TY - JOUR
AU - Knight, Ivor, T
AB - Abstract BACKGROUND High-resolution DNA melting analysis of small amplicons is a simple and inexpensive technique for genotyping. Microfluidics allows precise and rapid control of temperature during melting. METHODS Using a microfluidic platform for serial PCR and melting analysis, 4 targets containing single nucleotide variants were amplified and then melted at different rates over a 250-fold range from 0.13 to 32 °C/s. Genotypes (n = 1728) were determined manually by visual inspection after background removal, normalization, and conversion to negative derivative plots. Differences between genotypes were quantified by a genotype discrimination ratio on the basis of inter- and intragenotype differences using the absolute value of the maximum vertical difference between curves as a metric. RESULTS Different homozygous curves were genotyped by melting temperature and heterozygous curves were identified by shape. Technical artifacts preventing analysis (0.3%), incorrect (0.06%), and indeterminate (0.4%) results were minimal, occurring mostly at slow melting rates (0.13–0.5 °C/s). Genotype discrimination was maximal at around 8 °C/s (2–8 °C/s for homozygotes and 8–16 °C/s for heterozygotes), and no genotyping errors were made at rates >0.5 °C/s. PCR was completed in 10–12.2 min, followed by melting curve acquisition in 4 min down to <1 s. CONCLUSIONS Microfluidics enables genotyping by melting analysis at rates up to 32 °C/s, requiring <1 s to acquire an entire melting curve. High-speed melting reduces the time for melting analysis, decreases errors, and improves genotype discrimination of small amplicons. Combined with extreme PCR, high-speed melting promises nucleic acid amplification and genotyping in < 1 min. High-resolution DNA melting is a popular method for PCR product genotyping, variant scanning, sequence identity, methylation, and copy number analysis (1), and it is currently incorporated into almost all commercial real-time PCR instruments. When high-resolution DNA melting was first introduced, typical melting rates for genotyping were 0.3 °C/s and the process required only 2 min to perform (2, 3). Instruments today are typically slower, with some melting rates <0.01 °C/s, and thus, require up to 90 min after PCR (4, 5). Slowing down the melting rate is one way to improve the resolution by collecting more data and averaging. Intuitively, slower rates should allow more precise temperature measurements and better represent equilibrium conditions. Faster melting rates have been previously published for some applications. Allele-specific probes annealed to microbeads monolayered on a heater have allowed genotyping at 1 °C/s (6). Rates up to 20 °C/s have been reported for multiplex detection of PCR products using a single fluorescence channel with melting acquired each cycle (7). Genotyping in <7 s has been reported with molecular beacons annealed to artificial templates (8). However, none of these approaches are of high resolution nor do they investigate heteroduplex detection critical for solution genotyping used in high-resolution melting. Previously, we reported a microfluidic platform for rapid PCR integrated with high-speed melting (HSM)6 at an acquisition rate of 0.5 °C/s (9). This platform has been validated for genotyping by high-resolution melting (10, 11), including genotyping of heterozygotes by detection of heteroduplexes. To investigate the effect of melting rate on high-resolution melting, we studied melting rates over a 250-fold range from 0.13 to 32 °C/s. Materials and Methods Temperature cycling for rapid PCR and HSM were performed on a microfluidic genetic analyzer (Canon BioMedical) (Fig. 1). The instrument has been previously described (9), and genotyping accuracy studies have been performed (10, 11). Several components of the instrument have been modified or replaced to improve the reliability, usability, throughput, and data quality, as described in the Data Supplement under “Instrument Improvements” that accompanies the online version of this article at http://www.clinchem.org/content/vol63/issue10. The priming station and instrument used to perform high-speed melting. Fig. 1. Open in new tabDownload slide The priming station (31 cm × 18 cm × 10 cm) with the lid open, showing interfacing gaskets and the cartridge inserted (A). The instrument (79 cm × 79 cm × 79 cm) with the consumables drawer open and consumables inserted (B). The cartridge is the consumable on the right within the drawer. Fig. 1. Open in new tabDownload slide The priming station (31 cm × 18 cm × 10 cm) with the lid open, showing interfacing gaskets and the cartridge inserted (A). The instrument (79 cm × 79 cm × 79 cm) with the consumables drawer open and consumables inserted (B). The cartridge is the consumable on the right within the drawer. OLIGONUCLEOTIDES Primers, controls, and calibrators were synthesized by standard phosphoramidite chemistry (Integrated DNA