagnostics) and directly compared WGA and non-WGA samples. Sensitivity and specificity were 86% and 95% for WGA vs 100% and 99% for non-WGA samples, respectively. However, these calculations were based on only 7 mutants of 294 amplicon samples; in addition, the authors used a PCR-based WGA method, which may result in greater failure rate and sequence-dependent amplification than MDA (3). Our study is the largest direct comparison of performance of high-resolution DNA melting in WGA vs genomic DNA.
One limitation of our study is that we did not sequence all individuals; thus our sensitivity and specificities represent estimates, and use of incomplete genotyping data could have biased results. However, our choice of sequencing discordant melting patterns likely enriched our sample for finding errors and was otherwise not intentionally selective. In addition, our genotyping and sequencing data were able to verify genotypes in the majority of samples.
In summary, high-resolution DNA melting analysis using the LightScanner is a sensitive method for variant detection in genomic and whole-genome amplified DNA. Whole-genome amplified DNA appears to have a slightly higher failure rate and lower specificity than genomic DNA.
Further details on methods and results are available in the online Data Supplement file.
Acknowledgments
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors’ Disclosures of Potential Conflicts of Interest: Upon manuscript submission, all authors completed the Disclosures of Potential Conflict of Interest form. Potential conflicts of interest:
Employment or Leadership: None declared.
Consultant or Advisory Role: E.K. Silverman, GlaxoSmithKline and AstraZeneca.
Stock Ownership: None declared.
Honoraria: B.A. Raby, Novartis; E.K. Silverman, GlaxoSmithKline and AstraZeneca.
Research Funding: M.H. Cho, Idaho Technologies (use of LightScanner); E.K. Silverman, GlaxoSmithKline. This work was supported by US NIH grants R01 HL075478 (E.K. Silverman) and K08 HL74193 (B.A. Raby).
Expert Testimony: None declared.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.
Acknowledgments: The authors thank Glenn Oliveira for his technical work and all the study participants. We thank Idaho Technology, Salt Lake City, UT, for the use of the LightScanner.
Footnotes
1 Nonstandard abbreviations: WGA, whole-genome amplified; HRM, high-resolution DNA melting curve analysis; MDA, multiple displacement amplification; SNP, single nucleotide polymorphism. 
2 Human genes: LTBP4, latent transforming growth factor-β–binding protein 4; SFTPB, surfactant protein b.
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  R. van Eijk, M. van Puijenbroek, A. R. Chhatta, N. Gupta, R. H.A.M. Vossen, E. H. Lips, A.-M. Cleton-Jansen, H. Morreau, and T. van Wezel Sensitive and Specific KRAS Somatic Mutation Analysis on Whole-Genome Amplified DNA from Archival Tissues J. Mol. Diagn., January 1, 2010; 12(1): 27 - 34. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. H. Cho, D. M. Ciulla, B. J. Klanderman, C. P. Hersh, A. A. Litonjua, D. Sparrow, B. A. Raby, and E. K. Silverman Analysis of Exonic Elastin Variants in Severe, Early-Onset Chronic Obstructive Pulmonary Disease Am. J. Respir. Cell Mol. Biol., June 1, 2009; 40(6): 751 - 755. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