Technologies), and their sequences are shown in Table 1 in the online Data Supplement. The F27 primers yielded a 48 bp product, the F5 primers a 43 bp product, the MTHFR c.665 primers a 48 bp product, and the MTHFR c.1286 primers a 46 bp product. A 45 bp duplex internal temperature control was composed of 3′-phosphate terminated complementary oligonucleotides, and it was included in all reactions (12). Low- and high-melting temperature (Tm) calibrators for temperature calibration were used as previously described (9). Double-stranded DNA templates (gBlocks®, Integrated DNA Technologies) were synthesized for the F2, F5, and 2 MTHFR variant loci that included each primer pair specified above. Template sequences are provided in Table 2 in the online Data Supplement. Both wild-type and homozygous variant templates were synthesized for each locus, and heterozygous DNA samples were obtained by mixing equal amounts of wild-type and variant synthetic templates. The synthetic templates ranged in length from 200 to 201 bp. All oligonucleotides were quantified by ultraviolet absorbance at A260. POLYMERASE CHAIN REACTION (PCR) Genotyping assays for F2 c.*97G>A, F5 c.1601G>A, MTHFR c.665C>T, and MTHFR c.1286A>C were performed. A 384-well plate was first loaded with reagents manually, including a primer mixture for each assay and a template mixture for each sample analyzed. Each primer mixture included 2 primers, the 2 oligonucleotides making up the duplex internal temperature control (ITC), deoxynucleotide triphosphates, and common buffer reagents including Tris, KCl, MgCl2, betaine, DMSO, and Tween® 20. Each template mixture included a variant of TaqDNA polymerase with anti-Taq antibody (Titanium®Taq, Takara Bio USA), LCGreen® Plus dye (BioFire Defense), BSA, the common buffer components listed above, and template DNA. The template DNA was added last to each template mixture, consisting of either a synthetic template (homozygous wild-type, homozygous mutant, or heterozygous), human genomic DNA (wild-type at each of the 4 loci), or water (for the no template control). The primer mixture (primers, ITC, deoxynucleotide triphosphates) was separated from the template mixture (polymerase/antibody, BSA, dye, and template) until just before PCR was initiated to limit nonspecific amplification. The primer and template mixtures were combined robotically by the instrument just before amplification and analysis. The final mixed concentrations in the PCR were: 20 mmol/L of Tris, pH 8.3, 30 mmol/L of KCl, 1 mol/L of betaine, 2% DMSO, 0.05% BSA, 0.04% Tween® 20, 4.5 mmol/L of MgCl2, 1.5 mmol/L of total dNTPs, 0.5 μmol/L of ITC, 1.0 μmol/L of each primer, 1× LCGreen® Plus dye, 1× Titanium®TaqDNA polymerase including TaqStart® antibody, and a DNA template (either the synthetic template, genomic DNA, or water for the no template control). When synthetic templates were used, their final concentration was 0.002 pg/μL (approximately 10000 copies/μL). When genomic DNA was the template, 20 ng/μL was used (approximately 6400 haploid copies/μL). These concentrations produced similar quantification cycles (Cqs) with real-time PCR for each target, suggesting that the synthetic templates may not all have been full length and/or pure. The microchips were designed to run 8 samples at a time in the following positions on each microfluidic cartridge: Lane 1: Wild-type genomic DNA Lane 2: Wild-type synthetic template Lane 3: Heterozygous synthetic template Lane 4: Homozygous mutant synthetic template Lane 5: Heterozygous synthetic template Lane 6: Wild-type synthetic template Lane 7: Homozygous mutant synthetic template Lane 8: No template control Rapid temperature cycling included heating to 95 °C at a programmed melting rate of 50 °C/s with an initial denaturation hold of 30 s, followed by 40 cycles of cooling to X °C at a rate of 12.5 °C/s with a 2 s hold, heating to 72 °C at a rate of 1.8 °C/s with a 3 s hold, and heating to 95 °C at a rate of 50 °C/s with a 2 s hold. The annealing temperature (X °C) varied by assay: F2, X = 65 °C; F5, X = 62 °C; MTHFR c.665, X = 60 °C; and MTHFR c.1286, X = 62 °C. The time to complete PCR was 10 min for F2, 11.3 min for F5 and MTHFR c. 1286, and 12.2 min for MTHFR c. 665. Following 40 cycles of PCR, there was an additional denature/renature step that was completed in 8 s: heating at a programmed rate of 200 °C/s to 95 °C with a 1.5 s hold, followed by cooling at a programmed rate of 200 °C/s to 50 °C with a 2 s hold. HIGH-SPEED MELTING After PCR, the samples remained in the same microfluidic channel positions for HSM performed between 65 °C and 95 °C with a camera acquisition rate of 30 frames per second. Each product was melted 9 times, at 0.13, 0.25, 0.5, 1, 2, 4, 8, 16, and 32 °C/s, either beginning with the slowest speed progressing to the fastest or beginning at the fastest speed progressing to the slowest. Corresponding melting times and data acquisition densities are given in Table 3 in the online Data Supplement. Eight microfluidic cartridges were run, 4 at each site (Canon Virginia, Inc. and University of Utah). At each site, 2 cartridges were run with melting fast-to-slow and 2 with melting slow-to-fast. DATA ANALYSIS Two investigators, 1 from each site, analyzed all the data from the 8 cartridge runs by manually supervised, computer-assisted analysis using custom software written in LabView. The initial upper and lower temperature regions for background determination were automatically assigned by measuring the deviation of the melting signal from an exponential background. The inner temperatures for the melting curve region were set at 5% deviation and the outer limits defined a 2 °C interval for both the upper and lower regions. These regions were manually reviewed and adjusted if necessary. The ITCs included in each melting curve were used to control minor temperature variations between channels as previously described (12, 13). The distance between 2 curves was taken as the absolute value of the maximum vertical distance between the curves after background subtraction and normalization. To make this determination, all points collected within the melting region were used. This number varied with the melting rate (see Table 3 in the online Data Supplement). The ability to distinguish genotypes was determined manually (correct, incorrect, indeterminate) and quantified by the ratio of interclass to intraclass differences. For manual classification, unbiased hierarchal clustering was first automatically performed, followed by visual interpretation and modification of assignments as needed. For quantification, interclass differences were calculated by averaging all pairwise comparisons included in the interclass calculation. For example, the 4 pairwise differences between the 2 wild-type and 2 heterozygous samples on each 8-channel read were averaged to get the wild-type vs heterozygous interclass difference. For intraclass differences, the distance between all pairwise curves within each genotype involved were averaged. A total of 1728 melting curves were acquired (8 cartridge runs of 3 genotypes in duplicate at 4 loci at 9 melting rates) of which 5 (0.3%) were excluded from analysis (bubbles or irregular melting curves due to cartridge or fluidic control issues). The excluded samples were 1 curve at a 0.13 °C/s rate and 2 curves at both 1 °C/s and 2 °C/s rates. Custom software (available from author RAP) was used to perform the calculations. Results Four genetic loci related to coagulation—F2 c.*97, F5 c.1601, MTHFR c.665, and MTHFR c.1286—were amplified and melted in a high-speed genetic analyzer that performed rapid PCR followed by HSM. For each cartridge run, the melting of each PCR product was repeated 9 times at 0.13–32 °C/s, ordered either as accelerating or decelerating rates. The 9 rates from 8 microfluidic cartridges over 4 loci resulted in 288 data sets to analyze. Each 8-channel data set included 2 wild-type, 2 homozygous variants, 2 heterozygous variants, 1 negative control, and 1 genomic DNA sample. The negative control melting curves showed the internal temperature controls as expected but were otherwise negative. Synthetic templates were used to limit DNA amplification variance, to focus on the genotype resolution of melting, and to have 2 of each genotype per locus on each cartridge run. The results obtained with the synthetic templates were close to those obtained with genomic DNA. The similarity was quantified using the discrimination ratio of distances between the synthetic and genomic curves to distances between synthetic curves, which remained close to 1 (see Fig. 1 in the online Data Supplement). When all genotype data from the 4 cartridges using accelerating rates were compared to the 4 using decelerating rates, similar genotype discrimination vs melting rate curves were observed (see Fig. 2 in the online Data Supplement). Both the heterozygous (see Fig. 2A in the online Data Supplement) and homozygous (see Fig. 2B in the online Data Supplement) discrimination ratios were similar in shape, and peak around a melting rate of 8 °C/s. Hence, for further analysis of the effect of rate on genotyping, the accelerating and decelerating cartridges were combined. Similarly, the overall effect of the analyzer and location were combined because there were no apparent differences. The apparent Tm of each locus increased approximately 2.8 °C on average as the melting rate increased 250-fold from 0.13–32 °C (see Fig. 3 in the online Data Supplement). Increased variation occurred at 4 °C/s and higher, perhaps because decreased data density affected the precision of the negative derivative peak fit. Genotyping was not affected by this Tm shift because all 8 channels were affected equally; all genotype comparisons were made at specific rates. The effect of melting rate on genotyping the MTHFR c.1286A>C locus from 0.13 to 32 °C/s is shown in Fig. 2, taken from a representative cartridge run. Within each panel, a data set of 2 wild-type, 2 heterozygous, and 2 homozygous variant curves are shown. They are all aligned to the mean Tm of their ITCs, seen as the right-most peak in each curve. As the rates increased from 0.13 to 8 °C/s, the heterozygote heteroduplex peaks increased and approached the homoduplex peaks in height and area, and the homozygous peaks became taller and thinner. From 8 to 32 °C/s, the 2 heterozygous peaks merged and the homozygous peaks became shorter and thicker, apparently because of low data density (<2 points/°C at 16 °C/s). Even at 32 °C/s, the melting curves were still easily genotyped by visual inspection. Similar trends were seen for all 4 loci at melting rates of 0.13, 8, and 32 °C/s (Fig. 3), where the ITCs are not shown. For example, MTHFR c.665C>T genotyping at 0.13 °C/s had only a small heteroduplex peak. At 8 °C/s, the heteroduplex contribution was much larger, and although it had merged into the homoduplex peak, genotyping was visually easier, as confirmed by the quantitative genotyping ratios, which increased 1.9-fold between the 2 rates. That is, for heterozygote small amplicon genotyping, separation of the heteroduplex and homoduplex contributions into separate peaks was not as important as the separation of different genotypes. The effect of melting rate on PCR product melting curves containing a single nucleotide variant. Fig. 2. Open in new tabDownload slide A 46 bp PCR product encompassing the MTHFR c.1286A>C locus was amplified and repeatedly melted at different rates along with an internal temperature control. Melting data were processed by exponential background removal, normalization, and linear temperature adjustment to the internal temperature control to compensate for any temperature variation between channels. Negative derivative plots at each melting rate show 2 wild-type (WT) samples as black lines, 2 homozygous (HOM) variants as blue lines, and 2 heterozygotes (HET) as red lines. The PCR product melted at 70–83 °C, whereas the internal temperature control melted at higher temperatures, around 83–87 °C. The apparent melting temperatures increased with increasing melting rate. The duplicate genotypes cluster distinctly, and the lower temperature heteroduplex peaks of the heterozygotes become more pronounced as the melting rate increases. At 32 °C/s, heteroduplex and homoduplex peaks merge into a single, broad peak because of low data density. Nevertheless, genotyping is clearly possible at all rates. Fig. 2. Open in new tabDownload slide A 46 bp PCR product encompassing the MTHFR c.1286A>C locus was amplified and repeatedly melted at different rates along with an internal temperature control. Melting data were processed by exponential background removal, normalization, and linear temperature adjustment to the internal temperature control to compensate for any temperature variation between channels. Negative derivative plots at each melting rate show 2 wild-type (WT) samples as black lines, 2 homozygous (HOM) variants as blue lines, and 2 heterozygotes (HET) as red lines. The PCR product melted at 70–83 °C, whereas the internal temperature control melted at higher temperatures, around 83–87 °C. The apparent melting temperatures increased with increasing melting rate. The duplicate genotypes cluster distinctly, and the lower temperature heteroduplex peaks of the heterozygotes become more pronounced as the melting rate increases. At 32 °C/s, heteroduplex and homoduplex peaks merge into a single, broad peak because of low data density. Nevertheless, genotyping is clearly possible at all rates. Example melting curves of 4 SNV loci studied at slow (0.13 °C/s), fast (8 °C/s), and very fast (32 °C/s) melting rates. Fig. 3. Open in new tabDownload slide The normalized negative derivative melting curves are displayed at 3 melting rates after background removal, normalization, and temperature adjustment to the internal temperature control (internal control not shown). Each panel includes duplicates of the 3 genotypes with wild-type (black), homozygous variant (blue), and the heterozygote (red). Small heteroduplex peaks at 0.13 °C/s become larger and similar in height to the homoduplex peaks at 8 °C/s, whereas the homozygous peaks become taller and narrower. At 32 °C/s, heterozygous duplex peaks merge into a single broad peak as data acquisition rates limit homozygous peak sharpness, but all genotypes remain easily distinguishable. Fig. 3. Open in new tabDownload slide The normalized negative derivative melting curves are displayed at 3 melting rates after background removal, normalization, and temperature adjustment to the internal temperature control (internal control not shown). Each panel includes duplicates of the 3 genotypes with wild-type (black), homozygous variant (blue), and the heterozygote (red). Small heteroduplex peaks at 0.13 °C/s become larger and similar in height to the homoduplex peaks at 8 °C/s, whereas the homozygous peaks become taller and narrower. At 32 °C/s, heterozygous duplex peaks merge into a single broad peak as data acquisition rates limit homozygous peak sharpness, but all genotypes remain easily distinguishable. The genotype discrimination metric is a dimensionless ratio of interclass to intraclass differences (Fig. 4). Two comparisons were of interest: wild-type vs homozygote and wild-type vs heterozygote. There was better discrimination between wild-type and homozygote than between wild-type and heterozygote except at very high rates where the data density was low. Homozygote discrimination was maximal at 2–8 °C/s, whereas heterozygote discrimination was highest at slightly faster rates of 8–16 °C/s. Melting-rate dependence of genotype discrimination. Fig. 4. Open in new tabDownload slide Genotype discrimination ratios of interclass to intraclass differences were used to quantify the ease of distinguishing between genotypes. Wild-type vs homozygote (dashed line) and wild-type vs heterozygote (solid line) are shown. Each point in the figure displays the mean (dimensionless) discrimination ratio obtained from 2 users analyzing all 4 loci in 8 cartridge runs for each melting rate. All 4 genotyping loci are included to best display the effect of melting rate across loci, although absolute differences across the loci do increase the variance (error bars show the standard error of the mean). Using a one-tailed t-test and assuming unequal variance, genotyping discrimination between 0.13 °C/s and 8 °C/s is significantly different for homozygotes (P = 0.005) and heterozygotes (P = 0.0004). When individual loci are analyzed, the variance is less and significant differences are greater, but the trend is less clear (data not shown). The best discrimination of homozygotes occurs at rates of 2–8 °C/s, whereas heterozygotes are best discriminated at 8–16 °C/s. Fig. 4. Open in new tabDownload slide Genotype discrimination ratios of interclass to intraclass differences were used to quantify the ease of distinguishing between genotypes. Wild-type vs homozygote (dashed line) and wild-type vs heterozygote (solid line) are shown. Each point in the figure displays the mean (dimensionless) discrimination ratio obtained from 2 users analyzing all 4 loci in 8 cartridge runs for each melting rate. All 4 genotyping loci are included to best display the effect of melting rate across loci, although absolute differences across the loci do increase the variance (error bars show the standard error of the mean). Using a one-tailed t-test and assuming unequal variance, genotyping discrimination between 0.13 °C/s and 8 °C/s is significantly different for homozygotes (P = 0.005) and heterozygotes (P = 0.0004). When individual loci are analyzed, the variance is less and significant differences are greater, but the trend is less clear (data not shown). The best discrimination of homozygotes occurs at rates of 2–8 °C/s, whereas heterozygotes are best discriminated at 8–16 °C/s. Genotyping errors and indeterminate (analyst not willing to call) results were rare (Fig. 5). One analyst made 2 incorrect (0.1%) and 3 indeterminate calls (0.2%), whereas the other made no incorrect but 10 indeterminate (0.6%) calls, for an overall error rate of 0.06% and an indeterminate rate of 0.4%. All errors were made at rates <0.5 °C/s, whereas 77% of the indeterminate calls were at or below 1 °C/s, and 23% at rates at or above 16 °C/s. No incorrect or indeterminate calls were made at 2–8 °C/s. Manual genotyping accuracy vs melting rate. Fig. 5. Open in new tabDownload slide For each of the 9 melting rates, 2 investigators visually genotyped 192 melting curves or decided that they were not able to call the genotype (indeterminate). The percentage of indeterminate and incorrect genotype calls for the Utah (UT) and Virginia (VA) analyzers at each melting rate are shown. All combined, there were 13 (0.4%) indeterminate and 2 (0.06%) incorrect genotype calls. In general agreement with the genotyping discrimination ratio analysis, all genotyping calls at melting rates from 2 to 8 °C/s were correct; both incorrect calls and 10 of 13 indeterminate calls occurred at rates from 0.13 to 1 °C/s and the 3 remaining indeterminate calls occurred at rates from 16 to 32 °C/s, where data density could be a leading contributor to genotyping difficulty. Fig. 5. Open in new tabDownload slide For each of the 9 melting rates, 2 investigators visually genotyped 192 melting curves or decided that they were not able to call the genotype (indeterminate). The percentage of indeterminate and incorrect genotype calls for the Utah (UT) and Virginia (VA) analyzers at each melting rate are shown. All combined, there were 13 (0.4%) indeterminate and 2 (0.06%) incorrect genotype calls. In general agreement with the genotyping discrimination ratio analysis, all genotyping calls at melting rates from 2 to 8 °C/s were correct; both incorrect calls and 10 of 13 indeterminate calls occurred at rates from 0.13 to 1 °C/s and the 3 remaining indeterminate calls occurred at rates from 16 to 32 °C/s, where data density could be a leading contributor to genotyping difficulty. The influence of data density on genotyping was explored by “thinning” a high-density data set. Melting curves obtained at 0.13 °C/s were converted to a low-density data set equivalent to acquisition at 16 °C/s (Fig. 6). The thinned data did not group into genotypes as well as that acquired at 16 °C/s, demonstrating that the higher melting rate accounted for the improvement in genotyping rather than the change in data density. Isolating the effects of melting rate and data density. Fig. 6. Open in new tabDownload slide All melting data were acquired at 30 points/s, the maximum frequency supported by the hardware. At the slowest melting rate of 0.13 °C/s, approximately 230 points were obtained per degree Celsius, and the thick lines observed resulted from over-sampling. That is, the line thickness in all panels is equal, but it appears thicker in panel A only because of high data density and random error. At 32 °C/s, genotype discrimination was degraded by under-sampling melting curve features (not shown) (A). In (B), the 0.13 °C/s data from (A) is resampled at 2 points per degree Celsius, the same rate at which the 16 °C/s curves shown in (C) were acquired. At constant data density, the 16 °C/s melting rate allows higher confidence genotyping than at 0.13 °C/s. Fig. 6. Open in new tabDownload slide All melting data were acquired at 30 points/s, the maximum frequency supported by the hardware. At the slowest melting rate of 0.13 °C/s, approximately 230 points were obtained per degree Celsius, and the thick lines observed resulted from over-sampling. That is, the line thickness in all panels is equal, but it appears thicker in panel A only because of high data density and random error. At 32 °C/s, genotype discrimination was degraded by under-sampling melting curve features (not shown) (A). In (B), the 0.13 °C/s data from (A) is resampled at 2 points per degree Celsius, the same rate at which the 16 °C/s curves shown in (C) were acquired. At constant data density, the 16 °C/s melting rate allows higher confidence genotyping than at 0.13 °C/s. Additional blinded studies were performed to explore the generality of our observations to all classes of single nucleotide variants (SNVs) and to observe effects of guanine–cytosine (GC) content, homopolymer stretches, and amplicon length (see “Blinded Study of Additional Variants” and Tables 4 and 5 in the online Data Supplement). Our observations of improved small amplicon genotyping at faster melting rates appeared to hold true for all SNV classes and different GC contents from 39% to 65%, and it was not affected by homopolymer stretches. However, GC content and homopolymer stretches may affect the ease of PCR amplification. Once adequately amplified, GC content and homopolymer stretches did not appear to adversely affect HSM. In contrast, PCR product length did affect the melting-rate dependence of genotyping. Faster melting was most beneficial with small amplicons around 50 bp. At 100 bp, the melting rate did not appear to affect the ability to discriminate genotype, whereas at >200 bp, the trend was reversed and slower melting resulted in better genotyping. Discussion Microfluidics enables rapid sample processing and precise control of fluids, promising better and less expensive cellular and molecular assays (14). For example, rare circulating tumor cells can be enriched, manipulated, and assayed in microdevices (15). Throughput can be increased by increasing the number of parallel reactions, such as in massively parallel sequencing (16) or digital PCR (17). Throughput can also be increased by shortening the turnaround time, and speed is particularly important in point-of-care diagnostics. Very rapid, “extreme” PCR has recently been reported, with 35 cycles of specific, sensitive, high-yield amplification in <15 s (18). However, extreme PCR must be paired with rapid sample preparation and analysis methods to provide short sample-to-answer turnaround times. Here, we investigated rapid DNA melting on a microfluidic platform to determine the effect of melting rate on genotyping. Current real-time PCR instruments that claim high-resolution melting vary in the melting rates recommended. Rates from 0.005 °C/s to approximately 0.1 °C/s appear standard on currently available instruments, requiring 5–95 min to acquire a melting curve (5). The melting rates investigated here are faster, varying from 0.13 to 32 °C/s with acquisition times from approximately 4 min to <1 s. Interestingly, rare genotyping errors were only made at low rates rather than at faster rates. That is, faster melting of small amplicons appeared to achieve better melting, at least up to 8 °C/s. Even at the fastest rate, where the curves broadened because of low data density caused by a fixed camera frame rate, genotyping was still visually clear. What is the cause of better genotyping at faster rates, rates that are 10–100 times faster than those of “state of the art” instruments in current use? The melting technique used here detects heteroduplexes arising from the amplification of heterozygotes (19). As the melting rate increases, the heteroduplex peaks increase in height and area, making heterozygotes easier to identify. This can be rationalized by considering heteroduplexes in small amplicons as unstable, nonequilibrium duplexes that, over time, recombine to form the more stable, equilibrium homoduplexes. At slower rates, there is more time for recombination to occur at critical temperatures, and thus the observed heteroduplex peak diminishes. At faster rates, there is no time for recombination, and thus more heteroduplexes are observed. This explanation is consistent with prior observations (2, 5). However, heteroduplexes in longer PCR products are less prone to recombine. Rationalizing why faster rates improve genotyping of homozygotes in small amplicons is more difficult. As the rates increase above 0.13 °C/s, the apparent Tms increases and derivative peaks amplified from homozygotes become sharper (taller and narrower), increasing the vertical separation between samples of different genotypes and the resulting discrimination ratio, thereby resulting in more distinct visual and quantitative genotyping. Perhaps the mechanism is also a nonequilibrium effect of strand disassociation vs association. The best rates for homozygote genotyping were 2–8 °C/s, lower than the best rates for heterozygote genotyping (8–16 °C/s), suggesting a unique mechanism. Both homozygote and heterozygote discrimination degrade when the data density becomes low enough that melting features become obscured, particularly at 32 °C/s. Faster data acquisition during melting may remove this limit, perhaps enabling even faster rates with even better genotype discrimination. The genotyping improvement seen here with faster melting of small amplicons around 50 bp appears to generally apply to all types of SNVs. Although amplification of high or low GC content or homopolymer stretches may complicate PCR, if targets can be amplified, they can be genotyped by high-speed melting. However, better results with faster rates may not apply to other melting applications that use larger amplicons >100 bp such as heteroduplex scanning. Indeed, the genotyping improvement with faster rates decreased as the amplicons became larger, even reversing the trend with amplicons >200 bp. Whatever the mechanism for improved genotyping of small amplicons at faster rates, the optimal rate of 8 °C/s shared by both homozygous and heterozygous genotyping observed here will provide added value to extreme PCR. A protocol of 15–30 s of PCR followed by 4 s of melting would make point-of-care molecular diagnostics much more feasible (10× faster) than the same 15–30 s of PCR followed by 4 min of melting, even permitting reflex sequential testing on-site. For samples that do not require preparation before PCR, results should be available in <30 s. If sample preparation is required and such procedures can be completed in <30 s, a 1 min sample-to-answer molecular diagnostic solution is enabled. As a final consideration, whenever a limit has been proposed (and widely accepted) for how fast PCR or melting can be performed, those limits have become obsolete over time. Turnaround times are still limited by instrumentation, and instrumentation will continue to improve, both experimental and commercial. Molecules are fast; people and their instruments are slow but are gaining in speed. 6 Nonstandard abbreviations HSM high-speed melting SNV single nucleotide variants GC guanine–cytosine ITC internal temperature control. 7 Human Genes F2 coagulation factor II (thrombin) F5 coagulation factor V MTHFR methylenetetrahydrofolate reductase. " 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 or Potential Conflicts of Interest:Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest: " Employment or Leadership: J.T. Myrick, Canon Virginia, Inc.; S.O. Sundberg, Canon U.S. Life Sciences; J.Y. Paek, Canon Information and Imaging Solutions, Inc.; C.T. Wittwer, Clinical Chemistry, AACC; I.T. Knight, Canon U.S. Life Sciences, Inc. " Consultant or Advisory Role: None declared. " Stock Ownership: None declared. " Honoraria: None declared. " Research Funding: Canon US Life Sciences provided a grant to the University of Utah to enable this work. C.T. Wittwer, Canon US Life Sciences. " Expert Testimony: None declared. " Patents: All authors, provisional; C.T. Wittwer and R.A. Palais, US8296074, US9273346. " Role of Sponsor: The funding organization played no role in the design of study, choice of enrolled patients, review and interpretation of data, or final approval of manuscript. Acknowledgments We thank Lingxia Jiang for the design of the synthetic templates and Zach Dwight for discussions on analysis algorithms. References 1. Farrar JS , Wittwer CT . High-resolution melting curve analysis for molecular diagnostics . In: Patrinos GP, Danielson PB, Ansorge W , editors. Molecular diagnostics , 3rd Ed. London : Elsevier ; 2017 . p. 79 – 102 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 2. Wittwer CT , Reed GH, Gundry CN, Vandersteen JG, Pryor RJ . High-resolution genotyping by amplicon melting analysis using LCGreen . Clin Chem 2003 ; 49 : 853 – 60 . Google Scholar Crossref Search ADS PubMed WorldCat 3. Liew M , Pryor R, Palais R, Meadows C, Erali M, Lyon E, Wittwer C . Genotyping of single-nucleotide polymorphisms by high-resolution melting of small amplicons . Clin Chem 2004 ; 50 : 1156 – 64 . Google Scholar Crossref Search ADS PubMed WorldCat 4. Herrmann MG , Durtschi JD, Bromley LK, Wittwer CT, Voelkerding KV . Amplicon DNA melting analysis for mutation scanning and genotyping: cross-platform comparison of instruments and dyes . Clin Chem 2006 ; 52 : 494 – 503 . Google Scholar Crossref Search ADS PubMed WorldCat 5. Li M , Zhou L, Palais RA, Wittwer CT . Genotyping accuracy of high-resolution DNA melting instruments . Clin Chem 2014 ; 60 : 864 – 72 . Google Scholar Crossref Search ADS PubMed WorldCat 6. Russom A , Haasl S, Brookes AJ, Andersson H, Stemme G . Rapid melting curve analysis on monolayered beads for high-throughput genotyping of single-nucleotide polymorphisms . Anal Chem 2006 ; 78 : 2220 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat 7. Ahberg CD , Manz A, Neuzil P . Single fluorescence channel-based multiplex detection of avian influenza virus by quantitative PCR with intercalating dye . Sci Rep 2015 ; 5 : 11479 Google Scholar Crossref Search ADS PubMed WorldCat 8. Chen T , Jia Y, Dong C, Gao J, Mak PI, Martins RP . Sub-7-second genotyping of single-nucleotide polymorphism by high-resolution melting curve analysis on a thermal digital microfluidic device . Lab Chip 2016 ; 16 : 743 – 52 . Google Scholar Crossref Search ADS PubMed WorldCat 9. Cao W , Bean B, Corey S, Coursey JS, Hasson KC, Inoue H , et al. Automated microfluidic platform for serial polymerase chain reaction and high-resolution melting analysis . J Lab Autom 2016 ; 21 : 402 – 11 . Google Scholar Crossref Search ADS PubMed WorldCat 10. Sundberg SO , Wittwer CT, Howell RM, Huuskonen J, Pryor RJ, Farrar JS , et al. Microfluidic genotyping by rapid serial PCR and high-speed melting analysis . Clin Chem 2014 ; 60 : 1306 – 13 . Google Scholar Crossref Search ADS PubMed WorldCat 11. Langaee T , Stauffer L, Galloway C, Solayman MH, Cavallari L . Cross-validation of high-resolution melting analysis-based genotyping platform . Genet Test Mol Biomarkers 2017 ; 21 : 259 – 64 . Google Scholar Crossref Search ADS PubMed WorldCat 12. Seipp MT , Durtschi JD, Liew MA, Williams J, Damjanovich K, Pont-Kingdon G , et al. Unlabeled oligonucleotides as internal temperature controls for genotyping by amplicon melting . J Mol Diagn 2007 ; 9 : 284 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat 13. Palais R , Wittwer CT . Mathematical algorithms for high-resolution DNA melting analysis . In: Johnson ML, Brand L , editors. Methods in enzymology , Vol. 454 . New York : Academic Press ; 2009 . p. 323 – 43 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC 14. Sackmann EK , Fulton AL, Beebe DJ . The present and future role of microfluidics in biomedical research . Nature 2014 ; 507 : 181 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat 15. Antfolk M , Laurell T . Continuous flow microfluidic separation and processing of rare cells and bioparticles found in blood - a review . Anal Chim Acta 2017 ; 965 : 9 – 35 . Google Scholar Crossref Search ADS PubMed WorldCat 16. Heather JM , Chain B . The sequence of sequencers: The history of sequencing DNA . Genomics 2016 ; 107 : 1 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 17. Huggett JF , Cowen S, Foy CA . Considerations for digital PCR as an accurate molecular diagnostic tool . Clin Chem 2015 ; 61 : 79 – 88 . Google Scholar Crossref Search ADS PubMed WorldCat 18. Farrar JS , Wittwer CT . Extreme PCR: Efficient and specific DNA amplification in 15–60 seconds . Clin Chem 2015 ; 61 : 145 – 53 . Google Scholar Crossref Search ADS PubMed WorldCat 19. Liew M , Seipp M, Durtschi J, Margraf RL, Dames S, Erali M , et al. Closed-tube SNP genotyping without labeled probes: a comparison between unlabeled probe and amplicon melting . Am J Clin Pathol 2007 ; 127 : 1 – 8 . Google Scholar Crossref Search ADS WorldCat © 2017 The American Association for Clinical Chemistry This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - High-Speed Melting Analysis: The Effect of Melting Rate on Small Amplicon Microfluidic Genotyping
JF - Clinical Chemistry
DO - 10.1373/clinchem.2017.276147
DA - 2017-10-01
UR - https://www.deepdyve.com/lp/oxford-university-press/high-speed-melting-analysis-the-effect-of-melting-rate-on-small-Zv0EIsa0rh
SP - 1624
VL - 63
IS - 10
DP - DeepDyve
ER -